nature outlokk diabetes 2012

1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1

Cover art: Nik Spencer

Nature Outlooks are sponsored supplements that aim to stimulate interest and debate around a subject of interest to the sponsor, while satisfying the editorial values of Nature and our readers’ expecta-tions. The boundaries of sponsor involvement are clearly delineated in the Nature Outlook Editorial guidelines available at http://www.nature.com/advertising/resources/pdf/outlook_guidelines.pdf

CITING THE OUTLOOKCite as a supplement to Nature, for example, Nature Vol XXX, No. XXXX Suppl, Sxx–Sxx (2012). To cite previously published articles from the collection, please use the original citation, which can be found at the start of each article.

VISIT THE OUTLOOK ONLINEThe Nature Outlook Diabetes supplement can be found at http://www.nature.com/nature/outlook/diabetes

All featured articles will be freely available for 6 months.

SUBSCRIPTIONS AND CUSTOMER SERVICESFor UK/Europe (excluding Japan): Nature Publishing Group, Subscriptions, Brunel Road, Basingstoke, Hants, RG21 6XS, UK. Tel: +44 (0) 1256 329242. Subscriptions and customer services for Americas – including Canada, Latin America and the Caribbean: Nature Publishing Group, 75 Varick St, 9th floor, New York, NY 10013-1917, USA. Tel: +1 866 363 7860 (US/Canada) or +1 212 726 9223 (outside US/Canada). Japan/China/Korea:Nature Publishing Group — Asia-Pacific, Chiyoda Building 5-6th Floor, 2-37 Ichigaya Tamachi, Shinjuku-ku, Tokyo, 162-0843, Japan. Tel: +81 3 3267 8751.

CUSTOMER [email protected] Copyright © 2012 Nature Publishing Group

S2 STATISTICS Diabetes in numbers

Disease burden and economic impact

S4 IMMUNOMODULATORS Cell savers

Protect beta cells, treat the disease

S6 DEVICES Managed by machine

The bionic pancreas goes automatic

S9 PERSPECTIVE Rethink the immune connection

Carla Greenbaum

S10 PATHOLOGY Cause and effect

Solving the puzzle of how diabetes arises

S12 MICROBIOME The critters within

Evidence points to gut bacteria influencing the course of diabetes

S14 PUBLIC HEALTH The Indian time bomb

Modernization has spawned a massive epidemic

S17 PERSPECTIVE Testing failures

Thomas Mandrup-Poulsen

S18 PREVENTION Nipped in the bud

Diet and exercise work — could a vaccine be in the offing?

Editorial Herb Brody, Michelle Grayson, Tony Scully, Nick Haines

Art & Design Wes Fernandes, Alisdair Macdonald, Andrea Duffy

Production Karl Smart, Susan Gray, Leonora Dawson-Bowling

Sponsorship David Bagshaw, Yvette Smith, Gerard Preston

Marketing Elena Woodstock, Hannah Phipps

Project Manager Christian Manco

Art Director Kelly Buckheit Krause

Magazine Editor Tim Appenzeller

Editor-in-Chief Philip Campbell

Editorial Advisors Joana Osario, Randy Levinson

The distinct but biologically related disorders that share the name diabetes impose vast human and economic losses. The US Center for Disease Control and Prevention

estimates that medical expenses for people living with diabetes in the United States are, on average, 2.3 times higher than for non-diabetics. According to the World Health Organization, 346 million people worldwide — roughly the combined populations of the United States and Canada — have diabetes (page S2).

In the developing world, Type 2 diabetes is growing at an alarming rate as people gain access to the trappings of modernity — Western-style diets along with a more sedentary lifestyle. India, for example, is experiencing an alarming epidemic in T2D that threatens to sap the country’s economic potency (S14).

Advances in medicine and technology offer some hope to those with type 1 diabetes — an autoimmune disorder that requires routine insulin injections. Immunomodulator agents under development could stop the body’s misguided attack on the insulin-producing pancreatic cells (S4). And computer-controlled devices that monitor blood sugar levels and deliver insulin in response are taking some of the guesswork and inconvenience out of this vitally important task.

There is remarkably little certainty on how these conditions arise (S10). And although it remains unclear what triggers either T1D or T2D, the bacteria that live within us are implicated (S12).

Is diabetes preventable? On this question, the differences between T1D and T2D are perhaps most apparent (S18). Vaccines might one day be able to guard against T1D, but that day is still distant. T2D, on the other hand, appears to offer ample opportunity for individuals to manage their destiny through a healthy diet and exercise.

We acknowledge the financial support of Eli Lilly and Company in producing this Outlook. As always, Nature has full responsibility for all editorial content.

Herb BrodySupplements Editor

OUTLOOK

C O N T E N T S

17 May 2012 / Vol 485 / Issue No. 7398

DIABETES

COLLECTIONS21 The NLRP3 inflammasome

instigates obesity-induced inflammation and insulin resistance

B. Vandanmagsar et al.

S31 Solving the plot: early events are the key to diabetes intervention

Alexander V. Chervonsky

S33 Sleep and eating behavior in adults at risk for type 2 diabetesJ. M. Kilkus et al.

S39 The worldwide epidemiology of type 2 diabetes mellitus—present and future perspectives

Lei Chen, Dianna J. Magliano and Paul Z. Zimmet

© 2012 Macmillan Publishers Limited. All rights reserved

S 2 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

DIABETES IN NUMBERS

30

25

15

20

10

5

0 0

2

4

6

8

10

12

�e number of people living with, and dying of, diabetes across the world is shocking: 90 million Chinese live with diabetes and 1.3 million died in 2011; 23% of Qatari adults have developed diabetes. Here we chart the extent of the global epidemic and present some of the implications for national governments by Tony Scully.


R E A L P E O P L EPercentages and predictions can mask the enormity of the diabetes problem. Large numbers of people with diabetes are unaware they have the disease because they have not been diagnosed (shown as the shaded ridge in the country bubbles). �e imperative for public-health professional is to diagnose and treat people as soon as possible.


C H I N A1,133,918 related deaths in 2011

N I G E R I A5% of adults or 3 million people

U K

I N D I A983,203 related deaths in 2011

U S Aspends $8,468 per patient

The NHS spent $16 billion in 2011

U A E19% of adults have diabetes

S A U D I A R A B I A15,399 related deaths in 2011

Highest prev-alence in the world at 23%

Q ATA R

A U S T R A L I A

N Z

F R A N C E3 . 2 M I TA LY

4 M

T U R K E Y3 . 5 M I R A N

4 . 7 M

PA K I S TA N6 . 4 M

I N D O N E S I A7 M

I N D I A6 1 . 3 M

C H I N A9 0 M

B A N G L A D E S H8 . 4 M

E G Y P T7. 3 M

G E R M A N Y5 M

S PA I N3 M

N I G E R I A3 . 1 M

P O L A N D3 M R U S S I A

1 2 . 6 M

J A PA N1 0 . 1 M

1 M

1 . 3 M

2 . 7 M

1 . 3 M 1 . 4 M1 . 7 M

1 . 4 M

1 . 9 M

1 . 2 M

1 . 5 M

1 M

2 . 8 M

4 . 2 M

4 M

2 M

1 . 7 M

1 . 5 M

3 . 1 M1 M

P O R T U G A L

S W E D E ND E N M A R KN E AT H E R L A N D S

G R E E C E

C Z E C H R E P

N O R WAY

F I N L A N DI R E L A N DC A N A D A

I S R A E L

U A EQ ATA R

S A U D IA R A B I AM E X I C O

1 0 . 3 M

U K3 . 1 M

B R A Z I L1 2 M

1 . 7 M

1 . 5 M

1 M

2 . 6 M

A R G E N T I N A

C U B A

A LG E R I A

S E R B I A

L I B YA

M A L AY S I A

R O M A N I A

PA N A M A

C O S TA R I C A

P E R U

V E N E Z U E L A

C O LU M B I A

E C U A D O R

C A M B O D I A

Y E M E N

S U D A N

U K R A I N E

M O R O C C O S Y R I A

J O R D A N

P H I L I P P I N E S

S E N E G A L

V I E T N A M

B O L I V I APA R A G U AY

I R A Q

A F G H A N I S TA N

E T H I O P I AD R CG H A N A

Z I M B A B W E

S O U T H A F R I C A

T H A I L A N D

S O U T H K O R E A

N O R T H K O R E A

Underestimated until only recently, the Chinese diabetes epidemic is the largest in the world.

M I D D L E E A S TRapid economic development has led to soaring rates of diabetes, from around 6% in 1990 to over 20% in parts today.

A F R I C ADiabetes is relatively rare in sub-Saharan Africa, a�icting only 4.5% of adults. But prevalence is predicted to double over the next 20 years — the fastest rise of any region in the world.

I N D I ANationwide prevalence now tops 9%, and is as high as 20% in the relatively prosperous southern cities. The resulting healthcare costs and depletion of productivity threaten to undo recent economic development.

T S U N A M I O F D I A B E T E S Type 2 diabetes accounts for almost 90% of all cases of diabetes in adults worldwide. In general, as countries become richer, people eat a more sugar- and fat-rich diet and are less physical active — and the incidence of diabetes rises. On average, nearly 8% of adults living in high-income countries (see map for country classi�cation) have diabetes. It is, however, upper-middle and middle-income countries that have the highest prevalence of diabetes; over 10% of adults in these countries have the condition. In high-income countries, diabetes primarily a�icts people over 50 years of age. But in middle-income countries, the highest prevalence is in younger people — the most productive age groups. As these people age, and as life expectancies increase, prevalence in older age groups will rise further. �is trend will put a huge burden on healthcare systems and governments. �e mortality rate of diabetes varies sharply with the prosperity of the country. In 2011, the disease caused more than 3.5 million deaths in middle-income countries, of which more than 1 million were in China and just less than a million were in India. Approximately 1.2 adults die of a diabetes-associated illness per 1,000 cases in 2011 in low- and middle-income countries: more than double the mortality rate of high-income countries. Mortality rates are much lower in high-income countries with the greater healthcare recourses, but those tolls are still high: approximately 180,000 people died in the United States in 2011, for example. Unsurprisingly, high-income countries spent vastly more on diabetes-related costs in 2011 than lower-income countries. In developing countries, the looming costs in human lives, healthcare expenditure and lost productivity threatens to undo recent economic gains.


T H E I N V E S T M E N T G U L FFigures for a selection of countries detail national prevalence alongside total expenditure per patient and number of diabetes-related deaths. �e countries with the highest prevalence and rates of mortality spend far less per patient than some other countries. As epidemics mature, costs and mortality are estimated to rise.


25

20

15

10

5

020

500

400

300

200

100

0

Expen

ditu

re (

USD

$ b

illio

ns)

30 40 50 60 70 80

Total healthcare expenditures (2011)

Qatar

Saud

i Ara

bia

UAEEg

ypt

Mex

ico Iran

Bangla

desh

Brazil

Philip

pine

s

Russ

iaUSA

India

China

Japa

nIsr

ael

Austr

alia

Germ

any

UK

Nigeria

Sout

h Af

rica

Ukrain

e

Num

ber

of dea

ths

(×100,0

00)

Expen

ditu

re p

er p

atie

nt(U

S$ t

hous

ands)

Com

par

ativ

e pre

vela

nce

amon

g ad

ults

(%

)

U S A2 3 . 7 M

MHigh-income

Upper-middle

Lower-middle

Low-incomeProportion of cases that remain undiagnosed

Economic zones

Millions of adults with diabetes

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0

per

1,0

00 p

eople

(ag

ed 2

0–7

9)

Mortality ratesin adults (aged 20–79)

12

10

8

6

4

2

0

Pre

vale

nce

(%)

Pre

vale

nce

(%)

Prevalence of diabetes in adults (aged 20–79)

Prevalence of diabetes amongadults (aged 20–79)

people worldwide have diabetes. More than 80% of diabetes deaths occur in low- and middle-income countries, according to the WHO.346 M

20

30

20

11

Upper-middle

High-income

Lower-middle

Low-income

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 3

DIABETES IN NUMBERS

30

25

15

20

10

5

0 0

2

4

6

8

10

12





C H I N A1,133,918 related deaths in 2011

N I G E R I A5% of adults or 3 million people

U K

I N D I A983,203 related deaths in 2011

U S Aspends $8,468 per patient

The NHS spent $16 billion in 2011

U A E19% of adults have diabetes

S A U D I A R A B I A15,399 related deaths in 2011

Highest prev-alence in the world at 23%

Q ATA R

A U S T R A L I A

N Z

F R A N C E3 . 2 M I TA LY

4 M

T U R K E Y3 . 5 M I R A N

4 . 7 M

PA K I S TA N6 . 4 M

I N D O N E S I A7 M

I N D I A6 1 . 3 M

C H I N A9 0 M

B A N G L A D E S H8 . 4 M

E G Y P T7. 3 M

G E R M A N Y5 M

S PA I N3 M

N I G E R I A3 . 1 M

P O L A N D3 M R U S S I A

1 2 . 6 M

J A PA N1 0 . 1 M

1 M

1 . 3 M

2 . 7 M

1 . 3 M 1 . 4 M1 . 7 M

1 . 4 M

1 . 9 M

1 . 2 M

1 . 5 M

1 M

2 . 8 M

4 . 2 M

4 M

2 M

1 . 7 M

1 . 5 M

3 . 1 M1 M

P O R T U G A L

S W E D E ND E N M A R KN E AT H E R L A N D S

G R E E C E

C Z E C H R E P

N O R WAY

F I N L A N DI R E L A N DC A N A D A

I S R A E L

U A EQ ATA R

S A U D IA R A B I AM E X I C O

1 0 . 3 M

U K3 . 1 M

B R A Z I L1 2 M

1 . 7 M

1 . 5 M

1 M

2 . 6 M

A R G E N T I N A

C U B A

A LG E R I A

S E R B I A

L I B YA

M A L AY S I A

R O M A N I A

PA N A M A

C O S TA R I C A

P E R U

V E N E Z U E L A

C O LU M B I A

E C U A D O R

C A M B O D I A

Y E M E N

S U D A N

U K R A I N E

M O R O C C O S Y R I A

J O R D A N

P H I L I P P I N E S

S E N E G A L

V I E T N A M

B O L I V I APA R A G U AY

I R A Q

A F G H A N I S TA N

E T H I O P I AD R CG H A N A

Z I M B A B W E

S O U T H A F R I C A

T H A I L A N D

S O U T H K O R E A

N O R T H K O R E A

Underestimated until only recently, the Chinese diabetes epidemic is the largest in the world.

M I D D L E E A S TRapid economic development has led to soaring rates of diabetes, from around 6% in 1990 to over 20% in parts today.

A F R I C ADiabetes is relatively rare in sub-Saharan Africa, a�icting only 4.5% of adults. But prevalence is predicted to double over the next 20 years — the fastest rise of any region in the world.

I N D I ANationwide prevalence now tops 9%, and is as high as 20% in the relatively prosperous southern cities. The resulting healthcare costs and depletion of productivity threaten to undo recent economic development.





25

20

15

10

5

020

500

400

300

200

100

0

Expen

ditu

re (

USD

$ b

illio

ns)

30 40 50 60 70 80

Total healthcare expenditures (2011)

Qatar

Saud

i Ara

bia

UAEEg

ypt

Mex

ico Iran

Bangla

desh

Brazil

Philip

pine

s

Russ

iaUSA

India

China

Japa

nIsr

ael

Austr

alia

Germ

any

UK

Nigeria

Sout

h Af

rica

Ukrain

e

Num

ber

of dea

ths

(×100,0

00)

Expen

ditu

re p

er p

atie

nt(U

S$ t

hous

ands)

Com

par

ativ

e pre

vela

nce

amon

g ad

ults

(%

)

U S A2 3 . 7 M

MHigh-income

Upper-middle

Lower-middle

Low-incomeProportion of cases that remain undiagnosed

Economic zones

Millions of adults with diabetes

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0

per

1,0

00 p

eople

(ag

ed 2

0–7

9)

Mortality ratesin adults (aged 20–79)

12

10

8

6

4

2

0

Pre

vale

nce

(%)

Pre

vale

nce

(%)

Prevalence of diabetes in adults (aged 20–79)

Prevalence of diabetes amongadults (aged 20–79)

people worldwide have diabetes. More than 80% of diabetes deaths occur in low- and middle-income countries, according to the WHO.346 M

20

30

20

11

Upper-middle

High-income

Lower-middle

Low-incomeIn

tern

atio

nal D

iabe

tes

Fede

rati

on. T

he ID

F D

iabe

tes

Atl

as. F

ifth

Edi

tion

. Bru

ssel

s (2

01

1).

.

OUTLOOKDIABETES


S 4 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

B Y S A R A H D E W E E R D T

Heart attack, stroke, aneurysm: each is a potentially fatal event. Most people probably wouldn’t consider type 1 dia-

betes to be as urgent, but at least one researcher argues that it is.

Jean-Francois Bach, an immunologist at the University of Paris-Descartes, argues that type 1 diabetes (T1D) should be considered a “medical emergency”, and that the goal of treatment should be to reverse the disease as soon as possible after diagnosis.

T1D occurs when the immune system mistakenly attacks beta cells — the insulin- producing cells in the pancreas. The assault takes years before the disease manifests, but only one-third of beta cells survive in most patients by the time diabetes is diagnosed, so it is critical to save those beta cells that remain. If the immune attack could be halted and those remaining cells could be preserved, scientists believe that they could be sufficient to produce most of the body’s insulin on their own.

Over the past decade, numerous clinical tri-als have tried to use immune-modifying drugs, many borrowed from the treatment of other autoimmune diseases, to try to save beta cells. So far, none have worked especially well. And because the disease is most often diagnosed in children and young adults, any effective treat-ment would have to be tolerated over the course of many years. But researchers are continuing to hone the dosage, timing and perhaps combi-nation of drugs to a therapy that works.

HIGH HOPESIn the 1980s, studies demonstrated that suppress-ing the immune system of people recently diag-nosed with T1D reduced their insulin dependence and provided persuasive evidence that T1D is an autoimmune disease. “Back then it wasn’t so clear,” says immunologist Jeffrey Bluestone of the University of California, San Francisco.

The early immunomodulators, such as cyclo-sporine and antithymocyte globulin, were blunt instruments, targeting not just the immune cells responsible for killing the beta cells, but other

parts of the immune system as well. The drugs were too toxic for patients to take for extended periods, and any protective effect failed to persist after treatment. But even this limited effective-ness was enough to spur researchers to look for more specific immune modulators that would affect the precise mechanisms of beta-cell destruction in diabetes.

Much attention has focused on T cells, thought to be behind the targeted killing of beta cells in T1D. Antibodies against CD3, a receptor found on T cells, can prevent or even permanently reverse diabetes in non-obese dia-betic (NOD) mice and other mouse models of T1D. “What was exciting in the mouse model was that we gave short-term dosing and we got long-term effects,” says Bluestone. He and other researchers hope that anti-CD3 might attenuate the immune system rather than switch it off with anti-rejection drugs.

Two different anti-CD3 antibody drugs, tepli-zumab and otelixizumab, have been tested in humans. Studies in Europe and the United States showed that a short course of treatment — two weeks or less — in people recently diagnosed with diabetes can improve beta-cell function for as long as five years.

In contrast to the mouse results, the effect in humans is temporary; eventually, beta-cell depletion begins again and the disease pro-gresses. A bigger setback came in 2011, with the results of two large, phase III clinical trials of these agents on recently diagnosed patients. Although the design and endpoints of the stud-ies were slightly different, both showed that the anti-CD3 approach did not improve on standard insulin therapy after one year1.

That’s not an unfamiliar result. Indeed, Blue-stone says, showing improvement over standard therapy is “really tough in the first year after diagnosis”. That’s because insulin formulations and delivery systems are now so good that most patients can easily keep their diabetes well con-trolled at first. Comparing any immunosuppres-sion strategies to insulin treatment has been the undoing of a number of other drug treatments, he says.

The same story of high hopes dashed in phase III trials unfolded for glutamic acid decarboxy-lase (GAD), a molecule normally found in the pancreas and one of the antigens targeted for destruction in the immune attack. Delivering GAD in a different form might help the immune system regard the molecule as friend rather than foe. But a European study found that injections of GAD failed to either stem the loss of beta-cell function or improve diabetes control over the course of 15 months in recently diagnosed T1D patients2. A similar trial in the United States was halted because GAD didn’t seem to be effective.

Jay Skyler, chair of Type 1 Diabetes TrialNet, a National Institutes of Health-funded research consortium in the United States, says that the problem with GAD might have been its delivery. Animal studies showed some promise when the molecule was administered orally, nasally or by

I M M U N O M O D U L AT O R S

Cell saversIn type 1 diabetes, the immune system goes haywire and depletes insulin-producing cells. Drugs that interfere with this process could one day reverse the disease’s course.

Kerby Bennett is a participant in the NIH’s TrialNet diabetes prevention study.

PH

OTO

PR

OVID

ED B

Y VA

ND

ERB

ILT

UN

IVER

SIT

Y M

EDIC

AL

CEN

TER

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 5

injection into the abdominal cavity, but both the US and European clinical trials used an injection under the skin. “So I personally am not ready to give up on GAD,” says Skyler, who is also a profes-sor of medicine and paediatrics at the University of Miami in Florida. “I want to see more data.”

There is some optimism about other antigens, such as insulin and its precursor, proinsulin, which might form the basis of other immune-modulating therapies. This approach offers the advantage of specificity: rather than altering the function of whole classes of immune cells, which might render a patient susceptible to infection, antigen therapies affect only activated T cells that attack beta cells. This precision is “a very impor-tant safety net”, says Chantal Mathieu, an endo-crinologist at the Catholic University of Leuven in Belgium.

DIVERSE TWEAKSMathieu is investigating other, gentler strategies for immune modulation. She is coordinator of Natural Immunomodulators as Novel Immu-notherapies for Type 1 Diabetes (NAIMIT), an EU-funded project. For example, Mathieu’s group has previously shown that vitamin D can prevent diabetes in NOD mice. “The problem is that the doses we needed were huge,” she says — equivalent to 1,000 times what could be used in humans. Such massive amounts of vitamin D can cause heart arrhythmias, kidney stones and dangerously high levels of calcium.

But researchers have not written off vitamin D just yet. A team led by diabetes and immunology specialist Bart Roep at Leiden University in the Netherlands showed that when dendritic cells (a type of immune cell) mature in vitro in the pres-ence of vitamin D, they dampen T-cell responses. “This is a very elegant way of exploiting the immune-modulatory effect of vitamin D,” Mathieu says. The team hopes to begin human studies, infusing the dendritic cells back into T1D patients, sometime in 2012.

Another possible immunological tweak is to block costimulation — the final stage of T-cell activation. A TrialNet study taking this approach enrolled 77 people with recently diag-nosed diabetes. They each received 27 infusions of abatacept, a costimulation inhibitor used in rheumatoid arthritis, over 2 years. Results were

mixed. Those who received abatacept had better beta-cell func-tion than the control group3. However, the drug had its strongest effects during the first 6 months of treatment. After that, the abata-

cept group lost beta-cell mass and function at the same rate as controls.

TrialNet researchers found similar results in a study targeting B cells, which help T cells recog-nize antigens. They administered four doses of rituximab, an antibody against the CD20 recep-tor found on B cells, to 57 people with recently

diagnosed diabetes. A year later, those who had received rituximab had better beta-cell func-tion and required less insulin than controls4. But again, the effects were most dramatic early on, and by 6 months the rituximab group were also losing beta cells. “These drugs all seem to have similar effects, and they all seem to have effects only during this certain window of time — within that first 3–6 months,” says Carla Greenbaum, director of the diabetes programme at Benaroya Research Institute in Seattle, Washington, and vice-chair of TrialNet.

INNATE INTERPLAYThis research cul-de-sac has prompted research-ers to explore more than just the antigen-specific responses of B and T cells (that is, adaptive immunity) — and to look at the non-specific, or innate, immune responses such as inflam-mation, as well. As early as the mid-1980s, endocrinologist Mandrup-Poulsen, now at the University of Copenhagen in Denmark, showed that one molecule involved in innate immunity — the pro-inflammatory interleukin-1 (IL-1) — can kill insulin-producing beta cells. The recent revival of interest in innate immunity has prompted further study. In 2011, a preliminary study found that 15 children recently diagnosed with T1D who were given a 28-day course of the IL-1 blocker anakinra needed less insulin 4 months after diagnosis than did controls5.

Two phase II trials of IL-1 blockers in T1D are just wrapping up. A European study headed by Mandrup-Poulsen also uses anakinra; a Tri-alNet study in the United States uses a similar drug called canakinumab. Both groups plan to announce the results of their studies at the

American Diabetes Association meeting in Philadelphia, Pennsylvania, June 2012. Even if these studies are negative, Mandrup-Poulsen argues, researchers should consider investigat-ing IL-1 blockers as part of combination therapy.

In fact, a recent mouse study provides support for such a strategy. Researchers found that com-bining anti-CD3 antibody with anakinra — at doses too low for either drug to work alone — permanently reversed diabetes in NOD mice6. “This is really exciting because it would indicate that if you titrate these two immune modulatory agents you may obtain very potent effects on the disease process,” says Mandrup-Poulsen, a co-author of the study’s report.

The impact of immunotherapy could extend beyond treatment. It might also become the basis for diabetes prevention, because the immune-related destruction of beta cells is thought to begin several years before diabetes becomes clinically apparent. “In theory, stopping the immune attack before it’s fully blown ought to be more effective than stopping it when everything is underway,” says Skyler. In other words, the best strategy for dealing with an immune emergency would be to prevent it altogether. ■

Sarah DeWeerdt is a science writer based in Seattle, Washington.

1 Sherry, N. et al. Lancet 378, 487–497(2011).2. Ludvigsson, J. et al. N. Engl. J. Med. 366, 433–442 (2012).3. Orban, T. et al. Lancet 378, 412–419 (2011).4. Pescovitz, M. et al. N. Engl. J. Med. 361, 2143–2152

(2009).5. Sumpter, K. M. et al. Pediatric Diabetes 12, 656–667

(2011).6. Ablamunits, V. et al. Diabetes 61, 145–154 (2012).

“Immunotherapy could extend beyond treatment to become the basis for diabetes prevention.”

OUTLOOKDIABETES

Dendritic cell

B lymphocyte

T lymphocyte

Costimulation

Antigen-speci�ctherapies

Activated T lymphocyte

Vitamin D

IL-1 and other in�ammatory molecules

Anti-CD3(teplizumab,otelexizumab)

Anti-costimulation(abatacept)

Anti-IL-1(anakinra, canakinumab and other anti-in�ammatories)

Anti-CD20(Rituximab)

Pancreaticislet

β cells

CD3

CD20

IMMUNOTHERAPYThe object in each case is to prevent the immune system (T cells) from attacking the beta cells in the pancreatic islet (shown on the right).


S 6 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

B Y E L I E D O L G I N

Leah Moynihan lifts her shirt to reveal half a dozen devices strapped to her midriff: four glucose sensors and two hormone

pumps attached to her belly and a pack of remote controls slung across her chest. On this Saturday in February, Moynihan is wired up to test what could be a major advance in the treatment of type 1 diabetes (T1D): a bionic pancreas that automatically dispenses the right amount of insulin in response to fluctuations in blood glucose levels.

“It looks like chewing gum and paper clips right now,” admits Kendra Magyar, a research nurse at Massachusetts General Hospital

(MGH) in Boston who has type 1 diabetes herself and is helping to run the trial. “If it all works out, it’ll get smaller and less invasive.”

The current standard of care for treating T1D leaves much to be desired. Typically, people prick their fingers several times a day to monitor their blood sugar levels. They then try to regulate their blood glucose, either by eating sugary foods if blood sugar levels are low (hypoglycaemia) or by injecting them-selves with insulin when glucose levels spike. Two types of wearable devices that help peo-ple manage the condition have recently hit the market. One, the continuous glucose moni-tor, is a tiny sensor placed just under the skin that checks sugar levels automatically every

few minutes. The other is an insulin pump about the size of a mobile phone that attaches to a fine needle implanted under the skin to deliver the missing pancreatic hormone at the click of the button.

The trouble is that both systems still require people to decide for themselves if, when and how to get their blood sugar levels back into the normal range — and a wrong decision can be deadly. People with T1D suffer an average of two episodes of symptomatic hypoglycae-mia per week, and as many as 10% of deaths in this patient group are caused by insulin-related complications. “When you look at the care and the burden of type 1 diabetes — test-ing and correcting 24 hours a day — it really is unbelievable,” says Dana Ball, programme director for the T1D programme at the Helmsley Charitable Trust, a New York-based non-profit organization that is partly funding the MGH trial. “A more sophisticated device would be incredible.”

For Moynihan, a nurse practitioner at the nearby Mount Auburn Hospital in Cambridge, Massachusetts, who has lived with T1D for close to three decades, such a device could not come soon enough. Diabetes “interferes with my life every day, all day long”, she says. “So it gives me some hope that there will be more than my having to think about how much insu-lin I need to take or whether I need a snack.”

IN THE LOOPTo improve the quality of life for people with T1D and help prevent diabetes-related prema-ture death, several researchers are designing automated systems that close the loop between glucose monitors and insulin infusion devices by transmitting information wirelessly from a sensor to an insulin pump. These closed-loop artificial pancreases rely on advanced control algorithms — mathematical formulations run on software — to make therapeutic decisions and accurately regulate blood sugar in real time with minimal human input. “This is an unprecedented kind of technology in which you’re handing over therapeutic decisions to software,” says Ed Damiano, a biomedical engineer at Boston University in Massachu-setts involved with the MGH trial. “As soon as it’s available, it will make the current standard of care obsolete.”

Artificial pancreases trace their roots back more than 35 years to the Biostator, a device introduced by Indiana-based Miles Labora-tories in the late 1970s. The refrigerator-sized controller relied on intravenous blood read-ings and intravenous infusions of insulin, which meant its use was limited to the hospital. Still, the Biostator proved that such a closed-loop platform was pos-sible. Products that were more portable soon fol-lowed, but the surgical procedures needed to implant the sensors and

M E D I C A L D E V I C E S

Managed by machineArtificial pancreases promise to take the decision-making — and human mistakes — out of managing type 1 diabetes.

NATURE.COMNutrition & Diabetes, a new open-access journal : go.nature.com/oypqa7

CO

UR

TESY

OF

MED

TRO

NIC

, IN

C.

Software may soon close the loop between glucose sensors (left) and insulin pumps (right).

DIABETESOUTLOOK


go.nature.com/oypqa7

1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 7

pumps deep inside the body, among other safety and usability problems, prevented their broad commercialization.

Closed-loop devices for use in the home took a big step forwards about ten years ago, with the roll-out of glucose sensors that could be implanted beneath the skin. The first-generation devices provided only retrospec-tive data that could be analysed after the fact to inform disease management. But in 2005, Medtronic, a medical device company based in Minneapolis, Minnesota, began selling the Guardian RT, which relayed glucose results every five minutes. Subcutaneous devices had already been available for insulin delivery for decades. Now all that was missing was the link between the two.

IT TAKES TWOTo spur the development of algorithms that could make therapeutic decisions, the JDRF, a New York-based non-profit foundation, initiated the Artificial Pancreas Project in 2005. This multimillion dollar initiative brought together diabetes researchers and businesses determined to make the artificial pancreas a reality. Around the same time, the US Food and Drug Administration (FDA) identified the artificial pancreas as a top priority and, together with the US National Institutes of

Health, formed the Interagency Artificial Pan-creas Working Group to identify and work through any clinical and scientific challenges. Meanwhile, government funding bodies in the United States and Europe, as well as many med-ical device companies, started spending tens of millions of dollars to encourage the develop-ment of an artificial pancreas.

In the wake of rapid progress, a handful of independent research groups launched human clinical trials, and several algorithms are being tested (see ‘Control issue’). For the most part, studies have been conducted under the controlled confines of the hospital setting, often with participants hooked up to laptop computers and intravenous backup systems that limit their mobility, as Moynihan was. But some investigators have taken their devices to the next level.

At the Princess Margaret Hospital for Chil-dren in Perth, Australia, Medtronic is running its algorithm on a BlackBerry smartphone. In Italy and France, researchers are using mobile phones and tablet computers to conduct trials in hotels — not hospitals — with doctors and engineers in separate rooms in case safety prob-lems arise. “The patients wanted to go home with it,” says Eric Renard, a diabetes specialist at Montpellier University Hospital in France who is leading the hotel-based trial. “After only a few hours, they say they’re completely differ-ent. Never before have they had this feeling that they don’t have to think about their disease.” In March 2012, the FDA approved a similar trial using the same technology at the University of Virginia in Charlottesville and at the Sansum Diabetes Research Institute (SDRI) in Santa Barbara, California.

In the United States, some investigators have also started experimenting with systems that try to improve how the artificial pancreas works. For example, Damiano’s team and an independent group in Portland, Oregon, are using a pancreatic hormone called glucagon to help raise blood glucose when too much insu-

lin has been delivered and blood sugar levels start to plummet. At Yale University School of Medicine in New Haven, Connecticut, researchers are add-ing pramlintide, a synthetic version of another human hor-

mone called amylin, to help slow the absorp-tion of nutrients from the gut as glucose levels rise after mealtimes. The Yale group has also tested a patch that heats the skin before insu-lin release to increase blood flow to the site in order to speed up the hormone’s uptake. Given the inherent lag times associated with sub-cutaneous insulin absorption, “you’re going to have a problem with catch up”, says Stuart Weinzimer, a paediatric endocrinologist who is leading the Yale trials. “Anything you can do to

speed up insulin delivery or slow down glucose absorption will help.”

A RISKY PROPOSITIONAlthough developers of artificial pancreases have differing opinions about the best closed-loop design, all agree that safety must remain a top priority as more authority is handed over to the device. “Hypoglycaemia is extraordi-narily dangerous. You lose consciousness and then you have seizures and you die if someone doesn’t help you,” warns Steven Russell, a dia-betes specialist at MGH who is collaborating with Damiano on the trials in Boston. “Giving over control entirely to a machine is a high-risk proposition,” he says, making it imperative that the process be “done properly”.

To help make the safe transition to a fully closed-loop system that requires minimal human input, many experts and companies are advancing hybrid control algorithms that are only partly automated. “We want to take iterative steps to closing the loop,” says John Mastrototaro, vice-president of global medi-cal, scientific and health affairs at Medtronic’s diabetes division in Northridge, California.

The first such product could be Medtronic’s Paradigm Veo, an insulin pump that auto-matically turns off when a sensor reports that glucose levels have fallen below a certain level. Already available in Europe, this ‘low glucose suspend’ system is now undergoing in-home testing in the United States, and is expected to receive regulatory approval in 2013.

Subsequent partly automated systems will probably benefit from technological improve-ments. The next logical step is a predictive low-glucose sensor that anticipates declining

glucose levels, rather than relying on a hard cut-off point as the Paradigm Veo does. Then maybe there will be a device that automatically increases the insulin rate when blood glucose levels rise above a certain threshold, followed perhaps by a fully closed-loop system that only works when people are asleep, thereby avoiding

“Never before have patients had this feeling that they don’t have to think about their disease.”

Ed Damiano checks the readings on his son David’s continuous glucose monitor.

Closed-loop algorithms can run on smartphones; this version shows insulin delivery and glucose level.

MEL

OD

Y KO

MYE

RO

V F

OR

BO

STO

N U

NIV

ERS

ITY

UN

IVER

SIT

Y O

F VIR

GIN

IA P

ATEN

T FO

UN

DAT

ION

OUTLOOKDIABETES


S 8 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

the confounding factors of meals, stress and exercise, all of which can complicate blood glu-cose management. Importantly, all these hybrid devices would be automated some of the time but still maintain some degree of human input in the intervening periods.

“This is the Wright Brothers at Kitty Hawk when everyone wants to go to the Moon,” says endocrinologist David Klonoff, medical director of the Diabetes Research Institute at Mills-Peninsula Health Services in San Mateo, California, who has been involved in the Para-digm Veo’s in-hospital trials. “It’s a first step, but you’ve got to start somewhere.”

“We all have this goal of a fully auto-mated system,” says Howard Zisser, direc-tor of clinical research and diabetes technology at the SDRI. “But we need to har-vest some of this low-hanging fruit,” he says of semi-automated systems, which can be put into practice more easily. What’s more, he adds, “it will be easier to convince the regulatory authorities that an artificial pancreas can read-ily help people with type 1 diabetes.”

Getting to that point, however, could be a long and bumpy road, especially in the United States — as shown by the slow path to approval for low glucose suspend systems. To speed the approval process along, in October 2011 the JDRF launched a campaign to convince the FDA, which was drawing up guidelines on arti-ficial pancreases at the time, to create a clear and reasonable path to approval for closed-loop devices. “The reason we’re putting pressure on here is because there is a critical unmet medical need,” says Aaron Kowalski, research director of the JDRF’s Artificial Pancreas Project. “We all want safe and effective products, but we also appreciate that people with diabetes are strug-gling now and the technology exists to help

them do better.”The response to the JDRF appeal was over-

whelming. In only three weeks, more than 100,000 people signed a petition — and the FDA paid attention. In December 2011, the agency released draft guidelines in which it promised to be flexible on trial sizes, durations and clinical endpoints needed for approval of an artificial pancreas. “There is no magic number or glucose level that the FDA believes is necessary to approve these devices,” says Charles Zimliki, chair of the FDA’s Artificial Pancreas Critical Path Initiative. “We’re really trying to say: ‘Come in and talk to us as you’re developing these systems’.”

In February 2012, Damiano and Russell did just that. They met with FDA officials to discuss setting up five-day trials at MGH with their algorithm running on an iPhone. “Sub-jects will have free run of the entire hospital campus,” Damiano says. After that, they hope to run 12-day trials involving MGH staff with type 1 diabetes; these study volunteers would go about their jobs at the hospital as normal while wired up to the device, and would even be able to sleep at home while still connected. Then, the Boston team plans to conduct one- to two-week trials with children at diabetes camps, fol-lowed sometime in 2014 by pivotal long-term outpatient trials.

Through it all, Damiano remains confident that a fully closed-loop device will make it to market sometime before his son, who was diag-nosed with T1D in 2000 at just 11 months of age, graduates from high school. “Before my son goes to college, he has to wear one of these things,” he says. “Or else I’m going with him.” ■

Elie Dolgin is a news editor with Nature Medicine in Cambridge, Massachusetts.

Diabetes researches and clinicians generally agree that a safe and effective artificial pancreas should provide better treatment than the current standard of care. But the bioengineers behind the systems don’t agree on the best type of algorithm to control the closed-loop devices.

In one camp sit the advocates of so-called ‘proportional-integral-derivative’ (PID) controllers, a simple strategy widely used in feedback control in settings ranging from missile steering to automobile cruise control. For artificial pancreases, PID-based algorithms use glucose values and rates of change to make calculations of insulin dosing. Gary Steil, a former Medtronic engineer who is now developing his own algorithms and running clinical trials at Children’s Hospital Boston in Massachusetts, says that the PID approach best emulates how the body’s insulin-producing beta cells manage glucose naturally, as they simply react to blood glucose levels and then spit out hormones as needed. “Everything in this algorithm is linked to something that the beta cell does,” Steil says.

But others endorse a more predictive strategy to make up for the unavoidable time lags associated with subcutaneous glucose sensing and insulin release. Known as ‘model-predictive control’, this method tries to plan several moves ahead in someone’s glucose control based on past actions and responses. According to Boris Kovatchev, director of the University of Virginia Center for Diabetes Technology in Charlottesville, this level of built-in prediction is vital to ensure patient safety. “Safety cannot be reactive,” he says. “It’s too late to be reactive.”

Some researchers, however, say this whole debate around control theory techniques is a red herring. “Algorithms aren’t the issue,” says Ken Ward, an endocrinologist at Oregon Health and Science University in Portland. “If we had a really reliable sensor and reliably fast insulin, I think the artificial pancreas would work with any number of algorithms.” — E. D.

C O N T R O L I S S U EThe algorithm method

Leah Moynihan gives her bionic pancreas a test ride.

ELIE

DO

LGIN

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 9

Decades ago, investigators established the pathology of type 1 diabetes (T1D) — the adaptive immune system mistakenly attacks the insulin-producing beta cells in the pancreatic islets.

Long before the clinical onset of disease (defined by hyperglycaemia), the immune assault triggers a progressive process of beta-cell dys-function and cell death. As this process unfolds, diabetes-related autoantibodies begin to circulate through the body, and the secretion of insulin is impaired. The attack continues after diagnosis.

This model has served us well in predicting who will get the disease. For example, a relative of someone with T1D who has one of the diabe-tes-related antibodies has about a 3% chance of developing T1D over the next five years; those with two or more antibodies have a 35–85% chance. Although some beta cells remain when clinical symptoms appear, over time the beta cells are completely destroyed.

But an explosion of data about the immune system is yet to yield a cure or prevention strategy for T1D. And we now have the results of several clinical trials testing the hypothesis that it is the adaptive immune system that is wreaking havoc on beta cells. In individuals with recently diagnosed diabetes, altering com-ponents of the adaptive immune system, for example through anti-T-cell therapies or anti-B-cell therapies, seems to improve insulin secretion (an indication of beta-cell function) by roughly 25% compared to control subjects1. Better beta-cell func-tion is associated with important clinical benefits — less hypoglycaemia and fewer complications — but with limited clinical data, the long-term benefits to individual patients remain unknown.

UNSUSTAINABLE RESPONSEMoreover, attempts to use short-term treatments to induce long-term immune tolerance of beta cells in a bid to stop disease have not worked. The best interventions so far have slowed the rate of decline of beta-cell function within the first months of diag-nosis, but repeated or continued treatments failed to sustain this response. It has also been postulated that treating a T1D patient with antigen, such as insulin or GAD65 in alum, could safely induce tolerance (see ‘Cell savers’, page S4). However, beta-cell function continues to deteriorate in people with diagnosed diabe-tes who receive antigen. It is possible that antigen therapy might work at different doses, in different populations of people (particu-larly earlier in the disease course), and in conjunction with other therapies. And yet, clinical trials testing insulin as a prophylactic — whether delivered nasally or parenterally — also failed to prevent diabetes in those who were identified as at risk for type 1 diabetes2,3.

One interpretation of these clinical failures is that we have not been aggressive enough in our attempts to save beta cells in those with T1D. An uncontrolled trial of haematopoietic stem-cell therapy to save beta cells has had some success4. However, this approach has

unknown benefits and is fraught with risks, such as pneumonia and decreased gonad function; just because we can do it, it doesn’t mean we should — especially in a disease affecting children.

Rather than not being aggressive enough with therapies, an alternative explanation to the limited success seen to date is that we have narrowly defined therapeutic targets in our intervention trials — namely molecules and pathways of the adaptive immune system. There is undoubtedly a role of the innate immune system and inflammation in beta-cell destruction; clinical trials testing this hypothesis, including blocking the proinflammatory protein interleukin-1, are underway. Moreover, beta cells in T1D might not be the victim of an immune attack, but rather have defective responses to injury or stress. It is true that genome wide asso-ciation studies (GWAS) have implicated immune related genes.

But these same genes have other func-tions, including influencing beta-cell function and response. When we look at results from an immune-centric approach, we risk missing other fac-tors that can contribute to beta-cell dysfunction. For instance, several hypotheses suggesting that environ-mental and behaviours factors play a role in the climbing incidence of T1D world wide5 await further testing. Our next generation of trials must address multiple components — immunology, genetics, environment and behav-iour. Animal models alone will not be enough to guide our future endeavours.

UNSUSTAINABLE RESPONSE Before we embark on other large clinical trials, we need more basic research, particularly proof-of-mechanism studies6. Such clinical research entails testing new therapeutic approaches in a small number of individuals to measure a biological or mecha-nistic response. This is the way to examine how alterations in metabolic state or beta-cell stress affect immune function, or to assess off-target effects of combination therapy. Evaluating all data should allow us to guard against evidentiary conservatism (the tendency to base clinical inferences on narrow classes of evidence) and to design the next generation of studies with open minds. To change the course of diabetes, we might need to alter our course. ■

Carla Greenbaum is an endocrinologist and director of the diabetes program at the Benaroya Research Institute in Seattle, Washington. email: [email protected]

1. Orban,T. et al. Lancet 378, 412–419 (2011).2. Diabetes Prevention Trial Study Group. N. Engl. J. Med. 346, 1685–1691 (2002).3. Nanto-Salonen, K. et al. Lancet 372, 1746–1755 (2008).4. Voltarelli, J. C. et al. J. Am. Med. Assoc. 297, 1568–1576 (2007).5. Karvonen, M. et al. Diabetes Care 23,1516–1526 (2000)6. Kimmelman, J. & London, A. J. PLoS Med. 8, e1001010 (2011).

PERSPECTIVERethink the immune connection Recent research suggests that the fight against type 1 diabetes is focusing too narrowly on the adaptive immune system, says Carla Greenbaum.

BEFORE WE EMBARK ON OTHER LARGE CLINICAL

TRIALS, WE NEED MORE BASIC RESEARCH, PARTICULARLY

PROOF-OF-MECHANISM STUDIES.


S 1 0 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

B Y E R I K A J O N I E T Z

Type 1 and type 2 diabetes have long been viewed as two diseases, the first auto-immune with a large genetic com-

ponent, the second metabolic, linked to obe-sity and a sedentary lifestyle. “These are two very different dis orders,” says C. Ronald Kahn, a senior investigator at Joslin Diabetes Center and professor of medicine at Harvard Medical School in Boston, Massachusetts. “They lead to similar metabolic problems and similar long-term complications, but they have two very different pathogenic routes.”

Researchers, however, are showing that each type has more in common with the other than once believed: both involve a faulty immune system and share some mechanisms that ultimately kill the insulin-producing beta cells in the pancreatic islets. Yet in neither type 1 diabetes (T1D) nor type 2 diabetes (T2D) does genetics or behaviour fully explain why some people get the disease and others don’t.

Recent findings show that despite some common risk factors, the two are indeed sepa-rate conditions. In addition to the many genetic factors involved, scientists have implicated epi-genetic and environmental influence in each

type of diabetes. Researchers continue to search for certain causes in an effort to prevent both.

ALL IN THE FAMILY?Type 1 diabetes is an autoimmune disease in which the immune system kills insulin-produc-ing beta cells. It runs in families, the hallmark of any genetic disease. About 60% of the genetic risk comes from a few specific variants in the human leukocyte antigen (HLA) genes. These genes encode the proteins that present antigens to immune cells and are involved in the mis-guided immune response in T1D.

Better understanding HLA, therefore, could help unravel the origins of T1D. Over the past five years, George Eisenbarth, an endocrinolo-gist at University of Colorado Medical School in Denver, along with immunologist John Kap-pler at National Jewish Health, has been work-ing out the structure of a three-protein complex he believes is the crux of the disease. The com-plex consists of an antigen-presenting HLA molecule, the antigen itself (a specific insulin peptide) and a T-cell receptor that recognizes the HLA–antigen combination.

T cells are central to all autoimmune dis-eases, including type 1 diabetes. Normally, cytotoxic T cells destroy only infected cells; T cells that react to molecules native to the body are eliminated before they mature, thus endowing the immune system with tolerance to ‘self ’. In type 1 diabetes, however, things go awry: T cells primed to recognize beta cells enter circulation and go on to attack the cells. How these T cells escape destruction and reach maturity isn’t clear. A number of factors appear to be involved, including variations in the gene encoding insulin, diet, and the presence or absence of certain bacteria in the gut flora (see ‘The critters within’, page S12).

Eisenbarth was involved in much of the early work that identified the antigens that prime T cells against beta cells; besides insulin, the major autoantigens are ZnT8, GAD65 and IA-2. “By following the development of antibodies to these four antigens,” Eisenbarth says, “we can now predict diabetes.” He adds: “Whoever has two of those, they almost all get diabetes.”

Further insight into the origin of diabetes could come from a new technology that can track the development of the disease in humans and mice. Researchers at Harvard Medical School and Massachusetts General Hospital in Boston have used magnetic resonance imaging (MRI) of magnetic nanoparticles to visualize insulitis, the inflamed pancreatic tissue that is the earliest clinical manifestation of diabetes. They also used MRI to distinguish at just 6–10 weeks of age which non-obese diabetic (NOD) mice — a model of T1D — will develop full-blown diabetes; mice with the highest pan-creatic accumulation of the magnetic nanopar-ticles, used as a probe, got diabetes.

PAT H O L O G Y

Cause and effectDecades of study into the causes of diabetes have produced no definitive answers.

NATURE.COMIs it time to reclassify autoimmune disease? go.nature.com/bkxr2g

The Coxsackie virus has been linked to diabetes, but do viral infections trigger or stave off diabetes?

FLO

DST

RO

M, M

. ET

AL.

NAT

. IM

MU

NO

L. 3

, 373–3

82 (2002).

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1 1

“We now have a way to know very early whether [mice] will or won’t get diabetes and then compare them at the molecular level,” says Diane Mathis, an immunologist at Harvard Medical School. Performing these compari-sons has enabled her group to identify several previously unknown molecular and cellular elements associated with a lower chance of the mice developing diabetes.

As not everyone with a genetic susceptibil-ity to T1D actually develops the disease, some sort of trigger might be involved. Evidence sug-gests that a viral infection — possibly by entero-viruses such as the Coxsackie virus — causes the immune system to misbehave. There are two theories about viral exposure: one suggests that viruses and other microorganisms improve tolerance and may thus protect against T1D. The presence of such pathogens might help overcome a sort of “boredom of the immune system” resulting from fewer childhood infec-tions, says Matthias von Herrath, director of the Type 1 Diabetes Center at La Jolla Institute for Allergy and Immunology in California. The other theory is that a virus somehow exposes antigens on beta cells, causing the immune sys-tem to attack them.

To determine whether viruses or anything else in the environment trigger type 1 dia-betes, the Diabetes Auto Immunity Study in the Young (DAISY) began in July 1993. DAISY HLA-typed about 30,000 newborns and enrolled children with a parent or sibling with T1D or children in the general popula-tion with genetic markers that indicated they were at moderate or high risk for the disease. Researchers collected blood samples and interviewed parents about diet, health and other aspects of their children’s lives; as of February 2007, 61 of the children were diag-nosed with type 1 diabetes. According to the DAISY organizers, the team identified several autoantigens and genes associated with T1D over 15 years. They also linked diet to the onset or delay of diabetes, and disproved any asso-ciation between type 1 diabetes and the age of childhood vaccinations. And although a small prospective study found no link between enterovirus infection and T1Ds, the team noted the need for more studies.

IT’S COMPLICATEDThe search for the triggers of type 2 diabetes is not any easier. This condition occurs when muscle and fat tissue respond abnormally to insulin, together with a failure of beta cells to compensate by pumping out more insulin. The statistical connection between T2D and a high-calorie diet and sedentary lifestyle is well established, but researchers still debate how — or whether —these factors cause the initial resistance to insulin. After all, 75-80% of obese people never develop type 2 diabetes. Moreover, as with type 1 diabetes, type 2 diabetes seems to run in families. Together these data suggest genetic elements.

The data from genome-wide association studies (GWAS) are far from clear, however. Studies so far have identified more than 40 genes associated with T2D, most of them having to do with beta-cell function1. But added together, they account for only about 10% of the appar-ent genetic causes. To find the missing heritabil-ity, biochemist Alan Attie at the University of Wisconsin-Madison has crossbred two strains of mice used as models — one obese but non-diabetic, the other obese and prone to diabetes — to hunt down genes linked to intermediate processes involved in diabetes, such as those that govern beta-cell regeneration, insulin deg-radation and insulin secretion.

Some genes implicated by GWAS are expressed only in adipocytes (fat cells), which might help explain how overeating can lead to diabetes. Adipose tissue stores excess lipids, which are otherwise toxic to the body. When fat cells malfunction and aren’t able to store away the extra lipids generated by overeating, lipids begin to accumulate in muscle tissue and in the

liver. Philipp Scherer, a diabetes researcher at the University of Texas Southwestern Medical Center in Dallas, believes this aberrant build-up triggers insulin resist-

ance. When adipose tissue expands in a healthy way there is no insulin resistance — which for Scherer explains why some obese people never develop type 2 diabetes.

Another consequence of abnormal adipose growth is inflammation. Expanding fat mass produces proteins called cytokines and other substances that promote inflammation and recruit macrophages (killer immune cells). As macrophages accumulate in adipose tissue, they change and secrete even more cytokines and other inflammatory factors into the bloodstream.

This promotes inflammation in other tissues, including pancreatic islets. Researchers agree almost unanimously that insulitis plays a role in type 2 diabetes; the nature of that role, however, is still a matter of debate.

While Scherer sees the inflammation as a result of insulin resistance, other biologists believe inflammation is a primary cause of diabetes. Steven Shoelson, a doctor and structural biologist at Joslin Diabetes Center and Harvard Medical School, sees things the latter way: he believes that cytokines released in response to metabolic stress may directly lead to insulin resistance. Shoelson is involved in trials to assess whether salsalate, a non-steroidal anti-inflammatory drug, can lower levels of sugar and lipids in the blood of patients with T2D, and plans to present the results of the latest large-scale trial of salsalate at the American Diabetes Association meeting in June 2012 in Philadelphia, Pennsylvania.

But even genetics, diet and activity levels com-bined don’t completely explain the origins of type 2 diabetes. Other factors that might con-tribute include environment toxins and the gut microbiome. Another influence may be mater-nal diet: research in both mice and humans has shown that maternal caloric restriction during gestation increases the risk of T2D in offspring2. The mechanism might involve strong epigenetic programming, Kahn says. Rat and human studies, for example, show that poor diet during pregnancy may affect the expression of genes that influence fetal fat-cell development, making it harder for adipocytes to effectively store excess lipids.

In a novel effort to identify specific envi-ronmental factors associated with T2D, Atul Butte, a paediatric endocrinologist and medical informaticist at Stanford Univer-sity in California, created an environmen-tal-wide association study analogous to GWAS3. A pilot study found significant links between T2D and the pesticide derivative heptachlor epoxide, vitamin E and polychlorin-ated biphenyls (PCBs).

Despite experts’ increasing knowledge about both types of diabetes, much about the patholo-gies of both diseases and virtually everything about their aetiologies remains a mystery. Most researchers acknowledge that it’s unlikely there is a single trigger; some even suggest that different genes and environmental factors may lead to disease processes that differ from person to person. “Researchers in the field are confused and have different opinions,” says Shoelson. Whether scientists ultimately find the factors that cause diabetes or not, Scherer agrees, “the bottom line is, it’s complicated”. ■

Erika Jonietz is a science writer based in Austin, Texas.

1. Fu, W. et al. Nat. Immunol. 13, 361–368 (2011).2. Herder, C. & Roden, M. Eur. J. Clin. Invest. 41,

679–692 (2011).3. Patel, C. J., Bhattacharya, J. & Butt, A. J. PLoS ONE

5, e10746 (2010).

“Researchers in the field are confused and have different opinions.”

MRI of a mouse pancreas (colour) tracks disease progression at the cellular level.

G W

OJT

KIE

WIC

Z A

ND

R. W

EIS

SLE

DER

OUTLOOKDIABETES


S 1 2 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

B Y L A U R E N G R A V I T Z

In 2004, Fredrik Bäckhed and his colleagues at Washington University in St Louis, Mis-souri, noticed that gnotobiotic mice — born

and raised to be free of germs — tended to be slimmer than their conventional counterparts. After they transplanted the feces of normal mice to germ-free ones, the rodents gained weight and their insulin was less effective at lowering blood sugar levels1. Some of the same researchers later transplanted bacteria from the intestines of either lean or obese mice into the guts of gnotobiotic mice; those animals that received bacteria from obese mice gained nearly twice as much weight as mice on the same diet that received bacteria from lean donors2. These studies jump-started research that is transforming the way we think about obesity and diabetes.

The average human gut is home to trillions of bacteria. They outnumber the cells of their human host by a factor of ten to one, and collec-tively their genes outnumber human genes one hundred-fold. Together, they function as another

organ, complementing and interacting with human metabolism in ways not fully understood. But one thing is becoming clear: the composition of bacterial species in the gut can influence the course of diabetes and its treatment.

“I have been studying diabetes for the past 25 years, and this is the most important discov-ery that has been made in my field,” says Rémy Burcelin, research director at the French National Medical Research Institute (INSERM) in Tou-louse. “We’ve discovered a new organ. We know there is a brain, a pancreas, a liver. Now we also know there are microbiota.”

Humans and the microbiome — the bacteria that reside in and on us — have co-evolved for millennia. But lately we have been messing with the delicate balance between our flora and our-selves by eating more fats and sugars, by washing with antibacterial soap, and by taking antibiot-ics at the faintest hint of infection. This shift in behaviour has coincided with an increase in the incidence of type 1 and type 2 diabetes, both of which are rising at a pace that cannot be down to genetics alone (see ‘Cause and effect’ page S10).

“There’s an order of magnitude more bugs in our gut then there are cells in our bodies, so it’s not very difficult to imagine that they would have a profound impact on metabolic balance and metabolic activity,” says Christopher New-gard, a metabolism researcher at Duke Uni-versity in Durham, North Carolina. “But, as attractive and enticing as the theory may be, it has not yet been proven in a systematic way.”

FINDING A FOOTHOLDResearchers know that certain phyla of bacte-ria are more populous in obese mice, whereas others are more common in lean ones, and the same seems to hold true in people. Moreover, bacterial composition in the gut can improve or worsen insulin resistance in mice and, initial results suggest, in people. There also appears to be a connection between inflammation and the development of insulin resistance — some of the bacteria in obese and insulin-resistant people have the potential to trigger chronic, low-grade inflammation. What researchers don’t know is how all these pieces fit together.

Two questions loom large. First, what is cause and what is effect? That is, do altered bacterial populations trigger insulin resistance or are they the product of something else in the body — and to what extent does an atypical microbiome affect the metabolism of it human host?

And second, what mechanisms are involved in any metabolic change? The answers to these questions will ultimately inform research on both the prevention and treatment of diabetes.

At the moment, researchers are trying to figure out precisely how the gut microbiome is influencing the metabolism, and thus the devel-opment of diabetes, of its human host. Several theories exist. One, for instance, blames the metabolites and other chemicals excreted by the bacteria. Another theory implicates the immune system’s reaction to the bacterial cells themselves (see ‘Microbial influence’).

Whatever the mechanism, the bacterial changes that precede insulin resistance can often be attributed to changes in diet. In mice, it takes only one day after switching from a low-fat to high-fat diet for insulin resistance to be detecta-ble3. In type 2 diabetes, many researchers believe there is a web of complex interactions between a person’s genome and gut flora. Some people are genetically predisposed to have more ben-eficial bacteria, while others people’s guts may be hospitable to pathogenic strains and may be more likely to develop diabetes when they eat high-fat foods. “Your own human nuclear genome controls a considerable part of your individual gut microflora,” says Oluf Pedersen, head of diabetes genetics research at the Hage-dorn Research Institute in Gentofte, Denmark. “But if your microbiota go off kilter then they can be causative and, at least in rodent models, effect a major change in phenotype.” Such phe-notypic changes might include weight gain and the development of metabolic syndrome — a precursor to diabetes.

M I C R O B I O M E

The critters withinYour gut microflora might be aiding and abetting diabetes.

Segmented filamentous bacteria (SFB) in the terminal ileum of an 8-week old Taconic B6 mouse.

ALI

CE

LIA

NG

, DO

UG

WEI

, ER

IC R

OTH

, NYU

.

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1 3

If researchers can figure out which bacterial species of the mammalian gut are beneficial and which are pathogenic, they might be able to nudge the population away from diabetes or even cure it. But with such a vast number of species, many of them never before identified and nearly impossible to culture, developing an extensive profile of the bacteria associated with lean versus insulin-resistant individuals is prov-ing to be monstrously difficult.

Pedersen is tackling that task in his work with the MetaHIT (Metagenomics of the Human Intestinal Tract) consortium, a collaboration of 13 institutions working to understand how genes and intestinal microbiota interact to influence health and disease. As head of MetaHIT’s obe-sity effort, he is sorting people according to their metabolic traits, including insulin resistance, and trying to correlate those to their gut bacteria. In parsing the data, Pedersen and his colleagues are finding that they can sort people into two groups according to the quantity of bacterial genes they have. Roughly one-third of their obese subjects fall into the ‘low-gene-count’ group. These indi-viduals are more likely to have signs of inflamma-tion, such as high white-blood-cell counts and elevated levels of C-reactive protein. In the gen-eral population, about the same fraction of obese people, 30% to 40%, are at risk of developing dia-betes. “We seem to have identified a subgroup of obese individuals who have a greater risk of progressing to type 2 diabetes,” Pedersen says.

HARNESSING A MICROBEEstablishing that gut flora play a role in causing diabetes is a start, but until scientists can pin responsibility on specific bacterial species or genera, it will be difficult to apply this knowledge to developing diabetes treatments.

One research group took the fecal trans-plant method that Bäckhed and others had used in mice and adapted them for human testing. Max Nieuwdorp, an endocrinologist at the Academic Medical Center in Amster-dam, the Netherlands, led a team that tested fecal transplants in a trial of 18 men recently diagnosed with metabolic syndrome. Nine men received gut biota from lean donors, while the others had their own microbiota returned to them via a fecal transplant, similar to the procedure used in mice. Initial results provide tantalizing hints that manipulating gut microflora can improve health. After 6 weeks, the men who received transplants from lean donors showed improved insulin sensi-tivity — an indication that their road to type 2 diabetes had slowed or even halted. One year later, however, the subjects’ microbiomes, and their insulin sensitivity, had returned to their original states4.

Fecal transplants in their current form aren’t a practical cure for diabetes or obesity; there are too many risks, including the transfer of bac-terial infections from donor to recipient. But these transplants do confirm the impact of bac-terial composition on blood sugar regulation in

humans. And if researchers can figure out which bacteria are beneficial, and why, they might be able to develop drugs or bacterial supplements that mimic those effects. “No one knows whether there’s a causal relationship between bacteria and diabetes,” Nieuwdorp says. “We tried to show it with the fecal transplant, but the only thing we can say is that there seems to be a transmissible trait.” Nieuwdorp has already begun a longer trial of 45 people in conjunction with gene-chip testing to discover whether multiple transplants might produce a longer-lasting effect and to identify the bacterial species involved.

If scientists can determine which bacteria are associated with which metabolic profile (lean and insulin sensitive versus overweight and insulin resistant), they might be able to sup-plement accordingly. Probiotics (live bacteria) and prebiotics (which encourage the growth of beneficial bacteria) could be used to tune a per-son’s microbiome towards greater insulin sen-sitivity. Antibiotics could be designed to target pathogenic species, or prescribed in conjunc-tion with supplements of beneficial bacteria to

prevent irreparable harm. And if researchers can identify the mechanisms of action, they should be able to develop drugs that mimic the chemicals produced by the bacteria found in lean people’s guts, or inhibit the metabolites or other molecules that lead to insulin resistance and diabetes.

“We don’t think the gut microbes are acting by one mechanism but by a contribution of several,” says Bäckhed, now at the University of Gothenberg in Sweden. “We don’t know what we do when we change the microbiota yet — we might cure type 2 diabetes and predispose someone to type 1. I wouldn’t say that changing the microbiota could cure everybody, but I think that together with lifestyle changes it could help a lot of people.” ■

Lauren Gravitz is a science writer based in Los Angeles, California.

1. Backhed, F. et al. Proc. Natl. Acad. Sci. USA 101, 15718–15723 (2004).

2. Turnbaugh, P. J. et al. Nature 444, 1027–1031 (2006).3. Turnbaugh, P. J. et al. Sci. Transl. Med. 1, 6ra14 (2009).4. Vrieze, A. et al. Diabetologia 53, 606–613 (2010).

MICROBIAL INFLUENCEResearch by Patrice Cani, at the Université Catholique de Louvain in Brussels, has shown that, in mice, a decrease in the population of bi�dobacteria species in the gut causes the tight junctions between the cells of the gut lining to loosen. The loose junctions increase the gut's permeability and allow lipopolysaccharide (LPS) from these microbes to leak through the gut wall. The resulting metabolic endotoxaemia causes a low-grade in�ammation and can induce a number of metabolic disorders – including the insulin resistance that characterizes T2D.

Research by Harvard immunologist Diane Mathis suggests that certain bacteria may protect against T1D.

Gut wall

Liver• Lipogenesis• Inflammation• Oxidative stress• Steatosis• Insulin resistance

Fat• Inflammation• Macrophage infiltration• Oxidative stress• Insulin resistance

Muscle• Inflammation• Insulin resistance

Bi�dobacterium

LPS

1

1Gutpermeability

32

2Metabolicendotoxemia

Th17

T-cell

T-cell

Small intestinewall

Mucous membrane

Segmented �lamentous bacteria (SFB) can a�ect the maturation of T-helper cells.

The presence of SFB in the gut promotes development of a compartment in the lining of the small intestine in which T-helper cells di�erentiate and mature into Th17 cells.

An abundance of Th17 cells may prevent T1D by preventing pancreatic islet cell damage caused by Th1 cells

SFB

OUTLOOKDIABETES


S 1 4 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

Mumbai’s Linking road, a congested artery at the heart of the city’s eternal traffic jam, offers a disturbing snap-

shot of the way that growing wealth is com-promising India’s health. Roadside carts selling traditional fried sweets and samosas jostle with fast-food joints selling burgers and fries, and shopping malls are full of shops selling myriad labour-saving appliances.

India’s embrace of the worst of both Eastern and Western ways is sending lifestyle illnesses such as obesity and diabetes skyrocketing. In 2011, India had 62.4 million people with type 2 diabetes, compared with 50.8 million the previous year, according to the International Diabetes Federation (IDF) and the Madras Diabetes Research Foundation. As the econ-omy started growing, so did the incidence of diabetes. The nationwide prevalence of diabe-tes in India now tops 9%, and is as high as 20% in the relatively prosperous southern cities. By 2030, the IDF predicts, India will have 100 mil-lion people with diabetes.

Health experts are alarmed because, although the onset of type 2 diabetes tends to affect people in the West in their 40s and 50s, the disease strikes Indians much younger. Indi-ans as young as 25 are being diagnosed with the disease, a trend that threatens to seriously hamper the country’s economic development.

The rise of type 2 diabetes in India’s cities was to some extent expected. And in fact, until the 1980s, the urban prevalence of dia-betes was at least double the rural prevalence. But the recent surge in diabetes has spilled out of the cities into the countryside. The spike in rural areas has been shocking, says Nikhil Tandon, an endocrinologist at the All India Institute of Medical Sciences in New Delhi (see ‘India’s diabetes boom’). “Villages in wealthier southern states like Tamil Nadu and Kerala are seeing prevalence hit double digits, which is enormous,” he says. “If it was confined to afflu-ent India, you could still put a lid on it, but now it’s rising quickly all over the country.”

THRIFTY GENESHealth experts in countries like the United States have for years been lamenting the trend towards overeating and lack of physical exercise, and the resulting rise in obesity, diabetes and heart disease. Indians seem to be even more vulner-able to these lifestyle changes. The culprit may be what is called the ‘thrifty genotype’, whereby millennia of evolution have shaped the genetic profile to cope with hardship. According to this theory, some of the world’s populations, includ-ing Indians, are genetically adapted to an envi-ronment in which calories are scarce. As a result, their bodies can’t cope in times of over-indul-gence, and it takes only a small increase in daily calories (or a small drop in calorie expenditure) for their metabolism to tip over into diabetes.

P U B L I C H E A LT H

India’s diabetes time bombEpigenetics and lifestyle are conspiring to inflict a massive epidemic of type 2 diabetes in the subcontinent.B Y P R I Y A S H E T T Y

NATURE.COMvisit Nature India for latest and best Indian research: go.nature.com/znowpk

JAG

AD

EES

H N

.V/R

EUTE

RS

/CO

RB

IS

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1 5

But the idea of a thrifty genotype doesn’t fully explain the prevalence of type 2 diabetes among certain populations, says Antonio Gon-zalez-Bulnes, a geneticist at the National Insti-tute for Agriculture and Food Research and Technology (INIA), Spain. He points out that although this genotype has been identified in rat models and in particular ethnicities around the world, such as South Pacific islanders, these groups haven’t always had higher rates of metabolic disease. C. Ronald Kahn, direc-tor of the Joslin Diabetes Center in Boston, Massachusetts, emphasizes that “we still need to regard it as a hypothesis, since no genes have been specifically identified that contribute to the phenotype”.

Instead, researchers like Gonzales-Bulnes and Tandon are interested in a related idea of a ‘thrifty phenotype’: that being deprived of nutrients in the womb, but then exposed to a high-calorie and low-exercise life, leads to a person to develop diabetes. The supposed mechanism is epigenetic; the fetal environment triggers changes in DNA methylation, which is responsible for switching genes on or off. The environment in utero thereby affects the expression of genes that code for enzymes that regulate blood sugar or tell our brains when we have eaten enough. “The mother’s nutrition, or even her smoking or alcohol consumption, can change the way the baby’s genes react to the environment: a poor or excessive diet and sedentary lifestyle,” says Paul Zimmet, head of international research at the Baker IDI Heart and Diabetes Institute in Melbourne, Australia, and one of the first to predict the Indian diabe-tes epidemic.

Several epigenetic studies back up the idea that the in utero environment has a life-long influence on health. In one study of the Dutch Hunger Winter of 1944, in which thousands of people starved during a German blockade, children born to women who were pregnant during the famine were far more likely to develop obesity or diabetes; this finding was backed up by studies of children born during the Chinese famine of 1959–61.

IN THE WOMBThe problem of the thrifty phenotype begins before the child is born. A fetus growing in a malnourished mother will need to grab all the glucose it can for its development. It does this by making its muscles resistant to insulin; since insulin is responsible for allowing fat and mus-cles to store sugar, insulin resistance forces the sugar to circulate in the blood instead. But when food is freely available, this inability to store glu-cose can send blood sugar levels soaring and trigger the onset of type 2 diabetes.

The maternal link may help explain why the diabetes epidemic is being seen all over India, in both rural and urban areas, says Caroline Fall, a paediatric epidemiologist at the University of Southampton, UK. India already has a problem with babies being born underweight — 40%

of 20 million babies born weighing less than 2.5 kilograms in the developing world are born in India. Fall points out that low birthweight (a marker of poor maternal and fetal nutrition)

does not differ that much between cities and villages in India. “You don’t need to have severe maternal malnutrition to produce the problem of obesity and diabetes in later life,” Fall says.

If poor maternal nutrition could cause dia-betes, might improving it prevent the disease? Fall is investigating the effect of micronutrients such as folate and vitamin B12 in pregnancy on the child’s development of diabetes. This link has been proven in animal models, says Fall. She is now trying to see if there is a similar effect in humans, through a study of maternal nutrition in 5,000 women living in a Mumbai slum.

Emerging data lends further sup-port to the prenatal nutrition link. Sanjay Kinra, a chronic-dis-ease epidemiologist at the London School of Hygiene and Tropical Medicine, found that

children whose mothers took nutritional sup-plements during pregnancy had lower insulin resistance2. Similar results have been found in the Gambia, says Fall. And studies by Tandon and others of the New Delhi Birth Cohort, which is following people born between 1969

and 1972, reinforce the link. According to Tandon, “those who were born small relative to their peers and then gained weight rapidly —not necessarily becoming obese — are the ones who later in life had the highest risk of developing metabolic diseases3”.

Tandon’s findings tie in with previous obser-vations that Indians don’t need to be as over-weight as people of other ethnicities to develop diabetes. The reason lies in Indians’ natural body composition, says Fall. Pound for pound, she says, “Indians have less lean mass, more body fat, and more central fat than a white Caucasian. All of these very much increase the risk of diabetes.” This difference in body type affects standard measurements such as body mass index (BMI): according to Fall, an Indian with a BMI of 23 has the same amount of body fat as a British Caucasian person with a BMI of 25. Thus, the BMI threshold that serves as a warning sign for developing diseases like type 2 diabetes is much lower in Indians.

Moreover, says Fall, the Indian susceptibility starts before birth. Even without poor maternal nutrition, “a lower muscle growth but higher fat growth in utero makes babies more vulner-able,” she says. “This is why we think diabetes hits earlier in India — they are more vulnerable from the start. If babies were well nourished in the womb, it might mean that they were not so biologically susceptible to changes in diet and lifestyle, and therefore more immune to diseases like diabetes.”

In addition to improving maternal nutri-tion, Fall wants to see routine screening for gestational diabetes (high blood glucose in pregnant women) because the condition is known to prime the child to be insulin-resistant and significantly increases the chance that the child will develop diabetes later in life. She points out that gestational diabetes is 5 or 10 times more common in Indian cities than in the United Kingdom.

Tandon points out that the focus on mothers and babies has a corollary: it means that “the problem cannot just be solved by taking 30- or 40-somethings and getting them to exercise”, he says. “We’ve missed the boat if we do that.”

TOO LITTLE ACTIONScreening for gestational diabetes could be implemented as part of a national diabetes pre-vention plan in India, though that is still being developed slowly. Preventing diabetes should be a high priority for India but there is little evidence that any major initiatives are under-way in that direction,” says Zimmet. He adds that while “diabetes is now regarded as a very serious problem by the Indian government, the question is whether they are prepared to put in the resources that are needed to turn around the epidemic”.

The early signs are that it has at least set the wheels in motion. In its latest 5-year plan, the Indian government has dedicated a signifi-cant chunk of funding for non-communicable

“The question is whether the Indian government is prepared to put in the resources.”

AP

PH

OTO

/RA

JES

H K

UM

AR

SIN

GH

Doctors and students “walk together to keep diabetes away” at a rally on World Diabetes.

OUTLOOKDIABETES


S 1 6 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

diseases such as diabetes. On chronic dis-eases overall, it will spend 580 billion rupees (US$11.6 billion) — six times what was allo-cated in the previous 5-year plan4. Progress is slow, however. India’s National Programme for Prevention and Control of Diabetes, Car-diovascular Diseases and Stroke (NPCDS), launched in 2008, has made little headway

in either strengthen-ing infrastructure or implementing pre-vention plans, other than developing a website (healthy-india.org) to educate people about risk factors for chronic diseases such as dia-betes. (India’s Minis-try of Health did not

respond to Nature Outlook’s requests for com-ment on the diabetes epidemic.)

India’s national programme on diabetes might be at a nascent stage, says Tandon, but he nevertheless finds it “reassuring” that in 100 districts, the government will be targeting programmes at schoolchildren to help reduce diabetes and other health risks in later life. The department of health is also rolling out a much-needed nationwide prevalence study, he

says, which should help provide a cohesive pic-ture — most data currently available are from fragmented regional studies, predominantly in cities.

One big barrier to improving diabetes care, says Tandon, is India’s chaotic patient refer-ral system. “The lack of systemic processing of patients has been a bugbear of the Indian healthcare system,” he says. Patients aren’t first seen by primary care workers, and then referred to secondary or tertiary care. Because 80% of healthcare in India is private, many people bypass general practitioners and go straight to the specialists, a habit that over-burdens their clinics. “Unless we streamline patient flow in the future, it will be very dif-ficult to handle the growing numbers of people with chronic diseases,” says Tandon.

Tackling most epidemics starts with screen-ing, but this is difficult given India’s ailing healthcare system and the inability of most people to afford glucose tests. The World Health Organization has recommended HbA1c as a proxy for blood glucose level, as the test is cheaper and quicker than a glucose-tolerance test. Fall points out that this method tends to create a lot of false alarms in India because of the country’s unusually high preva-lence of iron-deficiency anaemia, a condition that elevates HbA1c levels.

Although better screening would be desir-able, Zimmet argues that the top priority needs to be prevention. He points out that India won’t have the resources to treat growing numbers. “Rather, India needs to look to the future, and that may be 20 or 30 years down the track, to reducing the burden. Attention to maternal and child health may be an important way of eventually stemming the epidemic.”

For India to effectively fight the onslaught of diabetes will require more than govern-ment programmes, however. Indian society’s nonchalant attitude towards the disease must change as well. “Almost 50% don’t follow any diet and exercise regime despite our advice, and 25% of the rest will follow it initially but then abandon it,” says Anoop Misra, head of diabetes and metabolic diseases at Fortis Hos-pital in New Delhi. People believe that since they don’t have symptoms from their high-sugar levels, “they don’t need to worry about something that won’t harm them until a decade later”, says Misra.

Tandon agrees, saying that studies of diabe-tes awareness, especially in urban areas, have shown “pathetic” results5. “If you’re only wor-ried about peeing a bit more in the night, with-out realizing that this is a disease that could blind you, knock your kidneys out or give you a heart attack, you won’t worry too much.”

Ignorance about diabetes can be lethal when combined with the fact that many patients first seek out alternatives such as homeopathy or the traditional Indian medicine known as ayurveda, says Misra. He estimates that about 10–15% of his patients first tried traditional medicines — a detour that can delay their treatment through conventional medicine by up to a year.

The Indian government has been trying to raise awareness with television advertisements and posters about diabetes in doctor’s clinics and hospitals. Tandon welcomes these efforts. But considering the momentous cultural and political shifts required, he’s cautious about how quickly change will come. “It’s still an incredibly long haul.”

The middle-aged, middle-class people passing through Mumbai’s shopping mecca consider themselves to be part of the new, prosperous India. But as the epidemic of type 2 diabetes takes hold, many of them face a chronic condition — and a fate that could signal a warning for other parts of the world that are starting to enjoy abundant food and freedom from labour. ■

Priya Shetty is a science writer based in London.

1. Hales, C. N. & Barker, D. J. Diabetologia 35, 595 (1992).2. Kinra, S. et al. Brit. Med. Journ. 337, 605 (2008).3. Sachdev, H. P. et al. Arch. Dis Child. 94, 768–774

(2009).4. http://articles.timesofindia.indiatimes.com/2012-

01-17/india/30634918_1_chronic-ncds-diseases-chronic-kidney

5. Mohan, D. et al. J. Assoc. Physicians India 53, 283–287 (2005).

“Unless we streamline patient flow, it will be very difficult to handle the growing number of people with chronic diseases.”

INDIA’S DIABETES BOOMThe Western diet and lifestyle that have accompanied India’s growing prosperity has brought an alarming rise in cases of type 2 diabetes. Nationwide, prevalence of T2D is more than 9%. The epidemic is not surprising in urban areas. However, the disease is now also becoming common in rural villages, especially in wealthy southern states.

Bangalore

Coimbatore

Delhi Dibrugarh

Comparative prevelance of type 2 diabetes

Hyderabad

100

90

80

70

60

50

40

30

20

10

0

Lucknow

Ludhiana

Nagpur

Pune

Trivandrum

2000 2007 2010 2011 2030

T H E R I S E O F T Y P E 2

Num

ber

of

India

ns

with

type

2 d

iabet

es (

mill

ions)

6 . 6 %

7. 7 %

4 . 2 %

1 0 . 9 %1 0 . 8 %

1 0 . 7 %

1 4 . 1 %

1 6 . 6 %

2 . 3 %

8 . 4 %

%

IDF

DIA

BET

ES A

TLA

S, 4

TH E

D. I

NTE

RN

ATIO

NA

L D

IAB

ETS

FED

ERAT

ION

(2009).

RED

DY,

K. S

. BU

LL W

OR

LD H

EALT

H O

RG

AN

. 84, 4

61–4

69 (2006).

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1 7

The development of certain diabetes drugs keep hitting a snag — phase III clinical trials. This final stage of clinical testing is designed to test the efficacy and safety of treatments in 300

to 1,000 or more patients to ensure that the results from earlier trial phases can be applied to a more general population. Recently, a strik-ing pattern has emerged: trials are failing to confirm encouraging results obtained in earlier trials. In particular, recent phase II stud-ies of short courses of immunomodulatory biologics have provided proof-of-principle that this strategy can at least transiently improve glycaemia, insulin sensitivity or beta-cell function in people with type 1 and type 2 diabetes (T1D and T2D). Four to six infusions of antibodies against the common T-cell surface marker CD3 (ref. 1) or the B-cell surface antigen CD20 (ref. 2) — both central determi-nants of adaptive immunity — preserved beta-cell function and/or reduced insulin needs after 12–18 months in groups of 80–90 patients with recent-onset T1D. In 70 long-term patients with T2D, a blocker of the receptor binding interleukin-1 (IL-1), the primary inflammatory medi-ator of innate immunity, resulted in an improvement in beta-cell function — an effect that lasted throughout the 39-week follow-up3,4. These trials created opti-mism for the success of these agents in later phase trials.

Disappointingly, the larger trials of these drugs have failed to meet their pri-mary clinical endpoints — the measure of a trial’s success. Careful analysis has pointed to important differences in the design of the phase II and III trials. In the case of anti-CD3 antibody, the Protégé phase III study of more than 500 patients with new-onset T1D used a dose regimen different from that of the companion phase II study5; moreover, this study, conducted by MacroGenics, selected glycaemia and insulin needs as primary endpoints, instead of beta-cell function (the phase II endpoint). Another anti-CD3 study, Defend-1, conducted by GlaxoSmithKline, used beta-cell function as an endpoint. Because the full study results have not been published, we do not know such important details as whether beta-cell function was measured during fasting or after meal stimulation as generally recommended. Further-more, the study used a 15-fold lower dose than that effective in phase II.

Similarly, a large phase IIb trial of IL-1 blockade conducted by XOMA, a firm in Berkeley, California, and not yet published, enrolled more than 400 patients with T2D. The trial subjects were on average 6 years post-diagnosis, and were maintaining a baseline glycaemia of 7.8% on a single oral antidiabetic (less than 6% is con-sidered a healthy level). In contrast, patients in the phase II trial were taking a combination of oral antidiabetics and insulin, and had a mean disease duration of 11 years and baseline glycaemia of 8.5% (ref. 3). So the subjects in the larger trial had a shorter disease duration and better glucose control than those enrolled in the proof-of-principle study.

This experience prompts the question: were the right drugs tested at a wrong dose or in the wrong patients? Post-hoc analysis of the Protégé study did find significantly improved glycaemia and reduced insulin needs in the cohort receiving the highest dose5, suggesting that patients with new-onset T1D are highly sensitive to the dosing of anti-CD3 antibody. This subgroup analysis also suggests that insufficient doses might account for the failure of the Defend-1 trial. Finally, there is preclinical evidence that IL-1 blockade is more effective at preserving insulin secretion when the glucose drive is high.

Changes in study rationale, dosage, patient selection and clinical end-points may compromise the ability to confirm phase II findings in larger trials. The implications for drug development are clear, and organizers of new trials would be well advised to consider the following:

1. Recognizing that certain therapies may only be effective in sub-sets of patients, phase III trials should use entry criteria and endpoints

as close as possible to those used in phase II, and generalization of the outcomes to the prescribed patient population could then be broadened by less restrictive exclusion criteria.

2. Phase III trials should include the doses and dosing regimens effective in phase II.

3. Negative results should be published to allow learning from failure.

4. Collaboration between academia and industry should be promoted to ensure that trial designs are based on the strongest experimental and empirical evidence.

These may be more general implica-tions for developers of drugs to treat

chronic degenerative diseases, for which current clinical classifications are too crude to discriminate between aetiologically and pathogenetically different populations of patients that may require different management.

Industry and academia are in this boat together. In these times of financial constraints, with growing rates of attrition in industry and funding sources drying up in academia, there has never been a greater need for trustworthy public–private partnerships. ■

Acknowledgements Marc Y. Donath and Charles A. Dinarello are thanked for comments.

Thomas Mandrup-Poulsen is an endocrinologist at the University of Copenhagen’s Institute of Biomedical Sciences, Denmark, and at the Karolinska Institute in Stockholm, Sweden. email: [email protected]

1. Keymeulen, B. et al. N. Engl. J. Med. 352, 2598–2608 (2005).2. Pescovitz, M. D. et al. N. Engl. J. Med. 361, 2143–2152 (2009).3. Larsen, C. M. et al. N. Engl. J. Med. 356, 1517–1526 (2007).4. Larsen, C. M. et al. Diabetes Care 32, 1663–1668 (2009).5. Sherry, N. et al. Lancet 378, 487–497 (2011).

PERSPECTIVETesting failuresPromising drugs to treat diabetes stumble in the latter stages of clinical testing. Thomas Mandrup-Poulsen explains why — and how to fix it.

THE DEVELOPMENT OF CERTAIN DRUGS KEEPS

HITTING A SNAG — PHASE III TRIALS RARELY CONFIRM

ENCOURAGING RESULTS

OUTLOOKDIABETES


S 1 8 | N A T U R E | V O L 4 8 5 | 1 7 M A Y 2 0 1 2

B Y S C O T T P. E D W A R D S

They share a name, are characterized by elevated blood glucose levels, and carry potentially devastating complications if

left uncontrolled. But beyond that, type 1 and type 2 diabetes could not be more dissimilar, says diabetes researcher John Buse, director of the Diabetes Care Center at the University of North Carolina. And perhaps the biggest dif-ference of all, he says, is in their preventability. “With type 1 diabetes, there’s an immune pro-cess at play,” says Buse, “and we don’t have effec-tive ways to prevent it. Type 2 diabetes (T2D) has classic risk factors that can be modified to either delay the onset of the disease or prevent it completely.”

In the United States, 90–95% of the 17.9 mil-lion people diagnosed with diabetes have T2D. Before developing the disease, most people almost always have a related condition, called prediabetes, in which their blood has higher concentrations of glucose than is consid-ered normal, but not high enough to signify

diabetes. The American Diabetes Association estimates that 79 million people in the United States have prediabetes and thus are at high risk for developing T2D. Similar to T2D, one of the top risk factors for prediabetes is excess body weight, especially when fat is carried around the abdomen, indicative of physical inactivity and overconsumption. In general, says Buse, people with prediabetes have sev-eral problems, including insulin resistance and impaired insulin secretion, so “the train has left the station, and unless these people make changes to reduce their risk, many will keep hurtling down the tracks” towards T2D.

CHANGING LIFESTYLESScientists and clinicians have long known that people who change their lifestyle, such as by eating a healthier diet, losing weight and exer-cising more, lower their chances of developing T2D. Ten years ago, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), part of the US National Institutes of Health, confirmed this point of view when

it released the findings of the Diabetes Preven-tion Program (DPP). This multi-centre clinical research study aimed to determine if lifestyle modifications or treatment with an oral diabe-tes drug (metformin) could prevent or delay the onset of T2D. The answer was an unequiv-ocal yes to both: according to the NIDDK, “millions of high-risk people can avoid devel-oping type 2 diabetes” by losing 7% of their body weight — and maintaining that loss — by eating less fat and fewer calories, and by exer-cising for at least 150 minutes per week. The study found that diet and exercise interven-tions reduced the risk of a person developing T2D by 58%. Lifestyle changes were shown to be particularly effective in older people; those 60 years and older reduced their risk by 71%. The study also found that metformin, an oral drug widely used for the treatment of T2D, can help forestall onset of the disease and reduced risk by 31%, most effectively in young, over-weight people.

More recently, researchers in China stud-ied the long-term effects of intensive lifestyle

P R E V E N T I O N

Nipped in the budWhile type 1 diabetes might be promising ground for a vaccine, the most effective way to avoid type 2 remains good old-fashioned diet and exercise.

TIM

CLA

YTO

N/C

OR

BIS

DIABETESOUTLOOK


1 7 M A Y 2 0 1 2 | V O L 4 8 5 | N A T U R E | S 1 9

interventions on the incidence of T2D in those with impaired glucose tolerance, another pre-cursor of T2D (ref. 1). Six years of consuming a diet rich in vegetables and low in alcohol and simple sugar, as well as 20 minutes of moderate exercise, delayed the onset of T2D for as long as 14 years, although the majority of partici-pants still developed T2D.

STOPPING THE UNSTOPPABLEThe evidence is clear on T2D prevention, but the picture with type 1 diabetes (T1D) is opaque. “At this point,” says Jay Skyler of the Diabetes Research Institute, part of the Uni-versity of Miami, Florida, “there is no known way to prevent or lower the risk of developing type 1 diabetes.” Studies in animals in which oral insulin is given before diabetes develops, however, have been shown to delay the onset of disease. Given that the immune system reacts differently to drugs whether given in oral form or injected subcutaneously (the normal route of delivery for insulin), Skyler and other scientists suspect that cells in the digestive tract might play an important role in delaying or mitigating the immune response. “When an antigen, in this case insulin, is presented across a mucosal barrier like the digestive tract,” Skyler says, “the immune system forms protective immu-nity [against disease] rather than destructive immunity [such as an autoimmune disorder].”

This hypothesis is supported by recent human research at the University of South Florida in Tampa. Treatment with oral insu-lin was found to delay the onset of disease in high-risk relatives of people with T1D (ref. 2). The study also showed that oral insulin could postpone the onset of T1D in those with insu-lin autoantibodies for as long as four years, but once treatment ceased, patients developed T1D at the same rate as those taking a placebo. The task now, says Skyler, is to determine the mechanisms involved. Once the process is bet-ter understood, clinicians can assess the proper dosage at which oral insulin becomes an effec-tive preventive therapy.

The South Florida study was a follow-up to the Diabetes Prevention Trial-Type 1, or DPT-1, which ran from 1994 to 2003 under Skyler’s direction. The original DPT-1 study helped to establish how to predict T1D risk and provided insight into the immune events that lead to the development of the disease, but it brought scientists no closer to a prevention strategy. One result, for example, was that low-dose insulin injections do not prevent T1D in people who have a high risk of developing the disease within five years.

Because T1D is an autoimmune disease, one option to prevent disease could be vac-cines. Many people with T1D have antibodies to an enzyme found in the brain and pancreas called glutamic acid decarboxylase (GAD). Among other things, GAD is an autoantigen, which activates a subset of T cells that react to GAD as an ally rather than an enemy. In

T1D, these friendly T cells can quell the attack against the beta cells. However, results of a recent trial using a GAD-based vaccine were less than promising: treatment with an alum-formulated GAD vaccine, which contains an adjuvant to boost the body’s response to antigens, did not preserve beta-cell function in patients with T1D (ref. 3). Nevertheless, a phase II trial of another GAD vaccine, called Diamyd (produced by a company of the same name in Stockholm, Sweden), is set to evalu-ate whether preventive treatment with the vac-cine can delay or halt the progression of T1D in children with a high risk of developing T1D.

Preliminary results of another diabetes vac-cine study show that it might be possible to reverse T1D using an inexpensive and long-used tuberculosis vaccine. In animal studies, the Bacillus Calmette-Guérin (BCG) vaccine prevented T cells from destroying insulin-pro-ducing cells and allowed the pancreas to once again ramp up insulin production. BCG is an attractive vaccine candidate because it raises levels of tumour-necrosis factor (TNF), an immune protein that can suppress the attack on the pancreas. Autoimmunity specialist Den-ise Faustman of Massachusetts General Hos-pital in Boston reported at a 2011 American Diabetes Association meeting in San Diego, California, that low doses of the BCG vaccine

temporarily increased insulin production in patients who have had T1D for more than 20 years. In a recent s tudy, Faustman showed that the pan-creas actually slowly declines over decades, rather than weeks or

months4.“This is the first clue we’ve got about how to kill bad T cells in humans with long-term disease and in which the pancreas kicked in to produce insulin,” Faustman says. In a phase II trial, which has begun pre-screening subjects, Faustman and her colleagues will see just how far they can encourage the pancreas to re- establish insulin secretion.

If Faustman’s vaccine can truly restore insu-lin production, it could in theory serve as the basis for a prophylactic T1D vaccine. “At all stages of diabetes,” she says, “we need to slow or prevent the deterioration of the pancreas’ ability to produce insulin or, even better yet, restore insulin secretion to higher levels as we have started to do.” ■

Scott P. Edwards is a freelance science writer based in Holliston, Massachusetts.

1. Li, G. et al. Lancet 371, 1783–1789 (2008).2. Vehik, K. et al. Diabetes Care 34, 1585–1590 (2011).3. Ludvigsson, J. et al. N. Engl. J. Med. 366, 433–442 (2012).4. Wang, L., Lovejoy, N. F. & Faustman, D. L. Diabetes

Care 35, 465–4670 (2012).5. Mingrone, G. et al. N. Engl. J. Med. 366,1577–1585 (2012).6. Schauer, P. R. et al. N. Engl. J. Med. 366, 1567–1576 (2012).

“Millions of high-risk people can avoid developing type 2 diabetes” by losing 7% of their body weight.

DR

. BA

RR

Y S

LAVEN

, VIS

UA

LS U

NLI

MIT

ED /

SC

IEN

CE

PH

OTO

LIB

RA

RY

Is weight-loss surgery the next step in diabetes prevention? Two new studies reported in the New England Journal of Medicine suggest it might be in obese people with type 2 diabetes (T2D).

A team at the Catholic University in Rome compared two weight-loss surgery procedures with typical diabetes treatment5. After two years, 95% of patients receiving a biliopancreatic diversion and 75% of those receiving a Roux-en-Y gastric bypass were in disease remission with normal blood glucose levels. Both procedures shrink the stomach to the size of a chicken’s egg and bypass portions of the small intestine, restricting food absorption; in addition, biliopancreatic diversion removes part of the stomach. None of the patients in a group receiving only medical treatment — consisting of oral diabetes drugs or insulin, modified diet, and increased physical activity — went into remission.

The second study, conducted at the Cleveland Clinic in Ohio, compared patients who had either gastric bypass or sleeve gastrectomy, which cuts the stomach to the size of a banana, to those who received intensive treatment with diet, exercise and medication6. One year after surgery, 42% of the gastric-bypass patients and 37% of those having sleeve surgery were in remission, compared with only 12% of the patients treated but not operated on.

Previous observational studies have shown a connection between weight-loss surgery and reduced incidence of T2D, but there was until now no solid evidence making the connection. “This is an important result,” says John Buse, a diabetes researcher at the University of North Carolina. “As a randomized study, it is the first proof of the clinical observations that had previously been made.” — S. P. E.

T H E S U R G I C A L S O L U T I O N Shrinking the stomach

OUTLOOKDIABETES


nature medicine VOLUME 17 | NUMBER 2 | FEBRUARY 2011 179

a r t i c l e s

Despite the absence of any overt infection or definitive evidence of autoantigen-mediated immune cell activation during obesity, chronic inflammation in this state is an established instigator of several dis-eases such as type 2 diabetes, defective immunity, atherosclerosis, certain cancers, central nervous system dysfunction and dementia1–4. Activation of adipose tissue macrophages within fat depots is coupled with the development of obesity-induced proinflammatory state and insulin resistance5,6. The activation of classically activated M1 macro-phages at the expense of anti-inflammatory M2 macrophages has been causally linked to the development of adipose tissue inflammation and metabolic syndrome7,8, a pathophysiological state aptly termed as ‘metainflammation’9. It is recognized that several proinflammatory cytokines10, including IL-1β, are implicated in disrupting insulin sig-naling11. Consistent with these data, randomized clinical trials have shown that blockade of IL-1β signaling by anakinra, a recombinant human IL-1 receptor antagonist, leads to a sustained reduction in systemic inflammation and improvement of type 2 diabetes12,13. Thus, while the specific release of proinflammatory cytokines by macrophages during infection is a critical mechanism for a protec-tive immune response14,15, the origin of inflammation during obesity and the underlying molecular mechanisms that explain its occurrence are not fully understood.

Innate immune cells such as macrophages discriminate between infectious agents and self proteins by detecting pathogen-associated

molecular patterns through expression of pattern recognition receptors such as Toll-like receptors (TLRs) and NLRs15–17. Emerging evidence suggests that macrophages can also recognize danger- associated molecular patterns (DAMPs) derived from injured or damaged cells and release proinflammatory cytokines such as IL-1β (refs. 18–20). The release of bioactive IL-1β and IL-18 from macrophages is in turn dependent on autocatalytic activation of procaspase-1 zymogen into enzymatically active 10-kDa and 20-kDa caspase-1 heterodimers21–23. The presence of a large N-terminal homotypic protein–protein interaction motif called caspase acti-vation recruitment domain (CARD) is crucial for formation of multiprotein scaffolds called inflammasomes wherein caspase-1 undergoes conformational change required for its cleavage and full activation18–20,24.

Structurally, the NLR proteins contain N-terminal CARD or pyrin domains needed for homotypic protein-protein interaction, an inter-mediate nucleotide binding self-oligomerization NACHT domain and a C-terminal domain containing leucine-rich repeats18–20,24. Among the NLR family members, the Nlrp3 inflammasome has been implicated in sensing the non–microbial-originated DAMPs such as extracellular ATP, urate crystals, asbestos, silica and β-amyloid25–28. The assembly of the Nlrp3 inflammasome requires interaction of the pyrin domain of ASC (apoptosis-associated speck-like protein containing carboxy-terminal CARD) with the pyrin domain of Nlrp3,

1Laboratory of Neuroendocrine-Immunology, Louisiana State University, Baton Rouge, Louisiana, USA. 2Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, Louisiana, USA. 3Human Physiology Laboratory, Louisiana State University, Baton Rouge, Louisiana, USA. 4Oxidative Stress and Disease Laboratory, Louisiana State University, Baton Rouge, Louisiana, USA. 5Department of Biological Sciences, Louisiana State University System, Baton Rouge, Louisiana, USA. Correspondence should be addressed to V.D.D. ([email protected]).

Received 18 May 2010; accepted 18 November 2010; published online 9 January 2011; doi:10.1038/nm.2279

The NLRP3 inflammasome instigates obesity-induced inflammation and insulin resistanceBolormaa Vandanmagsar1,2, Yun-Hee Youm1,2, Anthony Ravussin1,2, Jose E Galgani2,3, Krisztian Stadler2,4, Randall L Mynatt2, Eric Ravussin2,3, Jacqueline M Stephens5 & Vishwa Deep Dixit1,2

The emergence of chronic inflammation during obesity in the absence of overt infection or well-defined autoimmune processes is a puzzling phenomenon. The Nod-like receptor (NLR) family of innate immune cell sensors, such as the nucleotide-binding domain, leucine-rich–containing family, pyrin domain–containing-3 (Nlrp3, but also known as Nalp3 or cryopyrin) inflammasome are implicated in recognizing certain nonmicrobial originated ‘danger signals’ leading to caspase-1 activation and subsequent interleukin-1b (IL-1b) and IL-18 secretion. We show that calorie restriction and exercise-mediated weight loss in obese individuals with type 2 diabetes is associated with a reduction in adipose tissue expression of Nlrp3 as well as with decreased inflammation and improved insulin sensitivity. We further found that the Nlrp3 inflammasome senses lipotoxicity-associated increases in intracellular ceramide to induce caspase-1 cleavage in macrophages and adipose tissue. Ablation of Nlrp3 in mice prevents obesity-induced inflammasome activation in fat depots and liver as well as enhances insulin signaling. Furthermore, elimination of Nlrp3 in obese mice reduces IL-18 and adipose tissue interferon-g (IFN-g) expression, increases naive T cell numbers and reduces effector T cell numbers in adipose tissue. Collectively, these data establish that the Nlrp3 inflammasome senses obesity-associated danger signals and contributes to obesity-induced inflammation and insulin resistance.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

http://www.nature.com/naturemedicine/

http://www.nature.com/doifinder/10.1038/nm.2279

a r t i c l e s

180 VOLUME 17 | NUMBER 2 | FEBRUARY 2011 nature medicine

and a functional inflammasome complex is formed through CARD-CARD interaction of ASC with procaspase-1 (refs. 24,29).

It has been shown that obesity-induced elevation in specific satu-rated free fatty acids may activate the TLR4-mediated signaling in macrophages and participate in inducing insulin resistance30,31. Interestingly, it was recently reported that mice deficient in either Nlrp3, caspase-1, or IL-1β and fed a normal-chow diet have improved insulin sensitivity, suggesting a role of the Nlrp3 inflammasome in regulating glucose homeostasis32. However, it remains unclear whether cytosolic pattern recognition receptors such as Nlrp3 have a role in sensing obesity-related danger signals or whether the Nlrp3 inflammasome participates in immune dysfunction leading to chronic inflammation and insulin resistance in diet-induced obesity. We show that Nlrp3 inflammasome activation in obesity promotes macrophage-mediated T cell activation in adipose tissue and impairs insulin sensitivity.

RESULTSNlrp3 is associated with obesity-induced insulin resistanceOur initial studies evaluated the expression of IL-1β and Nlrp3 in adipose tissue. We found that in mice the mRNA expression of Il1b and Nlrp3 in the visceral adipose tissue (VAT) correlated with body weight and adiposity (Fig. 1a,b). Calorie restriction extends lifespan, reduces inflammation and enhances insulin sensitivity33, and there-fore we examined whether calorie restriction affects the inflamma-some transcriptional machinery in fat depots. Compared to ad libitum normal chow diet–fed control mice, chronic calorie restriction (40%

reduction in food intake) in age-matched female mice (12 months old) resulted in a significant reduction (P < 0.01) in Nlrp3, Asc (also called Pycard) and Il1b mRNA in both VAT and subcutaneous adi-pose tissue (SAT) (Fig. 1c–e) in parallel with a reduction in fat cell size (Fig. 1f,g).

To test the clinical relevance of data generated from mouse models, we investigated obese individuals with type 2 diabetes mellitus (T2DM) before and after weight loss achieved by intensive behavioral modifications such as calorie restriction and exercise (Supplementary Table 1). The intervention was designed to achieve and maintain weight loss through decreased caloric intake and increased physical activity with an expected 1-year weight loss of ≥7% of the initial value34. The weight loss in obese subjects with T2DM resulted in substantial reduction in fat cell size and improvement of insulin sensitivity (Supplementary Table 1). We collected abdominal SAT biopsies from obese male subjects of European descent with T2DM (n = 10) before and after 1 year of weight loss intervention. We conducted real-time PCR analysis to quantify the mRNA levels of NLRP3, ASC (PYCARD) and IL1B. To prevent observer bias, we did the analysis in a blinded fashion. The weight loss enhanced insulin sensitivity in obese individuals with T2DM, and this insulin sensiti-vity was associated with a significant reduction in IL1B and NLRP3 mRNA expression in SAT, with no change in PYCARD (Fig. 1h). Of note, the reduction in IL1B and NLRP3 expression in SAT was coupled with lower glycemia and an improvement in homeostasis model assessment–insulin resistance (HOMA-IR) in these subjects (Fig. 1i and Supplementary Table 1). Together, these experiments

a b

0

0.5

1.0

1.5

2.0

2.5

3.0

20 30 40 50 60

Rel

ativ

e N

lrp3

mR

NA

expr

essi

on

Body weight (g)

00.10.20.30.40.50.60.70.8

20 30 40 50 60 70

Rel

ativ

e Il1

b m

RN

Aex

pres

sion

Body weight (g)

00.10.20.30.40.50.60.70.8

Rel

ativ

e Il1

b m

RN

Aex

pres

sion

VAT SAT

*

**0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Rel

ativ

e N

lrp3

mR

NA

exp

ress

ion

VAT SAT0

1

2

3

4

5

Rel

ativ

e P

ycar

dm

RN

A e

xpre

ssio

n

******

**

ALCR

ALCR

ALCR

VAT SAT

Ad libitum control Calorie restricted Ad libitum control Calorie restricted

VA

T

SA

T

edc

gf

100 µm 100 µm 100 µm 100 µm

0

2

4

6

8

Rel

ativ

e IL

1Bm

RN

A e

xpre

ssio

n

ObeseT2D

ObeseT2DM

after WL

ObeseT2D

ObeseT2DM

after WL

ObeseT2D

ObeseT2DM

after WL

*

02468

1012

Rel

ativ

e N

LRP

3m

RN

A e

xpre

ssio

n

**

05

1015202530

Rel

ativ

e P

YC

AR

Dm

RN

A e

xpre

ssio

nh

-

i

–60.0

–50.0

–40.0

–30.0

–20.0

–10.0

0

–8.00 –6.00 –4.00 –2.00 0 2.00

Cha

nge

in g

lyce

mia

(yea

r 1

base

line)

Change IL1B mRNA in adipose tissue(year 1 baseline)

–60.0

–50.0

–40.0

–30.0

–20.0

–10.0

0

–10.00 –8.00 –6.00 –4.00 –2.00 0 2.00

Cha

nge

in g

lyce

mia

(yea

r 1

base

line)

Change NLRP3 mRNA in adipose tissue(year 1 baseline)

r = 0.604P < 0.0001

r = 0.69P = 0.03

r = 0.364P = 0.017

r = 0.53P = 0.12

Figure 1 Reduction of Nlrp3 and IL-1β expression is associated with improvement of insulin sensitivity. (a,b) Positive correlation of the visceral fat mRNA expression of Il1b (a) and Nlrp3 (b) with body weight of C57BL/6 mice (n = 32); Pearson’s correlations are r = 0.364, P = 0.0178 for Il1b and r = 0.672, P < 0.0001 for Nlrp3, respectively. (c–e) Il1b (c), Nlrp3 (d) and Pycard (e) mRNA in visceral and subcutaneous adipose tissue from ad libitum chow-fed control (AL) and 40% calorie-restricted (CR) 12-month-old mice, n = 6; *P < 0.01, **P < 0.005. (f,g) Representative H&E staining showing adipocyte size in visceral (f) and subcutaneous (g) fat tissue from ad libitum–fed control and calorie-restricted 12-month-old C57BL/6 mice. (h) IL1B (left), NLRP3 (middle) and PYCARD (right) gene expression, as examined by quantitative RT-PCR (qRT-PCR) in human SAT in obese individuals with T2DM before and after 1-year weight loss. *P = 0.01, **P = 0.001. (i) Positive correlation of changes in gene expression of IL1B and NLRP3 in human abdominal subcutaneous fat with changes in fasting glucose level from baseline to 1 year after intervention; Pearson’s correlations are r = 0.53, P = 0.12 for IL1B and r = 0.69, P = 0.03 for NLRP3. Relative gene expression levels are depicted as means ± s.e.m. n = 10.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


indicate that the expression of NLRP3 inflammasome components is associated with adiposity and insulin resistance.

Inflammasome activation in obesity regulates IL-1b and IL-18Nlrp3 is known to be present in several tissues and cell types35, but it is not known which cellular compartments in adipose tissue express the inflammasome components. Immunostaining of adipose tissue sections of obese mice revealed strong colocalization of Nlrp3 with the macrophage marker F4/80 (Fig. 2a) in crown-like structures. Consistent with the immunofluorescence data, examination of puri-fied F4/80+ adipose tissue macrophages (ATMs), stromal vascular fraction (SVF) and mature 3T3-L1 adipocytes revealed that both Nlrp3 and Pycard are highly expressed in ATMs and SVF cells with low expression in adipocytes (Fig. 2b,c). We also analyzed Nlrp3 expres-sion in enriched primary adipocytes from adipose tissue of mice. Normalization of Nlrp3 mRNA expression with that of the differen-tiated macrophage marker Cd11c in the adipocyte fraction revealed that almost all Nlrp3 expression in enriched primary adipocytes may be attributed to contaminating lipid-engorged macrophages (data not shown).

Notably, the development of progressive adiposity in high-fat diet (HFD)-fed mice led to strong caspase-1 autoactivation in adi-pose tissue (Fig. 2d and Supplementary Fig. 1a,b). Consistent with progressive caspase-1 activation in obesity, the highest expression of active IL-1β in adipose tissue was detected in 9-month-old diet-induced obese (DIO) mice (Fig. 2d). Since we detected the most caspase-1 and IL-1β activation in 9-month-old DIO mice (Fig. 2d and Supplementary Fig. 2a) and because of the high prevalence of

obesity and related comorbidities in the middle-aged population, we next studied the role of the Nlrp3 inflammasome in a chronic DIO model. This rationale is based on the hypothesis and our initial data (Fig. 2d) that 28 weeks of HFD feeding in mice is likely to induce a more severe disease than 6–8 weeks of HFD feeding. Furthermore, although it is established that the Nlrp3 inflammasome regulates post–translational processing of caspase-1 in vitro in bone marrow–derived macrophages27,28, the specificity of this pathway for caspase-1 activation in vivo in different cell types in HFD-induced obesity is not established. Of note, compared to healthy, lean mice, obese mice had markedly more caspase-1 autoactivation in VAT, SAT and liver (Fig. 2e), (Supplementary Fig. 1). Ablation of Nlrp3 partially blocked the obesity-induced caspase-1 autoactivation in VAT, SAT and liver (Fig. 2e and Supplementary Fig. 1) but did not affect caspase-1 cleavage in kidneys of chronic DIO mice (Fig. 2f). These findings indicate that in obesity caspase-1 is specifically activated in adipose tissue and liver through an Nlrp3 inflammasome–dependent mechanism, but in kidney this process is independent of Nlrp3 and is tissue specific.

The post-translational processing of IL-1β is complex and can be regulated through several inflammasomes, including the Ipaf and AIM2 inflammasomes19,20. Furthermore, pro–IL-1β can be processed through neutrophil-derived serine proteases in a caspase-1– and inflammasome-independent fashion21,36. Our data show that, com-pared to wild-type (WT) mice, in DIO mice the ablation of the Nlrp3 inflammasome reduced the expression of active IL-1β in adipose tissue (Fig. 2g and Supplementary Fig. 2b). Normal chow–fed WT mice and Nlrp3-deficient mice did not show a difference in IL-1β

b

e

160 *

WT

IL-1

8 (p

g m

l–1)

200

406080

100120140

<20 pg ml–1

below assay sensitivity

*

a F4/80 WT WT

WT Nlrp3–/–

Nlrp3

F4/80Nlrp3DAPI

F4/80Nlrp3DAPI

c1.0

0.2

0.6

0

Fol

d ch

ange

Nlrp

3 m

RN

A

1.4

Fol

d ch

ange

Pyc

ard

mR

NA

1.0

0.2

0.6

0

1.4 d

WT leanMW(kDa)

WT DIO

Procasp-1

p20

Actin

493825V

AT

17

Procasp-1

p20

Actin

Live

r

49382517

Procasp-1

p20

Actin

SA

T

49382517

eFat ProIL-1β

p17–IL-1β

p20eFat

Procasp-1

g

h

WT (6 months)

eFat

493825

17

fProcasp-1

p20Kid

ney

WT lean WT DIO

Actin

eFat

WT (7 months)

4 month chow4 month DIO7 month chow7 month DIO

Nlrp3–/– DIO

Nlrp3–/– (6 months)

Nlrp3–/– DIO

Nlrp3–/–

100 µm

ProIL-1β

p17–IL-1β

ProIL-1β

p17–IL-1β

3 4 6 9 Time (months)

ATMSVF

SVFAdip

ATMAdip

Nlrp3–/– (7 months)

VATSAT

Nlrp3–/

–

Asc–/

–

Figure 2 Elimination of Nlrp3 expression prevents obesity-induced caspase-1 cleavage and IL-1β and IL-18 activation. (a) Immunofluorescence staining of epididymal fat (eFat) tissue sections stained with antibodies against F4/80 (red) and Nlrp3 (green). Merge of images with nuclear staining with DAPI shows colocalization of Nlrp3 with ATM (yellow arrowheads). Negative control staining with antibody to Nlrp3 together with antibody to F4/80 in adipose tissue of Nlrp3−/− mice shows reduced Nlrp3-specific immunostaining. (b,c) The qRT-PCR analysis of Nlrp3 (b) and Pycard (c) mRNA in purified ATM and SVF cells derived from SAT and VAT of 6-month-old DIO mice, and mature 3T3-L1 adipocytes (Adip). (d) Immunoblot analysis showing the kinetics of caspase-1 (p20) cleavage and active IL-1β (p17) accumulation in adipose tissue of mice at various stages of diet-induced obesity. The age, in months, of the mice at the time of tissue harvesting is indicated. (e) Western blot analysis of activated caspase-1 (p20) in VAT, SAT and liver tissues from 9-month-old DIO-WT (C57BL/6) and DIO-Nlrp3−/− mice. Results shown are representative of three independent experiments. MW, molecular weight. (f) Western blot analysis of activated caspase-1 (p20) in kidney from 9-month-old DIO-WT and DIO-Nlrp3−/− mice. (g) Western blot analysis of IL-1β activation in adipose tissue of 6- and 7-month-old DIO mice. (h) Serum IL-18 concentration in age-matched WT and Nlrp3−/− mice fed normal chow or 60% HFD, starting at 2 months of age for all groups, for 4 months and 7 months. All data are presented as means ± s.e.m., n = 6–10 mice; *P < 0.05.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


processing in VAT (data not shown). Notably, compared to normal chow–fed mice, obese mice had significantly higher serum IL-18 con-centrations, which were blocked upon ablation of Nlrp3 (Fig. 2h). The serum concentrations of IL-1β in our WT and Nlrp3−/− mice were below the detection limit of the ELISA (data not shown). These data indicate that the Nlrp3 inflammasome is specifically activated in response to HFD and controls the production of IL-1β in adipose tissue and IL-18 in obesity.

Nlrp3 inflammasome impairs insulin sensitivity in obesityTo further understand the role of the Nlrp3 inflammasome in regula-ting insulin action, we performed insulin and glucose tolerance tests (ITTs and GTTs) in WT and Nlrp3−/− mice that had been fed a HFD, starting at 2 months of age, for 6 weeks, 4 months and 7 months. The ITTs and GTTs revealed that in early stages of obesity, the elimination of the Nlrp3 inflammasome affords substantial protection against HFD-induced insulin resistance (Fig. 3a). In 6-month-old normal chow–fed control male mice, ablation of Nlrp3 did not alter insulin action (Supplementary Fig. 3a,b) or affect glu-cose homeostasis, suggesting specific activation of the Nlrp3 inflam-masome in diet-induced obesity. Compared to 6-month-old WT DIO mice, Nlrp3−/− mice showed a considerable reduction in glucose after insulin injection (Fig. 3b), as well as an improvement in GTT (Fig. 3b). Furthermore, compared to age-matched control WT mice,

the 9-month-old obese Nlrp3−/− mice (again, on HFD for 7 months) showed a significant reduction in fasting glucose levels (Fig. 3c) as well as improved glucose homeostasis, as revealed by GTT (Fig. 3c). Because of the lower baseline glucose values of Nlrp3−/− mice (Fig. 3c), the percentage reduction in glucose after insulin injection was not significantly different. Estimation of the area under the curve of ITT in 6- and 9-month-old DIO mice indicated a significant (P < 0.05) reduction in glucose concentrations in Nlrp3−/− mice (Fig. 3d).

Consistent with an association of Nlrp3 with adipocyte size and insulin sensitivity in obese humans with T2DM (Supplementary Table 1), genetic ablation of Nlrp3 in mice was associated with a reduction in the fat cell size in VAT (Fig. 3e) but not in inguinal SAT (data not shown). Although the epididymal fat pad weight was lower in 3-month-old DIO Nlrp3−/− mice than DIO WT control mice, the final body weights of male WT and Nlrp3−/− mice after 6 weeks, 4 months and 7 months of 60% HFD (or normal chow diet) feeding did not show significant differences (Supplementary Fig. 4a,b). We did not detect any change in blood triglyceride and cholesterol concentrations in middle-aged obese Nlrp3−/− mice (Supplementary Fig. 4c,d).

Given that the metabolic effects of insulin are dependent on phospho-inositol 3-kinase (PI3K)-AKT signaling, we investigated the Ser473 phosphorylation of AKT as a readout of AKT activity after 5-min and 10-min insulin injection in 8-month-old obese WT (n = 9) and Nlrp3−/− mice (n = 9). Consistent with higher insulin sensitivity, the Nlrp3−/−

*

Crat

0

0.5

1.0

1.5

2.0

Fol

d ch

ange

Fol

d ch

ange

00.51.01.52.02.5

Ehhadh

**

0

0.5

1.0

1.5

2.0Cpt1a

*

Fol

d ch

ange

0

0.3

0.6

0.9

1.2

1.5Acaca

100 µm

200 µm

pAKT Ser473 pAKT Ser473

pAKT Ser473

pAKT Ser473

pAKT Ser473

pAKT Ser473

pAKT Ser473

pAKT Ser473

VA

TS

AT

Akt

Akt

VA

TS

AT

Akt

Akt

Mus

cle

Akt Mus

cle

Akt

5 min after insulin inj.

Live

r

Akt Live

r

Akt

VA

T

pErk-1/2

STAT5A

10 min after insulin inj.

SA

T

pErk-1/2

STAT5A

pIRS-1 Ser332/336

pIRS-1 Ser332/336

Live

r

IRS-1

IRS-1SA

T

f

h

200 µm 200 µm

d

e

g

8

9

10

11

12

13

14

*

8

10

12

14

16

*

WT

WT

WT

ITT ITT

Fol

d ch

ange

AU

C (

× 10

3 )

AU

C (

× 10

3 )

Nlrp3–/

–

WT WT Nlrp3–/– Nlrp3–/–

WT Nlrp3–/–WT Nlrp3–/–

Nlrp3–/–Nlrp3+/+

Nlrp3–/–

Nlrp3–/

–

WT-DIO WT-DIO Nlrp3–/– DIO Nlrp3–/– DIO

0

50

100

150

200

250

0 10 20 40 60 90 0 30 60Time (min)

100200300400500

0

**

*

cITT

90

GTT

Time (min)

Glu

cose

(m

g dl

–1)

Glu

cose

(m

g dl

–1)

a

020406080

100120

45 60 90Time (min)

0 3015

Glu

cose

(%

cha

nge) *

**

*ITT

90

100

GTT

200

300

400

500

015 30 45 60

Glu

cose

(m

g dl

–1)

Time (min)

* * * *

WT Nlrp3–/–

30 60 900

50

100

150

200

250

0Time (min)

900 30 60

100200300400500

0

**

bITT GTT

Time (min)

Glu

cose

(m

g dl

–1)

Glu

cose

(m

g dl

–1)

02468

101214

*

WT Nlrp3–/–

Cel

l siz

e(×

103 , m

m2 )

Figure 3 The Nlrp3 inflammasome regulates insulin sensitivity and steatohepatitis in obesity. (a) Insulin tolerance test (ITT) and glucose tolerance test (GTT) in male WT and Nlrp3−/− DIO mice fed 60% HFD for 6 weeks. (b,c) The ITT and GTT in 6-month-old (b) and 9-month-old (c) WT and Nlrp3−/− DIO mice showing the area under the curve (green for WT and blue for Nlrp3−/−) with trend lines (n = 5–7 per group). (d) The total area under the curve (AUC) for ITT in 6- (left) and 9-month-old (right) WT and Nlrp3−/− DIO mice. (*P < 0.01). (e) Adipocyte size in visceral fat tissue from 8-month-old DIO-WT (Nlrp3+/+) and Nlrp3−/− mice (H&E staining). Representative micrographs and the quantified results are shown. n = 6; *P < 0.001. (f) Akt, IRS-1 and Erk-1/2 signaling in vivo, as determined by western blotting for phosphorylation, in VAT, SAT, liver and muscle from the 8-month-old DIO Nlrp3+/+ and DIO Nlrp3−/− mice. The Ser322/338 antibody detects phosphorylation at Ser332 and Ser336. (g) H&E staining of sections from liver of 9-month-old DIO WT and Nlrp3−/− mice. (h) Fatty acid oxidation and fatty acid synthesis gene expression, as examined by qRT-PCR in the liver of 9-month-old DIO WT and Nlrp3−/− mice. mRNA expression was normalized to glyceraldehyde 3-phosphate dehydrogenase (Gapdh) expression and is shown as fold change (∆∆Ct) with the values for WT arbitrarily set to 1. All data are presented as means ± s.e.m. (n = 5), *P < 0.05, **P < 0.001.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


mice had significantly greater (P < 0.05) AKT activity in VAT, SAT, liver and muscle com-pared to the insulin-treated obese WT control mice (Fig. 3f and Supplementary Fig. 5), whereas vehicle-injected control mice did not show any AKT phosphorylation (Supplementary Fig. 6a). Consistent with enhanced insulin sensitivity in the absence of the Nlrp3 inflammasome, there was also a reduction in serine phos-phorylation of insulin receptor substrate-1 (IRS-1) in liver and fat of Nlrp3−/− obese mice (Fig. 3f and Supplementary Fig. 5e).

Apart from the PI3K-AKT pathways, insulin can also signal by pro-moting the interaction of the adaptor protein growth factor receptor bound protein 2 (Grb2) with the son-of-sevenless protein to activate Ras–mitogen-activated protein kinase (MAPK) signaling37. Considering that the Ras-MAPK pathway primarily regulates the nonmetabolic effects of insulin such as cell growth, differentiation and survival37, we next tested the specificity of insulin signaling in Nlrp3 inflammasome– deficient obese mice. Notably, compared to control mice, we found marked activation of the MAPKs extracellular signal–regulated kinases 1 and 2 (Erk1/2) specifically in the VAT of DIO Nlrp3−/− mice but not in SAT liver or muscle (Fig. 3f and Supplementary Fig. 6b). These data suggest that improvement of insulin sensitivity in Nlrp3-deficient mice is related to overall stimulation of the PI3K-AKT pathway, whereas the MAPK pathway is selectively activated in visceral fat.

Given that knockout of Nlrp3 results in a reduction in caspase-1 activation in the liver and also improvement in insulin signaling in liver, we also investigated hepatic steatosis in 9-month-old DIO Nlrp3−/− mice. The histological evidence suggests that, compared to 9-month-old DIO WT mice, obese Nlrp3−/− mice have reduced hepatic steatosis (Fig. 3g), which is consistent with recent findings that acti-vation of the Nlrp3 inflammasome induces hepatic fibrosis and liver injury38,39. Furthermore, the reduction in fatty liver disease in obese Nlrp3−/− mice was associated with an increase in fatty acid oxidation regulators enoyl–coenzyme A hydratase/3-hydroxyacyl coenzyme A dehydrogenase (encoded by Ehhadh), carnitine palmitoyltransferase 1a (Cpt1a) and carnitine acetyltransferase (Crat), with no change in mRNA expression of acetyl–coenzyme A carboxylase alpha (Acaca) and fatty acid synthase (Fasn) (data not shown), which are known to regulate fatty acid synthesis (Fig. 3h).

Ceramides activate Nlrp3 inflammasome and ATMs in obesityInhibition of obesity-induced caspase-1 activation and insulin resistance in Nlrp3−/− mice prompted us to investigate whether the Nlrp3 inflammasome senses specific inducers that activate innate immune cells such as macrophages. Recent evidence sug-gests that during progressive obesity the development of adipose tissue fibrosis restricts adipocyte expansion, which may ultimately cause lipid spillover in tissues and increases in circulating levels of free fatty acids40. Because ATMs can scavenge lipids, and generation of ceramide from fatty acids during obesity induces inflammation41–44, we tested whether the Nlrp3 inflammasome senses ceramide. We primed bone marrow–derived macrophages (BMDMs) with lipopolysaccharide (LPS) to induce transcriptional activation of IL-1β, and secondary inflammasome activation sig-nal was provided by ceramides as established previously27,28. The immunoblot analysis revealed that, together with the positive con-trol (LPS plus ATP), exposure of WT macrophages to ceramides causes activation of caspase-1 (Fig. 4a). Notably, ceramide-induced caspase-1 activation was blocked in the absence of Nlrp3 (Fig. 4a). Consistent with these data, we observed that ceramide-induced IL-1β secretion from macrophages was reduced in absence of Nlrp3 (Fig. 4b).

Next, we investigated the physiological relevance of these data and tested whether ceramides can induce caspase-1 activation within adipose tissue. We cultured epididymal adipose tissue explants from 9-month-old WT DIO and Nlrp3−/− DIO mice in the presence of LPS and ceramide. As in BMDMs, the LPS priming and ceramide-induced caspase-1 activation was reduced in adipose tissue explants of Nlrp3−/− mice (Fig. 4c). Considering that within adipose tissue, the ATMs highly express the Nlrp3 inflammasome components (Fig. 2b,c), we investigated whether reduction of Nlrp3-mediated sensing of obesity- associated proinflammatory inducers affects the proinflammatory profiles of ATMs in vivo. The ATMs were enriched from the stromal vascular fraction (SVF) through positive selection of F4/80-expressing cells from VAT and SAT of 9-month-old control WT and Nlrp3−/−

0

0.5

1.0

1.5

2.0 Il10

**

**

VAT SAT0

0.5

1.0

1.5

2.0

2.5

3.0Arg1

*

VAT SAT

0

0.2

0.4

0.6

0.8

1.0

*

*

VAT SAT

Fol

d ch

ange

Tnfa

0

0.2

0.4

0.6

0.8

1.0

1.2Nos2

VAT SAT

0

0.2

0.4

0.6

0.8

1.0

Fol

d ch

ange

VAT SAT

Ccl20

0

0.2

0.4

0.6

0.8

1.0

1.2

VAT SAT

Cxcl11

*

d

Procasp-1

p20

+/+ +/+ –/–+/+

IL-1

β (p

g m

l–1)

0

10

20

30

40

50

LPS

LPS +

cera

mide

LPS +

cera

mideLP

S

ba

c

Caspase-1(p20)

Actin

BMDM

Adipose tissue explants

–/–

F4/80+ adipose tissue macrophages

Fol

d ch

ange

Fol

d ch

ange

F

old

chan

ge

Fol

d ch

ange



Nlrp3+/+ Nlrp3–/–

WTNlrp3–/–

WTNlrp3–/–

WTNlrp3–/–

LPS +

cera

mide

LPS +

ATP

LPS

LPS +

ATP

LPS +

cera

mide

Ceram

ide

LPS +

cera

mide

LPS +

cera

mide

LPS

LPS

Vehicl

e

Ceram

ide

LPS +

cera

mide

LPS +

cera

mide

LPS

LPS

Vehicl

e

Figure 4 Nlrp3 senses ceramide to induce IL-1β and regulates adipose tissue macrophage activation in obesity. (a) The cell extracts of LPS-primed bone marrow–derived macrophages (BMDMs) from Nlrp3+/+ and Nlrp3−/− mice (n = 6–9) were analyzed for caspase-1 active form by western blotting as indicated. (b) The BMDM cells were stimulated with LPS and C2 ceramide (0.1 mM), and cell supernatants were analyzed for IL-1β. Results are representative of three separate experiments. (c) Epididymal adipose tissue explants from 9-month-old Nlrp3+/+ and Nlrp3−/− DIO mice were cultured for 24 h in the presence of LPS, C2 ceramide or both. Caspase-1 activation was determined by immunoblot analysis. (d) M1- and M2-associated gene expression, as examined by qRT-PCR in macrophages originating from VAT and SAT of 9-month-old DIO-Nlrp3+/+ and DIO-Nlrp3−/− mice. The mRNA expression was normalized to glyceraldehydes 3-phosphate dehydrogenase (Gapdh) and shown as fold change (∆∆Ct) with the values for WT arbitrarily set to 1. All data are presented as means ± s.e.m., *P < 0.01, **P < 0.001.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


DIO mice. We found that loss of Nlrp3 function increased the expres-sion of the M2 macrophage–associated transcripts interleukin-10 (encoded by Il10) and arginase (Arg1) in SAT ATMs but not in VAT macrophages (Fig. 4d). These results suggest that obese Nlrp3−/− mice specifically retain an M2-like macrophage phenotype in the inguinal SAT. In addition, we found that M1 macrophage–associated expres-sion of tumor necrosis factor (Tnfa), chemokine (C-C motif) ligand 20

(Ccl20) and chemokine (C-X-C motif) ligand 11 (Cxcl11) were speci-fically reduced in the visceral fat–derived ATMs but not in SAT macrophages (Fig. 4d). Also, the expression of Il10 was increased in SAT of Nlrp3−/− mice, whereas Tnfa expression was reduced in obese Nlrp3-deficient mice (Supplementary Fig. 7). Compared to WT obese mice, the ATMs derived from the SAT of Nlrp3−/− obese mice showed lower inducible nitric oxide synthase 2 (Nos2) expression

CD

8

CD4

CD

62L

CD

62L

CD44 CD44

28.6 ± 1.2

26.3 ± 3.1

27.3 ± 2.3

32.6 ± 4.5 0.5 ± 0.1

36.8 ± 3.5

1.2 ± 0.4

43.8 ± 5.7 76.3 ± 1.5

79.6 ± 4.9

8.4 ± 2.0

6.7 ± 2.8

81.6 ± 5.8

74.3 ± 2.1

6.1 ± 2.1

3.7 ± 1.2

0

0.05

0.10

0.15

Cel

l num

ber

(× 1

06 )C

ell n

umbe

r (×

106 )

0.20

0.25

CD4+

0

0.05

0.10

0.15

0.20

CD4+ effectormemory

0

0.005

0.010

0.015

0.020

*

CD4+ naive

CD8+ CD8+ effectormemory

CD8+ naive

00.050.100.150.200.250.30

0

0.05

0.10

0.15

0.20

0.25

00.0020.0040.0060.0080.0100.012

*

CD4

CD

8

CD

62L

CD44

CD

62L

CD44

FSC

SS

CW

T (

VA

T)

SS

C

FSC

*

*

a

b

CD11c CD11c

CD

206

F4/80 F4/80

28.2%

35.3%

35.34

28.20

4.1% 4.10

4.9%4.94

24.09%24.09

14.8%

35.1%

4.2%

4.5%

0

20.19%

FSC FSCS

SC

WT VAT Nlrp3–/– VAT

0.2%

0.20

0.3%

Gate 1

Gate 2

Gate 1

Gate 2

24%

11%

VAT SAT0

10

20

30

40

50

60

Tre

g ce

lls (

% g

ated

)

c

d e f

0

1

2

3

4

5

6

7

P = 0.004

P = 0.005

SV

C (

mili

on p

er g

fat)

P = 0.024

9 month DIO3 month DIO

Fox

P3

CD4CD25

SS

C

FSC

28.7%28.71

12.97

3.9%

35.5%

Fox

P3

CD4

12.9%

6.4%

6.40

32.1%

1.08 32.11

11.5255.28

CD25

SS

C

FSC

WT DIO Nlrp3–/– DIO

WT DIO

CD4+CD25+FoxP3+

Nlrp3–/– DIO

WT DIONlrp3–/– DIO

CD

206

SS

C

100 101 102 103 104

100 101 102 103 104 100 101 102 103 104

100100

101

101

102

102

103

103

104

100 101 102 103 104

100 101 102 103 104100 101 102 103 104

104

100

101

102

103

104

100

101

102

103

104

100

101

102

103

104

0.35

0

0.35

20.19

35.01

4.54

4.27

14.88

24.91

11.87

Nlrp

3–/– (

VA

T)

Nlrp

3–/– V

AT

WT

VA

T

35.50

22.2840.34

1.89

3.91

Inguin

al

Perire

nal

Epidid

ymal

Epidid

ymal

1.04

Figure 5 Ablation of the Nlrp3 inflammasome reduces adipose tissue effector T cells without affecting Treg cells in visceral fat of obese mice. (a) FACS plot of SVF cells isolated from VAT of 9-month-old WT and Nlrp3−/− obese mice. The dot plots depict FSC and SSC (left) and the sequential gating strategy for analysis of ATMs and T cells. Gate 1 (top) of larger cells shows the presence of F4/80+ cells (histogram) and expression of macrophage markers CD206 and CD11c on ATMs. Gating of smaller cells (gate 2, bottom) reveals the absence of ATMs in this population of SVF. (b) The gate 2 (lymphoid gate) showing CD4+ and CD8+ T cells in SVF. CD4+ and CD8+ cells were evaluated for naive T cells (CD62L+CD44−, blue boxes) and effector memory (CD62L−CD44+, red boxes) CD4+ and CD8+ T cells. The FACS analysis was repeated in three independent pooled SVF fractions from a total of 12–14 mice and percent gated cell frequencies are indicated in each representative dot plot. (c) Gated percentage and absolute numbers (in million cells) of naive (CD62L+CD44−) and effector memory (CD62L−CD44+) CD4+ and CD8+ T cells. (d) Number of stromal vascular cells per gram of fat tissue (n = 4–6) in 3- and 9-month-old WT and Nlrp3−/− DIO mice. (e) Representative FACS plots showing CD4+CD25+Foxp3+ T regulatory cells in VAT of 9-month-old WT and Nlrp3−/− DIO mice. (f) Gated percentage of Treg cells in VAT and SAT of 9-month-old WT and Nlrp3−/− DIO mice (n = 6 per group). All data are presented as means ± s.e.m., *P < 0.05.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


(Fig. 4d). Collectively, these data suggest that ablation of Nlrp3 pro-tects from obesity-associated proinflammatory ATM activation.

Nlrp3 inflammasome regulates adipose-T cells in obesityIn addition to macrophages, the expanded adipose tissue during obesity also harbors activated T cell populations that are thought to participate in local adipose tissue inflammation and insulin resistance45–48. A reduction in adipose inflammation and improved insulin action due to Nlrp3 deficiency prompted us to investigate

ATMs and adipose tissue T cell subsets. We found that in VAT of 9-month-old DIO mice, the loss of Nlrp3 function did not affect the frequency of M1 (F4/80+CD11c+CD206−) or M2 (F4/80+CD11c−CD206+) ATMs (Fig. 5a). Notably, we observed reduced numbers of smaller cells (gate 2, on forward scatter (FSC) and side scatter (SSC), Fig. 5a) in SVF of obese Nlrp3−/− mice. Gating this population in SVF (gate 2) revealed that these cells are not ATMs (which are present in gate 1), as they lack expression of F4/80, CD11c and CD206.

c

a b

CD11c

CD

206

6 month WT SAT 9 month WT SAT6 month Nlrp3–/– SAT 9 month Nlrp3–/– SAT

0

2

4

6

8

10

12

14

16 WT

6 monthDIO

9 monthDIO

*11.9%

4.9% 3.9% 3.6% 4.6%

12.7% 7.5% 15.1%15.08

4.263.66

7.5612.79

3.984.93

11.98

d

25.26%25.26

27.3 ± 1.2

27.3 ± 0.643.5 ± 0.8

0.8 ± 0.5

48.3 ± 6.4

20 ± 4

WT

SA

T

SS

C

20.39%20.39

23 ± 2.6* 0.3 ± 0.1

54.7 ± 4.7 68.6 ± 8.3* 64.3 ± 9.421 ± 0.2 **

53 ± 4.3

e f

00.20.40.60.81.01.21.41.6

* *

VAT SAT0

200400600800

1,0001,200

0

50

100

150

200

250

*

*

**

Serum

Serum

CC

L2 (

pg m

l–1)

Actin

WT DIO

gWT

CD4+

CD8+

00.050.100.150.200.250.300.35 **

00.050.100.150.200.250.300.35

*

CD4+ effector memory *

CD4+ naive

CD8+ effector memory

*

CD8+ naive

0

0.03

0.06

0.09

0.12

0.15

0

0.02

0.04

0.06

0.08

0.10

0

0.05

0.10

0.15

0.20

00.010.020.030.040.050.060.070.08

WT

Cel

l num

ber

(×10

6 )C

ell n

umbe

r (×

106 )

Nlrp3–/–

Nlrp3–/–

InflammationIFN-γ

Effector T cellsCD44

+ CD62L

–

Insulin resistance

Lean state

No endogenous danger signals

Obese state

Danger signals: ceramides, LDL, necrosis, hypoxia, ROS, hydrophobic moieties

Sensing by Nlrp3Inflammasome

IL-18

IL-1β

(M1)Macrophage activation

Adipocyte death

Tissue–resident (M2)macrophage

Antigen presentation,tissue repair and

scavenging

No Nlrp3 inflammasomeactivation

Resident T cellsEffector cell

Naive T cell

Maintenance of tissue function

Chronic inflammation

h Adipose tissueNlrp3–/–

Nlrp3–/– DIO

Nlrp

3–/– S

AT

CD

11c_ C

D20

6+

(%

gat

ed)

FSC

CD

8

CD4

CD

62L

CD44

CD

62L

CD44

SS

C

FSC

CD

8

CD4

CD

62L

CD44

CD

62L

CD44

Fol

d ch

ange

Ifnγ

IFNγ

CX

CL1

0 (p

g m

l–1)

WT le

an

WT D

IO

Nlrp3–/

– DIO

WT le

an

WT D

IO

Nlrp3–/

– DIO

Cas

pase

1ac

tivat

ion

104

103

102

101

100

100 101 102 103 104

104

103

102

101

100

100 101 102 103 104

104

103

102

101

100

100 101 102 103 104

104

103

102

101

100

100 101 102 103 104

15.5 ± 6.8* 12.6 ± 6.2

29.6 ± 4.7

Figure 6 Elimination of the Nlrp3 inflammasome reduces obesity-induced macrophage–mediated T cell activation in adipose tissue. (a) Representative FACS plots of SVF cells from SAT of 6- and 9-month-old WT and Nlrp3−/− DIO mice stained with CD206 and CD11c. (b) The gated percentage of CD11c−CD206+ M2 cells in SAT of 9-month-old obese mice. (c) Dot plots showing naive (CD62L+CD44−, blue boxes) and effector memory (CD62L−CD44+, red boxes) CD4+ and CD8+ T cells in SVF from 9-month-old obese WT and Nlrp3−/− mice. The percent gated cell frequencies are indicated in each representative dot plot (d) Gated percentage and absolute numbers (in million cells) of naive (CD62L+CD44−) and effector memory (CD62L−CD44+) CD4+ and CD8+ T cells. (e) mRNA level of the TH1 cytokine Ifnγ in VAT and SAT of 9-month-old obese WT and Nlrp3−/− mice as determined by qRT–PCR. (f) IFN-γ (19-kDa) in VAT of 9-month-old WT and Nlrp3−/− DIO mice, as examined by western blotting. (g) IP10 and MCP-1 levels in the serum of 9-month-old lean WT and obese WT and Nlrp3−/− mice (n = 5). The data shown are means ± s.e.m., *P < 0.05. (h) Hypothetical model of Nlrp3 inflammasome activation in obesity. In the absence of danger signals in the healthy lean state, tissue-resident macrophages and T cells may participate in the maintenance of adipose tissue function. In obesity, the Nlrp3 inflammasome senses the obesity-associated danger signals such as ceramides leading to caspase-1 autoactivation and IL-1β and IL-18 production from ATMs. Secondary signals from activated ATMs to effector adipose T cells (defined as CD44+CD62L−) sustain the reciprocal proinflammatory feed-forward cascade during obesity, leading to insulin resistance. ROS, reactive oxygen species.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


Further examination of smaller cells in gate 2 within SVF revealed that in WT mice approximately 60% of these cells are T lymphocytes (Fig. 5b). Compared to WT controls, 9-month-old Nlrp3-deficient mice did not show significant differences in the frequencies of CD4+ and CD8+ cells (Fig. 5b). Furthermore, in 9-month-old obese mice, elimination of Nlrp3 inflammasome signaling did not affect the overall ratio (percentage gated) of CD4+ and CD8+ effector memory (CD62L−CD44+) or CD4+ and CD8+ naive T cell subsets (CD62L+CD44−) (Fig. 5b). Compared to WT control mice, the ablation of Nlrp3 in 9-month-old DIO mice reduced the overall per-centage of lymphocytes (in FSC and SSC gate 2) within the SVF of VAT (Fig. 5b). The estimation of T cell numbers within SVF revealed that loss of Nlrp3 led to a significant (P < 0.05) reduction in the overall numbers of CD4+ T cells, CD4+ effector memory cells, CD8+ T cells and CD8+ effector memory cells without significant difference in naive T cells (Fig. 5c). Consistent with these data, compared to WT DIO mice, Nlrp3-deficient obese mice had a significant reduction in the total number of SVF cells in the visceral fat at both early (3 month olds) and chronic (9 month olds) stages of obesity (Fig. 5d).

Among various T cell subsets, the CD4+CD25+FoxP3+ T regula-tory (Treg) cells constitute a major defense mechanism to dampen the proinflammatory response in several autoimmune and infec-tious diseases45. Recent studies have shown that the number of anti- inflammatory Treg cells in adipose tissue are reduced in obesity, and increasing the Treg cell number improves insulin action45. Analysis of Treg cells in visceral fat (epididymal and perirenal) and subcutaneous fat (inguinal) in middle-aged DIO mice revealed no significant differ-ences between WT and Nlrp3-deficient mice (Fig. 5e,f), suggesting that the Nlrp3-dependent effects on adipose tissue effector T cell subsets are specific.

Together with the VAT, the SAT also constitutes a substantial pro-portion of adipose mass in obese mice. Considering that obesity-associated caspase-1 autoactivation in SAT is Nlrp3 dependent, we next investigated whether the reduction in macrophage activation and insulin sensitivity is associated with changes in distribution of naive and effector memory adipose resident T cell subsets in sub-cutaneous fat. Notably, elimination of the Nlrp3 inflammasome in 9-month-old obese mice increased the number of M2 macrophages (F4/80+CD11c−CD206+) without affecting the M1 macrophage (F4/80+CD11c+CD206−) frequency (Fig. 6a,b). Unexpectedly, com-pared to WT obese mice, the adipose tissue T cell number of both CD4+ and CD8+ cells was significantly higher in Nlrp3−/− mice (Fig. 6c,d). Notably, examination of total CD4+ and CD8+ cells revealed that there was a preponderance of naive cells (CD4+CD62L+CD44−, CD8+CD62L+CD44−) within the adipose T cell compartment of Nlrp3-deficient DIO mice (Fig. 6c,d), suggesting a lower inflamma-tory profile in fat.

Considering that macrophage-derived IL-18 induces a T helper1 (TH1) response for host defense against specific infections, we also investigated the amount of IFN-γ expression, a classical TH1-derived cytokine in the context of non-infectious sterile obesity–induced inflammation in Nlrp3-deficient mice. Consistent with the reduc-tion in expression of obesity-induced IL-18 in Nlrp3−/− mice (Fig. 2h), ablation of the Nlrp3 inflammasome lowered the expression of Ifng mRNA, as well as protein, in adipose tissue during obesity (Fig. 6e,f). Furthermore, in the absence of the Nlrp3 inflammasome, the obesity-induced increase in circulating interferon-γ–inducible protein (IP-10) and monocyte chemoattractant protein-1 (MCP-1) were significantly reduced (Fig. 6g) without any change in IL-17 abundance (data not shown). Collectively, the data suggest that blocking the activation of

the Nlrp3 inflammasome in response to obesity-related danger signals may lower macrophage–T cell activation that participate in sustaining chronic inflammation (Fig. 6h).

DISCUSSIONObesity is associated with self-directed tissue inflammation where local or systemic factors other than infectious agents activate the cells of the innate immune system. Despite the evidence of skewed T cell receptor repertoire in obesity4,47,48, the evidence that sterile inflammation during obesity results from underlying autoimmune processes or due to the presence of organ-specific autoantibodies has so far not been definitively shown. Therefore, obesity-associated inflammation may qualify as autoinflammation, which was defined by McGonagle and McDermott as self-directed tissue inflammation, where local factors at disease-prone sites determine activation of the innate immune system49. The influx of macrophages as well as T cells in adipose tissue upon chronic caloric excess–driven obesity and release of proinflammatory mediators by these cells causes insulin resistance6–9,31,45–48. However, an upstream initiating event for obesity-induced immune cell activation in adipose tissue is not established. Here we have shown that the Nlrp3 inflammasome has a substantial role in sensing obesity-associated inducers of caspase-1 activation and therefore regulates development and the magnitude of inflammation and its downstream effects on insulin signaling.

Our findings that the adipose tissue expression of IL-1β and Nlrp3 inflammasome components are coupled with the development of insulin resistance and severity of T2DM in obese individuals raised several questions. First, can caspase-1 undergo autoactivation in obesity, and does eliminating the signaling through Nlrp3 reduce adipose inflammation and improve insulin action? We found that induction of HFD-induced obesity caused marked caspase-1 activa-tion in adipose tissue and liver. Such a steady-state in vivo activation of caspase-1 in obesity has so far not been reported in the context of self-directed tissue inflammation. The loss of Nlrp3 function reduced, but did not eliminate, the caspase-1 activation in visceral fat, subcutaneous fat and liver, suggesting that other inflammasomes may contribute to the pathophysiology of obesity. In addition, Nlrp3 inflammasome acti-vation was tissue specific, as obesity-induced caspase-1 activation in kidney was not affected in Nlrp3-deficient mice. Consistent with these data, obese Nlrp3−/− mice were more insulin sensitive.

What is being sensed by the Nlrp3 inflammasome to induce caspase-1 activation in obesity? This question is difficult to answer, particularly in a complex disorder such as diet-induced obesity where alterations in several cell stress–associated metabolites could potentially activate Nlrp3 or other inflammasomes (Fig. 6h). It has been shown that saturated fatty acids such as oleate and palmitate can induce inflammation by TLR4 (ref. 30) and loss of TLR4 func-tion can partially protect against obesity-induced insulin resistance31. Notably, obesity-related increased levels of lipotoxic ceramides can induce cell death and also trigger inflammation41–44. Our data show that ceramides induce caspase-1 activation in an Nlrp3-dependent mechanism. However, it is possible that in diet-induced obesity several other inducers may participate in caspase-1 activation. For example, the Nlrp3 inflammasome has recently been shown to be activated by oxidized low-density lipoprotein and cholesterol crystals in models of HFD-induced atherogenesis, leading to macrophage activation and IL-1β secretion50. Cells undergoing necrosis can also activate the Nlrp3 inflammasome in macrophages51. Considering that the development of obesity is also associated with hypoxia52 and adipocyte death53, it is likely that ATMs that form crown-like structures can be activated

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

a r t i c l e s


via the Nlrp3 sensing pathway. Furthermore, visceral adiposity and obesity are associated with an increase in monosodium urate (MSU) levels and increased risk of diabetes and atherosclerosis54,55. Given that MSU is sensed by the Nlrp3 inflammasome and causes increased inflammation28, it is possible that MSU may be one of the danger sig-nals sensed by the Nlrp3 inflammasome, thus contributing to obesity-induced disease. Further, it is also likely that reduced sensing of such DAMPs as a result of Nlrp3 deficiency protects from long-term obesity and hyperglycemia-associated insult (Fig. 6h).

Lastly, which mediators downstream of the Nlrp3 inflammasome participate in the mechanism of obesity-induced inflammation and insulin sensitivity? Although the Nlrp3 inflammasome specifically regulates caspase-1 activation, which allows the release of IL-1β and IL-18, both of these cytokines in turn can initiate several events that may amplify inflammatory responses21. Our prior studies have found that obesity is associated with an increase in the number of IFN-γ+ T cells in the adipose tissue48. It is well known that IFN-γ alone or together with microbial stimuli (such as LPS) can cause induction of classically activated M1 macrophages17. In the context of obesity- associated inflammation, we provide the additional mechanistic insight that a reduction in IL-1β and IL-18 processing is regulated by the Nlrp3 inflammasome, which may deliver secondary signals to adipose tissue T cells and induce an effector TH1 proinflammatory profile in adipose tissue. Consistent with our data of reduced IFN-γ expression in Nlrp3−/− mice and improved insulin signaling in these mice, the ablation of IFN-γ is associated with improved metabolic outcomes in obesity56. Furthermore, increases in M1 macrophage–derived cytokines and effector T cells in adipose tissue are causally linked to insulin resistance8,45–48. Together with our previous studies that show that purified adipose T cells produce several cytokines that further activate macrophages48, our present data show that elimination of the Nlrp3 inflammasome reduces M1-like macrophage gene expression and increases the expression of M2-like cytokines. Together these data suggest that the Nlrp3 inflammasome sensing pathway participates in the origin of inflammation in obesity by inducing macrophage and subsequent T cell activation (Fig. 6h).

Notably, blocking the obesity-induced gain of Nlrp3 inflamma-some function did not affect Treg cell homeostasis in adipose tissue but specifically reduced the numbers of effector memory cells and increased the naive CD4+ and CD8+ T cell populations. Consistent with our data, reduction in effector memory cells counts in VAT may be partly linked to a decrease in CCL20 and CXCL11 chemokine expression57. The improvement in overall metabolic profiles upon elimination of Nlrp3 signaling may also be related to less hepatic steatosis and regulation of genes that control hepatic fatty acid oxidation in obese mice. It has recently been reported that uric acid–induced release of reactive oxygen species facilitates the interaction of thioredoxin-interacting protein to Nlrp3, leading to IL-1β release58. Consistent with our findings, these studies also support a role of the Nlrp3 inflammasome in regulating glucose homeostasis32,58.

It was recently reported that inflammasome-mediated caspase-1 processing in adipocytes impairs insulin sensitivity32. In our studies, we found that Nlrp3 and ASC expression in adipocytes, as compared to adipose tissue macrophages, was very low. In addition, we found that the high expression of Nlrp3 and ASC in primary adipocyte fractions of enzymatically digested adipose tissue may be attributable in large part to lipid-laden macrophages that contaminate enriched adipocyte fractions. Whether 3T3-L1 adipocytes or primary adipocytes can pro-cess procaspase-1 into active p20 and p10 heterodimers in response to specific signals via inflammasomes remains to be established.

Although TNF induces insulin resistance in vitro and in mice, the attempts to block TNF signaling in obese individuals with T2DM have so far yielded disappointing results59. In contrast, inhibiting IL-1β signaling by the IL-1 receptor antagonist anak-inra can dampen systemic inflammation, including by decreas-ing circulating C-reactive protein and IL-6 and reducing type 2 diabetes in humans12,13. Furthermore, widely used antidiabetic sulfonylurea drugs such as glyburide can also block Nlrp3 inflam-masome activation and reduce mortality due to septic shock60. Taken together, our data establish that Nlrp3 inflammasome–dependent post-translational processing of IL-1β and IL-18 in response to obesity-associated danger signals participates in the development of a chronic proinflammatory state that impairs insu-lin sensitivity. These findings highlight the potential of targeting the molecular pathways regulating caspase-1 activation in obesity for management of insulin resistance and chronic inflammation– induced comorbidities.

METhODSMethods and any associated references are available in the online version of the paper at http://www.nature.com/naturemedicine/.

Note: Supplementary information is available on the Nature Medicine website.

AcKnowLEDGMEntSWe thank Vishva M. Dixit at Genentech for providing caspase-1–specific antibody and Nlrp3−/− mice and J. Suttles from the University of Louisville for L929 media. We also thank S. Bond for expert technical assistance and D.H. Ryan, C. Bouchard and J.M. Salbaum for helpful discussions. This work was supported in part by pilot grants to B.V. and V.D.D. from the Nutrition and Obesity Research Center (US National Institutes of Health (NIH) National Institute of Diabetes and Digestive and Kidney Diseases grant P30 DK072476). The research in the Dixit laboratory is supported in part by the NIH (R01AG31797), the Coypu Foundation and the Pennington Biomedical Research Foundation. K.S. was partially supported by the NIH (DK083615). This work used the facilities of the Genomics and Cell Biology & Bioimaging Core supported by NIH grant 1 P20 RR02/1945.

AUtHoR contRIBUtIonSB.V. performed real-time PCRs, flow cytometry assays, adipose tissue macrophage selections, some western blots, ITT, GTT and cytokine assays, managed the transgenic animal colony and participated in experimental design, data analysis and manuscript preparation. Y.-H.Y. performed all caspase-1 and IL1β western blots, adipose and liver histologies, and macrophage culture experiments and analyzed the data. A.R. Performed body composition analysis, tissue collections, lipid analysis, animal husbandry, genotyping, ITT, GTT and adipocyte size measurement. J.E.G. and E.R. designed and supervised the human studies, analyzed the glucose and insulin sensitivity data in obese T2DM subjects and discussed the hypotheses. K.S. performed the caspase-1 western blot in the kidneys of HFD fed WT and Nlrp3 null mice. R.L.M. participated in standardizing the ITT and GTT assays and advised on the design of experiments for liver fatty acid synthesis and oxidation gene expression. J.M.S. performed the western blots for some of the insulin-signaling experiments, helped with data interpretation, discussed the hypotheses and participated in manuscript preparation. V.D.D. conceived the project, designed the experiments, performed some of the cytokine assays and flow cytometry, helped with data interpretation, participated in data analysis, directed the project and wrote the manuscript.

coMPEtInG FInAncIAL IntEREStSThe authors declare no competing financial interests.

Published online at http://www.nature.com/naturemedicine/. Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/.

1. Dixit, V.D. Adipose-immune interactions during obesity and caloric restriction: reciprocal mechanisms regulating immunity and health span. J. Leukoc. Biol. 84, 882–892 (2008).

2. Hotamisligil, G.S. & Erbay, E. Nutrient sensing and inflammation in metabolic diseases. Nat. Rev. Immunol. 8, 923–934 (2008).

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

http://www.nature.com/naturemedicine/

a r t i c l e s


3. Luchsinger, J.A. & Gustafson, D.R. Adiposity, type 2 diabetes and Alzheimer’s disease. J. Alzheimers Dis. 16, 693–704 (2009).

4. Yang, H. et al. Obesity accelerates thymic aging. Blood 114, 3803–3812 (2009).

5. Weisberg, S.P. et al. Obesity is associated with macrophage accumulation in adipose tissue. J. Clin. Invest. 112, 1796–1808 (2003).

6. Odegaard, J.I. & Chawla, A. Mechanisms of macrophage activation in obesity-induced insulin resistance. Nat. Clin. Pract. Endocrinol. Metab. 4, 619–626 (2008).

7. Lumeng, C.N., Bodzin, J.L. & Saltiel, A.R. Obesity induces a phenotypic switch in adipose tissue macrophage polarization. J. Clin. Invest. 117, 175–184 (2007).

8. Patsouris, D. et al. Ablation of CD11c-positive cells normalizes insulin sensitivity in obese insulin resistant animals. Cell Metab. 8, 301–309 (2008).

9. Hotamisligil, G.S. Inflammation and metabolic disorders. Nature 444, 860–867 (2006).

10. Arkan, M.C. et al. IKK-β links inflammation to obesity-induced insulin resistance. Nat. Med. 11, 191–198 (2005).

11. Jager, J., Grémeaux, T., Cormont, M., Le Marchand-Brustel, Y. & Tanti, J.F. Interleukin-1β–induced insulin resistance in adipocytes through down-regulation of insulin receptor substrate-1 expression. Endocrinology 148, 241–251 (2007).

12. Larsen, C.M. et al. Interleukin-1–receptor antagonist in type 2 diabetes mellitus. N. Engl. J. Med. 356, 1517–1526 (2007).

13. Osborn, O. et al. Treatment with an Interleukin 1 β antibody improves glycemic control in diet-induced obesity. Cytokine 44, 141–148 (2008).

14. Medzhitov, R. Recognition of microorganisms and activation of the immune response. Nature 449, 819–826 (2007).

15. Barton, G.M. & Kagan, J.C. A cell biological view of Toll-like receptor function: regulation through compartmentalization. Nat. Rev. Immunol. 9, 535–542 (2009).

16. Agostini, L. et al. NLRP3 forms an IL-1β–processing inflammasome with increased activity in Muckle-Wells autoinflammatory disorder. Immunity 20, 319–325 (2004).

17. Medzhitov, R. Origin and physiological roles of inflammation. Nature 454, 428–435 (2008).

18. Lamkanfi, M. & Dixit, V.M. Inflammasomes: guardians of cytosolic sanctity. Immunol. Rev. 227, 95–105 (2009).

19. Franchi, L., Eigenbrod, T., Muñoz-Planillo, R. & Nuñez, G. The inflammasome: a caspase-1–activation platform that regulates immune responses and disease pathogenesis. Nat. Immunol. 10, 241–247 (2009).

20. Martinon, F., Mayor, A. & Tschopp, J. The inflammasomes: guardians of the body. Annu. Rev. Immunol. 27, 229–265 (2009).

21. Dinarello, C.A. Immunological and inflammatory functions of the interleukin-1 family. Annu. Rev. Immunol. 27, 519–550 (2009).

22. Kostura, M.J. et al. Identification of a monocyte specific pre–interleukin 1 β convertase activity. Proc. Natl. Acad. Sci. USA 86, 5227–5231 (1989).

23. Kuida, K. et al. Altered cytokine export and apoptosis in mice deficient in interleukin-1 beta converting enzyme. Science 267, 2000–2003 (1995).

24. Martinon, F., Burns, K. & Tschopp, J. The inflammasome: a molecular platform triggering activation of inflammatory caspases and processing of proIL-β. Mol. Cell 10, 417–426 (2002).

25. Dostert, C. et al. Innate immune activation through Nlrp3 inflammasome sensing of asbestos and silica. Science 320, 674–677 (2008).

26. Halle, A. et al. The NLRP3 inflammasome is involved in the innate immune response to amyloid-β. Nat. Immunol. 9, 857–865 (2008).

27. Mariathasan, S. et al. Cryopyrin activates the inflammasome in response to toxins and ATP. Nature 440, 228–232 (2006).

28. Martinon, F., Pétrilli, V., Mayor, A., Tardivel, A. & Tschopp, J. Gout-associated uric acid crystals activate the NLRP3 inflammasome. Nature 440, 237–241 (2006).

29. Mariathasan, S. et al. Differential activation of the inflammasome by caspase-1 adaptors ASC and Ipaf. Nature 430, 213–218 (2004).

30. Shi, H. et al. TLR4 links innate immunity and fatty acid–induced insulin resistance. J. Clin. Invest. 116, 3015–3025 (2006).

31. Saberi, M. et al. Hematopoietic cell-specific deletion of Toll-like receptor 4 ameliorates hepatic and adipose tissue insulin resistance in high-fat-fed mice. Cell Metab. 10, 419–429 (2009).

32. Stienstra, R. et al. The inflammasome-mediated caspase-1 activation controls adipocyte differentiation and insulin sensitivity. Cell. Metab. 12, 593–605 (2009).

33. Bishop, N.A. & Guarente, L. Genetic links between diet and lifespan: shared mechanisms from yeast to humans. Nat. Rev. Genet. 8, 835–844 (2007).

34. Galgani, J.E. et al. Metabolic flexibility in response to glucose is not impaired in people with type 2 diabetes after controlling for glucose disposal rate. Diabetes 57, 841–845 (2008).

35. Kummer, J.A. et al. Inflammasome components NALP 1 and 3 show distinct but separate expression profiles in human tissues suggesting a site–specific role in the inflammatory response. J. Histochem. Cytochem. 55, 443–452 (2007).

36. Joosten, L.A. Inflammatory arthritis in caspase 1 gene–deficient mice: contribution of proteinase 3 to caspase 1–independent production of bioactive interleukin-1 β. Arthritis Rheum. 60, 3651–3662 (2009).

37. Virkamäki, A., Ueki, K. & Kahn, C.R. Protein-protein interaction in insulin signaling and the molecular mechanisms of insulin resistance. J. Clin. Invest. 103, 931–943 (1999).

38. Imaeda, A.B. et al. Acetaminophen-induced hepatotoxicity in mice is dependent on Tlr9 and the Nlrp3 inflammasome. J. Clin. Invest. 119, 305–314 (2009).

39. Watanabe, A. et al. Inflammasome-mediated regulation of hepatic stellate cells. Am. J. Physiol. Gastrointest. Liver Physiol. 296, G1248–G1257 (2009).

40. Khan, T. et al. Metabolic dysregulation and adipose tissue fibrosis: role of collagen VI. Mol. Cell. Biol. 29, 1575–1591 (2009).

41. Håversen, L., Danielsson, K.N., Fogelstrand, L. & Wiklund, O. Induction of proinflammatory cytokines by long-chain saturated fatty acids in human macrophages. Atherosclerosis 202, 382–393 (2009).

42. Park, T.S. et al. Ceramide is a cardiotoxin in lipotoxic cardiomyopathy. J. Lipid Res. 49, 2101–2112 (2008).

43. Prieur, X., Roszer, T. & Ricote, M. Lipotoxicity in macrophages: evidence from diseases associated with the metabolic syndrome. Biochim. Biophys. Acta 1801, 327–337 (2010).

44. Shah, C. et al. Protection from high fat diet-induced increase in ceramide in mice lacking plasminogen activator inhibitor 1. J. Biol. Chem. 283, 13538–13548 (2008).

45. Feuerer, M. et al. Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters. Nat. Med. 15, 930–939 (2009).

46. Nishimura, S. et al. CD8+ effector T cells contribute to macrophage recruitment and adipose tissue inflammation in obesity. Nat. Med. 15, 914–920 (2009).

47. Winer, S. et al. Normalization of obesity-associated insulin resistance through immunotherapy. Nat. Med. 15, 921–929 (2009).

48. Yang, H. et al. Obesity increases the production of proinflammatory mediators from adipose tissue T cells and compromises TCR repertoire diversity: implications for systemic inflammation and insulin resistance. J. Immunol. 185, 1836–1845 (2010).

49. McGonagle, D. & McDermott, M.F. A proposed classification of the immunological diseases. PLoS Med. 3, e297 (2006).

50. Duewell, P. et al. NLRP3 inflammasomes are required for atherogenesis and activated by cholesterol crystals. Nature 464, 1357–1361 (2010).

51. Li, H., Ambade, A. & Re, F. Cutting edge: Necrosis activates the NLRP3 inflammasome. J. Immunol. 183, 1528–1532 (2009).

52. Halberg, N. et al. Hypoxia-inducible factor 1α induces fibrosis and insulin resistance in white adipose tissue. Mol. Cell. Biol. 29, 4467–4483 (2009).

53. Strissel, K.J. et al. Adipocyte death, adipose tissue remodeling and obesity complications. Diabetes 56, 2910–2918 (2007).

54. So, A. & Thorens, B. Uric acid transport and disease. J. Clin. Invest. 120, 1791–1799 (2010).

55. Feig, D.I., Kang, D.H. & Johnson, R.J. Uric acid and cardiovascular risk. N. Engl. J. Med. 359, 1811–1821 (2008).

56. Strissel, K.J. et al. T-cell recruitment and TH1 polarization in adipose tissue during diet-induced obesity in C57BL/6 mice. Obesity 18, 1918–1925 (2010).

57. Duffaut, C. et al. Interplay between human adipocytes and T lymphocytes in obesity: CCL20 as an adipochemokine and T lymphocytes as lipogenic modulators. Arterioscler. Thromb. Vasc. Biol. 29, 1608–1614 (2009).

58. Zhou, R., Tardivel, A., Thorens, B., Choi, I. & Tschopp, J. Thioredoxin-interacting protein links oxidative stress to inflammasome activation. Nat. Immunol. 11, 136–140 (2010).

59. Dominguez, H. et al. Metabolic and vascular effects of tumor necrosis factor-alpha blockade with etanercept in obese patients with type 2 diabetes. J. Vasc. Res. 42, 517–525 (2005).

60. Lamkanfi, M. et al. Glyburide inhibits the Cryopyrin/Nlrp3 inflammasome. J. Cell Biol. 187, 61–70 (2009).

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature medicinedoi:10.1038/nm.2279

ONLINE METhODSHuman subjects. The participants (ten obese males of European descent with type 2 diabetes) were examined to study the effects of 1-year inten-sive lifestyle intervention to promote weight loss33. The intervention was designed to achieve and maintain a weight loss of at least 7% through decreased caloric intake and increased physical activity. The inclusion and exclusion criteria have been previously described33. We measured the fat mass in all subjects by dual-energy X-ray absorptiometry (Hologic QDR 4500A) and the analysis of fasting levels of glucose, free fatty acids, insulin and adiponectin and HOMA-IR calculations were performed as described previously12,33. We determined the insulin-stimulated glucose disposal rate with a euglycemic-hyperinsulinemic clamp (90 mg dl−1, 80 mU m2 min−1). The NLRP3 and IL1β mRNA in the abdominal subcutaneous adipose tis-sue was determined by qRT-PCR, and PPIB (encoding cyclophilin B) was used as a housekeeping-gene control. All studies were performed under Pennington institutional review board–approved protocol with informed consent from participants.

Mice. Nlrp3−/− mice have been described previously35. The mice were fed with ad libitum HFD consisting of 60% calories from fat (D12492i; Research Diets) starting at 8 weeks of age and control mice with a normal chow diet consisting of 4.5% fat (5002; LabDiet). We purchased the 12-month-old female calorie-restricted mice and ad libitum–fed, C57BL/6 mice from National Institute on Aging (NIA) rodent colony. We performed all experiments in compliance with the NIH Guide for the Care and Use of Laboratory Animals, and they were approved by the Institutional Animal Care and Use Committee at Pennington Biomedical Research Center.

Insulin tolerance test and glucose tolerance test. We performed ITT and GTT on 4-h fasted mice by giving an intraperitoneal injection of insulin (0.83 mU per g body weight, Sigma) and D-glucose (1.8 mg per g body weight, Sigma). Glucose concentrations in tail blood were measured with a glucometer (Breeze, Bayer Health Care).

Preparation of stromal vascular fraction cells and adipose tissue macro-phages isolation. The subcutaneous (inguinal fat pad) and visceral adipose tissue (epididymal fat pad) were digested as described previously54. We isolated the macrophages from SVF cells by positive selection with an F4/80-specific antibody (eBiosciences), followed by biotin-conjugated secondary rat IgG, and we purified the cells were with Streptavidin-labeled magnetic beads (Dynabeads, Invitrogen).

Flow cytometry. We stained the SVF cells from specific adipose depots with antibodies to F4/80, CD206 and CD11c (eBiosciences and Biolegend) to iden-tify macrophage subsets. We labeled the T cell subpopulations in SVF with antibodies to CD4, CD8, CD25, CD62L, FoxP3 and CD44 (eBiosciences) as described previously54. We conducted the FACS analysis on a FACSCalibur (BD Pharmingen), and the FACS data were analyzed by post-collection compensation with FlowJo (Treestar) software.

Macrophage culture and cytokine analysis. We prepared the mouse BMDMs and cultured them according to previously established protocols26. We primed the BMDMs with 100 ng ml−1 ultrapure LPS (E coli serotype 0111:B4; Sigma) and, after 12 h, ATP was added in a control well (1 mM) and ceramides were added in treatment wells with a final concentration of 0.1 mM. The IL-1β con-centrations in BMDM supernatants were measured by ELISA (eBioscience), and cell lysates were used for western blot analyses.

Adipose tissue explants culture. We prepared and cultured the adipose tissue explants from epididymal fat pad as described previously58. After 24 h of the treatment with LPS and ceramides, we prepared the protein lysates from adipose explants and analyzed them for caspase-1 activation by immunoblot analysis.

Immunohistochemistry. We formalin-fixed and paraffin-embedded the adipose tissue from mice and stained the tissue sections with H&E. The cross-sectional size (area, µm2) of adipocytes (100 cells per mouse) was determined in VAT using Image J software. The immunofluorescence analy-sis of Nlrp3 and F4/80 in adipose tissue sections was performed as described previously54, and images were acquired on a Zeiss 510 Meta multiphoton confocal microscope.

Cytokine measurement. We measured the concentrations of IL-1β (eBio-science) and IL-18 by ELISA (MBL) and the serum levels of IP10 and MCP-1 with the Milliplex bead assay (Millipore).

qRT-PCR. We isolated the total RNA from human SAT biopsy tissue, mouse fat, liver and positively selected macrophages with the RNeasy Lipid Tissue Mini Kit (Qiagen). All samples were DNase digested to remove potential genomic DNA contamination. We conducted the qRT-PCR as described previously (Bio-Rad)4,54.

Western blot analysis. For insulin-signaling experiments, we collected the adipose tissue, liver, kidney and muscle (gastrocnemius) and snap froze them in liquid nitrogen 5 and 10 min after intraperitoneal insulin (0.8 mU per g body weight) injections. We conducted the immunoblot analysis for AKT, IRS-1 and MAPK as described previously54. For caspase-1 cleavage, we collected the organs of lean and DIO mice from additional cohorts of mice and snap froze them for western blot analysis. The protein immune complexes were detected with specific fluorescent secondary antibodies conjugated with IRDye 800CW (Rockland), and membranes were imaged with the Odyssey Infrared Imaging System (LI-COR Odyssey BLOT).

Statistical analyses. We used a two-tailed Student’s t test to determine signifi-cance in differences between genotypes or treatments; *P < 0.05 and P < 0.01, **P < 0.005 and P < 0.001. We expressed the results as the mean ± s.e.m. The differences between means and the effects of treatments were also analyzed by one-way analysis of variance with Tukey’s test (Sigma Stat), which protects the significance (P < 0.05) of all pair combinations.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.

nature immunology volume 13 number 4 APrIl 2012 311

alexander v. Chervonsky is in the department of

Pathology, The University of Chicago, Chicago,

Illinois, Usa.

e-mail: [email protected]

In the opening scene of Quentin Tarantino’s movie Pulp Fiction, the petty criminals

Pumpkin and Honey Bunny rob the Hawthorne Grill diner and face the hardcore gangsters Jules Winnfield and Vincent Vega. What happens to the robbers depends on previous events that, luckily for them, prompt Jules to restrain from violence. The chain of those events is presented in retrospect, and it takes some effort on the part of the viewer to appreciate them with clarity. A similar problem is faced in research into type 1 diabetes: the disease can be read-ily diagnosed at the point of no return (overt diabetes), but it is difficult to accurately predict which person will or will not develop the dis-ease early enough for a potentially successful intervention. However, in this issue of Nature Immunology, Mathis and colleagues use a strat-egy of magnetic resonance imaging (MRI) that has been confirmed in patients to non invasively visualize the local effects of pancreatic-islet inflammation to predict diabetes onset in mice of the nonobese diabetic strain1.

There is a need for biomarkers that can be used to predict diabetes onset with high prob-ability of accuracy. High-fidelity biomarkers must satisfy the following criteria: they should be detectable at the early stages of disease devel-opment when its course is reversible (or, more accurately, ‘potentially reversible’); their use should be simple, reliable and reproducible; and they should ideally be available at a low cost and should be easily applicable (like the Mantoux test for tuberculosis). Most importantly, a good bio-marker should link to a pathogenic or preventive mechanism that in turn leads to new ideas for prophylactic or therapeutic intervention.

Are reasonable biomarkers now available and are new and better ones needed? The answer is ‘yes’ to both parts of that question. As type 1 diabetes is an autoimmune disease with a distinct organ-specific adaptive immune response to pancreatic antigens, most of the past effort to find predictive tests was con-centrated on specific antibodies and T cells as a measure of the ongoing attack on the islets

solving the plot: early events are the key to diabetes interventionAlexander V Chervonsky

Type 1 diabetes is usually diagnosed after most insulin-producing islets of Langerhans have already been destroyed. However, magnetic resonance imaging can be used to predict the onset of type 1 diabetes, and a benign prognosis correlates with the presence of anti-inflammatory tissue-resident macrophages.

and on the C-peptide of insulin as a measure of the residual beta-cell function. Antibodies are a likely venue: they are specific and they are in the blood, which is easy to obtain and test. There are several problems associated with antibodies, however. It turns out that the best predictive power for type 1 diabetes in humans is the presence of antibodies to multiple pan-creatic antigens at the same time2. That fact by itself suggests that some damage has been done to the pancreas by the time this antibody ‘cock-tail’ is detected. That seems to be the case, as it was has been shown in a mouse model of type 1 diabetes that the ability of anti bodies to CD20 to preempt the development of type 1 diabetes drops precipitously at the time that antibodies to insulin show up3. That finding indicates that at least some types of intervention address spe-cific steps in the development of type 1 diabetes that antibodies cannot be used to predict.

With the development of techniques that allow complexes of major histocompatibility complex molecules and peptide to be used for detection of antigen-specific T cells, these cells also became candidates as biomarkers of ongo-ing progression of type 1 diabetes4,5. The nature of the T cell response (low initial frequency and accumulation in the regional lymph nodes and in the affected tissues) makes the use of the detection of pancreas-specific T cells difficult at the early stages of disease development. It is not a problem to detect specific T cells in islet infil-trates, but this is a problem in human patients. There are two separate issues here. How does the detection of certain specificities corre-late with disease progression (is it too late for intervention), and how does the presence of these T cells in peripheral blood correlate with what is happening in the pancreas itself ? It is hard to address such questions in humans, but there is an indication that in mice with trans-genic expression of the human major histo-compatibility complex molecule HLA-A2, the presence of islet antigen–specific T cells in the peripheral blood and islets does correlate6.

It is also very likely that some sort of inflam-matory (nonadaptive) process takes place in the pancreas before activation of ‘adaptive auto immunity’ to pancreatic antigens7. This process does not happen in healthy organisms and thus must have two features: the response

must be genetically or environmentally induced (or both), and there must be mechanisms in place that prevent or block such induction. How can these mechanisms be elucidated? Can they help to identify new biomarkers of the devel-opment of type 1 diabetes? Can knowledge

MRI

MNP

Pancreas

Diabetic Nondiabetic

Protective genesexpression signature

Gene-expression

analysis

Vsig4 CRIg-Fc

Protection

Figure 1 The predictive power of MRI leads to the development of a prophylactic treatment for type 1 diabetes. Mice are assessed by the MRI-MNP approach at various ages and images of their pancreata are archived, then the mice are monitored for the development of type 1 diabetes. A correlation is then established between the destiny to develop type 1 diabetes and MRI features (signs of edema and retention of MNPs in the pancreas). Thus, the prediction of type 1 diabetes can be accurately made at 10 weeks of age. Comparison of gene expression in the pancreata of both types of mice (those destined to develop type 1 diabetes and those destined to remain normoglycemic) shows that only the ‘protective’ gene-expression signature is statistically significant. That protective signature is attributed by bioinformatics to tissue-resident macrophages. One of the genes identified, Vsig4, which encodes CRIg, is selected for further analysis. A chimeric fusion of CRIg and Fc that is able to inhibit T cell responses is shown to protect mice from type 1 diabetes. Heat map (bottom): blue, low expression; red, high expression.

NEWS AND v IEWSnp

g©

201

2 N

atur

e A

mer

ica,

Inc.

All

right

s re

serv

ed.

312 volume 13 number 4 APrIl 2012 nature immunology

activation of adaptive autoimmunity in suscep-tible mice12. Several genes encoding molecules of the complement system are also upregulated in resistant mice. Those molecules could be involved in the control of phagocytosis and regulation of the complement cascade that itself diminishes inflammation. An important finding stemming from the gene-expression profiling and antibody-staining data is that tissue-resident macrophages must be put on the list of important participants in the devel-opment or prevention of type 1 diabetes.

The translation of these results to humans is the next logical step and a considerable chal-lenge. Human disease is probably heterogenic, and some underlying mechanisms may vary greatly, but the outcome (loss of insulin pro-duction) is always the same. The stratification of human type 1 diabetes is very important and it must be possible (on the basis of genet-ics and other features) to identify the group of ‘nonobese diabetic humans’ that may have the disease relevant to type 1 diabetes in mice. That group would be ideal for the translation of the findings by Mathis and colleagues1.

COMPETING FINANCIAL INTERESTSThe author declares no competing financial interests.

1. Fu, W., Wojtkiewicz, G., Benoist, C. & Mathis, D. Nat. Immunol. 13, 361–368 (2012).

2. Winter, W.E. & Schatz, D.A. Clin. Chem. 57, 168–175 (2011).

3. Serreze, D.v. et al. Diabetes 60, 2914–2921 (2011). 4. Fierabracci, A. Diabetes Metab. Res. Rev. 27, 216–229

(2011).5. DiLorenzo, T.P. Diabetes Metab. Res. Rev. 27,

778–783 (2011).6. Enée, E. et al. J. Immunol. 180, 5430–5438

(2008).7. Jansen, A. et al. Diabetes 43, 667–675 (1994).8. Turvey, S.E. et al. J. Clin. Invest. 115, 2454–2461

(2005).9. Gaglia, J.L. et al. J. Clin. Invest. 121, 442–445 (2011).10. vogt, L. et al. J. Clin. Invest. 116, 2817–2826 (2006).11. Miyanishi, M. et al. Nature 450, 435–439 (2007).12. Mathis, D., vence, L. & Benoist, C. Nature 414,

792–798 (2001).

immune process is a case of ‘mistaken iden-tity’ in which the organism mistakes its own pancreas for a pathogen and that multiple mechanisms (some interdependent and some self-sufficient) are deployed to destroy the organ, as it would be in the case of the elimina-tion of pathogens. Negative, anti-autoimmune regulatory mechanisms, in contrast, are prob-ably less dependent on the specific pathogen and should be more uniform. The findings by Mathis and colleagues support that idea1.

The second, and clearly more practically impor-tant, outcome is that some anti-autoimmune mechanisms can be artificially enhanced and used for prophylactic-therapeutic approaches. One of the ~100 genes upregulated in mice that do not progress to diabetes is Vsig4, which encodes the complement receptor CRIg. In addition to its participation in regulating complement, CRIg negatively regulates T cell responses10, probably through the blockade of costimulation. Mathis and colleagues find that the administration of an agonist consisting of a chimera of CRIg and the Fc portion of immunoglobulin results in less development of type 1 diabetes1.

Thus, the paper describes a full cycle from identifying a noninvasive method of inspection of the pancreas in mice predisposed to type 1 diabetes, to determining the fate of those mice, to finding new markers for disease progression based on the predictable changes in the first test and, finally, using that knowledge to verify the importance of the new biomarker and to use it for treatment of the mice with type 1 diabetes. The next goal is obviously to expand studies of the newly found biomarkers to establish their importance. One candidate highlighted in the paper is the phosphatidylserine receptor Timd4, which is involved in the scavenging of apoptotic cells11 and is potentially important because the developmentally regulated death of insulin-producing cells may be involved in the

of the newly identified mechanisms be used for therapeutic purposes?

Mathis and colleagues have now taken on those difficult tasks1 (Fig. 1). The technique of MRI of injected magnetic nanoparticles (MNPs), which is laborious (and expensive, and thus is definitely far from a Mantoux-type approach), can detect differences between normal and diabetic pancreata in nonhuman animals8 and in humans9. MRI assesses the properties of water molecules and thus detects changes in blood flow and edema. MNPs can be engulfed by phagocytic macrophages, and their retention is indicative of macrophage infil-tration. Pursuing a longitudinal study of mice, Mathis and colleagues find that quantitative changes in MRI are present early in the develop-ment of type 1 diabetes and that individual mice differ in the severity of changes1. They monitor a reasonably sized cohort of mice to 40 weeks of age and determine which mice do or do not develop diabetes. The MRI scans show that the early changes in MRI can be used to perfectly predict the development of full-blown disease. That finding puts the authors in a good posi-tion to search for biomarkers associated with the dichotomy in MRI-MNP results. Gene-expression analysis of the pancreatic infiltrates identifies many changes that correlate with the MRI results. Notably, only changes in gene expression that negatively correlate with dis-ease progression are significant. Those genes, therefore, encode proteins that participate in protection against autoimmunity.

Although it is possible that with the obser-vation of a much larger group, the markers of pathogenic processes could be established, two ramifications emanate from the present find-ings. First, it is not too surprising (especially if the answer is already known) that destructive autoimmune mechanisms vary from mouse to mouse. The point can made that the auto-

Mübeccel akdis and Cezmi a. akdis are with the

swiss Institute of allergy and asthma Research,

University of Zurich, davos, switzerland.

e-mail: [email protected]

low-affinity Fc receptors and in the patho-genesis of allergic disease and host defense against helminth infection1,2. Class switching of the B cell immunoglobulin isotype to IgE production is a tightly regulated process, but the location and kinetics of the responses of IgE-producing (IgE+) memory B cells and plasma cells has been rather enigmatic3,4.

Ige class switching and cellular memoryMübeccel Akdis & Cezmi A Akdis

After class switching in naive B cells, memory B cells and plasma cells that produce immunoglobulin E (IgE+ cells) develop through a germinal-center IgE+ intermediate cell without an IgG1 phase. In addition, cellular IgE memory resides in IgE+ memory B cells, and IgG1+ memory B cells are not an important source of IgE memory.

In large part, the difficulty of studying IgE+ B cells has been a consequence of their van-ishingly small numbers under physiological conditions. In the present issue of Nature Immunology, Talay et al. investigate IgE switch-ing and memory in a reporter mouse in which a bicistronic reporter gene encoding enhanced green fluorescent protein (GFP) is inserted

Immunoglobulin E (IgE) antibodies are involved in type 1 hypersensitivity reactions

and allergen capture via their high-affinity and

NEWS AND v IEWSnp

g©

201

2 N

atur

e A

mer

ica,

Inc.

All

right

s re

serv

ed.

mailto:[email protected]

112 VOLUME 20 NUMBER 1 | JaNUaRy 2012 | www.obesityjournal.org

articles nature publishing group

Behavior and Psychology

IntroductIonThe rise in obesity-related morbidity in modern society reflects a number of environmental and behavioral changes that facili-tate overeating and physical inactivity. Several aspects of eating behavior, including patterns of food intake driven by hunger, cognitive restraint, loss of voluntary control and emotional distress, are thought to modify individual propensity to over-eat (1–3). Better understanding of the relationship between these eating patterns and other obesity-promoting behaviors may facilitate the development of improved strategies for suc-cessful weight maintenance and metabolic risk reduction.

Sleep is an important health-related factor that may influ-ence human eating behavior (4). Today, many Americans sleep less than 6 h per night and some (5), but not other (6,7), epide-miologic studies have associated such short sleep duration with increased incidence of obesity. Studies of healthy volunteers in the laboratory indicate that short-term sleep restriction can modify energy intake and expenditure, fuel metabolism, and concentrations of circulating hormones that affect hunger and appetite (8–16). However, the contribution of such acute exper-imentally-induced changes in hunger and energy metabolism

to the association between self-reported short sleep and obes-ity in free-living adults is poorly understood (4,17). There is also concern that the relationship between self-reported short sleep and obesity in epidemiologic studies may be confounded by factors, such as undiagnosed sleep problems (e.g., sleep apnea, insomnia) (18,19), poor physical and emotional health (18,19), and systemic bias in the subjective recall of sleep and physical activity (20,21).

Adults with parental history of type 2 diabetes have a high risk for developing the disease, which is exacerbated by physi-cal inactivity and excessive weight gain (22). Prevention of obesity in the offspring of diabetic patients is accompanied by a reduction in excess risk of type 2 diabetes by nearly 40% (22). Thus, it is important to understand the relationship between food intake, physical activity, and sleep in this high-risk popu-lation (23). Recent data from our laboratory suggest that urban adults with parental history of type 2 diabetes who habitually curtail their sleep have reduced everyday physical activity (24). Whether habitual sleep patterns could be related to potentially problematic patterns of eating behavior in this high-risk popu-lation is not known. Therefore, we examined the association

Sleep and Eating Behavior in adults at Risk for Type 2 DiabetesJennifer M. Kilkus1, John N. Booth2, Lindsay E. Bromley2, Amy P. Darukhanavala2, Jacqueline G. Imperial1 and Plamen D. Penev2

Insufficient quantity and quality of sleep may modulate eating behavior, everyday physical activity, overall energy balance, and individual risk of obesity and type 2 diabetes. We examined the association of habitual sleep quantity and quality with the self-reported pattern of eating behavior in 53 healthy urban adults with parental history of type 2 diabetes (30 F/23 M; mean (s.d.) age: 27 (4) years; BMI: 23.9 (2.3) kg/m2) while taking into consideration the amount of their everyday physical activity. Participants completed 13 (3) days of sleep and physical activity monitoring by wrist actigraphy and waist accelerometry while following their usual lifestyle at home. Overnight laboratory polysomnography was used to screen for sleep disorders. Subjective sleep quality was measured with the Pittsburgh Sleep Quality Index. Eating behavior was assessed using the original 51-item and the revised 18-item version of the Three-Factor Eating Questionnaire including measures of cognitive restraint, disinhibition, hunger, and uncontrolled and emotional eating. In multivariable regression analyses adjusted for age, BMI, gender, race/ethnicity, level of education, habitual sleep time measured by wrist actigraphy and physical activity measured by waist accelerometry, lower subjective sleep quality was associated with increased hunger, more disinhibited, uncontrolled and emotional eating, and higher cognitive restraint. There was no significant association between the amount of sleep measured by wrist actigraphy and any of these eating behavior factors. Our findings indicate that small decrements in self-reported sleep quality can be a sensitive indicator for the presence of potentially problematic eating patterns in healthy urban adults with familial risk for type 2 diabetes.

Obesity (2012) 20, 112–117. doi:10.1038/oby.2011.319

1General Clinical Research Center, University of Chicago, Chicago, Illinois, USA; 2Department of Medicine, University of Chicago, Chicago, Illinois, USA. Correspondence: Jennifer M. Kilkus ([email protected])

Received 29 June 2011; accepted 14 September 2011; published online 13 October 2011. doi:10.1038/oby.2011.319

http://www.nature.com/doifinder/10.1038/oby.2011.319


http://www.nature.com/doifinder/10.1038/oby.2011.319

obesity | VOLUME 20 NUMBER 1 | JaNUaRy 2012 113

articlesBehavior and Psychology

of habitual sleep quantity and quality with patterns of eating behavior characterized by cognitive restraint, hunger, loss of voluntary control, and emotional eating in urban adults with parental history of type 2 diabetes, while taking into considera-tion the amount of their everyday physical activity.

Methods and ProceduresParticipantsHealthy men and women between the ages of 21 and 40 years with BMI between 19 and 27 kg/m2, who lived in the greater Chicago area and had at least one parent with type 2 diabetes, were recruited through local advertisements. Volunteers who passed a brief telephone interview were invited for screening in our Clinical Research Center. Body weight and height were measured after an overnight fast using a calibrated medi-cal scale (Scale-Tronix, Wheaton, IL) and a stationary Harpenden sta-diometer (Holtain, Crymych, Wales) with participants dressed in light clothing without shoes. Individuals were excluded from participation if they had abnormal findings on medical history, physical examination, and routine screening tests (complete blood counts, comprehensive metabolic and thyroid function panels, 12-lead electrocardiogram, 75-g oral glucose tolerance test); depressed mood (Center for Epidemiologic Studies of Depression, CES-D (25), score >15 confirmed by clinical interview); pregnancy or childbirth during the last year; night-shift work, frequent travel across time zones, or self-reported sleep problems (Pittsburgh Sleep Quality Index (PSQI) (26), global score >5); use of tobacco, excess alcohol (>14 drinks/week for men; >7 for women), or prescription, over-the-counter and illegal drugs and supplements that can effect sleep and eating behavior. Research volunteers gave written informed consent and were paid for their participation.

study protocolThe study protocol was approved by the Institutional Review Board of the University of Chicago. Enrolled participants were asked to complete 14 days of sleep monitoring while following their usual lifestyle at home. A small accelerometer equipped with an event marker (Actiwatch-64, Mini-Mitter Respironics, Bend, OR) was attached to a wrist band on their nondominant arm and actigraphy data were collected continu-ously in 1-min epochs to measure sleep duration under free-living conditions (27). Since some prior epidemiologic studies have found an association between self-reported sleep and physical activity (28–31), we also measured the amount of body movement of each participant using a small waist accelerometer (Actical, Mini-Mitter-Respironics, Bend, OR). Physical activity data from a subset of the participants in this study have been reported elsewhere (24).

After the 2-week home monitoring period, participants were scheduled to complete one night of laboratory polysomnography (Neurofax-1100 EEG Acquisition System, Nihon-Kohden, Foothill Ranch, CA) including electroencephalography, electrooculography, electromyography, airflow, thoracic and abdominal respiratory effort, electrocardiography, and pulse oximetry to exclude the presence of primary sleep pathology, sleep move-ment disorder, or sleep disordered breathing (respiratory disturbance index >10 or sleep apnea index >3). Sleep was scheduled between 2300–2400 h and 0730–0830 h with a fixed time-in-bed of 8.5 h. Records were scored in 30-s epochs of wake, movement, stage 1, 2, 3, 4, and rapid-eye-movement sleep according to standard criteria. Respiratory events, periodic leg move-ments, and arousals were scored using current clinical guidelines.

data analysis and statisticsHome activity records were analyzed using version 2.12 of the software provided with the Actical device. The total number of activity counts during each 24-h period was averaged across all recorded days to obtain a measure of individual physical activity. Nighttime sleep was scored automatically with Actiware Sleep version 3.4 provided with the Actiwatch using a medium sensitivity setting of 40. Habitual sleep duration was calculated as the average number of minutes scored as

sleep across all recorded nights. Subjects with less than six nights of data were not included in the analysis. The average sleep fragmentation index of each individual wrist actigraphy data set was used as a measure of habitual sleep quality. The index is calculated during the nighttime sleep period by adding the percentage of time spent in nonsleep epochs containing above-sleep-threshold amounts of movement and the per-centage of time spent in brief periods with sub-sleep-threshold wrist movement that last only 1 min.

The sum of all epochs scored as sleep was used to measure the amount of overnight sleep in the laboratory. Measures of laboratory sleep quality included the number of arousals per hour of sleep (arousal index), the number of awakenings and the amount of wake time during the night. The PSQI score of global sleep disturbance was used as a measure of self-reported sleep quality (26). Higher scores on this scale reflect lower subjective sleep quality.

During the initial screening visit, all participants completed the 51-item Three-Factor Eating Questionnaire, which measures three dimensions of eating behavior: cognitive restraint of eating, disinhibition, and hunger (32). Karlsson et al. have revised and abbreviated the original 51-item Three-Factor Eating Questionnaire to improve its scaling properties and construct validity and we used their 18-item Three-Factor Eating Ques-tionnaire version (33) to derive additional scores of cognitive restraint (tendency to consciously restrict food intake to control body weight), uncontrolled eating (tendency to eat more because of loss of control over food intake), and emotional eating (tendency to overeat related to dys-phoric mood) (34).

All statistical analyses were performed using SPSS version 18.0 (SPSS, Chicago, IL). Multivariable linear regression models adjusted for age, gender, BMI, race/ethnicity, and level of education (as a surrogate of socioeconomic status) were used to examine the relationship between each eating behavior factor as a dependent variable and each of three predictor variables: habitual sleep duration (measured by wrist actig-raphy), subjective sleep quality (PSQI score), and free-living physical activity (measured by waist accelerometry). Since self-reported sleep quality and physical activity emerged as significant predictors of several aspects of eating behavior, the role of each of these three predictor vari-ables was re-examined after control for the other two was added to the initial regression models. Finally, partial correlation analysis, controlling for age, gender, race/ethnicity, BMI, level of education, and objectively-measured sleep duration and physical activity, was used to explore the relationship of subjective sleep quality (PSQI score) with actigraphy- and polysomnography-based measures of sleep structure and quality and self-ratings of depressed mood (CES-D score).

resultsFifty three participants completed an average of 13 (s.d. 3) days of home sleep monitoring. Participant characteristics are sum-marized in Table 1. The average sleep time of the participants measured by home actigraphy ranged between 4 h 33 min and 8 h 14 min per night. Thirty-eight percent of the participants habitu-ally slept <6 h/night. All subjects had good subjective sleep qual-ity with PSQI scores ranging between 0 and 5 (26). Overnight laboratory polysomnography in 48 (91%) of the participants who kept their study appointments showed no sleep pathology. The sleep architecture of the participants was typical for healthy individuals monitored by full polysomnography under labora-tory conditions without prior habituation (Table 1).

Reduced subjective sleep quality (higher PSQI score) was associated with eating behaviors characterized by increased hunger, uncontrolled and emotional eating, and more cog-nitive restraint (Table 2; Model 1). In contrast, there was no significant association between sleep duration measured by wrist actigraphy and any of these eating behavior factors



(Table 2). Higher levels of habitual physical activity were associated with a pattern of more uncontrolled and hun-ger-dependent eating (Table 2). The association of reduced subjective sleep quality (higher PSQI score) with eating behaviors characterized by increased hunger and cognitive restraint, and uncontrolled and emotional eating remained qualitatively and quantitatively similar when regression analyses controlled for habitual sleep duration measured by wrist actigraphy and physical activity measured by waist accelerometry (Table 2; Model 2).

Partial correlation analysis to explore the correlates of self-rated sleep quality showed that higher PSQI scores were asso-ciated only with more depressed mood (CES-D score; R = 0.33; P = 0.036). Using CES-D instead of PSQI as a predictor in our fully-adjusted regression analysis (see Model 2 in Table 2) showed that the score for depressed mood was associated with hunger (B: 1.0; 95%CI: 0.2–1.7; P = 0.014) and uncontrolled eating (B: 0.6; 95%CI: 0.1–1.1; P = 0.020). The association of subjective sleep quality with restrained, disinhibited, uncon-trolled, and emotional eating did not change qualitatively or quantitatively when control for CES-D was included in Model 2, whereas the association with hunger was attenuated (B: 1.5; 95%CI −0.2–3.2; P = 0.084). There was no significant relation-ship between subjective sleep quality and wrist actigraphy- or laboratory polysomnography-based measures of sleep quality (sleep fragmentation index, arousal index, number of awaken-ings, and wake time during the night) or other sleep architec-ture indices.

dIscussIonThis study examined the relationship of habitual duration and quality of sleep with several dimensions of eating behavior in free-living urban adults with parental history of type 2 diabe-tes. Our results show that decrements in self-reported sleep quality are associated with eating patterns characterized by increased hunger, uncontrolled and emotional eating, and cog-nitive restraint. In contrast, there was no significant associa-tion between the amount of habitual sleep measured by wrist actigraphy and any of these eating behavior factors.

It has been argued that cognitively restrained eating is an adaptive behavior for individuals prone to gain weight in the setting of easy access to abundant and palatable food (35). However, young adults are highly susceptible to failure of restraint when it is associated with other potentially prob-lematic eating behaviors, such as uncontrolled and emotional eating (36). Our findings that lower subjective sleep quality reflected by PSQI scores at the higher end of the typical range seen in good sleepers (26) are associated with increased hun-ger, uncontrolled and emotional eating, and more cognitive restraint suggest that small decrements in self-reported sleep quality can be a sensitive indicator for the presence of poten-tially problematic eating patterns in young urban adults at high risk for type 2 diabetes. Self-reported sleep quality did not reflect specific changes in the architecture, consolidation and efficiency of sleep in the laboratory (assessed by polysomnog-raphy) or the habitual amount and fragmentation of sleep at home (measured by wrist actigraphy), but was related in part to individual ratings of depressed mood. Consistent with our findings, population-based observations in the Penn State (19) and MONICA/KORA (18) cohort studies indicate that com-plaints of poor sleep quality and psychological distress could be important determinants of the association between self-reported short sleep and chronic metabolic morbidity. In the Penn State cohort for example, complaints of poor sleep and measures of psychological distress were the primary predictors of self-reported short sleep among obese participants: those

table 1 Participant characteristics and measures of eating behavior, sleep, and activity

Participant characteristics

Number of participants 53 (30 F/23 M)

White/African American/Asian/Hispanic 29/13/7/4

Age (years) 27 (4)

BMI (kg/m2) 23.9 (2.3)

Level of education (years) 17 (2)

Depressed mood (CES-D score) 4 (2; 8)

Eating behavior factors

Cognitive restraint score (51-item TFEQ) 9 (5; 12.5)

Disinhibition score (51-item TFEQ) 4 (3; 6)

Hunger score (51-item TFEQ) 3 (1.5; 5)

Cognitive restraint score (18-item TFEQ) 2 (1; 4)

Uncontrolled eating score (18-item TFEQ) 1 (0; 2)

Emotional eating score (18-item TFEQ) 0 (0; 2)

Subjective sleep quality

Global sleep disturbance (PSQI score) 2 (1; 3)

Free-living sleep and activity monitoring

Measured sleep duration (min/day) 379 (53)

Sleep fragmentation index (%) 31 (10)

Total activity counts (thousands/day) 217 (117)

Laboratory polysomnography

Number of participants 48 (27 F/21 M)

Sleep onset latency (min) 32 (30)

Total sleep time (min) 439 (48)

Stage 1 sleep (min) 30 (17)

Stage 2 sleep (min) 262 (41)

Slow wave sleep (stages 3 + 4, min) 51 (33)

Rapid-eye-movement sleep (min) 100 (32)

Wake after sleep onset (min) 43 (27)

Sleep efficiency (%) 87 (8)

Arousal index (events/h) 13 (7)

Number of awakenings 4 (3)

Respiratory disturbance index (events/h) 3 (4)

Data are reported as mean (s.d.) for continuous variables and median (inter-quartile range) for questionnaire based scores.CES-D, Center for Epidemiologic Studies of Depression scale; PSQI, Pittsburgh Sleep Quality Index; TFEQ, Three-Factor Eating Questionnaire.



with insomnia reported having the shortest sleep duration averaging 5.9 h per night, whereas the subjective sleep duration of obese individuals without sleep complaints (7.0 h) was simi-lar to that of nonobese good sleepers (6.9 h (19)). Therefore, future investigations of the relationship between sleep and the control of human eating behavior, hunger, and body weight regulation should include formal screening for the presence of sleep pathology and assessment of the psychological well being of the study participants (19).

Consistent with previous population-based data (37), nearly 40% of our subjects habitually slept <6 h/night. Prior labora-tory experiments have found that young men exposed to sleep

curtailment of 4 h/night and caloric restriction (1,500 kcal/day for the average 75 kg study participant) at the time of sam-pling have lower circulating concentrations of the anorexigenic hormone, leptin, higher concentrations of the orexigenic hor-mone, ghrelin, and increased subjective hunger (8). Supported by observations from the Wisconsin Sleep Cohort showing a positive association between leptin and self-reported sleep time, and an inverse association between ghrelin and polysom-nographic sleep time (4), these findings have given rise to the widespread notion that insufficient sleep triggers key hormo-nal signals of “famine in the midst of plenty” to cause excessive food intake and weight gain.

table 2 subjective sleep quality and measured sleep duration and physical activity as predictors of eating behavior

Model 1 Model 2

Sleep qualitya Sleep durationb Physical activityc Sleep qualitya Sleep durationb Physical activityc

Restraint (TFEQ-51)

B 3.1 −1.0 −0.3 3.5 −1.1 0.0

95% CI −0.1 to 6.3 −2.8 to 0.7 −1.5 to 1.0 0.3 to 6.7 −2.8 to 0.6 −1.3 to 1.2

R2 change 0.067 0.026 0.004 0.093 0.032 0.000

P 0.053 0.235 0.675 0.032 0.198 0.948

Disinhibition

B 1.9 −0.4 0.0 1.9 −0.5 0.1

95% CI 0.5 to 3.4 −1.2 to 0.4 −0.6 to 0.6 0.5 to 3.3 −1.3 to 0.3 −0.4 to 0.7

R2 change 0.119 0.017 0.000 0.122 0.031 0.003

P 0.010 0.341 0.949 0.011 0.190 0.658

Hunger

B 1.9 0.5 1.0 2.1 0.1 1.0

95% CI 0.2 to 3.7 −0.4 to 1.5 0.3 to 1.7 0.4 to 3.7 −0.8 to 1.0 0.4 to 1.6

R2 change 0.087 0.024 0.169 0.097 0.001 0.158

P 0.034 0.278 0.004 0.018 0.776 0.003

Restraint (TFEQ-18)

B 1.3 −0.1 −0.3 1.4 0.0 −0.3

95% CI 0.1 to 2.5 −0.8 to 0.5 −0.8 to 0.2 0.2 to 2.6 −0.7 to 0.6 −0.7 to 0.2

R2 change 0.088 0.002 0.037 0.104 0.000 0.031

P 0.029 0.745 0.184 0.023 0.921 0.204

Uncontrolled eating

B 1.4 0.2 0.5 1.6 0.0 0.5

95% CI 0.3 to 2.5 −0.4 to 0.9 0.1 to 0.9 0.5 to 2.7 −0.6 to 0.6 0.1 to 0.9

R2 change 0.112 0.012 0.101 0.151 0.000 0.103

P 0.017 0.451 0.030 0.005 0.921 0.019

Emotional eating

B 0.8 −0.1 0.0 0.7 −0.1 0.0

95% CI 0.2 to 1.4 −0.4 to 0.3 −0.2 to 0.2 0.1 to 1.3 −0.4 to 0.2 −0.2 to 0.3

R2 change 0.119 0.002 0.001 0.100 0.008 0.003

P 0.010 0.761 0.846 0.025 0.501 0.682

Model 1 was adjusted for age, gender, race/ethnicity, BMI, and years of education. Model 2 controlled for the remaining two predictors in addition to the variables included in Model 1. B, regression coefficient reflecting the change in the dependent variable for: a1-point increase in the square-root transformed PSQI score; b1-h increase in measured sleep duration; c100,000-count increase in average daily body movement (bold numbers show significant associations). 95% CI, 95% confidence interval for B.



More recent experiments have exposed human volunteers to sleep restriction in the presence of adequate or excess amounts of self-selected calories. In these studies, short-term sleep restriction was accompanied by increased leptin concentra-tions in women (9,13–15) and had no independent effect on leptin in men (10,16). Other experiments combining 2 weeks of sleep restriction with over- or underfeeding found that sleep loss did not interfere with the expected physiological rise or fall in leptin concentrations, while sleep-loss-related increases in ghrelin levels and hunger were seen only in the presence of negative, but not positive, energy balance (11,12). These experimental findings and some newer epidemiological data (17) suggest that prior reports of increased hunger, lower lep-tin and higher ghrelin concentrations related to short-term sleep restriction (8) did not reflect the presence of “famine in the midst of plenty”, but rather the ability of sleep loss to amplify the human behavioral and neuroendocrine response to caloric restriction (12). In agreement with these newer data on the relationship of hunger-regulating hormones with sleep loss (9–11,13–17), there was no significant association between measured sleep duration and hunger-related eating in the present study. Similarly, the amount of habitual sleep was not a significant predictor of any other self-reported eat-ing behaviors in free-living adults with familial risk for type 2 diabetes.

Physical activity can modulate eating behavior and the rela-tionship of Three-Factor Eating Questionnaire scores with measures of adiposity (38). Some epidemiologic (28,29,31) and experimental studies (10) also indicate that short sleep is accompanied by lower levels of habitual physical activity. In the present study, higher amounts of objectively-measured physical activity were associated with a more uncontrolled and hunger-dependent eating behavior (Table 2). As reported else-where (24), there was also a positive association between habit-ual sleep duration and the amount of everyday physical activity in the participants of this study. However, inclusion of sleep duration and physical activity measured by wrist actigraphy and waist accelerometry as independent variables in the mul-tiple regression analysis did not attenuate the significant asso-ciation of self-reported sleep quality with important aspects of eating behavior (Table 2). Given the association between sleep time and free-living physical activity (10,24,28,29,31), and between free-living physical activity and uncontrolled and hunger-dependent eating behavior (Table 2), future studies of the link between sleep and eating behavior should take into consideration the large individual differences in the amount of everyday physical activity.

Our study has several strengths and limitations. We col-lected a set of exploratory data using a carefully screened sample of healthy individuals at high risk for type 2 dia-betes, while avoiding the potentially confounding effects of obesity and its comorbid conditions on various aspects of eating behavior. The use of laboratory polysomnography and continuous ambulatory monitoring of habitual sleep and free-living physical activity also allowed us to exclude the presence of sleep pathology and avoid assessments based

on unreliable self-reports of these behaviors. Finally, it was important to study a population with high risk for type-2 diabetes which may inform future behavioral research on sleep and metabolic risk reduction. Despite its strengths, this was an exploratory study which included a relatively small number of subjects who were not randomly selected and the results may not be entirely representative of the relation-ship between sleep and eating behavior in this population. Furthermore, all assessments of eating behavior were based on self-report and we do not know what are the implications of the link between subjective sleep quality and key eating behavior factors for the long-term regulation of energy intake and body weight in this high-risk population.

In conclusion, our results suggest that small decrements in self-reported sleep quality can be a sensitive marker for the presence of eating patterns characterized by increased hunger, uncontrolled and emotional eating, and cognitive restraint—a potentially problematic combination (36)—in young urban adults with increased risk of developing type 2 diabetes. Lower self-reported sleep quality did not reflect changes in sleep structure and consolidation assessed by wrist actigraphy and polysomnography, but was correlated in part with individ-ual ratings of depressed mood. Since emotional distress and problematic eating patterns can be ameliorated, complaints of reduced sleep quality in such susceptible individuals may war-rant special attention when considering specific lifestyle modi-fication strategies for metabolic risk reduction. Additional studies are needed to explore the association of reduced sleep quality, emotional well being, and problematic eating behav-iors in individuals with increased susceptibility to obesity and type 2 diabetes.

acknowledgMentsThis work was supported by NIH grants R01-HL089637, CTSa-RR024999, and P60-DK020595. We thank Luis alcantar in the Department of Medicine at the University of Chicago and the staff of the University of Chicago Clinical Research Center for their excellent technical assistance.

dIsclosure The authors have no conflict of interest.

© 2011 The Obesity Society

reFerences1. Tholin S, Rasmussen F, Tynelius P, Karlsson J. Genetic and environmental

influences on eating behavior: the Swedish Young Male Twins Study. Am J Clin Nutr 2005;81:564–569.

2. Keskitalo K, Tuorila H, Spector TD et al. The Three-Factor Eating Questionnaire, body mass index, and responses to sweet and salty fatty foods: a twin study of genetic and environmental associations. Am J Clin Nutr 2008;88:263–271.

3. Keränen AM, Strengell K, Savolainen MJ, Laitinen JH. Effect of weight loss intervention on the association between eating behaviour measured by TFEQ-18 and dietary intake in adults. Appetite 2011;56:156–162.

4. Taheri S, Lin L, Austin D, Young T, Mignot E. Short sleep duration is associated with reduced leptin, elevated ghrelin, and increased body mass index. PLoS Med 2004;1:e62.

5. Cappuccio FP, Taggart FM, Kandala NB et al. Meta-analysis of short sleep duration and obesity in children and adults. Sleep 2008;31:619–626.

6. Stranges S, Cappuccio FP, Kandala NB et al. Cross-sectional versus prospective associations of sleep duration with changes in relative weight and body fat distribution: the Whitehall II Study. Am J Epidemiol 2008;167:321–329.



7. Lauderdale DS, Knutson KL, Rathouz PJ et al. Cross-sectional and longitudinal associations between objectively measured sleep duration and body mass index: the CARDIA Sleep Study. Am J Epidemiol 2009;170: 805–813.

8. Spiegel K, Tasali E, Penev P, Van Cauter E. Brief communication: Sleep curtailment in healthy young men is associated with decreased leptin levels, elevated ghrelin levels, and increased hunger and appetite. Ann Intern Med 2004;141:846–850.

9. Bosy-Westphal A, Hinrichs S, Jauch-Chara K et al. Influence of partial sleep deprivation on energy balance and insulin sensitivity in healthy women. Obes Facts 2008;1:266–273.

10. Schmid SM, Hallschmid M, Jauch-Chara K et al. Short-term sleep loss decreases physical activity under free-living conditions but does not increase food intake under time-deprived laboratory conditions in healthy men. Am J Clin Nutr 2009;90:1476–1482.

11. Nedeltcheva AV, Kilkus JM, Imperial J et al. Sleep curtailment is accompanied by increased intake of calories from snacks. Am J Clin Nutr 2009;89:126–133.

12. Nedeltcheva AV, Kilkus JM, Imperial J, Schoeller DA, Penev PD. Insufficient sleep undermines dietary efforts to reduce adiposity. Ann Intern Med 2010;153:435–441.

13. Omisade A, Buxton OM, Rusak B. Impact of acute sleep restriction on cortisol and leptin levels in young women. Physiol Behav 2010;99:651–656.

14. Pejovic S, Vgontzas AN, Basta M et al. Leptin and hunger levels in young healthy adults after one night of sleep loss. J Sleep Res 2010;19:552–558.

15. Simpson NS, Banks S, Dinges DF. Sleep restriction is associated with increased morning plasma leptin concentrations, especially in women. Biol Res Nurs 2010;12:47–53.

16. Benedict C, Hallschmid M, Lassen A et al. Acute sleep deprivation reduces energy expenditure in healthy men. Am J Clin Nutr 2011;93:1229–1236.

17. Hayes AL, Xu F, Babineau D, Patel SR. Sleep duration and circulating adipokine levels. Sleep 2011;34:147–152.

18. Meisinger C, Heier M, Löwel H, Schneider A, Döring A. Sleep duration and sleep complaints and risk of myocardial infarction in middle-aged men and women from the general population: the MONICA/KORA Augsburg cohort study. Sleep 2007;30:1121–1127.

19. Vgontzas AN, Lin HM, Papaliaga M et al. Short sleep duration and obesity: the role of emotional stress and sleep disturbances. Int J Obes (Lond) 2008;32:801–809.

20. Shephard RJ. Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med 2003;37:197–206; discussion 206.

21. Lauderdale DS, Knutson KL, Yan LL, Liu K, Rathouz PJ. Self-reported and measured sleep duration: how similar are they? Epidemiology 2008;19: 838–845.

22. Sargeant LA, Wareham NJ, Khaw KT. Family history of diabetes identifies a group at increased risk for the metabolic consequences of obesity and physical inactivity in EPIC-Norfolk: a population-based study. The European Prospective Investigation into Cancer. Int J Obes Relat Metab Disord 2000;24:1333–1339.

23. Harrison TA, Hindorff LA, Kim H et al. Family history of diabetes as a potential public health tool. Am J Prev Med 2003;24:152–159.

24. Booth JN, Bromley L, Darukhanavala A et al. Adults at risk for type 2 diabetes who curtail their sleep have reduced physical activity. Diabetes 2011;60(Suppl. 1).

25. Radloff LS. The CES-D Scale: a self-report depression scale for reseach in the general population. Applied Psychological Measurement 1977;1: 385–401.

26. Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res 1989;28:193–213.

27. Ancoli-Israel S, Cole R, Alessi C et al. The role of actigraphy in the study of sleep and circadian rhythms. Sleep 2003;26:342–392.

28. Liu X, Uchiyama M, Kim K et al. Sleep loss and daytime sleepiness in the general adult population of Japan. Psychiatry Res 2000;93:1–11.

29. Ohida T, Kamal AM, Uchiyama M et al. The influence of lifestyle and health status factors on sleep loss among the Japanese general population. Sleep 2001;24:333–338.

30. Chaput JP, Després JP, Bouchard C, Tremblay A. The association between sleep duration and weight gain in adults: a 6-year prospective study from the Quebec Family Study. Sleep 2008;31:517–523.

31. Patel SR, Blackwell T, Redline S et al.; Osteoporotic Fractures in Men Research Group; Study of Osteoporotic Fractures Research Group. The association between sleep duration and obesity in older adults. Int J Obes (Lond) 2008;32:1825–1834.

32. Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 1985;29: 71–83.

33. Karlsson J, Persson LO, Sjöström L, Sullivan M. Psychometric properties and factor structure of the Three-Factor Eating Questionnaire (TFEQ) in obese men and women. Results from the Swedish Obese Subjects (SOS) study. Int J Obes Relat Metab Disord 2000;24:1715–1725.

34. de Lauzon B, Romon M, Deschamps V et al.; Fleurbaix Laventie Ville Sante Study Group. The Three-Factor Eating Questionnaire-R18 is able to distinguish among different eating patterns in a general population. J Nutr 2004;134:2372–2380.

35. de Lauzon-Guillain B, Basdevant A, Romon M et al.; FLVS Study Group. Is restrained eating a risk factor for weight gain in a general population? Am J Clin Nutr 2006;83:132–138.

36. van Strien T. The concurrent validity of a classification of dieters with low versus high susceptibility toward failure of restraint. Addict Behav 1997;22:587–597.

37. Lauderdale DS, Knutson KL, Yan LL et al. Objectively measured sleep characteristics among early-middle-aged adults: the CARDIA study. Am J Epidemiol 2006;164:5–16.

38. Riou ME, Doucet E, Provencher V et al. Influence of physical activity participation on the associations between eating behaviour traits and body mass index in healthy postmenopausal women. J Obes 2011; e-pub ahead of print 14 September 2010.

228 | APRIL 2012 | VOLUME 8 www.nature.com/nrendo

Baker IDI Heart and Diabetes Institute, 99 Commercial Road, Melbourne, VIC 3004, Australia (L. Chen, D. J. Magliano, P. Z. Zimmet)

Correspondence to: P. Z. Zimmet paul.zimmet@ bakeridi.edu.au

The worldwide epidemiology of type 2 diabetes mellitus—present and future perspectivesLei Chen, Dianna J. Magliano and Paul Z. Zimmet

Abstract | Over the past three decades, the number of people with diabetes mellitus has more than doubled globally, making it one of the most important public health challenges to all nations. Type 2 diabetes mellitus (T2DM) and prediabetes are increasingly observed among children, adolescents and younger adults. The causes of the epidemic of T2DM are embedded in a very complex group of genetic and epigenetic systems interacting within an equally complex societal framework that determines behavior and environmental influences. This complexity is reflected in the diverse topics discussed in this Review. In the past few years considerable emphasis has been placed on the effect of the intrauterine environment in the epidemic of T2DM, particularly in the early onset of T2DM and obesity. Prevention of T2DM is a ‘whole-of-life’ task and requires an integrated approach operating from the origin of the disease. Future research is necessary to better understand the potential role of remaining factors, such as genetic predisposition and maternal environment, to help shape prevention programs. The potential effect on global diabetes surveillance of using HbA1c rather than glucose values in the diagnosis of T2DM is also discussed.

Chen, L. et al. Nat. Rev. Endocrinol. 8, 228–236 (2012); published online 8 November 2011; doi:10.1038/nrendo.2011.183

IntroductionThe global prevalence of diabetes mellitus is rapidly increasing as a result of population ageing, urbanization and associated lifestyle changes.1 The number of people with diabetes mellitus worldwide has more than doubled over the past three decades.2 In 2010, an estimated 285 million people worldwide had diabetes mellitus,3 90% of whom had type 2 diabetes mellitus (T2DM).1 The number of people globally with diabetes mellitus is projected to rise to 439 million by 2030, which represents 7.7% of the total adult population of the world aged 20–79 years (Figure 1a,b).3 This Review explores current trends in the epidemic of T2DM and the associated major risk factors (Box 1). In particular, the proposed role of genetic and epigenetic predispositions in the T2DM epidemic is discussed and the potential effect on global diabetes surveillance of the use of HbA1c rather than glucose values as an alternative diagnostic approach is addressed.

New trends in the T2DM epidemicT2DM was relatively rare in developing countries some decades ago; for example, the prevalence of the disease was <1% in China in 1980.4 However, higher rates observed in Asian Indian and Chinese populations in Mauritius,5 as well as in Asian immigrants in Western countries6,7 strongly predicted the potential epidemic of T2DM that has now emerged in mainland China and India.

The major burden of diabetes mellitus is now taking place in developing rather than in developed countries. 80% of cases of diabetes mellitus worldwide live in less

developed countries and areas.3 Asia has emerged as the ‘diabetes epicenter’ in the world, as a result of rapid economic development, urbanization and nutrition transition over a relatively short period of time.4 Among the 10 countries with the largest numbers of people predicted to have diabetes mellitus in 2030, five are in Asia (China, India, Pakistan, Indonesia and Bangladesh).3 In particular, the latest figures derived from a national survey in China between 2007 and 2008 suggest that China has overtaken India and become the global epicenter of the diabetes epidemic with more than 92 million adults (9.7% of the total population) with diabetes mellitus and another 148.2 million adults (15.5% of the total population) with prediabetes, which includes individuals with impaired glucose tolerance (IGT) and/or impaired fasting glucose (IFG).8 In addition to Asia, the Gulf region in the Middle East3 and Africa9,10 are other hot spots for dia betes mellitus. A higher prevalence of diabetes mellitus in immigrants from the Middle East living in Sweden than in native Swedes has also been reported.11

Compared with developed countries, the proportion of young to middleaged individuals with T2DM is higher in developing countries.3 Furthermore, T2DM is not necessarily less prevalent in rural than in urban areas of developing countries, as is generally believed. The rural–urban difference in prevalence is predicted to narrow owing to urbanization, rural to urban migration and its associated lifestyle changes. A study from India showed a significant increase in diabetes mellitus prevalence in both urban (from 13.9% in 2000 to 18.2% in 2006) and rural areas (from 6.4% in 2000 to 9.2% in 2006).12 Similar findings have been reported from other Asian countries.4

Competing interestsThe authors declare no competing interests.

REVIEWS




http://www.nature.com/doifinder/10.1038/nrendo.2011.183

NATURE REVIEWS | ENDOCRINOLOGY VOLUME 8 | APRIL 2012 | 229

Among Chinese individuals aged 35–74 years, between 2001–2002 and 2006, the rural prevalence of diabetes mellitus increased from 5.3% to 14.2% in men and from 8.9% to 13.8% in women, compared with an increase in urban regions in men from 11.3% to 19.2% and in women from 11.3% to 16.1%.13

T2DM and prediabetes in youthT2DM, traditionally considered a metabolic disorder exclusively of adults, has become more common not only in young adults but also in adolescents and, occasionally, in children.14 For example, the crude prevalence of T2DM among North American youth aged 10–19 years in 2001 was estimated to be 42 cases per 100,000 youth.15 T2DM constitutes an increasing percentage of all incident cases of pediatric diabetes mellitus, with less than 4% reported two decades ago and up to more than 80% of newonset cases among adolescents in some ethnic groups, such as American Indian, Asian and Pacific Islander populations.16,17

The prevalence and incidence of T2DM in youth varies dramatically by ethnicity, with higher rates observed among highrisk ethnic groups, such as Native American and Australian Indigenous populations, and African American, Hispanic, Pacific Islander and Asian populations.15,17 For example, the incidence of T2DM in youth is six times higher in Australian Indigenous indivi duals than in the general population.18 In the SEARCH for Diabetes in Youth Study in the USA, a higher incidence rate was observed among youths aged 15–19 years in minority populations (17.0 to 49.4 per 100,000 personyears) compared with 5.6 per 100,000 personyears in nonHispanic whites.17 Further analyses from the same study suggest that the contributions of genetic and/or environmental factors to earlyonset of T2DM may differ in various ethnic groups.19

In addition to the rising prevalence of T2DM in youth, the trend of prediabetes among adolescents is increasing. On the basis of the latest data from the US National Health and Nutrition Examination Survey (NHANES), a 87.1% increase in prevalence of IFG has occurred, from 7% in 1999–2000 to 13.1% in 2005–2006, among US adolescents aged 12–19 years.20,21 Furthermore, an estimated 16.1% of US adolescents had IFG and/or IGT in 2005–2006.21 The prevalence of prediabetes is even higher in pediatric populations who have other risk factors, such as obesity, hyperinsulinemia or a family history of diabetes mellitus.21–23 The continuous increase in the prevalence of obesity in youth in Asian countries, such as China and India, portends to increasing numbers of individuals developing T2DM at younger ages if no effective in tervention strategies slow down the obesity epidemic.24

The fall in the age of onset of T2DM25 and the unfavorable metabolic control in youth with the disease26 will substantially influence the future burden of T2DM. Youth with T2DM represent a population at increased risk of development of early complications, and the occurrence of lifelong chronic complications is likely to be higher in this age group owing to the long duration of the disease.

Key points

■ The prevalence of type 2 diabetes mellitus (T2DM) and prediabetes has been rapidly rising worldwide over the past three decades, particularly in developing countries

■ In addition to the early onset of T2DM in young adults, an increasing trend of T2DM and prediabetes is noticeable among children and adolescents

■ The epidemic of T2DM is attributable to a mixture of genetic and epigenetic predispositions and a variety of behavioral and environmental risk factors

■ An integrated approach, taking into account genetic and epigenetic determinants, is required for the effective prevention of T2DM beginning from the start of life

Risk factors for T2DM: new insightsOverweight and obesityThe global epidemic of T2DM is tied to rising rates of overweight and obesity in adults as well as in youth. The prevalence of overweight (BMI of 25–30 kg/m2) or

300 –

250 –

200 –

150 –

100 –

50 –

0 –

2010 2030 2010 2030 2010 2030 2010 2030 2010 2030 2010 2030 2010 2030

Africa South andCentralAmerica

IGTDiabetes mellitus

Middle Eastand North

Africa

NorthAmerica

Europe South-East Asia

WesternPaci c

Num

ber

of p

eopl

e (m

illlio

ns)

a

b

37.453.242%

18.029.665%

World2010 = 285 million2030 = 439 million

Increase 54%

12.123.998%

55.466.520%

26.651.794%

58.7101.072%

76.7112.847%

Figure 1 | Global projections for the diabetes epidemic: 2010–2030. a | In each box, the top and middle values represent the number of people with diabetes mellitus (in millions) in each of seven world regions (depicted with different colors) for 2010 and 2030, respectively; the bottom value is the percentage increase from 2010 to 2030. The number of people globally with diabetes mellitus is projected to rise from 285 million in 2010 to 439 million by 2030, a 54% increase. b | The number of people with diabetes mellitus and IGT (in millions) by region among adults aged 20–79 years for the years 2010 and 2030. Data courtesy of the International Diabetes Federation Diabetes Atlas.114 Abbreviation: IGT, impaired glucose tolerance.

REVIEWS



obesity (BMI of ≥30 kg/m2) in the world’s adult population is predicted to rise from 33% in 2005 to 57.8% in 2030, if recent secular trends of obesity continue.27 Overweight or obesity are the single most important predictors of T2DM,28 and the effect of obesity on lifetime risk of T2DM is stronger in younger adults.29 Moreover, in the European Prospective Investigation into Cancer and Nutrition–Potsdam Study, weight gain in early adulthood (25–40 years) was found to be associated with a higher risk and earlier onset of T2DM than was weight gain after the age of 40 years.30

The concept of being metabolically obese despite a normal weight has been proposed to explain the high risk of T2DM in some normalweight individuals.31 Studies have shown that the incidence of T2DM is higher among individuals with normal weight but with insulin resistance or the metabolic syndrome than in overweight individuals who do not have insulin resistance or the metabolic syndrome.32,33

The ‘metabolically obese’ phenotype might also explain the mismatch between the rates of obesity and T2DM in Asia.34 Although the prevalence of overweight or obesity is generally lower in most Asian than white populations, Asian individuals tend to develop T2DM at a lower BMI level than Europids,34 and the risk of T2DM tends to be higher in Asian populations compared with people of Europid origin for any given BMI levels.35 Asian indivi duals are more likely to have a higher fat percentage36 or visceral adiposity37,38 at a given BMI or waist circumference than Europids. Data from the Obesity in Asia Collaboration, which includes information on >263,000 individuals from 21 studies in the AsiaPacific region, have shown that measures of central adiposity, such as waist circumference, have a stronger association with prevalent T2DM than BMI.35

Visceral adiposity is an independent risk factor for insulin resistance, T2DM and other cardiovascular risk factors.39 Some investigators argue that nonalcoholic fatty liver disease (NAFLD), a phenotype of ectopic fat accumulation in the liver, is a better indicator of T2DM risk than excessive accumulation of visceral adipose tissue.40 NAFLD is found to have a stronger association with peripheral insulin resistance than abdominal fat content.41 In a metaanalysis of 21 cohort studies, ultrasonographydiagnosed NAFLD and associated elevation in its surrogate markers, such as alanine amino transferase and γglutamyltransferase, are consistently found to independently predict the development of T2DM.42 Moreover, Taylor has proposed that ectopic fat deposition in the liver and islets underlies the development of hepatic insulin resistance and βcell dysfunction.43

Developmental origins of T2DMIntrauterine development is a critical and sensitive period during which an adverse intrauterine milieu can affect fetal development by modifying epigenetic gene expression. Epigenetic modifications have been defined as herit able alterations in gene expression that are not associated with changes in DNA sequence, but instead involve DNA methylation and histone modification.44–46

Low birth weight has been consistently found to be associated with an increased risk of the development of T2DM in later life. In a metaanalysis that included 28 populations from different ethnicities, a 1 kg increase in birth weight was associated with a 20% risk reduction of T2DM.47 Low birth weight due to nutritional deprivation in utero influences later susceptibility to obesity, T2DM and other metabolic abnormalities through the acquisition of a ‘thrifty phenotype’; the thrifty phenotype hypothesis postulates that poor fetal and infant nutrition lead to permanent changes in glucose metabolism.48 The link between fetal malnutrition and later T2DM risk in humans is elegantly illustrated by the observations from the Dutch Hunger Famine birth cohort study, which reported that adults who had been exposed to famine during fetal life had a worse glucose tolerance status than unexposed individuals.49 Similar findings have been repli cated in populations from other regions that have suffered nutritional hardship, such as the Chinese Famine (1959–1961) Study.50

Fetal undernutrition is proposed to predispose individuals to insulin resistance49,51,52 and reduced βcell mass and function,53 which in turn increases their susceptibility to the development of T2DM in later life. The risk of T2DM owing to inadequate fetal nutrition is likely to be exacerbated in people who are exposed to an affluent nutritional environment in adult life and who have excess and rapid weight gain in early adulthood.49,50,54 The role of mismatch between intrauterine and adult life environment in this proposed mechanism might explain the current diabetes epidemic in some developing countries. For example, in Cambodia, where severe undernutrition occurred during the political upheaval some three decades ago, economic development and

Box 1 | Modifiable and nonmodifiable risk factors for T2DM

Modifiable risk factors

■ Overweight or obesity

■ Physical inactivity

■ Sedentary behavior

■ Dietary factors

■ Smoking

■ Previously identified glucose tolerance (IGT and/or IFG)

■ Abnormal lipids (elevated triglycerides, low HDL cholesterol levels)

■ Hypertension

■ Inflammation ■ Intrauterine environment

Non-modifiable risk factors ■ Age

■ Sex

■ Ethnicity

■ Family history of T2DM

■ History of gestational diabetes

■ Polycystic ovary syndrome

Abbreviations: IFG, impaired fasting glucose; IGT, impaired glucose tolerance; T2DM, type 2 diabetes mellitus.

REVIEWS



improved nutrition have coincided with the emergence of T2DM at prevalence rates comparable with those of developed nations.55

Instead of a simple reverse relationship between birth weight and the risk of T2DM, a few studies, such as those in Pima Indians,51 Taiwanese schoolchildren56 and Australians,57 have suggested a Ushaped or reversed Jshaped association between birth weight and T2DM, in which high birth weights (>4.0 kg) are also associated with an elevated risk of T2DM. Exposure to intrauterine hyperglycemia predisposes the fetus to produce additional insulin, which acts as a fetal growth hormone leading to excess fetal growth. Abnormal glucose tolerance during pregnancy is, therefore, a twoedged sword, because it predisposes both mother and offspring to the development of obesity, the metabolic syndrome and T2DM in later life.

Women with gestational diabetes mellitus have a more than sevenfold increased risk of subsequently developing T2DM compared with women who experience a normoglycemic pregnancy.58 Exposure to intrauterine hyper glycemia is an important determinant of diabetes mellitus in adult offspring in addition to genetic susceptibility and independent of maternal diabetes type.59–62 Fetal exposures to maternal hyperglycemia lead to increased overall and abdominal obesity in youth63 and predispose to an earlier onset of T2DM.64 Furthermore, a strong association between maternal glycemia during pregnancy and risk of T2DM in Pima Indian offspring was observed even among mothers who had a normal range for glucose tolerance during pregnancy.65

Given the earlier onset of T2DM in female offspring with intrauterine exposure to hyperglycemia and the increasing prevalence of T2DM that predates pregnancy66 or gestational diabetes,67,68 it is essential that the vicious cycle of diabetes begetting diabetes over generations is broken. This intergenerational cycle of T2DM highlights the importance and need for effective prevention or intervention strategies to improve the management of glucose tolerance during pregnancy.

Genetic susceptibilityT2DM is a complex, multifactorial disease fuelled by interactions between multiple susceptible genetic loci and various environmental and behavioral factors (Box 1). Disparity in the risk of T2DM between dif ferent ethnic groups after controlling for diverse environmental attributes indicates a genetic predisposition in the develop ment of T2DM. For example, the common variants of the TCF7L2 gene are significantly associated with risk of T2DM, with a pooled odds ratio of 1.46 for the rs7903146 variant.69 However, substantial differences exist in the locations of risk allele and frequencies of occurrence of particular risk alleles across different ethnic groups.4

Despite multiple genetic loci being associated with the risk of T2DM, the discriminative ability of genetic scores based on a number of risk alleles is unsatisfactory.70 Furthermore, the addition of risk alleles only slightly improved the prediction of future T2DM compared with

risk models based on clinical risk factors or family history of T2DM (commonly considered to represent heritable genetic risk).71–77 However, the use of genetic markers might be a much more valuable addition in children and younger adults, as other commonly used phenotypic risk factors such as family history or hypertension may not yet be expressed. A report from the Framingham Offspring study with 34 years of followup showed that knowledge of common genetic variation appropriately improved T2DM risk reclassification after accounting for common clinical risk factors among people <50 years of age but not older people.78 In addition, the cost of potential use of genetic testing would be another important practical issue in the evaluation of its value, particularly in the countries and areas with limited budgets for health care.

Diabetes genetic testing can be used not only to ascertain an individual’s risk but also to motivate highrisk individuals to change their lifestyle and adhere to necessary preventive measures prior to the onset of clinical phenotypes. Some evidence suggests that genetic factors interact with the environment to affect risk of T2DM. In the Health Professionals Followup Study, a Western dietary pattern characterized by high consumptions of red and processed meats as well as refined foods, was significantly associated with an elevated risk of T2DM among men with a high genetic risk score but not among those with a low genetic risk score.79 Furthermore, findings from the Finnish Diabetes Prevention Study suggest that individual tailoring of lifestyle interventions based on genetic predisposition may maximize the benefits of the interventions.80

In the US Diabetes Prevention Program81 and Finnish Diabetes Prevention Study,82 lifestyle intervention significantly reduced the risk of T2DM among highrisk participants as determined by genetic polymorphisms. However, direct evidence that perception of T2DM genetic risk improves the motivation level of highrisk individuals towards behavioral change is scarce. In a survey from 152 patients without T2DM, over 70% of patients reported that an appraisal of highrisk status would motivate them much more to adhere to healthy lifestyle changes.83 On the other hand, concerns also exist that less motivated individuals would use the lowrisk assessment results to further reduce their engagement in health prevention. Interviews with 22 overweight individuals with high phenotypic risk of T2DM suggested that individuals’ knowledge and healthy behaviors prior to genetic testing would lead to a positive response of genetic testing results.84

Other lifestyle and environmental risk factorsApart from overweight and obesity, genetic and epigenetic factors and other major factors (Box 1), a number of novel factors have been identified to be independently associated with the risk of T2DM, such as sleeping disorders,85 depression86 and antidepressant medication use.87 Some studies have also suggested a potential role of environmental toxins, such as endocrine disruptors (for example, bisphenol A)88 and particulate matter in air pollution, in the development of T2DM.89,90

REVIEWS



Move to HbA1c for diabetes diagnosisPlasma (or blood or serum) glucose concentrations have been used in the diagnosis of T2DM and, therefore, estimates of T2DM prevalence and incidence have been primarily dependent on glucose measures. In the past few years, HbA1c, a measure of average glycemia over the previous 8–12 weeks91 has been recommended as an alternative means for the diagnosis of T2DM by the American Diabetes Association (ADA)92 and the WHO.93 Among individuals without T2DM, increasing HbA1c level is associated with not only future risk of T2DM94 but also a substantially increased risk of in cident cardiovascular events and deaths.95,96

The cutoff of HbA1c of ≥6.5% for the diagnosis of diabetes mellitus, as recommended by the ADA and WHO, was derived based on the association between HbA1c and prevalent retinopathy.97 The decision to use this threshold was based on data from the DETECT2 project, which pooled data from ~45,000 participants from five countries and showed a narrow thres hold range for HbA1c at which risk of diabetesspecific retino pathy (moderate nonproliferative and more severe retinopathy) increases significantly.98

Although HbA1c is a convenient diagnostic test, as it can be performed in the nonfasting state and has greater preanalytic stability and less biological perturbations compared with fasting or 2 h postload glucose testing, the discordant diagnosis of diabetes mellitus with the use of HbA1c and glucose criteria is concerning (Table 1).99–107 Research conducted in the USA and India indicates a substantially overlapping distribution of HbA1c level among individuals with different glucose tolerance status.108,109 Furthermore, of the 16 studies that have reported the prevalence of undiagnosed diabetes mellitus detected by either HbA1c ≥6.5% or oral glucose toler ance test (OGTT), 13 have found that HbA1c cri teria lead to a lower prevalence of undiagnosed diabetes mellitus than OGTT criteria. HbA1c criteria have high specificity (>90%) in detecting diabetes mellitus as defined by OGTT. However, the probability of having an HbA1c ≥6.5% among cases of diabetes mellitus based on OGTT criteria varies dramatically across ethnicities (from 17.0% among Australians to 78.0% in Asian Indians).99–107

A number of studies have proposed alternative HbA1c cutoff points for detecting undiagnosed diabetes mellitus and suggested the potential use of ethnic specific

Table 1 | Accuracy of using HbA1c ≥6.5% for detecting undiagnosed diabetes as defined by OGTT

Study Study period Age (years) Newly diagnosed diabetes by OGTT (%)

Newly diagnosed diabetes by HbA1c ≥6.5% (%)

Probability (sensitivity) of having HbA1c ≥6.5% among diabetes cases defined by OGTT (%)

Specificity of using HbA1c ≥6.5% for detecting diabetes cases defined by OGTT (%)

Qingdao, China99 2006 35–74 11.9 10.8 28.0 (men), 21.9 (women)

90.5 (men), 91.2 (women)

Shanghai, China107 2007–2008 ≥20 6.2 3.1 50.5 98.1

Inter99, Denmark100 1999–2001 46.2 ± 7.9 4.2 6.7 42.6 NA

Whitehall II, UK100 2002–2004 60.5 ± 5.9 3.7 1.0 25.0 NA

The Australian Diabetes, Obesity and Lifestyle Study, Australia100

1999–2000 ≥25 4.0 0.7 17.0 NA

Inuit Health in Transition Study, Greenland100

2005–2009 44.1 ± 14.6 7.0 3.9 29.6 NA

Kenya100 2005–2006 37.6 ± 10.6 3.4 1.4 20.0 NA

Chennai Urban Rural Epidemiology Study, India100

2001–2004 38.8 ± 12.6 10.2 12.9 78.0 NA

NHANES, USA101 2003–2006 ≥20 5.1 1.6 25.5 95.6

New Hoorn Study, Netherland102 2006–2007 40–65 4.0 1.0 24.0 99.0

Minority Americans, USA103 1995–1999, 1997–2000

54.2 15.5 8.9 40.0 96.8

Screening for Impaired Glucose Tolerance Study, USA104

2005–2008 ≥18 4.6 2.2 33.3 99.3

Non-Hispanic white or black adults in NHANES III, USA104

1988–1994 >40 7.6 2.8 37.9 98.5

Non-Hispanic white or black adults in NHANES 2005–2006, USA104

2005–2006 ≥18 5.2 1.8 29.2 99.6

Telde Study, Spain105 NA ≥30 6.4 2.9 38.7 99.6

Leicester Ethnic Atherosclerosis and Diabetes Risk study, UK106

2002–2004, 2004–2008

40–75 3.3 5.8 69.7 NA

Abbreviations: NA, not available; NHANES, National Health and Nutrition Examination Survey; OGTT, oral glucose tolerance test.

REVIEWS



cut points. However, these alternative thresholds are pri marily suggested on the basis of the sensitivity and specificity values for comparisons with glucosedefined diabetes mellitus cases, rather than on the basis of the association of different HbA1c values and longterm diabetes complications.107,109,110 The International Expert Committee report on the role of HbA1c in the diagnosis of diabetes mellitus in 2009 states that “establishing identical prevalences should not be the goal in defining a new means of diagnosing diabetes. The ultimate goal is to identify individuals at risk for diabetes complications so that they can be treated”.97 Among individuals with normal glucose levels, elevated HbA1c levels are associated with less favorable cardiovascular risk profiles, such as with older age, abdominal obesity and dyslipidemia.108 Interestingly, the Danish arm of the ADDITION trial (AngloDanishDutch Study of Intensive Treatment of People with Screen Detected Type 2 Diabetes in Primary Care) has shown that the combined use of HbA1c ≥6.0% with cardiovascular risk assessment could identify a similar proportion of people who might benefit from preventive interventions (96.7%) as those detected using glucose measures in combination with cardiovascular risk assessment (97.6%).111

In addition to the use of HbA1c as an alternative diagnostic option, the ADA has also proposed that persons with an HbA1c of 5.7–6.4% should be classified as being at high risk of diabetes mellitus, requiring lifestyle or pharmacological interventions.92 However, this criterion dramatically reduces the number of people identified to be at high risk compared with glucose measures.101,104,112 No statement on a specific HbA1c threshold for the presence of intermediate hyperglycemia has been released from the WHO.93 The selection of a threshold at which to initiate preventive strategies should take into account not only the sensitivity and specificity but also the cost and feasibility of the program in the target population.

Whilst considerable improvement in assay standardization of HbA1c has occurred, the cost and availability of HbA1c would be another concern for the wide adoption of the measure as a diagnostic approach. Finally, the potential shift to the use of HbA1c as diagnostic criteria would lead to a different prevalence estimate, because a remarkable discrepancy in diabetes prevalence defined by glucose measures and HbA1c exists. This change would have a considerable effect on the capacity to compare longitudinal changes in populations in future studies given the different methods used to diagnose diabetes mellitus—namely, glucose before and now HbA1c. A recommendation that diabetes prevalence be defined both by HbA1c and glucose criteria for a period of time is, therefore, appropriate as it will enable comparison of national and international historical data.

Prevention of T2DMIndividuals with blood glucose levels higher than normal but not high enough for a diagnosis of T2DM, such as those with IGT and/or IFG, are usually con sidered to have a high risk of future T2DM.113 Global estimates suggest that the number of people with IGT will increase from

344 million in 2010 to 472 million in 2030 (Figure 1a),114 which represents a large pool of people who are likely to develop T2DM in the near future.

Epidemiological studies have shown than nearly 90% of cases of incident T2DM can be attributed to five major lifestyle factors: diet, physical activity, smoking, overweight or obesity and alcohol consumption.28,115 The importance of adoption of a healthier lifestyle is supported by robust data from diabetes prevention trials. Several major trials unequivocally show that intensive lifestyle interventions, specifically aimed at weight loss and increased physical activity in highrisk indivi duals, can prevent or at least delay the progression to overt T2DM by 50% and that they are as effective as pharmacological interventions.116 Moreover, intensive lifestyle intervention in individuals with IGT is found to be most effective among those with a high baseline T2DM risk, as determined by the presence of multiple risk factors.117

A number of risk assessment tools for predicting incident T2DM, based on selfassessed, biological measures and even genetic markers, have been derived from different ethnicities,118 and they provide a more practical and valuable approach for detecting those at risk of developing T2DM compared with universal population screening using a blood glucose test.119 Some selfassessment tools and lifestyle intervention strategies have been incorporated into the diabetes screening and prevention program in the primary care setting in the past few years, such as the Finnish National Dia betes Prevention Program120 and the DEPLAN (Diabetes in Europe–Prevention using Lifestyle, Physical Activity and Nutritional Intervention) project.121 The 1year followup data from the last two studies strongly support the implementation of communitybased lifestyle intervention programs in highrisk populations, which will complement the populationbased approach,122 such as provision of supportive environments for physical activity.

ConclusionsOver the last three decades, we have experienced a spectacu lar rise in the prevalence of obesity and T2DM in nearly every nation of the world and the resulting heavy health burden associated with these disorders and their related complications. The causes of the diabetes epidemic are embedded in an extremely complex combination of genetic and epigenetic predispositions interacting within an equally complex combination of societal factors that determine behavior and environmental risks.

On the other hand, considerable improvements in the detection of individuals with undiagnosed T2DM and effective interventions in highrisk populations have taken place. Furthermore, some evidence now suggests that the prevalence of obesity is going to be stable or level off in adults as well as in youth in some developed countries,24 which might lead to the reduction of incident T2DM in these countries, as obesity remains a key driver of T2DM. However, a number of other factors are attributable to the diabetes epidemic other than obesity. They include fetal and early life nutrition status, as well

REVIEWS



as some factors associated with rapid socioeconomic development, such as depression,86 sleeping disorders85 and environmental pollutants.88

Tackling and curbing the escalating diabetes epidemic has been a long and twisted journey, and it requires efforts from each level of society, including scientists, medical practitioners, public health professionals, healthcare providers and policymakers, and most importantly, the awareness of the general population. Further research is required to better understand the potential role of the remaining risk factors, such as fetal and genetic predisposition, to help shape prevention programs.

Review criteria

Articles were identified from searches in PubMed. The search terms used were “diabetes”, “impaired glucose tolerance”, “impaired fasting glucose”, “glycated haemoglobin”, “prevalence”, “incidence”, “epidemiology”, “ethnicity”, “gestational diabetes”, “obesity”, “birth weight” and “genetic susceptibility”. Articles published between 1995 and 2011 were searched with a special focus on papers published since the year 2000. All articles identified were English-language, full-text papers. Additional references were selected from the reference lists of identified articles.

1. Zimmet, P., Alberti, K. G. & Shaw, J. Global and societal implications of the diabetes epidemic. Nature 414, 782–787 (2001).

2. Danaei, G. et al. National, regional, and global trends in fasting plasma glucose and diabetes prevalence since 1980: systematic analysis of health examination surveys and epidemiological studies with 370 country-years and 2.7 million participants. Lancet 378, 31–40 (2011).

3. Shaw, J. E., Sicree, R. A. & Zimmet, P. Z. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res. Clin. Pract. 87, 4–14 (2010).

4. Chan, J. C. et al. Diabetes in Asia: epidemiology, risk factors, and pathophysiology. JAMA 301, 2129–2140 (2009).

5. Dowse, G. K. et al. High prevalence of NIDDM and impaired glucose tolerance in Indian, Creole, and Chinese Mauritians. Mauritius Noncommunicable Disease Study Group. Diabetes 39, 390–396 (1990).

6. Simmons, D., Williams, D. R. & Powell, M. J. Prevalence of diabetes in a predominantly Asian community: preliminary findings of the Coventry diabetes study. BMJ 298, 18–21 (1989).

7. McNeely, M. J. & Boyko, E. J. Type 2 diabetes prevalence in Asian Americans: results of a national health survey. Diabetes Care 27, 66–69 (2004).

8. Yang, W. et al. Prevalence of diabetes among men and women in China. N. Engl. J. Med. 362, 1090–1101 (2010).

9. Mbanya, J. C., Motala, A. A., Sobngwi, E., Assah, F. K. & Enoru, S. T. Diabetes in sub-Saharan Africa. Lancet 375, 2254–2266 (2010).

10. Abubakari, A. R. et al. Prevalence and time trends in diabetes and physical inactivity among adult West African populations: the epidemic has arrived. Public Health 123, 602–614 (2009).

11. Wändell, P. E. et al. Estimation of diabetes prevalence among immigrants from the Middle East in Sweden by using three different data sources. Diabetes Metab. 34, 328–333 (2008).

12. Ramachandran, A., Mary, S., Yamuna, A., Murugesan, N. & Snehalatha, C. High prevalence of diabetes and cardiovascular risk factors associated with urbanization in India. Diabetes Care 31, 893–898 (2008).

13. Ning, F. et al. Risk factors associated with the dramatic increase in the prevalence of diabetes in the adult Chinese population in Qingdao, China. Diabet. Med. 26, 855–863 (2009).

14. Pinhas-Hamiel, O. & Zeitler, P. The global spread of type 2 diabetes mellitus in children and adolescents. J. Pediatr. 146, 693–700 (2005).

15. Liese, A. D. et al. The burden of diabetes mellitus among US youth: prevalence estimates

from the SEARCH for Diabetes in Youth Study. Pediatrics 118, 1510–1518 (2006).

16. Kitagawa, T., Owada, M., Urakami, T. & Yamauchi, K. Increased incidence of non-insulin dependent diabetes mellitus among Japanese schoolchildren correlates with an increased intake of animal protein and fat. Clin. Pediatr. (Phila.) 37, 111–115 (1998).

17. Dabelea, D. et al. Incidence of diabetes in youth in the United States. JAMA 297, 2716–2724 (2007).

18. Craig, M. E., Femia, G., Broyda, V., Lloyd, M. & Howard, N. J. Type 2 diabetes in Indigenous and non-Indigenous children and adolescents in New South Wales. Med. J. Aust. 186, 497–499 (2007).

19. Dabelea, D. et al. Association testing of TCF7L2 polymorphisms with type 2 diabetes in multi-ethnic youth. Diabetologia 54, 535–539 (2011).

20. Williams, D. E. et al. Prevalence of impaired fasting glucose and its relationship with cardiovascular disease risk factors in US adolescents, 1999–2000. Pediatrics 116, 1122–1126 (2005).

21. Li, C., Ford, E. S., Zhao, G. & Mokdad, A. H. Prevalence of pre-diabetes and its association with clustering of cardiometabolic risk factors and hyperinsulinemia among U.S. adolescents: National Health and Nutrition Examination Survey 2005–2006. Diabetes Care 32, 342–347 (2009).

22. Sinha, R. et al. Prevalence of impaired glucose tolerance among children and adolescents with marked obesity. N. Engl. J. Med. 346, 802–810 (2002).

23. Goran, M. I. et al. Impaired glucose tolerance and reduced beta-cell function in overweight Latino children with a positive family history for type 2 diabetes. J. Clin. Endocrinol. Metab. 89, 207–212 (2004).

24. Rokholm, B., Baker, J. L. & Sorensen, T. I. The levelling off of the obesity epidemic since the year 1999—a review of evidence and perspectives. Obes. Rev. 11, 835–846 (2010).

25. Mokdad, A. H. et al. Diabetes trends in the U.S.: 1990–1998. Diabetes Care 23, 1278–1283 (2000).

26. Lawrence, J. M. et al. Diabetes in Hispanic American youth: prevalence, incidence, demographics, and clinical characteristics: the SEARCH for Diabetes in Youth Study. Diabetes Care 32 (Suppl. 2), S123–S132 (2009).

27. Kelly, T., Yang, W., Chen, C. S., Reynolds, K. & He, J. Global burden of obesity in 2005 and projections to 2030. Int. J. Obes. (Lond.) 32, 1431–1437 (2008).

28. Hu, F. B. et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N. Engl. J. Med. 345, 790–797 (2001).

29. Narayan, K. M., Boyle, J. P., Thompson, T. J., Gregg, E. W. & Williamson, D. F. Effect of BMI on lifetime risk for diabetes in the U. S. Diabetes Care 30, 1562–1566 (2007).

30. Schienkiewitz, A., Schulze, M. B., Hoffmann, K., Kroke, A. & Boeing, H. Body mass index history and risk of type 2 diabetes: results from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. Am. J. Clin. Nutr. 84, 427–433 (2006).

31. Ruderman, N., Chisholm, D., Pi-Sunyer, X. & Schneider, S. The metabolically obese, normal-weight individual revisited. Diabetes 47, 699–713 (1998).

32. Meigs, J. B. et al. Body mass index, metabolic syndrome, and risk of type 2 diabetes or cardiovascular disease. J. Clin. Endocrinol. Metab. 91, 2906–2912 (2006).

33. Arnlöv, J., Sundström, J., Ingelsson, E. & Lind, L. Impact of BMI and the metabolic syndrome on the risk of diabetes in middle-aged men. Diabetes Care 34, 61–65 (2011).

34. Yoon, K.-H. et al. Epidemic obesity and type 2 diabetes in Asia. Lancet 368, 1681–1688 (2006).

35. Huxley, R. et al. Ethnic comparisons of the cross-sectional relationships between measures of body size with diabetes and hypertension. Obes. Rev. 9 (Suppl. 1), 53–61 (2008).

36. Deurenberg, P., Deurenberg-Yap, M. & Guricci, S. Asians are different from Caucasians and from each other in their body mass index/body fat per cent relationship. Obes. Rev. 3, 141–146 (2002).

37. Kadowaki, T. et al. Japanese men have larger areas of visceral adipose tissue than Caucasian men in the same levels of waist circumference in a population-based study. Int. J. Obes. (Lond.) 30, 1163–1165 (2006).

38. Lear, S. A. et al. Visceral adipose tissue accumulation differs according to ethnic background: results of the Multicultural Community Health Assessment Trial (M-CHAT). Am. J. Clin. Nutr. 86, 353–359 (2007).

39. Lebovitz, H. E. & Banerji, M. A. Point: visceral adiposity is causally related to insulin resistance. Diabetes Care 28, 2322–2325 (2005).

40. Perseghin, G. Lipids in the wrong place: visceral fat and nonalcoholic steatohepatitis. Diabetes Care 34 (Suppl. 2), S367–S370 (2011).

41. Hwang, J. H. et al. Increased intrahepatic triglyceride is associated with peripheral insulin resistance: in vivo MR imaging and spectroscopy studies. Am. J. Physiol. Endocrinol. Metab. 293, E1663–E1669 (2007).

42. Fraser, A. et al. Alanine aminotransferase, γ-glutamyltransferase, and incident diabetes: the British Women’s Heart and Health Study and meta-analysis. Diabetes Care 32, 741–750 (2009).

REVIEWS



43. Taylor, R. Pathogenesis of type 2 diabetes: tracing the reverse route from cure to cause. Diabetologia 51, 1781–1789 (2008).

44. Bird, A. Perceptions of epigenetics. Nature 447, 396–398 (2007).

45. Gluckman, P. D., Hanson, M. A., Cooper, C. & Thornburg, K. L. Effect of in utero and early-life conditions on adult health and disease. N. Engl. J. Med. 359, 61–73 (2008).

46. Pinney, S. E. & Simmons, R. A. Epigenetic mechanisms in the development of type 2 diabetes. Trends Endocrinol. Metab. 21, 223–229 (2010).

47. Whincup, P. H. et al. Birth weight and risk of type 2 diabetes: a systematic review. JAMA 300, 2886–2897 (2008).

48. Hales, C. N. & Barker, D. J. The thrifty phenotype hypothesis. Br. Med. Bull. 60, 5–20 (2001).

49. Ravelli, A. C. et al. Glucose tolerance in adults after prenatal exposure to famine. Lancet 351, 173–177 (1998).

50. Li, Y. et al. Exposure to the Chinese famine in early life and the risk of hyperglycemia and type 2 diabetes in adulthood. Diabetes 59, 2400–2406 (2010).

51. Dabelea, D. et al. Birth weight, type 2 diabetes, and insulin resistance in Pima Indian children and young adults. Diabetes Care 22, 944–950 (1999).

52. Bavdekar, A. et al. Insulin resistance syndrome in 8-year-old Indian children: small at birth, big at 8 years, or both? Diabetes 48, 2422–2429 (1999).

53. Fowden, A. L. & Hill, D. J. Intra-uterine programming of the endocrine pancreas. Br. Med. Bull. 60, 123–142 (2001).

54. Hyppönen, E., Power, C. & Smith, G. D. Prenatal growth, BMI, and risk of type 2 diabetes by early midlife. Diabetes Care 26, 2512–2517 (2003).

55. King, H. et al. Diabetes and associated disorders in Cambodia: two epidemiological surveys. Lancet 366, 1633–1639 (2005).

56. Wei, J. N. et al. Low birth weight and high birth weight infants are both at an increased risk to have type 2 diabetes among schoolchildren in Taiwan. Diabetes Care 26, 343–348 (2003).

57. Al Salmi, I. et al. Disorders of glucose regulation in adults and birth weight: results from the Australian Diabetes, Obesity and Lifestyle (AUSDIAB) Study. Diabetes Care 31, 159–164 (2008).

58. Bellamy, L., Casas, J. P., Hingorani, A. D. & Williams, D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet 373, 1773–1779 (2009).

59. Pettitt, D. J. et al. Congenital susceptibility to NIDDM. Role of intrauterine environment. Diabetes 37, 622–628 (1988).

60. Dabelea, D. et al. Intrauterine exposure to diabetes conveys risks for type 2 diabetes and obesity: a study of discordant sibships. Diabetes 49, 2208–2211 (2000).

61. Stride, A. et al. Intrauterine hyperglycemia is associated with an earlier diagnosis of diabetes in HNF-1alpha gene mutation carriers. Diabetes Care 25, 2287–2291 (2002).

62. Sobngwi, E. et al. Effect of a diabetic environment in utero on predisposition to type 2 diabetes. Lancet 361, 1861–1865 (2003).

63. Crume, T. L. et al. Association of exposure to diabetes in utero with adiposity and fat distribution in a multiethnic population of youth: the Exploring Perinatal Outcomes among Children (EPOCH) Study. Diabetologia 54, 87–92 (2011).

64. Pettitt, D. J. et al. Association between maternal diabetes in utero and age at offspring’s diagnosis of type 2 diabetes. Diabetes Care 31, 2126–2130 (2008).

65. Franks, P. W. et al. Gestational glucose tolerance and risk of type 2 diabetes in young Pima Indian offspring. Diabetes 55, 460–465 (2006).

66. Lawrence, J. M., Contreras, R., Chen, W. & Sacks, D. A. Trends in the prevalence of preexisting diabetes and gestational diabetes mellitus among a racially/ethnically diverse population of pregnant women, 1999–2005. Diabetes Care 31, 899–904 (2008).

67. Dabelea, D. et al. Increasing prevalence of gestational diabetes mellitus (GDM) over time and by birth cohort: Kaiser Permanente of Colorado GDM Screening Program. Diabetes Care 28, 579–584 (2005).

68. Getahun, D., Nath, C., Ananth, C. V., Chavez, M. R. & Smulian, J. C. Gestational diabetes in the United States: temporal trends 1989 through 2004. Am. J. Obstet. Gynecol. 198, 525.e1–525.e5 (2008).

69. Cauchi, S. et al. TCF7L2 is reproducibly associated with type 2 diabetes in various ethnic groups: a global meta-analysis. J. Mol. Med. 85, 777–782 (2007).

70. McCarthy, M. I. Genomics, type 2 diabetes, and obesity. N. Engl. J. Med. 363, 2339–2350 (2010).

71. Balkau, B. et al. Predicting diabetes: clinical, biological, and genetic approaches: data from the Epidemiological Study on the Insulin Resistance Syndrome (DESIR). Diabetes Care 31, 2056–2061 (2008).

72. Schulze, M. B. et al. Use of multiple metabolic and genetic markers to improve the prediction of type 2 diabetes: the EPIC-Potsdam Study. Diabetes Care 32, 2116–2119 (2009).

73. Lango, H. et al. Assessing the combined impact of 18 common genetic variants of modest effect sizes on type 2 diabetes risk. Diabetes 57, 3129–3135 (2008).

74. Lyssenko, V. et al. Clinical risk factors, DNA variants, and the development of type 2 diabetes. N. Engl. J. Med. 359, 2220–2232 (2008).

75. Meigs, J. B. et al. Genotype score in addition to common risk factors for prediction of type 2 diabetes. N. Engl. J. Med. 359, 2208–2219 (2008).

76. van Hoek, M. et al. Predicting type 2 diabetes based on polymorphisms from genome-wide association studies: a population-based study. Diabetes 57, 3122–3128 (2008).

77. Talmud, P. J. et al. Utility of genetic and non-genetic risk factors in prediction of type 2 diabetes: Whitehall II prospective cohort study. BMJ 340, b4838 (2010).

78. de Miguel-Yanes, J. M. et al. Genetic risk reclassification for type 2 diabetes by age below or above 50 years using 40 type 2 diabetes risk single nucleotide polymorphisms. Diabetes Care 34, 121–125 (2011).

79. Qi, L., Cornelis, M. C., Zhang, C., van Dam, R. M. & Hu, F. B. Genetic predisposition, Western dietary pattern, and the risk of type 2 diabetes in men. Am. J. Clin. Nutr. 89, 1453–1458 (2009).

80. Laaksonen, D. E. et al. Physical activity, diet, and incident diabetes in relation to an ADRA2B polymorphism. Med. Sci. Sports. Exerc. 39, 227–232 (2007).

81. Florez, J. C. et al. TCF7L2 polymorphisms and progression to diabetes in the Diabetes Prevention Program. N. Engl. J. Med. 355, 241–250 (2006).

82. Wang, J. et al. Variants of transcription factor 7-like 2 (TCF7L2) gene predict conversion to type 2 diabetes in the Finnish Diabetes Prevention Study and are associated with impaired glucose regulation and impaired insulin secretion. Diabetologia 50, 1192–1200 (2007).

83. Grant, R. W. et al. The clinical application of genetic testing in type 2 diabetes: a patient and physician survey. Diabetologia 52, 2299–2305 (2009).

84. Markowitz, S. M., Park, E. R., Delahanty, L. M., O’Brien, K. E. & Grant, R. W. Perceived impact of diabetes genetic risk testing among patients at high phenotypic risk for type 2 diabetes. Diabetes Care 34, 568–573 (2011).

85. Shaw, J. E., Punjabi, N. M., Wilding, J. P., Alberti, K. G. & Zimmet, P. Z. Sleep-disordered breathing and type 2 diabetes: a report from the International Diabetes Federation Taskforce on Epidemiology and Prevention. Diabetes Res. Clin. Pract. 81, 2–12 (2008).

86. Mezuk, B., Eaton, W. W., Albrecht, S. & Golden, S. H. Depression and type 2 diabetes over the lifespan: a meta-analysis. Diabetes Care 31, 2383–2390 (2008).

87. Kivimäki, M. et al. Antidepressant medication use, weight gain, and risk of type 2 diabetes: a population-based study. Diabetes Care 33, 2611–2616 (2010).

88. Alonso-Magdalena, P., Quesada, I. & Nadal, A. Endocrine disruptors in the etiology of type 2 diabetes mellitus. Nat. Rev. Endocrinol. 7, 346–353 (2011).

89. Brook, R. D. et al. Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association. Circulation 121, 2331–2378 (2010).

90. Krämer, U. et al. Traffic-related air pollution and incident type 2 diabetes: results from the SALIA cohort study. Environ. Health Perspect. 118, 1273–1279 (2010).

91. Nathan, D. M., Turgeon, H. & Regan, S. Relationship between glycated haemoglobin levels and mean glucose levels over time. Diabetologia 50, 2239–2244 (2007).

92. American Diabetes Association. Standards of medical care in diabetes–2010. Diabetes Care 33 (Suppl. 1), S11–S61 (2010).

93. World Health Organization. Use of glycated haemoglobin (HbA1c) in the diagnosis of diabetes mellitus: abbreviated report of a WHO consultation [online], http://www.who.int/diabetes/publications/report-hba1c_2011.pdf (2011).

94. Zhang, X. et al. A1C level and future risk of diabetes: a systematic review. Diabetes Care 33, 1665–1673 (2010).

95. Selvin, E. et al. Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. N. Engl. J. Med. 362, 800–811 (2010).

96. Santos-Oliveira, R. et al. Haemoglobin A(1c) levels and subsequent cardiovascular disease in persons without diabetes: a meta-analysis of prospective cohorts. Diabetologia 54, 1327–1334 (2011).

97. The International Expert Committe. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 32, 1327–1334 (2009).

98. Colagiuri, S. et al. Glycemic thresholds for diabetes-specific retinopathy: implications for diagnostic criteria for diabetes. Diabetes Care 34, 145–150 (2011).

99. Zhou, X. et al. Performance of an A1C and fasting capillary blood glucose test for screening newly diagnosed diabetes and pre-diabetes defined by an oral glucose tolerance test in Qingdao, China. Diabetes Care 33, 545–550 (2010).

100. Christensen, D. L. et al. Moving to an A1C-based diagnosis of diabetes has a different impact on prevalence in different ethnic groups. Diabetes Care 33, 580–582 (2010).

REVIEWS


http://www.who.int/diabetes/publications/report-hba1c_2011.pdf

http://www.who.int/diabetes/publications/report-hba1c_2011.pdf


101. Cowie, C. C. et al. Prevalence of diabetes and high risk for diabetes using A1C criteria in the U.S. population in 1988–2006. Diabetes Care 33, 562–568 (2010).

102. van ‘t Riet, E. et al. Relationship between A1C and glucose levels in the general Dutch population: the new Hoorn study. Diabetes Care 33, 61–66 (2010).

103. Araneta, M. R., Grandinetti, A. & Chang, H. K. A1C and diabetes diagnosis among Filipino Americans, Japanese Americans, and Native Hawaiians. Diabetes Care 33, 2626–2628 (2010).

104. Olson, D. E. et al. Screening for diabetes and pre-diabetes with proposed A1C-based diagnostic criteria. Diabetes Care 33, 2184–2189 (2010).

105. Boronat, M. et al. Differences in cardiovascular risk profile of diabetic subjects discordantly classified by diagnostic criteria based on glycated hemoglobin and oral glucose tolerance test. Diabetes Care 33, 2671–2673 (2010).

106. Mostafa, S. A. et al. The potential impact of using glycated haemoglobin as the preferred diagnostic tool for detecting type 2 diabetes mellitus. Diabet. Med. 27, 762–769 (2010).

107. Bao, Y. et al. Glycated haemoglobin A1c for diagnosing diabetes in Chinese population: cross sectional epidemiological survey. BMJ 340, c2249 (2010).

108. Selvin, E., Zhu, H. & Brancati, F. L. Elevated A1C in adults without a history of diabetes in the U.S. Diabetes Care 32, 828–833 (2009).

109. Mohan, V. et al. A1C cut points to define various glucose intolerance groups in Asian Indians. Diabetes Care 33, 515–519 (2010).

110. Bennett, C. M., Guo, M. & Dharmage, S. C. HbA1c as a screening tool for detection of type 2 diabetes: a systematic review. Diabet. Med. 24, 333–343 (2007).

111. Lauritzen, T., Sandbaek, A., Skriver, M. V. & Borch-Johnsen, K. HbA1c and cardiovascular risk score identify people who may benefit from preventive interventions: a 7 year follow-up of a high-risk screening programme for diabetes in primary care (ADDITION), Denmark. Diabetologia 54, 1318–1326 (2011).

112. Mann, D. M. et al. Impact of A1C screening criterion on the diagnosis of pre-diabetes among U.S. adults. Diabetes Care 33, 2190–2195 (2010).

113. Gerstein, H. C. et al. Annual incidence and relative risk of diabetes in people with various categories of dysglycemia: a systematic overview and meta-analysis of prospective studies. Diabetes Res. Clin. Pract. 78, 305–312 (2007).

114. International Diabetes Federation. IDF Diabetes Atlas (4th Edn). Diabetes estimates Excel tables [online]. http://www.idf.org/node/23640 (2009).

115. Mozaffarian, D. et al. Lifestyle risk factors and new-onset diabetes mellitus in older adults: the cardiovascular health study. Arch. Intern. Med. 169, 798–807 (2009).

116. Gillies, C. L. et al. Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes

in people with impaired glucose tolerance: systematic review and meta-analysis. BMJ 334, 299 (2007).

117. Lindström, J. et al. Determinants for the effectiveness of lifestyle intervention in the Finnish Diabetes Prevention Study. Diabetes Care 31, 857–862 (2008).

118. Buijsse, B., Simmons, R. K., Griffin, S. J. & Schulze, M. B. Risk assessment tools for identifying individuals at risk of developing type 2 diabetes. Epidemiol. Rev. 33, 46–62 (2011).

119. Chen, L. et al. Maximizing efficiency and cost-effectiveness of type 2 diabetes screening: the AusDiab study. Diabet. Med. 28, 414–423 (2011).

120. Saaristo, T. et al. Lifestyle intervention for prevention of type 2 diabetes in primary health care: one-year follow-up of the Finnish National Diabetes Prevention Program (FIN-D2D). Diabetes Care 33, 2146–2151 (2010).

121. Makrilakis, K., Liatis, S., Grammatikou, S., Perrea, D. & Katsilambros, N. Implementation and effectiveness of the first community lifestyle intervention programme to prevent type 2 diabetes in Greece. The DE-PLAN study. Diabet. Med. 27, 459–465 (2010).

122. Rose, G. Sick individuals and sick populations. Int. J. Epidemiol. 14, 32–38 (1985).

Author contributionsAll authors contributed equally to all aspect of article preparation.

REVIEWS


http://www.idf.org/node/23640

nature outlokk diabetes 2012

Documents