Breast Cancer in the Women’s Health Initiative Trial of Estrogen Plus Progestin

For the WHI Investigators

Rowan T Chlebowski, MD., Ph.D.

Menopausal Hormone Therapy and Breast Cancer (Background)

Preponderance of observational studies suggest long duration estrogen plus progestin increases breast cancers which have:

- Low stage and favorable prognosis

- Receptor positive preponderance

- More lobular histologyHolli J Clin Oncol 1998; 16: 3115 Gapstur JAMA 1999; 281: 2021

Delgado Maturitas 2001; 38: 147 Lower Breast Cancer Res Treat 1999; 58: 205

Chen, JAMA 287: 734, 2002 Coldity Am JEpid 147 (5): 645, 1998

Descriptive Characteristics in WHI Participants

Age Age at menarche Relatives with breast ca (n)

Ethnicity Term pregnancies (n) Benign breast disease

Education Age at first birth Prior estrogen (E) alone use

Gail Risk Children breastfed (n) Prior E + progestin (P) use

BMI Oral Contraceptive use NSAID use

Alcohol Use % Energy from fat Physical Activity

None of these characteristics differed significantly between treatment groups

Descriptive Characteristics By Treatment Group

Prior Menopausal

Hormone Use

E+P (n, %) Placebo

Never 6280 (73.9) 6024 (74.4)

Prior 1674 (19.7) 1588 (19.6)

Current 548 (6.4) 487 (6.0)

“Current” users required three month washout

• Baseline mammogram and clinical breast exams required for eligibility

• Annual mammograms and clinical breast exams required when on study

• Study medications withheld if safety procedures not performed

WHI Estrogen+Progestin TrialBreast Safety

Breast Cancers by Category and Treatment Group

Breast Ca E+P Placebo HR (95% CI)1 P-Value2

Total 245 185 1.24 (1.02-1.50)

0.0004

Invasive 199 150 1.24 (1.01–1.54)

0.003

In situ 47 37 1.18 (0.77-1.82)

0.086

1 Hazard ratios (HR) from unweighted Cox proportional hazards regression models

2 P values from weighted Cox proportional hazards regression models

Hazard ratios (HR) from unweighted Cox proportional hazards regressions models

Z Statistics and p values from weighted Cox proportional hazards regression models

Invasive Breast Cancer By Group

Sensitivity Analysis of Adherent Participants Invasive Breast Cancers by Group

Hazard ratios (HR) from unweighted Cox proportional hazards regressions models

Z Statistics and p values form weighted Cox proportional hazards regression models

Participants were censored 6 months after becoming non-adherent (taking < 80% study meds or taking non-protocol hormones)

Breast Cancers (Annualized Percentage) by Age

Age E+P Placebo HR P-value

50-59 y 52 (0.31) 40 (0.26) 1.20 (0.80-1.82)

60-69 y 94 (0.44) 72 (0.31) 1.22 (0.90-1.66)

.20

70-79 y 53 (0.54) 38 (0.41) 1.34 (0.88-2.04)

P-value tests interactions of E+P and age

Breast Cancer (Annualized Percentage) by BMI

BMI E+P Placebo HR P-value

25 45 (0.31) 32 (0.23) 1.35 (0.86-2.13)

25-30 72 (0.42) 49 (0.31) 1.40 (0.97-2.01)

.12

> 30 82 (0.50) 68 (0.45) 1.08 (0.78-1.49)

P-value tests for interaction of E+P with BMI

Breast Cancers (Annualized Percentage) by Prior Menopausal Hormone Therapy (MHT) Use

Prior MHT E+P Placebo HR (95% CI)

P value

None 141 (0.40) 121 (0.36) 1.09 (0.86-1.39)

.10

Ever 58 (0.46) 29 (0.25) 1.86

(1.19-2.91)

1 P value tests for interaction with E+P and prior MHT

More breast cancers on E+P in both groups

Non-significant trend, no interaction

Ever users at somewhat lower risk

Cumulative exposure versus selection bias

Breast Cancer Incidence by Prior MHT Use and Randomization Assignment

No prior MHT Prior MHT

Year E+P Placebo HR E+P Placebo HR

1 7 14 .48 5 5 .90

2 15 22 .65 11 10 1.1

3 19 19 .96 10 3 3.09

4 35 23 1.45 9 4 2.16

5 28 17 1.61 15 4 3.56

6+ 37 26 1.24 8 3 1.99

Z for trend 2.31 1.62

Breast Cancer Characteristics by Group

E+P Placebo P-ValueHistology

Ductal 67.8 % 67.3 %

Lobular 11.1 % 10.3 % 0.885

Ductal + Lobular 7.5 % 5.3 %

Grade

Well 25.0 % 20.3 %

Moderately 43.3 % 47.7 % 0.609

Poor 31.7 % 32.0 %

Similar histology and grade on E+P and placebo

1 P value tests association with treatment groups

Receptor Status E+P Placebo P-Value1, 2

Estrogen receptor

Positive 158 (86.8) 112 (88.2) 0.720

Negative 24 (13.2) 15 (11.8)

Missing 17 (8.5) 23 (15.4) 0.049

Progesterone receptor

Positive 135 (75.0) 86 (69.9) 0.328

Negative 45 (25.0) 37 (30.0)

Missing 19 (9.5) 27 (18.0) 0.021

Both receptor positive and negative breast cancers greater on E+P

1 The first P value tests association with treatment groups

2 P-value for “missing” rows test the association of % missing with treatment group

Breast Cancer Characteristics by Group

E+P Placebo P-Value

Tumor size, cm1 1.7 (1.1) 1.5 (0.9) 0.0382

Nodes Positive2 25.9 % 15.8% 0.033

SEER Stage

Regional / Mets

25.4% 16.0% 0.041

1 mean (SD) for tumor with known tumor size

2 P-values from weighted Cox proportional hazards models

More advanced stage on E+P

Year 1 Mammogram Findings by Group

Mammogram Findings E+P (n, %) Placebo

Negative / Benign finding 6940 (90.7) 6912 (94.6)

Abnormal (total) 716 (9.4) 398 (5.4)1

Short interval f/u 625 (8.2) 332 (4.5)

Suspicious abnormality 85 (1.1) 59 (0.8)

Highly suggestive 6 (0.1) 7 ( 0.1)

1 p < .0001 comparing E+P versus placebo

Increased abnormal mammograms after 1 year on E+P

SummaryMammogram Findings by Group and Time

Baseline Year 1 CumulativeE+P Placebo E+P Placebo E+P Placebo

Mammogram Performed1

100% 100% 90.3% 90.5% 97.3% 97.8%

Mammogram

Abnormal (total)2

5.2% 5.0% 9.4%1 5.4% 31.5%1 21.2%

1 % of women due for visit with mammogram in study period who had mammogram

2 % of women with any category of abnormal mammogram

3 p < 0.0001 E+P versus placebo

Abnormal Mammograms: Associated with Short Duration E+P Use

4 % absolute increase in abnormal mammograms after one year on E+P

10% absolute increase in abnormal mammograms after about 5 years on E+P

Recent Results from the UK Million Women Study

National Health Service Breast Cancer Screening Program

Inform Interpretation of WHI Results

Million Women Study• NHSBSP in the UK invites women 50-69 for

mammography screening q 3 years by letters

• A questionnaire regarding HT use was added to the screening invitation letter

• HT use data was linked to NHS central registries for breast cancer and death outcomes

• 1,084,110 women flagged

• 9,364 incident invasive breast cancers seen

Lancet 2003; 362: 419-27.

Relative Risk of Fatal Breast Cancer by HT Use at Baseline in the Million Women Study

Based on 517 deaths after 4.1 years

HT use RR

(95% CI)

Never 1.00 (0.88-1.14)

Current 1.22 (1.05-1.41)*

* P = 0.05 for current versus never Lancet 2003; 362; 419-27.

HT associated with increased breast cancer mortality in “short term” users

Relative Risk of Breast Cancer in the Million Women Study By E+P Duration

Duration Cases/Population RR

(95% CI)

< 1 yr 97/9771 1.45 (1.19-1.78)

1-4 yr 582/49240 1.74 (1.60-1.89)

5-9 yr 850/56912 2.17 (2.03-2.33)

> 10 yr 362/23673 2.31 (2.08-2.56)

E+P associated with increased breast cancers in < 1 year

Relative Risk of Breast Cancer in the Million Women Study By Hormone Type

Hormone Type RR (95% CI)

Estrogen only 1.30 (1.21-1.40)

Equine Estrogens 1.29 (1.16-1.43)

Ethinyloestradiol 1.24 (1.12-1.40)

Estrogen + Progestin 2.00 (1.88-2.12)

Medroxyprogesterone 1.60 (1.33-1.93)

Norethisterone 1.53 (1.35-1.75)

Conclusions

Combined E+P use increases breast Ca,

diagnosed at more advanced stage and

increases abnormal mammograms

These results suggest: Use of E+P may

stimulate breast cancer growth and hinder

breast cancer diagnosis