harnessing evidence to prevent breast cancer now

Harnessing Evidence to Prevent Breast Cancer Now

Graham A. Colditz, MD, DrPH

Niess-Gain ProfessorDepartment of Surgery

Division of Public Health Sciences

Department of SurgeryDivision of Public Health Sciences

Disclosure Information

I have no financial relationships to disclose

I will not discuss off-label use or investigational useof drugs or devices in my presentation

Slides available online slideshare.net/grahamcolditz


Overview

I. Burden – growing & global

II. Barriers to breast cancer prevention

III.Drivers of breast cancer

IV. Focusing breast cancer prevention

V. Steps we can take now

2012 – 14 M cases

1.7 M breast


Global burden, 2012

14 million new cases of cancer diagnosed

7.4M men; 6.6M women

1.7 Million new cases of breast cancer

25% of all cancer diagnosed in women

Globocan 2012, IARC


Success!

Breast Cancer Mortality rate has decreased30% decrease.

Age adjusted breast cancer mortality declined by 30% from 1994 to 2010

31 deaths per 100,000 women in 1994 to

22 per 100,000 women in 2010

1.9% per year decrease

Edwards BK, et al Cancer 2014

Breast Cancer Prevention

Cancer burden

Barriers

Drivers

Focusing prevention

Steps we can take now


US Population Structurechanging from2010 to2030 to2050


I. Burden of cancer – US costs

Total direct cost – billions ($US 125b in 2010)

• Breast cancer: $US 16.6 b, or 13% in 2010

Increasing burden also of indirect costs of cancer

• Lost productivity due to breast cancer $US 10.9b

Hassett & Elkin, 2013, and Cancer Australia, 2014

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


II. Barriers to breast cancer prevention

1. Skepticism that cancer can be prevented

2. Short-term focus of research

3. Timing: Interventions too late in life

4. Research focused on treatment not prevention

5. Debates among scientists

6. Societal factors ignored

7. Lack of transdisciplinary training

8. Complexity of implementationColditz et al. Sci Transl Med 2012: March 28

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

http://tinyurl.com/l4a4g2j

http://tinyurl.com/l4a4g2j



Cancer burden

Barriers

1. Skepticism

Drivers

Focusing prevention


Generally accepted breast ca. prevention strategies

Strategy Risk group % US pop Risk

reduction

Bilateral

oophorectomy

BRCA1/2 <1% 50%

Tamoxifen /

Raloxifene

>1.67%

5-yr risk

10-40% 50%

Weight loss (22lb) Overweight +

obese

60% 50%*

Stopping estrogen

& progestin Rx

Past vs. current 1-5% 10%

* Loss after menopause based on Eliassen et al. JAMA, 2006; Colditz & Bohlke 2014


Further evidence that breast cancer is preventable

Migrant studies

No US lifestyle,

Substantial changes in risk over generations, but disentangling the complexity of lifestyle changes makes interpretation difficult

Within-country changes

Remove E&P HT, Korea rapid increase, China, etc.

Cancer burden

Barriers:

1. Skepticism

Drivers

Focusing prevention


Breast Cancer

Prevention



Cancer burden

Barriers

1. Skepticism

Drivers

Focusing prevention


Within-country change:Menarche in Korea, 1920-85

Cho Eur J Pediatr 2009

30 years



Cancer burden

Barriers

1. Skepticism

Drivers

Focusing prevention


Within-country change: Fertility, 1960-2004

Calendar year Ito et al. NEBR, 2008

• .



Cancer burden

Barriers

1. Skepticism

Drivers

Focusing prevention


Age at first birth

www.oecd.org/social/family/database



Cancer burden

Barriers

1. Skepticism

Drivers

Focusing prevention


Within-country change: Breast cancer incidence, Korea

1998 40, born 19582008 40, born 1968

Jung et al. J Breast Ca, 2011


Lei Fan , et al. Breast cancer in China The Lancet Oncology, Volume 15, Issue 7, 2014, e279 - e289


Cancer burden

Barriers

1. Skepticism

Drivers

Implications for prevention


Within-country change: Incidence

of breast cancer, China, 2009.

Urban

Rural



Cancer burden

Barriers

1. Skepticism

2. Short-term

Drivers

Focusing prevention


Barrier 2: Short-term focus

Colditz et al. Sci Transl Med 2012: March 28

• Time required for disease prevention does not match funding periods

• Long-term benefits, e.g., smoking cessation takes decades to show at population level

• Funded studies focus too late in disease development process

• In contrast, the natural history or time-course of cancer shows development over decades


Life-course development of risk accumulation

• Long-term effects of early radiation

• Menopause – cessation of menstrual cycles leads to slower risk accumulation

• Menopause tells us hormones or accumulation through premenopausal years must be important

• Menarche and height point to early life as important


Cancer burden

Barriers

1. Skepticism

2. Short-term

3. Timing

Drivers

Focusing prevention



Model of breast cancer developmentWellings-Jensen Model (JNCI 55:231, 1975)

Adapted from Allred

Time (decades)

TDLU

ADH DCIS IBC

Growth

CCH

s Adhesion& Polarity

Diversity Invasion

Gertig et al 10+ years



Cancer burden

Barriers

1. Skepticism

2. Short-term

3. Timing

Drivers

Focusing prevention


Long-term effects of radiation in early life

Land et al. Radiation Research 2003

Atomic bomb survivors, 70,16540 year follow-up1059 cases


Moolgavkar et al JNCI 1979


Cancer burden

Barriers

1. Skepticism

2. Short-term

3. Timing

Drivers

Focusing prevention


24% of all breast cancer diagnosed before age 50

Early life clues


III. Drivers of breast cancer• Progression through premalignant

lesions as risk accumulates

• Faster rise in risk before menopause than after

• Hormonal cycling during premenopausal years

• Peak height growth velocity

• Diet and physical activity

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

Colditz & Bohlke npj Breast Cancer 2015


Using attained age as amarker of risk is a poor proxyAttained age as marker of risk masks the

underlying risk accumulation

• Accumulated exposure up to an age

• DNA damage

• Some other function of age, e.g., time between menarche and first birth

• One attained year of age may not be the same at age 10 as at age 60

• Look within age across life course to better understand drivers

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


Pike model – Nature 1983

To accommodate the higher incidence with late first birth, we add a constant representing an increase in risk with FFTP

(+b) in figurePike, et al. Nature 1983


Accumulating risk, multiple birth model

Rosner, Colditz, Willett, Am J Epidemiology 1994;139:826

Ra

te o

f t

issu

e

ag

ing

menarche birth birth birth menopause


Accumulating risk, multiple birth model

Rosner, Colditz, Willett Am J Epidemiology 1994;139:826

9% /yr

2.5% /yr

Menarche 1st birth Menopause Current age

Rate

of

tis

sue a

gin

g


Understanding growth velocity (Stuart study)Harvard Growth Study (HGS)

Females born in 1930s and 1940s, followed to age 18

Age at menarche recorded to month

Annual height measurements (identify year in which girl experienced most rapid adolescent height growth)

Mother interviewed annually on dietary intake over past 6 months while child being examined/measured, etc.

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


Growth curve of one girl, HGSPeak height velocity = amount of greatest height gain in a single year


Cancer burden

Barriers

Drivers

Focusing prevention


Berkey et al. Cancer 1999


Peak height velocityBreast Cancer

Prevention

Cancer burden

Barriers

Drivers

Focusing prevention


Berkey et al. Cancer 1999


Predictors measured from birth through age 5, HGS

Age at menarche =

12.8 (0.12) – 0.38 (0.12)height at age 3 to 5yr

+ 2.19 (0.91) vegetable protein ages 3 to 5 yr.

Peak height growth velocity =

14.2 + 4.25 (1.07) calories – 0.39 BMI ages 3 to 5yr + 2.08 (0.95) animal protein ages 3 to 5yr

Results consistent when repeated for exposures at age 10

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

Berkey, …, Colditz AJE 2000; 152:446-52


Applying peak height growth velocity: Nurses’ Health Study and adolescent cohort (GUTS)Higher peak height growth velocity (PHGV) associated with increased risk of pre and post menopausal breast cancer

• Highest vs. lowest quintile of PHGV; 8.9cm/yr vs. <7.6 cm/yr;

RR=1.31 premenopausal breast cancer

RR=1.40 postmenopausal breast cancer

For Benign Breast Disease same range in PHGV gave RR = 2.10

Cancer burden

Barriers

Drivers

Focusing prevention


Berkey et al. Cancer 1999 & 2011

Breast Cancer

Prevention


Danish Cohort confirms growth velocity finding• 141,393 girls, measured height annually in

school

• Linked to cancer registry

• 3340 cases of breast cancer

• Height increase age 8 to 14 positively related to risk of breast cancer before age 50 and after RR per 5 cm = 1.15 (1.05, 1.27): age <50 yr.

RR per 5 cm = 1.18 (1.07, 1.30): age 50+ yr.

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

Ahlgren et al. NEJM 2004; 351:1619-26


Childhood adiposityEarlier menarche & Lower growth height velocity

• Significant inverse relation with all molecular subtypes of breast cancer/ premenopausal and postmenopausal breast cancer Baer et al Tamimi et al

• Adiposity at age 7 associated with lower odds of dense breast. Anderson et al Breast Ca Res 2014


Cancer burden

Barriers

Drivers

Implications for prevention


Effects of obesity on breast cancer

• Obesity is well-established as a risk factor for post-menopausal breast cancer.

• Obesity is protective for pre-menopausal breast cancer

37

Cross-TREC Project CA155626, U54 CA155435, U54 CA155496, and PO1 CA 087969)

See Rosner,…., Colditz Br Ca Res Treat 2015 and forthcoming

Short-term effects of weight gain on breast cancer incidence

• Short-term effects of weight change on breast cancer risk are largely unknown

• We investigated short-term effects of obesity on breast cancer using NHS data

38

Study Population – Nurses’ Health Study

39

• 77, 232 women followed over 1,445,578 person-years from 1980-2008

• 4,196 incident cases of invasive breast cancer

Rosner et al Breast Cancer Res Treat 2015

Analytical approach

• Log incidence model of breast cancer (Colditz, et al AJE 2000) used to predict incident breast cancer as a function of

1) Average BMI before menopause

2) Average BMI after menopause

3) Other breast cancer risk factors

4) Short-term weight gain ≡ weight gain over previous 2 cycles (≈ 4 years) added as an additional risk factor

------------------------------------------------------------------------------0 10 18 40 44

Long term shortincidence

42Birth

Age, years

4-year weight change, breast ca. incidence

41

GroupNumber of

CasesLoss of > 5lbs

No change

(ref)++

Gain of 5.1-

9.9lbs

Gain of 10.0-

14.9lbs

Gain of ≥

15.0lbsRR** P_trend

overall 41960.97*

(0.89-1.07)1.0

1.05

(0.94-1.16)

1.09

(0.98-1.20)

1.20

(1.09-1.33)

1.13

(1.06-1.21)<0.001

Pre-

menopausal+736

1.02

(0.79-1.32)1.0

0.98

(0.77-1.25)

1.10

(0.88-1.38)

1.38

(1.13-1.69)

1.26

(1.08-1.48)0.004

Post-

menopausal+3443

0.98

(0.88-1.08)1.0

1.05

(0.93-1.19)

1.06

(0.94-1.20)

1.10

(0.97-1.25)

1.08

(1.00-1.16)0.063

*HR (95% CI)

** per 25 lbs.

+ at both time points

++ weight change of ≤ 5lbs.Rosner et al Breast Cancer Res Treat 2015

Overall there is a significant association of short-term weight gain with breast cancer incidence, particularly during pre-menopause.

42

Weight gain and risk according to starting BMI

Rosner et al Breast Cancer Res Treat 2015

Associations of short-term weight gain by tumor subtype, pre-menopausal women

43

GroupNumber

of casesLoss of > 5lbs

No

change

gain of 5.1-

9.9lbs

gain of 10.0-

14.9lbs

gain of ≥

15.0lbsRR ** P_trend P_het+

ER+/PR+ 3160.94*

(0.63-1.40)1.0

0.92

(0.63-1.34)

1.16

(0.83-1.61)

1.27

(0.93-1.73)

1.13

(0.89-1.42)0.32 ---

ER+/PR- 421.34

(0.50-3.58)1.0

1.11

(0.41-2.95)

0.60

(0.18-1.99)

1.77

(0.80-3.89)

2.19

(1.33-3.61)0.002 0.019

ER-/PR- 1001.07

(0.50-2.29)1.0

1.13

(0.57-2.26)

2.10

(1.23-3.56)

2.06

(1.21-3.51)

1.61

(1.09-2.38)0.016 0.12

*HR (95% CI)

**per 25lbs+vs ER+/PR+

Associations of pre-menopausal weight gain are stronger for ER+/PR- and ER-/PR- breast cancer, although case counts are small

Long-term weight change (since age 18) and breast cancer incidence

44

Group overall RR* P-value

Number of cases 4182

Weight at age 180.89

(0.85-0.94)<0.001

Long-term weight change†

Pre-menopause+1.09

(1.06-1.13)<0.001

Post-menopause++1.15

(1.09-1.21)<0.001

*per 25 lbs+ pre-menopausal weight change ≡ weight at min (age, age at menopause) – weight at age 18++ post-menopausal weight change ≡ weight at current age – weight at menopause if post-menopausal,= 0 if pre-menopausal† controlling for menopausal statusThere are protective effects of weight at age 18 and positive effects of long-term

weight change both pre- and post-menopause

Colditz et al Cancer Res 2016;76(4 Suppl):Abstract nr P1-07-05.

Long-term weight change (since age 18) and breast cancer incidence

45

Group ER+/PR+ P-value ER+/PR- P-value ER-/PR- P-value

Number of cases 2027 512 594

Weight at age 18

RR*0.98

(0.91-1.04)0.48

0.80(0.70-0.93)

0.0030.79

(0.69-0.90)<0.001


Pre-menopause+ RR1.14

(1.07-1.20)<0.001

0.90(0.81-1.00)

0.0471.07

(0.98-1.17)0.13

Post-menopause++ RR1.17

(1.08-1.25)<0.001

1.20(1.04-1.39)

0.0151.04

(0.90-1.21)0.56

*per 25 lbs+ pre-menopausal weight change ≡ weight at min (age, age at menopause) – weight at age 18++ post-menopausal weight change ≡ weight at current age – weight at menopause, if post-menopausal= 0 if pre-menopausal† controlling for menopausal status

Inverse association of weight at age 18 only present for ER+/PR- and ER-/PR- breast cancer.

Long-term weight gain is related to risk of breast cancer both pre- and post-menopause for ER+/PR+

breast cancer and for post-menopause for ER+/PR- breast cancer.

Long-term weight change (since age 10), breast ca incid.

46

Group overall RR per 25 lb P-value

Number of cases 3602

Weight at age 10⁰0.80

(0.74-0.86)<0.001


Age 10 to 180.97

(0.92-1.03)0.30

Pre-menopause+1.09

(1.06-1.13)<0.001

Post-menopause++1.15

(1.08-1.20)<0.001

⁰ estimated from weight somatotypes from the GUTS Study+ weight at min (age, age at menopause) – weight at age 18++ weight at current age – weight at menopause if post-menopausal, = 0 if pre-menopausal† controlling for menopausal status

.The protective effect of weight at age 18 seems to be derived from a protective effect at age 10

with no effect of weight change between age 10 to age 18.

Colditz et al Cancer Res 2016;76(4 Suppl):Abstract nr P1-07-05.

Summary – short-term weight gain

47

• Overall, there is a significant positive association of short-term weight gain on breast cancer incidence

• Associations are stronger among pre-menopausal women and women with BMI < 25

• Results are consistent with EPIC /PANACEA study (Emaus, et al., Int. J. Cancer 2014: 10.1002/ijc28926)

Summary – long-term weight change

48

• Overall, there is a protective effect of weight at age 10 on breast cancer incidence, which is present for all breast cancer subtypes.

• Premenopausal weight change since age 18 positively related to breast cancer incidence for ER+/PR+ tumors.

• Postmenopausal weight change related to breast cancer incidence for both ER+/PR+ and ER+/PR- breast cancer.


Model of breast cancer developmentWellings-Jensen Model (JNCI 55:231, 1975)

Adapted from Allred

Time (decades)

TDLU

ADH DCIS IBC

Growth

CCH

s Adhesion& Polarity

DiversityInvasion


Intermediate markers: benign breast disease (BBD)

RR = 1.8 = 3 to 5

London et al. JAMA et al 1989


IV. Focusing breast cancer preventionUnderstanding what predicts incidence of benign

breast disease

NHSII – incident BBD (RO1-CA50385)

• Central pathology review

• Components of adolescent lifestyle:

Diet including alcohol

Physical activity

GUTS, Growing Up Today Study

Prospective data collected

Self-report benign breast disease confirmed by breast biopsy

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

A known breast carcinogen

IARC 2007


Alcohol intake, ages 18-22, incident proliferative BBD

Alcohol intake(grams/day)

Cases

(678)

Person-year

RR (95% CI)

None 155 64,827 1.0 reference

0.1-4.9 193 78,365 1.11 (0.89, 1.38)

5.0-14.9 236 88,310 1.36 (1.09, 1.69)

>15 30 9519 1.35 (1.01, 1.81)

p, trend <0.01


Cancer burden

Barriers

Drivers

Focusing prevention


Liu et al. – Pediatrics, 2012


Alcohol before first pregnancy, NHSII

Liu, Colditz, Tamimi JNCI 2013


Cancer burden

Barriers

Drivers

Focusing prevention


Surrogates – soy, fiber, fruit and vegetables


Soy Asia vs. USABreast Cancer

Prevention

Cancer burden

Barriers

Drivers

Focusing prevention


The highest soy intake in the United States (measured by isoflavone intake) is well below the lowest intake in Asia

Wu et al. 2008


Adolescent fiber & proliferative

BBD: NHSIIBreast Cancer

Prevention

Cancer burden

Barriers

Drivers

Focusing prevention


Su et al. Cancer Causes Control 2010


BBD, GUTS cohort112 women confirmed biopsy for benign breast lesion

6950 free from biopsy

FFQ at entry 1996, then 1997, and 1998

• Peanut consumption (peanut butter and bags of peanuts) significantly inverse

• Total servings of vegetable per day inversely related to risk of BBD

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


Cumulative dietary intake 1996 to 1998 and risk of BBDDietary intake Mean

intakeOR 95%

Confidence Interval

Vegetable protein (10gm/d) 24 g 0.86 0.55-1.34

Vegetable fat (10gm/d) 33 g 0.72 0.52-0.98

Peanut butter and bags of peanuts (servings/3d)

0.52 0.56 0.35-0.87

Beans, lentils, soybeans (servings/3d) 0.24 0.95 0.55-1.62

Corn (servings/3d) 0.40 0.73 0.37-1.43

Total servings per day of peanut butter, bags of nuts, beans, lentils, soybeans, and corn

0.38 0.33 0.13-0.82


Cancer burden

Barriers

Drivers

Focusing prevention


Multivariable adjusted models include childhood adiposity, age at menarche, adolescent alcohol intake, and pregnancy (ever). Berkey et al. Breast Ca Res Treat, 2013


Ontario, Canada: Population-based case-control study

Recall of adolescent diet

(55 food items)

High participation

(2865 cases, 3299 controls)

Top vs. bottom quintile of intake

Fiber mvOR = 0.66 (0.55 - 0.78)

Vegetable protein mvOR = 0.80 (0.68 – 0.95)

Nuts mvOR = 0.76 (0.61 - 0.95)

Cancer burden

Barriers

Drivers

Focusing prevention


Liu, Y., Colditz, et al. Breast Cancer Res Treat 2014.

Breast Cancer

Prevention


0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Relative Risk (95% CI)

Lutein/zeaxanthin

Total carotenoids

Lycopene

Carotenoids Breast Ca. & BBD

• 3055 prospective cases breast and controls with bloods stored

• Total carotenoids: 19% lower risk

GUTS Prospective diet data (Boeke Peds 2014)

• Beta-carotene, Alpha-carotene, Lutein all inversely related to risk of incident BBD

Eliassen, et al. J Natl Cancer Inst 2012


Food consumption trends in China (ages 2-18), 1991 to 2011

178.6

87.6

59.5

236.4

41.859.5

266.2

25.446.1

0

50

100

150

200

250

300

Total animal-sourcefoods

Coarse grains Legumes andproducts

Kcal/

day

1991

2000

2011

F. Y. Zhai et al., Dynamics of the Chinese diet and the role of urbanicity, 1991-2011. Obes Rev 15 Suppl 1, 16 (Jan, 2014).


Physical activityBreast Cancer Prevention

Cancer burden

Barriers

Drivers

Focusing prevention


Total activity (MET-h/wk) during different ages and breast cancer

Maruti et al. JNCI 2008 100:728-737



Cancer burden

Barriers

Drivers

Focusing prevention


Summary of evidence:Adolescent exposures -- BBD

Lifestyle Relative Risk BBD

Alcohol

Peak Growth Velocity

height

Nuts

Fiber

Carotenoids

Vegetable protein

Family history

Physical activity


V. Steps we can take now to prevent breast cancer

Target prevention early in life

• Eat mostly a plant-based diet

• Limit alcohol before first pregnancy

• Increase and maintain physical activity

Work globally and locally

Refine messaging and social strategy


Cancer burden

Barriers

Drivers

Focusing prevention



-13%Regular physical activity

-25%

Having children (4)

-18%Avoiding alcohol

-16%Breast feeding

No children

Breast Cancer Risk Reduction

Starting in Early Life

Ris

k

Accum

ula

tion

Births

(20, 23, 26, 29)

Menopause

Age

10 20 30 40 50 60 70

Colditz and Bohlke 2014


No post-menopausal hormones

Physical activity No alcohol

Medications to lower risk

(for high risk women)

Healthy weight

Average woman – overweight

Obesity

Breast Cancer Risk Reduction

Starting in Mid Life

Ris

k

Accum

ula

tion

Births

(20, 23, 26, 29)

Menopause

Age

10 20 30 40 50 60 70

Colditz & Bohlke 2014


A global prevention imperativeTiming matters

• To maximize benefits we must focus on biologically relevant periods

• Identify lifestyle factors that limit the impact of drivers

• Tap potential benefit from childhood and adolescent plant diet and physical activity

What intermediate marker can we measure?

Must identify strategies to counter adverse effect of alcohol

Cancer burden

Barriers

What drives breast cancer?

Focusing prevention

Step to take now

Breast Cancer

Prevention


Implement prevention

Develop and refine strategies for delivery of prevention through:

Health care providers

Regulatory approaches

Families and communities

Refine prevention messages and our societal response to improve childhood and adolescent environment

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


Messages for 16 to 30 year old women and their families and communities1) Go big with plant based foods – fruits,

vegetables, nuts, and whole grains

2) Think before you drink

3) Put on those Dancing – and Walking and Running and Cycling shoes

4) Don’t obsess – but watch your weight

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention


Thank you

• Bernie Rosner & Cathy Berkey (statisticians)

• Stu Schnitt, Laura Collins, Jim Connolly, Craig Allred (pathologists)

• NHS/WUSTL/Cancer Care Ontario investigators and trainees and participants

• American Cancer Society Clinical Research Professorship

• NCI & Breast Cancer Research Foundation for funding

Cancer burden

Barriers

Drivers

Focusing prevention


Breast Cancer

Prevention

harnessing evidence to prevent breast cancer now

Health & Medicine