diabetes and risk of parkinson disease · 2011-12-28 · diabetes and risk of parkinson’s disease...

Diabetes and Risk of Parkinson’sDiseaseA systematic review and meta-analysis

EMANUELE CEREDA, MD, PHD1,2

MICHELA BARICHELLA, MD2

CARLO PEDROLLI, MD3

CATHERINE KLERSY, MD, MSC4

ERICA CASSANI, MD2

RICCARDO CACCIALANZA, MD1

GIANNI PEZZOLI, MD2

OBJECTIVEdDiabetes has been associated with chronic neurodegeneration.We performed asystematic review and meta-analysis to assess the relationship between pre-existing diabetes andParkinson’s disease (PD).

RESEARCHDESIGNANDMETHODSdOriginal articles in English published up to 10May 2011 were searched for in electronic databases (PubMed, Embase, and Scopus) and byreviewing references of eligible articles. Prospective cohort and case-control studies providingrisk and precision estimates relating to pre-existing diabetes and PD were considered eligible.

RESULTSdNine studies/1,947 citations (cohort,N = 4; case-control,N = 5) fulfilled inclusioncriteria for meta-analysis. In prospective studies, the onset of diabetes before onset of PD wasfound to be a risk factor for future PD (relative risk [RR] = 1.37 [95%CI 1.21–1.55]; P, 0.0001).This association was confirmed by secondary analyses based on estimates derived after theexclusion of participants who had vascular disease at baseline and/or who developed vasculardisease during follow-up (RR = 1.34 [1.14–1.58]; P, 0.001) and by sensitivity analyses addressingthe association with diabetes at baseline or during follow-up. However, the association found forcase-control studies was not significant (odds ratio [OR] 0.75 [95%CI 0.50–1.11]; P = 0.835).Sensitivity analysis based on estimates adjusted for BMI confirmed the lack of a relationshipbetween PD and diabetes (OR 0.56 [0.28–1.15]; P = 0.089).

CONCLUSIONSdAlthough data from cohort studies suggest that diabetes is a risk factor forPD, there is no conclusive evidence on this association. Further prospective studies focused onputative pathogenic pathways and taking a broad range of confounders into account is requiredto clarify this relationship.

Diabetes Care 34:2614–2623, 2011

Parkinson’s disease (PD) is a progressiveand disabling motor disorder in whichtypical symptoms(bradykinesia, resting

tremor, rigidity, and postural instability)are mainly a result of a reduction in dopa-minergic activity of the substantia nigra inthe midbrain. It is the second most com-mon chronic neurodegenerative disease ofaging people with an approximate ratio of1:800–1,000 in subjects aged over 60 years(1,2). PD is regarded mainly as a sporadic

disorder. In the last 30 years, research on itspathogenesis has identified several poten-tial contributing factors, suggesting that it isof multifactorial origin. Besides a numberof genetic mutations, nutritional and met-abolic factors appear to be involved (2–4).The relationship between diabetes and PDhas been the subject of several studies thathave yielded conflicting results (5–8).Along with this, various disease-relatedmechanisms have been suggested that are

involved in both the promotion of PD andprotection against it (9,10).

Because information on the etiologyof PD may improve clinical practice interms of both prevention and treatment,the aim of this study was to systematicallyreview the published literature relating tothe role of diabetes as a risk factor for PD.

RESEARCH DESIGN ANDMETHODSdThe review and meta-analysis procedures were planned, con-ducted, and reported following thePRISMAguidelines (11).

Literature search and studyinclusion criteriaA literature search for all English languagemanuscripts published up to 10 May 2011and focused on the relationship betweendiabetes and PD was performed indepen-dently by E.Ce. and C.P. Queried databaseswere as follows: PubMed (accessed 10 May2011), Scopus (accessed 21 April 2011),and Embase (accessed 21 April 2011).

The search strategy included terms(free text and/or MeSH terms adapted tothe requirements of each database) for di-abetes (diabetes mellitus, diabetes mellitustype-2, diabetes mellitus type-1, glucoseintolerance, and insulin resistance), and PD.We also searched the references of includedarticles for potential additional reports.Searching queries are listed in the Supple-mentary Data.

Both prospective (cohort) and case-control original studies investigating therole of diabetes as a risk factor for PDwerepotentially eligible for inclusion. Manu-scripts were initially selected on the basisof the title and abstract. To be included inthe quantitative analyses (meta-analysis),articles had to report at least a risk (relativerisk [RR], or odds ratio [OR]) and a precisionestimate (95% CI) relating pre-existingdiabetes to subsequent incident PD orenough data to calculate them.

Thereafter, selected studies were as-sessed for quality (risk of bias) accordingto the following items: 1) for cohort studies,the systematic lack of case identification orbias in case ascertainment (for PD) and 2)for case-control studies, the exclusion of

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the 1Nutrition and Dietetics Service, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy; the2Parkinson Institute–Istituti Clinici di Perfezionamento, Milan, Italy; the 3Dietetic and Clinical NutritionUnit, Trento Hospital, Trento, Italy; and the 4Biometry and Clinical Epidemiology Service, FondazioneIRCCS Policlinico San Matteo, Pavia, Italy.

Corresponding author: Emanuele Cereda, [email protected] 22 August 2011 and accepted 20 September 2011.DOI: 10.2337/dc11-1584This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10

.2337/dc11-1584/-/DC1.© 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited, the use is educational and not for profit, and thework is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

2614 DIABETES CARE, VOLUME 34, DECEMBER 2011 care.diabetesjournals.org

R e v i e w s / C o m m e n t a r i e s / A D A S t a t e m e n t sM E T A - A N A L Y S I S

Table

1dListan

dfeatures

oftheprospectivecohort

stud

iesinclud

edin

quan

titative

analyses

Source

[reference]

Coh

ortand

features

(nation)

Follo

w-up

(years)

Age

atinclusion

(years)

Age

atdiagno

sis

ofPD

(years)

Riskof

PD(95%

CI)

Adjustment

variables

Findings

Potential

estimationbias

Huetal.

2007

[6]

51,552

subjects

witho

utPD

,and

type1diabetes

(Finland

)

18.0

[mean]

25–74

[range]

Wom

en:6

5.8

Men:6

4.3

[mean]

1.83

(1.21–

2.76

)Age,sex,smok

ing,

BMI,alcoho

luse,

coffeeandtea

consumption,

education,

physicalactivity,

systolicbloo

dpressure,and

totalcholesterol

(baseline)

Typ

e2diabetes

was

associated

withan

increasedrisk

ofPD

.In

sensitivity

analysis

(exclusion

ofthose

who

hadvascular

diseases

atbaselin

eor

who

developed

stroke

during

follo

w-up

[N=47

,353

])risk

ofPD

was

1.94

[1.21–

3.11

].

Onlybaselin

eself-repo

rted

diabetes

and

confou

nders

wereincludedin

risk

analyses.

Surveillancebias

might

account

forhigh

erratesin

diabetes.F

inally,

asaresultof

case

ascertainment

procedure,itcould

notbe

excluded

thatfewcases

(mild

untreatedPD

patients)werelost

toidentification

.Simon

etal.

2007

[5]

171,87

9subjects

witho

utprevalent

stroke

andPD

atbaselin

e(121

,046

wom

en[N

urses’

HealthStudy

]and

50,833

men

[Health

Profession

als

Follo

w-upStud

y]);

participants

developing

stroke

before

PDon

set

werecensored

throug

hout

the

follo

w-up(U.S.)

Wom

en:2

2.9

Men:1

2.6

[mean]

Wom

en:3

0–55

Men:4

0–75

[range]

Wom

en:6

3.5

Men:6

9.7

[mean]

1.04

(0.74–

1.46

)Age,sex,and

smok

ing.The

adjustmentfor

multipleup

dated

(every

2years)

covariates

[BMI,

physicalactivity,

alcoho

l,caffeine

andenergy

intake,

andcomorbidities]

prod

uced

similar

results(datano

tshow

n)

Precedingdiabetes

(bothtype

1and

type2)

was

not

relatedto

increased

risk

ofPD

.The

associationof

baselin

ediabetes

was

also

nonsignificant

(RR=1.12

[0.69–

1.81

]).

Dataon

diabetes

wereself-reported.

care.diabetesjournals.org DIABETES CARE, VOLUME 34, DECEMBER 2011 2615

Cereda and Associates

Tab

le1d

Con

tinu

ed

Source

[reference]

Coh

ortand

features

(nation)

Follo

w-up

(years)

Age

atinclusion

(years)

Age

atdiagno

sis

ofPD

(years)

Riskof

PD(95%

CI)

Adjustment

variables

Findings

Potential

estimationbias

Driveretal.

2008

[12]

21,841

malesubjects

free

ofcancer,

vasculardisease,

dementia,and

PDenrolledin

the

Physicians’Health

Stud

y(U.S.)

23.1

[median]

40–84

[range]

73.1

[median]

1.34

(1.01–

1.77

)Age,smok

ing,

alcoho

luse,

BMI,ph

ysical

activity,

hypertension

,andhigh

serum

cholesterol

(updatedyearly)

Upd

ated

historytype

2diabetes

was

associated

with

increasedrisk

ofPD

.Sensitivity

analyses

show

edthatPD

was

more

associated

with

shortdu

ration

andun

complicated

diabetes,andlow

BMI.In

sensitivity

analysis(exclusion

ofthosedeveloping

vascular

disease

during

thefollo

w-up

[N=16

,423

])risk

ofPD

was

1.46

[1.00–

2.13

].

Dataon

diabetes

andPD

were

self-reported.

The

increasedrisk

forthosewith

shorterdu

ration

ofdiabetes

could

beexplained

bydetectionbias

from

increased

medicalsurveillance.

Xuetal.

2011

[13]

288,66

2subjects

witho

utprevalent

PDenrolledin

the

NationalInstitutes

ofHealth-AARP

DietandHealth

Stud

y(U.S.)

15[m

ean]

50–71

[range]

66.7

[7.3]

(mean[SD])

1.41

(1.20–

1.66

)Age,sex,race,BM

I,ph

ysicalactivity,

smok

ing,coffee

intake,and

education

Precedingdiabetes

(bothtype

1and

type2)

was

associated

withan

increasedrisk

ofPD

.In

sensitivity

analysis

(exclusion

ofthose

with

stroke,h

eart

disease,cancer

and

poor/fairhealth

[N=21

5,72

3])risk

ofPD

was

1.34

[1.06–

1.69

].

Dataon

diabetes

were

self-reported.O

nly

baselin

ediabetes

andconfoun

ders

wereincludedin

risk

analyses.T

he

increasedrisk

ofPD

could

bepartly

explained

bydetectionbias

from

increasedmedical

surveillancein

diabeticparticipants.


Diabetes and Parkinson’s disease

Table

2dListan

dfeatures

ofcase-control

stud

iesinclud

edin

qualitativean

alyses

Source

[reference]

Ethnicity

PD(N

)

Age

atinclusion

(years)*

Prevalence

ofdiabetes

(%)

Con

trol

subjects(N

)andmatching

variables

OR**

(95%

CI)

forcases

Adjustment

variables

Potentialestim

ationbias

andotherob

servations

Leibsonetal.

2006

[14]

U.S.(Olm

sted

Cou

nty;

Minnesota)

197incidentcases

70(11)

9.1

197subjects

matched

forage

(61year),sex,

andgeograph

ical

location

0.7(0.4–1.4)

Non

ePresence

ofother

neurologic

disease(e.g.,stroke

ordementia)was

notan

exclusion

criterion.

Atrendtowardahigh

erprevalence

ofstroke

and

dementia

was

presentin

thePD

grou

p.B

MIwas

notincludedam

ongthe

potentialconfou

nders

andadjustingvariables.

Powersetal.

2006

[15]

U.S.(Group

Health

Coo

perative

database;

Washington)

352newly

diagnosed

casesof

idiopathicPD

witho

utcognitive

impairment

69[35–

88]

7.4

484subjects

matched

forage

(indecades),sex,

year

ofenrollm

ent,

andgeograph

ical

location

0.62

(0.32–

1.01

)Age,ethnicity,

education,

and

smok

inghabit

Medicalcond

itions

were

self-reported.B

MIwas

not

includedam

ongpo

tential

confou

ndersand

adjustingvariables.

Sciglianoetal.

2006

[16]

Italy

157newly

diagnosed

cases

(duration

ofPD

,6mon

ths)

ofidiopathicPD

58.1

(11.4)

3.4

533subjects

matched

forage

(63years)and

sex

0.30

(0.13–

0.72

)Age

andsex

Con

trol

subjectswere

recruitedin

aho

spital

setting,andahigh

erprevalence

ofvascular

risk

factors(diabetes,

hypertension

,or

dyslipidem

ia)might

have

occurred.B

MI,althou

ghsimilarin

both

grou

ps,

was

notincludedam

ong

potentialconfou

nders

andadjustingvariables.

Finally,instepwise

multiv

ariableanalysis,

diabetes

was

nolonger

associated

withreduced

risk

ofPD

.



Tab

le2d

Con

tinu

ed

Source

[reference]

Ethnicity

PD(N

)

Age

atinclusion

(years)*

Prevalence

ofdiabetes

(%)

Con

trol

subjects(N

)andmatching

variables

OR**

(95%

CI)

forcases

Adjustment

variables

Potentialestim

ationbias

andotherob

servations

Becker

etal.

2008

[7]

U.K.(General

Practice

Research

Database)

3,63

7newdrug-free

casesof

PD(90%

withan

ageat

onset.60

years)

Not

repo

rted

(90%

aged

.60

years)

8.0%

3,63

7subjects

matched

forage

(sam

eyear

ofbirth),sex,and

generalpractice

0.95

(0.80–

1.14

)BM

I,sm

oking,

andmultip

lecomorbidities

(several

neurologic

disorders,

hypertension

,cardiovascular

diseases,and

dyslipidem

ia)

Detectio

nbias

deriving

from

increasedmedical

surveillancerelated

tosomemedical

cond

itions

could

notbe

excludedatall.

D’Amelio

etal.

2009

[8]

Italy(Italian

region

ofSicily)

318newly

diagnosed

cases

66.7

(-)

4.1

318subjects

matched

for

age(6

2year)

andsex

0.4(0.2–0.8)

Age,sex,

education,

BMI,

occupatio

nal

status,alcoho

landcoffee

consum

ptio

n,andsm

oking

habit

Ascertainmentof

diabetes

was

basedon

self-reporteddata.

Miyakeetal.

2010

[17]

Japan

(Osaka,

Kyoto,and

Wakayam

aPrefectures)

249caseswitha

diseasedu

ration

,6years

68.5

(8.6)

4.0

368in-patientsand

out-patientsno

t,individually

orin

larger

grou

ps,

matched

tocases

0.38

(0.17–

0.79

)Age,sex,smok

ing,

area

ofresidence,

BMI,education,

leisure-tim

eexercise,intake

ofenergy,

cholesterol,

vitamin

E,

alcoho

land

coffee,andthe

dietaryglycem

icindex

Ascertainmentof

diabetes

was

basedon

self-reported

data.A

ltho

ugh

control

subjectswererecruitedin

aho

spitalsetting,and

higher

ratesof

comorbidities

couldbe

expected,p

revalenceof

diabetes

was

comparable

withthatof

thegeneral

populationin

thesame

area.M

oreover,PD

cases

werethinner(P

=0.01

)andolder(P

=0.00

6),

butmultivariablemod

els

accoun

tedfortheeffectof

theseconfoun

ders.



Tab

le2d

Con

tinu

ed

Source

[reference]

Ethnicity

PD(N

)

Age

atinclusion

(years)*

Prevalence

ofdiabetes

(%)

Con

trol

subjects(N

)andmatching

variables

OR**

(95%

CI)

forcases

Adjustment

variables

Potentialestim

ationbias

andotherob

servations

Schernhammer

etal.2

011

[18]

Denmark

1,93

1PD

cases

admittedto

hospitalwith

afirst-time

diagnosisand

identified

throug

hthenation

wide

DanishHospital

Register

72.2

(10.5)

6.5

9,65

1free-living

individuals

(5foranycase)

selected

from

the

CentralPo

pulation

Registry,matched

forage(sam

eyear

ofbirth)

andsex

1.36

(1.08–

1.71

)Age,sex,and

chronic

obstructive

pulm

onary

disease

(lagged5

yearsas

surrogates

ofsm

oking)

Altho

ugh

caseswere

identified

throug

hthe

DanishHospitalregistry,

onlythosewho

were

registered

forthefirsttime

withaprim

arydiagno

sis

ofPD

wereincluded.

Thiscriterionshou

ldhave

sign

ificantlyreduced

therisk

ofhigher

ratesof

vascularfactors(diabetes,

hypertension

,or

dyslipidem

ia)in

hospitalized

patients.

How

ever,B

MIwas

not

includedam

ongpo

tential

confou

ndersandadjusting

variables.In

sensitivity

analysisperformed

after

theexclusion

ofcasesand

controlsubjectsdiagnosed

withdementia

orcerebrovasculardisease

2yearsbefore

the

indexing,risk

estimates

werechangedon

lyminim

ally.F

inally,in

analyses

restricted

toPD

casesaged

.60

yearsat

diagnosis,d

iabeteswas

nolonger

associated

with

increasedrisk

ofPD

(OR=1.16

[0.85–

1.57

]).

*Dataarepresentedas

median[range]or

asmean(SD);**OR(95%

CI)fortheassociationbetw

eenPD

andprecedingdiabetes.



secondary causes of parkinsonism (inclu-sion bias) and/or adequate comparability(selection bias) between PD cases and con-trol subjects.

Data extraction and analysesTitles, abstracts, and articles were reviewedindependently by two authors, E.Ce. andR.C. Abstracted data included study pop-ulation characteristics, risk and precisionestimates, adjustment variables, and po-tential for bias. Fully adjusted estimateswere preferably included and analyzed.Results from sensitivity analyses, particu-larly those taking the potential confound-ing by concomitant vascular diseases intoaccount, were also collected. Any discrep-ancies in data extraction were resolved bydiscussion with a third reviewer (C.P.).Authors of investigations were also con-tacted for further information as necessary.

Data fromprospective cohort and case-control studies were analyzed separately.Potential sources of heterogeneity betweenstudies were assessed using Cochran’s Qand I2 statistics. To compensate for poten-tial between-study heterogeneity, wecalculated a pooled RR or OR using arandom-effect model (DerSimonian-Lairdmethod); otherwise, we used fixed effectmodels. Sensitivity analyses were also con-sidered, and specific criteria were providedalong with the results. Forrest plots wereused to summarize results, and funnel plotswere used to assess publication bias. Theprevalence of diabetes in case-controlstudies included/not included in thismeta-analysis was computed, togetherwith 95% CI, and compared by means ofa loglinear model. Frequency weights forstudy size were used.

All analyses were performed usingSTATA 11 (StataCorp, College Station,TX), establishing the level of significanceat a two-tailed P value of ,0.05.

RESULTSdThe literature search yielded1,947 citations (electronic databases, N =1,824; reviewing of references, N = 123),of which 48 were thoroughly assessed foreligibility. Of these, 11 (cohort,N = 4; case-control,N = 7) (5–8,12–18) were includedin the present systematic review of diabetesas a risk factor for PD. Descriptive data andmain results from these studies are presen-ted in Table 1 (cohort) and Table 2 (case-control). In particular, we excluded 13case-control studies, mainly because theauthors did not evaluate incident cases ofPD (Supplementary Data). In these studies(for available data), the mean prevalence of

diabetes in idiopathic PD was 12.4% (95%CI 12.2–12.7%; range, 6.5–16.2%).

Finally, out of the 11 studies initiallyincluded for qualitative assessment, 9(cohort, N = 4; case-control studies, N = 5)met the additional inclusion criteria formeta-analysis.

Prospective cohort studiesAmong the cohort studies reviewed (Table1), two addressed the risk associated withboth type 1 and type 2 diabetes (with type 2diabetes accounting for .90% of cases)(6,13) and two estimated the risk of type2 diabetes (5,12). Simon et al. found thatthe risk of PD was similar, independentlyof diabetes status (5), whereas the otherstudies reported a significant risk of PDfor participants with pre-existing diabetes(6,12,13). In all cohorts a positive historyof diabetes was based on self-reporteddoctor-diagnosed diabetes either aloneor confirmed through the use of nationalregisters.

In the primary analysis (Fig. 1; plot A),based on fully adjusted estimates, thepooled risk for diabetes was 1.37 [95%CI 1.21–1.55] (z = 5.01; P , 0.0001)with limited variation attributable to studyheterogeneity (I2 = 34.3%; P = 0.207).Because cerebrovascular disorders andcoronary heart disease were among the im-portant causes of secondary parkinsonism

(vascular) (19), a secondary analysis wascarried out, based on the estimates ob-tained through sensitivity analyses per-formed after the exclusion of participantswho had vascular disorders at baseline orwho had had a stroke during follow-up(Fig. 1; plot B). The association remainedsignificant also in this analysis (RR = 1.34[1.14–1.58]; z = 3.58, P , 0.001) with asimilar variation attributable to study het-erogeneity (I2 = 36.1%; P = 0.195). Theexclusion of the only study performed outof the U.S. area brought similar findings(RR = 1.28 [1.08–1.52]; z = 2.81, P =0.005; heterogeneity, I2 = 2.3%, P = 0.359).Finally, sensitivity analyses for the risk as-sociated with diabetes at baseline and dia-betes that developed during follow-upwere also considered. For baseline diabetesthe RR was 1.38 [1.14–1.68] (z = 3.32, P =0.001; heterogeneity, I2 = 28.1%, P =0.249). However, for diabetes duringfollow-up the relationship approachedsignificance (RR = 1.21 [0.94–1.56] (z =1.48, P = 0.139; heterogeneity, I2 = 41.5%,P = 0.191).

Case-control studiesAbout half of the case-control studies(Table 2) included in qualitative assess-ment consistently concluded that preced-ing diabetes was inversely associated withPD (8,16,17). In one study this association

Figure 1dMeta-analysis and pooled RRs of PD in diabetic participants from prospective cohortstudies (Forrest plot A [A], association found for fully adjusted estimates; Forrest plot B [B], asso-ciation found for fully adjusted estimates obtained after the exclusion of participants who had vas-cular disease at baseline and/or who developed them during the follow-up). d.f., degrees of freedom.



was marginally significant (15), whereasin another a trend toward a lower preva-lence of diabetes in PD patients was de-scribed (14). Only Becker et al. (7) andSchernhammer et al. (18) found that incases the prevalence of diabetes was simi-lar to or even higher than control subjects.It is noteworthy that these investigationswere also the ones with the largest popu-lations. The prevalence of diabetes in PDpopulations was mildly heterogeneous,ranging between 3.4 and 9.1% (weightedmean 9.5% [95% CI 9.0–9.9%]; for ex-cluded case-control studies, RR = 1.62[95% CI 1.54–1.70], P , 0.001). Ascer-tainment was frequently based on self-re-ported data (8,15,17). A distinctionbetween type 1 and type 2 diabetes wasmade only in one study (16), but it is likelythat most cases refer to the latter. With theexception of one case (14), adjusted ORswere available.

According to quality assessment pro-cedures, we excluded the data of twostudies (14,16) fromquantitative synthesis.In the study by Leibson et al. (14), thepresence of other neurologic diseases(e.g., stroke or dementia) was not anexclusion criterion. A trend toward ahigher prevalence of stroke and dementiawas present in the PD group, and it is

reasonable to think that secondary formsof parkinsonism (vascular) may havebeen included. Moreover, only crudeORs were provided. In the study byScigliano et al. (16), control subjects wererecruited in a hospital setting, and theparticularly high prevalence of vascularrisk factors (diabetes, hypertension, ordyslipidemia) was recognized as sourceof estimation bias. Therefore, the meta-analysis was based on five studies (Fig. 2).When compared with matched controlsubjects, PD patients at diagnosis werecharacterized by a similar prevalence of di-abetes: OR = 0.75 [95% CI 0.50–1.11] (z =0.21; P = 0.835) with a high variation inrisk attributable to study heterogeneity(I2= 82.3%, P , 0.001). The sensitivityanalysis including studies providing esti-mates adjusted for BMI, an importantconfounder of the effect of diabetes, con-firmed the lack of association with PD(OR = 0.56 [0.28–1.15]; z = 1.70; P =0.089; I2 = 80.5%, P = 0.006).

For both the study designs consid-ered it was not possible to calculate sex-specific estimates of risk.

CONCLUSIONSdAccording to thismeta-analysis diabetes appears to be a riskfactor for PD. Higher weight was given to

the evidence derived from prospective in-vestigations. In etiologic epidemiology theuse of cohort studies is preferable, becauseit enables researchers to formulate a cause-effect hypothesis. However, the incidenceof PD is low and shifted toward advancedage; large populations and a long follow-upare required for adequately powered riskanalyses and, up to now, only one studyhas investigated a large number of cases(13). Thus the evidence collected is stilllimited. In this perspective, case-controlstudies could gain value and, since theyhave the advantage of being less expensiveand time-consuming, enable the collectionof a larger mass of data. However, they suf-fer mainly from recall (risk factors and du-ration of exposure) and inclusion (for bothcase subjects and control subjects) bias andonly associations, not real causal relation-ships, can be inferred.

Heterogeneity among the studies re-trieved and limitations of the literaturesearch deserve comment and should betaken into account for a correct interpre-tation of the results. Although we usedstrict inclusion criteria and performedsensitivity analyses, several residuals ofconfounding and estimation bias couldbe present.

Diabetes is a metabolic disorder con-tributing to both large and small vesseldiseases (20), and we were not sure thatthe study populations were confined toidiopathic PD. Although vascular parkin-sonism possibly relating to major vasculardisorders was generally taken into ac-count during the recruitment phase orthrough adjusted analyses, the exclusionof parkinsonism related to microvascularcomplications should probably be basedalso on imaging techniques (21), evenwhen the diagnosis of PD is made by aneurologist or a specialist in movementdisorders. Nevertheless, sensitivity analy-ses confirmed the role of diabetes as a riskfactor for future PD.

The lack of an establishedmechanismlinking diabetes to PD is a major limita-tion for the interpretation of our findings.Recent studies have shown that, indepen-dently of hyperglycemia, which is mainlyresponsible for endothelial dysfunction(20), distinct pathologic pathways maycontribute to neurodegeneration. Regard-ing PD, this hypothesis appears to be sup-ported by the various associations foundbetween PD and diabetes duration. Bothinsulin deficiency and insulin resistancewith compensatory hyperinsulinemiamight play a role (22–24). Insulin can actas a growth factor in the brain and can

Figure 2dForrest plot (pooled OR) for the association between pre-existing diabetes and PD incase-control studies: plot A (A), primary analysis (all studies); plot B (B), sensitivity analysis(studies providing estimates adjusted for BMI). I-V, inverse variance fixed-effect model; D+L,DerSimonian-Laird random-effect model; d.f., degrees of freedom.



reduce oxidative stress. Insulin resistancehas been shown to result in decreased in-sulin transport into the brain (9,10). Insu-lin receptors are densely represented in thesubstantia nigra and insulin increases do-pamine transporter mRNA in the substan-tia nigra. It also regulates dopamineconcentrations in the brain (9,10). Duringthe natural history of diabetes, bothchanges in insulin secretion and in insulinactivity may occur and have never been in-vestigated in terms of the risk of PD. How-ever, both insulin resistance and PD maybe consequences of the aging process(1,2,25), and the clustering of PD and di-abetes with age observed by Driver et al.(12) could also suggest the existence ofcommon pathogenic processes leading firstto diabetes and then to PD. In the Physi-cians’ Health Study, the prevalence of dia-betes, despite being similar to that reportedby Simon et al. (5) at study entry (;2%),was more than twice as high at the end ofthe follow-up (11 vs. 4.5%). Similarly, alsothe age at PD diagnosis appeared to behigher as well as the prevalence of PD(2.5%; about eightfold higher than in theother cohorts included in themeta-analysis).However, the increased risk for those withdiabetes of shorter duration may also be ex-plained by detection bias as a result of in-creased medical surveillance (12). Recentlythis clustering with age was not confirmedby the other prospective investigation re-porting an increased risk of PD in diabeticsubjects (13). In the cohort followed by Xuet al. (13) only a history of diabetes longerthan 10 years was significantly associatedwith PD risk. However, only diabetes atbaseline was included in the analyses, anddifferences in monitoring of diabetic andnondiabetic patients during follow-up couldhave contributed to estimation bias (13). Itis noteworthy that the prevalence of PDat the end of follow-up was similar to thatreported by Simon et al. (0.3 vs. 0.5%) (5),whereas the prevalence of diabetes at base-line was about 3.5-fold higher than in theother cohorts (7.4 vs.;2%) (5,6,12).

Other pathogenic mechanisms sharedby diabetes and PD involve the inflam-matory and mitochondrial dysfunctionpathways. The contribution of chroniclow-grade inflammation to the develop-ment of insulin resistance is generallyaccepted (26). Similarly, recent researchhas indicated that neuroinflammationplays a role in the pathogenesis of PD(27,28). Furthermore, recent studies sup-port the hypothesis that the link betweenthe two diseases may be the bioenergeticsimilarity of the midbrain substantia nigra

and the pancreatic islet b-cells. In partic-ular, the higher vulnerability to oxidativestress and toxins may result in lower re-spiratory capacity (ATP production) andcellular vitality (9,10).

Looking at prospective studies otherlimitations should be taken into account.First, no information on glucose controland related management was consideredduring the follow-up. Data from the liter-ature suggest that insulin and insulin-sensitizers, such as thiazolidinediones,could be a viable therapeutic approach tothe treatment of neurodegenerative disor-ders, since they would also improve sys-temic low-grade inflammation (10).Second, most of the information collectedwas self-reported, and recall bias cannotnot be excluded with advancing age, par-ticularly in the presence of concomitantdeterioration in cognitive status. This ap-plies at least to diabetes since blood glu-cose was not measured in prospectivestudies. Third, changes in the initial diag-nosis of parkinsonism may also occur(29). Fourth, the potential confoundingeffect of other cardiovascular risk factorsfrequently clustered with diabetes withinthe spectrum of the metabolic syndromehas been considered only to some extent.Although their role still needs to be eluci-dated as a result of inconsistent evidence,some investigators have suggested that hy-pertension and high blood cholesterol areassociated with an increased risk of PD(30,31). Conversely, high serum uratehas been reported to be a protective factor(32,33) as well as the use of statins (34).Finally, among the several specific fooditems investigated (antioxidants, macro-nutrients, dairy products, etc.) (35), onlythe effect of caffeine and alcohol intakewas taken into account. In regard to this,it should also be mentioned that the con-sumption of some foods could alterna-tively increase the risk of PD and reducethat of diabetes, as is the case for dairyproducts (36,37).

From a diagnostic point of view case-control studies probably offer some ad-vantages, at least in terms of an accuratediagnosis of idiopathic PD. However, PDitself may influence the risk of diabetes,thus confounding the association. At leasttwo factors should be considered. First,PD is associated with generalized sympa-thetic denervation (38,39), a conditionthat appears to contribute to reducedmetabolic activity within adipose tissueand to an improved cardiometabolic pro-file (15,40). However, the timing of theseeffects is unclear, and only one study

among those reviewed provided enoughinformation on the length of diabetes his-tory (8). Second, although some studieshave reported that body weight does notchange before the onset of PD, a recentliterature overview reported that weightloss, above all in terms of adipose tissue,commonly occurs (3). This would be acontinuous process, starting several yearsbefore the diagnosis, to which both re-duced calorie intake and increased energyexpenditure, secondary to motor symp-toms, can contribute (3). Although BMIhas been considered among the potentialconfounders, this parameter, togetherwith weight loss, should probably betaken into account during control selec-tion and inclusion. Finally, survival biasas a result of high mortality among dia-betic patients could contribute to the in-verse relationship between diabetes andPD in case-control studies.

Conclusion and perspectivesLiterature analysis suggests that the etio-logic role of diabetes in the developmentof PD requires further prospective inves-tigation. In future cohort studies attentionshould be paid to putative pathogenicpathways, such as abnormal insulin pro-duction, insulin resistance, therapies forglucose control as well as inflammationand other cardiovascular risk factors thatfrequently cluster with diabetes withinthe spectrum of the metabolic syndrome.Nonetheless, case ascertainment shouldprobably include the use of neuro-imagingtechniques.

AcknowledgmentsdThis work was sup-ported by the “Fondazione Grigioni per ilMorbo di Parkinson.”No potential conflicts of interest relevant to

this article were reported.E.Ce. contributed to literature review, data

extraction, and manuscript drafting. E.Ce. hadfull access to all of the data in the study andtakes responsibility for the integrity of the dataand the accuracy of the data analysis. M.B.contributed to literature review and critique.C.P. contributed to literature review and dataextraction. C.K. contributed to data analysis.E.Ca. contributed to data extraction. R.C. con-tributed to literature review and critique. G.P.provided supervision and contributed to re-view and critique. All authors significantlycontributed to the work and approved themanuscript for submission.The authors thank Professor Tobias Kurth

andDr. JaneAnnDriver (Brigham andWomen’sHospital, Boston, MA) and Professor Gang Hu(Pennington Biomedical Research Center,Louisiana State University, Baton Rouge, LA)



for kindly providing additional data required tocomplete the current study database. The au-thors also thank Jennifer S. Hartwig, MD(Parkinson Institute, Istituti Clinici di Per-fezionamento, Milan, Italy) for assistance indrafting the manuscript.

References1. Schapira AHV. Science, medicine, and the

future: Parkinson’s disease. BMJ 1999;318:311–314

2. Lees AJ, Hardy J, Revesz T. Parkinson’s dis-ease. Lancet 2009;373:2055–2066

3. Barichella M, Cereda E, Pezzoli G. Majornutritional issues in the management ofParkinson’s disease. Mov Disord 2009;24:1881–1892

4. Elbaz A, Moisan F. Update in the epide-miology of Parkinson’s disease. Curr OpinNeurol 2008;21:454–460

5. Simon KC, Chen H, Schwarzschild M,Ascherio A. Hypertension, hypercholesterol-emia, diabetes, and risk of Parkinson disease.Neurology 2007;69:1688–1695

6. Hu G, Jousilahti P, Bidel S, Antikainen R,Tuomilehto J. Type 2 diabetes and therisk of Parkinson’s disease. Diabetes Care2007;30:842–847

7. Becker C, Brobert GP, Johansson S, Jick SS,Meier CR. Diabetes in patients with idio-pathic Parkinson’s disease. Diabetes Care2008;31:1808–1812

8. D’Amelio M, Ragonese P, Callari G, et al.Diabetes preceding Parkinson’s diseaseonset. A case-control study. ParkinsonismRelat Disord 2009;15:660–664

9. Craft S, Watson GS. Insulin and neurode-generative disease: shared and specific mech-anisms. Lancet Neurol 2004;3:169–178

10. Cardoso S, Correia S, Santos RX, et al.Insulin is a two-edged knife on the brain.J Alzheimers Dis 2009;18:483–507

11. Moher D, Liberati A, Tetzlaff J, AltmanDG; PRISMA Group. Preferred reportingitems for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med2009;6:e1000097

12. Driver JA, Smith A, Buring JE, GazianoJM, Kurth T, Logroscino G. Prospectivecohort study of type 2 diabetes and therisk of Parkinson’s disease. Diabetes Care2008;31:2003–2005

13. Xu Q, Park Y, Huang X, et al. Diabetes andrisk of Parkinson’s disease. Diabetes Care2011;34:910–915

14. Leibson CL, Maraganore DM, Bower JH,Ransom JE, O’brien PC, RoccaWA. Comor-bid conditions associated with Parkinson’sdisease: a population-based study. MovDisord 2006;21:446–455

15. Powers KM, Smith-Weller T, FranklinGM, Longstreth WT Jr, Swanson PD,Checkoway H. Diabetes, smoking, andother medical conditions in relation toParkinson’s disease risk. ParkinsonismRelat Disord 2006;12:185–189

16. SciglianoG,MusiccoM, Soliveri P, Piccolo I,Ronchetti G, Girotti F. Reduced risk factorsfor vascular disorders in Parkinson diseasepatients: a case-control study. Stroke 2006;37:1184–1188

17. Miyake Y, Tanaka K, FukushimaW, et al.;Fukuoka Kinki Parkinson’s Disease StudyGroup. Case-control study of risk ofParkinson’s disease in relation to hyperten-sion, hypercholesterolemia, and diabetesin Japan. J Neurol Sci 2010;293:82–86

18. Schernhammer E, Hansen J, Rugbjerg K,Wermuth L, Ritz B. Diabetes and therisk of developing Parkinson’s disease inDenmark. Diabetes Care 2011;34:1102–1108

19. Thanvi B, Lo N, Robinson T. Vascularparkinsonismdan important cause ofparkinsonism in older people. Age Ageing2005;34:114–119

20. Haratz S, Tanne D. Diabetes, hyperglyce-mia and the management of cerebrovas-cular disease. Curr Opin Neurol 2011;24:81–88

21. Kalra S, Grosset DG, Benamer HT. Dif-ferentiating vascular parkinsonism fromidiopathic Parkinson’s disease: a system-atic review. Mov Disord 2010;25:149–156

22. Peila R, Rodriguez BL,White LR, Launer LJ.Fasting insulin and incident dementia in anelderly population of Japanese-Americanmen. Neurology 2004;63:228–233

23. Rönnemaa E, Zethelius B, Sundelöf J, et al.Impaired insulin secretion increases therisk of Alzheimer disease. Neurology 2008;71:1065–1071

24. Rönnemaa E, Zethelius B, Sundelöf J,et al. Glucose metabolism and the riskof Alzheimer’s disease and dementia: apopulation-based 12 year follow-up studyin 71-year-old men. Diabetologia 2009;52:1504–1510

25. Karakelides H, Irving BA, Short KR,O’Brien P, Nair KS. Age, obesity, and sexeffects on insulin sensitivity and skeletalmuscle mitochondrial function. Diabetes2010;59:89–97

26. Xu H, Barnes GT, Yang Q, et al. Chronicinflammation in fat plays a crucial role in thedevelopment of obesity-related insulin re-sistance. J Clin Invest 2003;112:1821–1830

27. Chen H, O’Reilly EJ, Schwarzschild MA,Ascherio A. Peripheral inflammatory bio-markers and risk of Parkinson’s disease.Am J Epidemiol 2008;167:90–95

28. Song IU,Kim JS,ChungSW,LeeKS. Is therean association between the level of high-sensitivity C-reactive protein and idio-pathic Parkinson’s disease? A comparisonof Parkinson’s disease patients, diseasecontrols and healthy individuals. EurNeurol2009;62:99–104

29. Caslake R, Moore JN, Gordon JC, HarrisCE, Counsell C. Changes in diagnosiswith follow-up in an incident cohortof patients with parkinsonism. J NeurolNeurosurg Psychiatry 2008;79:1202–1207

30. Qiu C, Hu G, Kivipelto M, et al. Associa-tion of blood pressure and hypertensionwith the risk of Parkinson disease: theNational FINRISK Study. Hypertension2011;57:1094–1100

31. Hu G, Antikainen R, Jousilahti P, KivipeltoM, Tuomilehto J. Total cholesterol and therisk of Parkinson disease. Neurology 2008;70:1972–1979

32. Chen H, Mosley TH, Alonso A, Huang X.Plasma urate and Parkinson’s disease inthe Atherosclerosis Risk in Communities(ARIC) study. Am J Epidemiol 2009;169:1064–1069

33. Weisskopf MG, O’Reilly E, Chen H,Schwarzschild MA, Ascherio A. Plasmaurate and risk of Parkinson’s disease. AmJ Epidemiol 2007;166:561–567

34. Dolga AM, Culmsee C, de Lau L, et al.Statinsdincreasing or reducing the riskof Parkinson’s disease? Exp Neurol 2011;228:1–4

35. Wirdefeldt K, Adami HO, Cole P,Trichopoulos D, Mandel J. Epidemiologyand etiology of Parkinson’s disease: a re-view of the evidence. Eur J Epidemiol2011;26(Suppl. 1):S1–S58

36. ChenH,O’Reilly E, McCulloughML, et al.Consumption of dairy products and riskof Parkinson’s disease. Am J Epidemiol2007;165:998–1006

37. Tong X, Dong JY, Wu ZW, Li W, Qin LQ.Dairy consumption and risk of type 2diabetesmellitus: ameta-analysis of cohortstudies. Eur J Clin Nutr 2011;65:1027–1031

38. Carmichael SW,Wilson RJ, BrimijoinWS,et al. Decreased catecholamines in theadrenal medulla of patients with parkin-sonism. N Engl J Med 1988;318:254

39. Wakabayashi K, Takahashi H. Neuropa-thology of autonomic nervous system inParkinson’s disease. Eur Neurol 1997;38(Suppl. 2):2–7

40. Cereda E, Cassani E, Barichella M,Caccialanza R, Pezzoli G. Anthropometricindices of fat distribution and cardio-metabolic risk in Parkinson’s disease. NutrMetab Cardiovasc Dis. In press



diabetes and risk of parkinson disease · 2011-12-28 · diabetes and risk of parkinson’s disease...

Documents