Mar 16

HRT and breast cancer incidence and mortality

Journal reference: Chlebowski R, Anderson G, Gass M, et al.  Estrogen plus progestin and breast cancer incidence and mortality in postmenopausal women. JAMA. 2010; 304 (15):1684-1692. [1]


Published: 20 October 2010

Evidence cookie says...

Combined oestrogen and progestin HRT is associated with increased risk of invasive breast cancer, node-positivity, and all cause mortality compared to placebo.

  • ↑ 25% risk of invasive breast cancer
  • NNH = 114; 11 years follow up with 5.6 years of HRT
  • ↑ 78% risk of node-positivity at time of diagnosis of breast cancer

The associated higher risk of breast cancer death and all cause mortality had wide confidence intervals.  They need to be viewed with caution.

HRT should be used only when there is clear benefit.  GPs should discuss the increased risks of breast cancer and mortality. Short-term use is prudent.

More details:

Article details

Study design:

follow up of placebo-controlled randomised controlled trial

Study aim:

to determine the effects of therapy with oestrogen plus progestin on cumulative breast cancer incidence and mortality after a total mean follow-up of 11 years

Methods summary:

  • follow up of the the participants in the Women’s Health Initiative (WHI) trial
  • brief history of the WHI:
    • November 1993:
      • initial randomisation begins
    • July 2002:
      • net harm for combined hormone therapy use was identified
      • participants instructed to stop their study medication
    • July 2002 – March 2005:
      • clinical visits and follow up continued as per protocol to the original trial completion date
    • April 2005 – August 2009:
      • study extension
      • subsequent follow up for additional breast cancer incidence results
      • 83% of the surviving participants consented to be part of the study extension


  • aged 50 to 79 years
  • were post-menopausal
  • provided written informed consent


  • prior hysterectomy
  • prior breast cancer
  • any condition that precluded a 3-year survival
  • women already using post-menopausal hormones unless have had at least a 3-month washout period
  • baseline mammograms and clinical breast examinations did not suggest cancer

Randomisation and interventions:

  • randomisation and double blinding process described
  • randomisation stratified by clinical centre and age group; co-ordinated centrally and implemented locally with a bar-code dispensing procedure
  • treatment group:
  • placebo control group:
    • identical appearing placebo pill
  • 6 monthly contact and annual clinical review including clinical examination and mammogram


  • breast cancer
    • verified by centrally trained, locally based physician adjudicators after medical record and pathology report review
    • final adjudication and coding of histology, hormone receptor status, and ERBB2 (HER2) status based on pathology report review performed at WHI Clinical Coordinating Centre
  • breast cancer death
    • attribution of death based on medical record review by physician adjudicators, blinded to randomisation allocation
    • National Death Index was cross checked with all clinical trial participants


  • original participant sample size calculated on basis of coronary heart disease
  • as a result, to detect a 15% increase in breast cancer:
    • it has a power of 55% after 9 years follow up
    • it has a power of 87% after 14 years follow up
  • results for breast cancer incidence and deaths were assessed based on intention to treat principle
  • hazard ratios estimated from Cox regression models
  • potential effect of women who did not reconsent to the study extension examined

Results summary:

  • 373 092 women screened for trial participation, 18 845 provided consent
    • 16 608 (5%) were eligible to participate and were randomised to treatment or placebo arms
  • excellent retention rates of participants until March 31, 2005 (original trial completion date)
    • 8056 out of 8506 (97%) participants for the treatment arm continued until March 31, 2005
    • 7682 out of 8102 (95%) participants for the placebo arm continued until March 31, 2005
  • for extension phase 6545 (78%) in treatment arm and 6243 (77%) continued to participate
    • majority of those not continuing refused consent
    • participants reconsenting more likely to be younger and of Caucasian background, compared with those not consenting
  • Follow-up and treatment:
    • mean follow up: 11.0 years, SD 2.7 years
    • mean follow up during the treatment phase: 5.6 years, SD 1.3 years


  • incidence of invasive breast cancer
    • oestrogen plus progestin group: 385 cases (0.42%) per year
    • placebo group: 293 cases (0.34%) per year
    • hazard ratio (HR) = 1.25 (95% CI 1.07-1.46), P = 0.04
    • that is, oestrogen plus progestin is associated with an increased incidence of breast cancer
    • Calculation by Morsels of Evidence:
      • number needed to harm (NNH) for an additional case of invasive breast cancer at 11 years follow up with treatment for 5.6 years: ~ 114
  • fraction of breast cancer presenting with positive lymph nodes
    • oestrogen plus progestin group: 81 cases (23.7% of those with breast cancer)
    • placebo group: 43 cases (16.2%)
    • HR = 1.78 (95% CI 1.23-2.58), P = 0.03
    • that is, oestrogen plus progestin is associated with a higher likelihood of positive lymph nodes at breast cancer diagnosis
  • receptor-positive and receptor-negative disease:
    • no difference between the two groups observed
    • note: this information was unavailable in 25-35% of breast cancer cases so the result may be unreliable
  • breast cancer mortality
    • oestrogen plus progestin group: 25 deaths (0.03% per year)
    • placebo group: 12 deaths (0.01% per year)
    • HR = 1.96 (95% CI 1.00-4.04), P = 0.49
    • represents 2.6 vs 1.3 deaths per 10 000 women per year, favouring placebo
    • note: this is not a statistically significant results
    • Calculation by Morsels of Evidence:
      • NNH for an additional breast cancer death at 11 years follow up with treatment for 5.6 years: ~ 700
  • all cause mortality
    • oestrogen plus progestin group: 51 deaths (0.05% per year)
    • placebo group: 31 deaths (0.03% per year)
    • HR = 1.57 (95% CI 1.01-2.48),  P = 0.045
    • represents 5.3 vs 3.4 deaths per 10 000 women per year, favouring placebo
    • note: this result just reaches statistical significance
    • Calculation by Morsels of Evidence:
      • NNH for an additional death at 11 years follow up with treatment for 5.6 years: ~ 450

Study conclusion:

Estrogen plus progestin was associated with greater breast cancer incidence, and the cancers are more commonly node-positive.  Breast cancer mortality also appears to be increased with combined use of estrogen plus progestin.


  • 16 608 postmenopausal women aged 50-79 years with no prior hysterectomy, breast cancer or conditions precluding 3 year mortality
  • selected from 40 US clinical centres from November 15, 1993
  • selected baseline characteristics:
    • age:
      • 50-59: 33.2%
      • 60-69: 45.2%
      • 70-79: 21.5%
    • ethnicity:
      • White: 84.0%
      • Black: 6.8%
      • Hispanic: 5.3%
      • Asian/Pacific Islander: 2.2%
    • years since menopause:
      • < 5 years: 16.7%
      • ≥ 15 years: 42.9%
  • extensive baseline demographic data was provided in the supplement to this study

Methodological weaknesses

  • as the study was designed from a cardiovascular outcomes perspective, the breast cancer incidence and mortality outcomes are relatively underpowered
    • the hazard ratio for breast cancer mortality did not reach statistical significance with a wide confidence interval but the results are suggestive
    • most subgroup analyses lack sufficient power for a statistically significant result
  • the loss of study participants at the extension phase of the study (which required reconsenting) may have introduced a bias
  • the majority of the study participants were older women and it has been 15 years or great since menopause
  • possible funding bias (though the results and conclusion do not favour hormone replacement therapy)

Methodological strengths

  • quality data from a randomised controlled trial; previous data largely from observational studies
  • well designed study
  • ethnically diverse population
  • few participants lost to follow up (apart from the loss at the beginning of the extension phase)
  • intention to treat analyses
  • long follow up period
  • reasonable number of women from a broad range of post-menopausal age groups
  • cautious interpretation of results

Biases and conflicts of interests

  • authors declare financial payment from a variety of pharmaceutical firms including: AstraZeneca, Novartis, and Pfizer, Procter and Gamble , Wyeth Laboratories, Upsher-Smith laboratories, Meditrina, Merck, Moehringer Ingelheim and Organon
  • one author (Rohan TE, seventeenth author) receives consultancy fees from legal firms regarding hormone therapy issues
  • the study was funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services
  • the study sponsor had input in to the design and conduct of the study, and had input into the report review but did not participate in preparation of the manuscript

Clinical relevance to primary health care

This study provides some clarity to the prescribing of hormone replacement therapy (HRT), specifically, oestrogen with progestin in post-menopausal women.  Results from observational studies have suggested that although there may be increased risk of breast cancer with HRT, these cancers are more favourable in terms to lower stage and longer survival compared with breast cancer diagnosed with non-hormone use.  Observation studies unfortunately are limited by confounding factors; the results from the randomised design of the Women’s Health Initiative study is a higher level of evidence and should be considered more reliable.

This study demonstrated that breast cancers diagnosed in women on HRT were 78% more likely to be node positive compared to women on placebo treatment, refuting the previous findings from observation studies.

The study also confirmed the association between the use of HRT and the incidence of invasive breast cancers with a 25% higher rate compared to placebo. The number needed to harm (NNH) is 1250 (per year) or only 114 cumulatively for the length of the study (note: where treated with HRT for 5.6 years and followed up for 11.0 years).

Moreover, women in the HRT group were associated with a higher all cause mortality and higher breast cancer mortality.  Caution should be taken, however, in interpreting these results as there are wide confidence intervals and the results for breast cancer mortality does not reach statistical significance.

Although most of the participants at baseline were older and it had been a decade and greater since menopause, there were sufficient numbers of younger women within 5 years of menopause for analysis. A significant association with higher risk of invasive breast cancer was detected: a 41% higher incidence rate.

This study does not provide information on the risk of breast cancer and breast cancer mortality for short-term use of HRT.

In the primary health care setting, hormonal replacement therapy in post-menopausal women should only be prescribed when there is a clear evidence of benefit to the patient.  Discussion about the increased risk of breast cancer and mortality should take place. Short-term use rather than longer-term use is prudent.


  1. Chlebowski R, Anderson G, Gass M et al.  Estrogen plus progestin and breast cancer incidence and mortality in postmenopausal women. JAMA. 2010; 304 (15):1684-1692.

Editor: Michael Tam

Print Friendly, PDF & Email

Leave a Reply

%d bloggers like this: