Breast cancer prevention Graham Colditz, MD, DrPH Niess-Gain Professor, Dept. of Surgery Washington University School of Medicine, ACS Clinical Research.

Breast cancer prevention

Graham Colditz, MD, DrPHNiess-Gain Professor, Dept. of Surgery

Washington University School of Medicine,ACS Clinical Research Professor, and

Associate Director, Prevention and Control

Long history of studying causesLong history of studying causes

1850’s family history1920’s reproductive risk factors

– Lane-Claypon, 1926 case-control study1950’s menopause1970 – onwards oral contracpetives,

postmenopausal hormones, diet, physical activity, obesity, endogenous hormones, SERMs

Do we know causes of breast Do we know causes of breast cancer?cancer?

How do we frame this question?Individual cases?At the population level?Does epidemiology guide prevention for

individual women or inform population strategies?

PreventionPrevention

Prevention today refers mainly to lowering the risk of disease.

Risk of most chronic diseases can't be totally eliminated, it can still be significantly reduced.

If everyone in the US led a healthy lifestyle, 80% of the cases of heart disease and diabetes could be avoided, as could 70% the cases of stroke and over 50% of cancer.

RiskRisk

Risk is a person's chance of getting a disease over a certain period of time.

There are many different ways to present risk.

Can we prevent breast CA?Can we prevent breast CA?

YES

International variationMigrationCommon claim we do not know causes

“ much of breast cancer epidemiology is not explained by known risk factors”

Breast Cancer Average Annual Breast Cancer Average Annual Incidence per 100,000 by age, 1982Incidence per 100,000 by age, 1982

1

10

100

1000

San Francisco

Japan

China

Age (years/female)

Seow A, et al Int J Epi 1996

1930 1990

Singapore breast cancer incidence by age and birth cohort

Seow 1996

1908

1948

GoalsGoalsReview risk factors in context of natural

history/biology of the breast,

- focus on reproductive factors

- contribution of postmenopausal hormones Potential for prevention

- SERMS (Selective Estrogen Receptor Modulators)

- diet, activity, weight loss (or control), breast feeding

Risk factorsRisk factors

Age Gender Family history Benign breast

disease Reproductive factors Endogenous

hormones Exogenous hormones

Adiposity Diet Physical activity Alcohol Radiation

Models of disease incidenceModels of disease incidence

Can summarize risk factors and take account of temporal relations between risk factors and disease

Temporal relations often ignored in standard risk estimation and interpretation

Offers one approach to summarizing a range of etiologic pathways

- predict population or individual risk

Pike modelPike model

Factors associated with reduced risk of breast cancer were considered to lower the rate of breast tissue aging– Pike et. al., Nature 1983;303:767-70

We translated this to mean the rate of cell division and accumulation of molecular damage on the pathway to breast cancer

One Birth ModelOne Birth ModelR

ate

of t

issu

e ag

ing

Rat

e of

tis

sue

agin

g

Rosner, Colditz, Willett, Am J Epidemiology 1994;139:824

Menarche First birth Menopause Age

Extensions to modelingExtensions to modeling

Includes time from birth to menarche Allows the impact to the first birth to vary with

age at first birth Fits log incidence (Poisson regression) model

giving terms that are interpretable Contrast contribution of risk factors for

receptor positive and negative breast cancer

Multiple Birth ModelMultiple Birth Model

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

Rat

e of

tis

sue

agin

g R

ate

of t

issu

e ag

ing

Application of models to NHSApplication of models to NHS

Observed that spacing of births was significantly related to reduced risk of breast cancer – the closer the births the lower the subsequent risk

A transient increase in risk was observed with first birth, but not subsequent births

Risk prediction and stratification now more accurate than Gail and other models

Pregnancy history

050

100150200250300350400450

Age

Inci

den

ce/1

00,0

00

Age birth35Nulliparous

Age birth20,23,26,29

Colditz and Rosner, Am J Epidemiology 2000;152:950-64

16%

27%

Age at menarcheAge at menarche

Later age - lower riskAge 15 vs age 11 gives 30% lower risk

to age 70Lack of physical activity associated with

earlier menarcheDiet may play a role as might fewer

childhood infections

Norway

Finland

Sweden

Impact of Menarche on Hormone Impact of Menarche on Hormone levelslevels

Singapore dataBreast cancer rates doubled144 post menopausal womenLate menarche (after 17) 24% lower

estradiol (circulating female hormone) than women with menarche before 17

Wu et al CEBP 2002

Age at menopause

0

50

100

150

200

250

300

350

400

30 35 40 45 50 55 60 65 70

Age

Inci

den

ce/1

00,0

00

Menopauseage 45

Menopauseage 55

Colditz and Rosner, Am J Epidemiology 2000;152:950-64

44%

MenopauseMenopause

Early menopause reduces riskHigh circulating hormones levels after

menopause increase risk, as does use of postmenopausal hormones

Anti-estrogens may have a role– who is target population– how are they identified, counseled, etc.– balance risks vs. benefits

Hormonal exposure after Hormonal exposure after menopausemenopause

Obesity is related to poor survivalTamoxifen reduces mortality among

women with breast cancerTamoxifen and Raloxifene reduce risk

of breast cancer in randomized controlled trials of breast cancer prevention

Risk of breast cancer by plasma estradiol levels: By tumor receptor status

0

1

2

3

4

5

6

1 2 3 4 1 2 3 4 1 2 3 4

ER+/PR+ ER-/PR- ER+/PR-

P for heterogeneity = < 0.001

Missmer et al, 2004 (case n = 152 ER+/PR+, 38 ER-/PR+, 33 ER-/PR-)

Body Mass Index and estrone Body Mass Index and estrone sulfatesulfate

0

50

100

150

200

250

<21 21-22.9 23-24.9 25-28.9 30+

Estrone sulfate

Hankinson et a, JNCI 1995;87:1297-1302l

Weight and weight gainWeight and weight gain

Adult weight gain increases risk of breast cancer

Relation seen most clearly among postmenopausal women who never have used hormones

20 kg gain from age 18 associated with doubling in risk of breast cancer vs. stable weight

Eliassen JAMA 2006

Schairer et alSchairer et al

BCDDP cohort followed 46,355 postmenopausal women

2082 cases of breast cancerRelative risk increased 0.01 (0.0002-

0.03) per year of use for estrogen aloneRR increased 0.08 (0.02-0.16) for E & PIncrease in RR stronger among women

with BMI < 24.4 kg/m2

JAMA 2000

Ross et al.Ross et al.

1879 postmenopausal cases and 1637 controls in LA county

Estrogen alone associated with RR 1.06 (0.97-1.15) for 5 years of use

E & P gave RR = 1.24 (1.07-1.45) per 5 years of use

Among E & P sequential therapy gave higher risk than continuous therapy

JNCI 2000

Women’s Health Initiative DesignWomen’s Health Initiative Design

A randomized controlled primary prevention trial

Planned duration 8.5 years16,608 postmenopausal women 50 – 79

years of age with intact uterus at baseline were recruited by 40 clinical centers in 1993-1998

InterventionIntervention

Conjugated equine estrogen 0.625 mg/d, plus medroxyprogesterone acetate, 2.5 mg/d, in 1 tablet (n=8506)

Placebo (n=8102)

Results at termination of trialResults at termination of trial

Mean duration of follow-up 5.2 years 290 cases of breast cancer Risk increased with duration of use (sig. trend

over time) Overall RR vs placebo = 1.26 (1.00-1.59) But, substantial noncompliance will bias results

to null:– 42% E&P and 38% placebo stopped study medication– RR in compliers = 1.49, p<0.001

International Agency for research on International Agency for research on Cancer (IARC)Cancer (IARC)

Classify agents as carcinogens after rigorous review of evidence, laboratory, animal, and human studies

Vol. 91 classifies combination estrogen plus progestin as carcinogenic to humans

Large drop in breast cancerLarge drop in breast cancer

US SEER (national tumor registry program) California state New Zealand Germany

US drop in prescribing Contribution of a decrease in screening has

been debated and ruled out as a cause for drop

Dispensed outpatient PMH Dispensed outpatient PMH prescriptionsprescriptions

Wysowski et al 200534.5M ’92 to high of 87.5M 2000

Clarke et al, CaliforniaClarke et al, California

Kaiser data on prescribing– 68% drop in E&P prescribing following release of WHI

results

10% drop in breast cancer incidence

For US women 50 to 69 (26 million women), this is 8,200 fewer cases of breast cancer, each year

J Clin Oncology Nov 2006

Further SEER analysisFurther SEER analysis

Jemal et al used state of art analysis (joint point analysis) to evaluate trends in breast cancer over time

1975 to 2003 – 394,891 invasive cancers Decrease in breast cancer largely confined to

ER+ tumors in the 2003 downturn Trend down strongest in women 55 to 64 In situ rates stable from 2000 to 2003 Rules out substantial screening impact

Jemal Breast Cancer Res May 2007

Further analysis of California dataFurther analysis of California data

California health interview survey California tumor registry breast cancer Classified CA counties into 3 levels based on

2001 E&P use Breast cancer incidence declined

– 8.8% in counties with smallest decline– 13.9% intermediate– 22.6% largest E&P decline– No change in proportion of women having

mammograms

Robbins and Clarke JCO 2007 (August)

Risk accumulationRisk accumulation

Overall evidence points to accumulation of risk through the life course

SERMs may offer some potential to inhibit final stages of progression to cancer - prevention greatest among those with high estrogen levels

Lifestyle contributes to cumulative riskNo one intervention for prevention

Physical activityPhysical activity

Evidence from more than 30 studiesTypical reduction in risk with 4 hours

per week = 20% decrease in riskEvidence present for pre and post-

menopausal women– Barriers to physical activity include

neighborhood safety, time and family responsibilities, social pressures

Cumulative rates of invasive and noninvasive Cumulative rates of invasive and noninvasive breast cancers occurring in participants breast cancers occurring in participants

receiving placebo or tamoxifen. The P value receiving placebo or tamoxifen. The P value are two-sidedare two-sided

Fisher et al, 1998; 90:1371-88Fisher et al, 1998; 90:1371-88

JAMA 1999, JNCI 2004

PreventabilityPreventability

International variation in ratesVariation in reproductive characteristicsGrowth and obesityPrimary prevention randomized trials

Social strategy to prevent breast Social strategy to prevent breast cancercancer

Provider– counseling on diet, activity, weight gain/loss– identify “higher risk” for preventive interventions

Balance risks and benefits

Regulation – facilitate lactation, physical activity, ?diet

Community– lactation, physical activity, access to care

Goals for PreventionGoals for Prevention

Reduce exposure to hormones after menopause– Avoid postmenopausal hormones– Weight loss– Anti estrogens for those at high enough

riskPromote increase in physical activityManage alcohol intake

No significant difference in No significant difference in adverse eventsadverse events

Placebo Raloxifene

Deaths 2.3% 1.7%

Thromboembolic disease

1.01% 1.72%

Endometrial cancer

0.39% 0.32%

Endometrial hyperplasia

0.29% 0.37%

Risk vs. benefit: who should get a Risk vs. benefit: who should get a SERMSERM

35.6M women 50 to 79 134,000 incident cases/yr Raloxifene would prevent 80,872 cases/yr Raloxifene would cause 67,649

thromboembolic events – Based on 19/10,000 per year treated

For benefit (reduced breast cancer) to exceed harm (thromboembolic events) incidence must be greater than 380/100,000

Age and risk decile for Age and risk decile for benefits to exceed risksbenefits to exceed risks

Age D4 D6 D8 D9 D10

50-4 185 233 302 357 504

55-9 246 309 401 474 668

60-4 278 349 453 536 756

65-9 339 425 552 654 921

Incidence per 100,000 women per year

50 to 64 year old population 5.1M eligible, 25%<65

Incidence based on Rosner/Colditz model

Incidence, Number Needed to Incidence, Number Needed to Treat (for 5 years) to prevent Treat (for 5 years) to prevent

1 cases (top decile)1 cases (top decile)Age Incidence NNT

50-4 504 79

55-9 668 60

60-4 756 53

65-9 921 43

Incidence per 100,000/yr.Assumes 50% reduction in Breast CA risk

Chen W, Rosner B, Colditz G. Cancer 2007

Next steps to prevention of Next steps to prevention of breast cancerbreast cancer

Refine assessment of risk– Stratify and provide appropriate counseling

Balance risk and benefits– Provide tools to aid weighing risks and benefits for

women

Implement population strategies to change behavior of providers, community, and even regulations to reinforce behavior changes (e.g., physical activity, weight control)