HRT and Cancer Risk: How the WHI Study Misled a Generation

The Women's Health Initiative, published in 2002, enrolled 16,608 postmenopausal women aged 50 to 79 and randomised them to conjugated equine oestrogen plus medroxyprogesterone acetate versus placebo.¹ The trial was stopped early at a mean of 5.6 years due to a nominally elevated breast cancer hazard ratio of HR 1.26 (95% CI, 1.00 to 1.59) in the combined hormone arm.¹ What followed was not a careful communication of absolute risk to prescribers and patients. It was a collapse in prescribing. HRT use in the United States dropped by more than 50% within two years of publication, and similar declines were recorded across the UK and Australia.² The problem was not that the data were fabricated. The problem was how they were framed, disseminated, and absorbed into clinical habit.

The WHI population skewed older than the women most likely to be candidates for HRT. The mean age at enrolment was 63 years, meaning a large proportion of participants were initiating hormones more than a decade after menopause, a window in which cardiovascular and oncological risk profiles differ materially from those of recently menopausal women.¹ The "timing hypothesis," subsequently articulated through secondary analyses of WHI data and supported by the Nurses' Health Study, holds that oestrogen's vascular and cellular effects differ depending on the stage of atherosclerosis and tissue oestrogen-receptor status at the time of exposure.² Initiating therapy close to menopause, in a window sometimes defined as the first ten years or before age 60, is associated with a more favourable cardiovascular risk profile than late initiation.²

What the data actually show

When WHI data are stratified by age, women aged 50 to 59 in the combined hormone arm showed a non-significant trend toward lower all-cause mortality during the intervention period, in contrast to the elevated hazard seen in older age groups.¹ In the oestrogen-only arm, which enrolled 10,739 post-hysterectomy women, breast cancer incidence was actually lower in the hormone group at trial completion: HR 0.77 (95% CI, 0.59 to 1.01), a finding that received a fraction of the media and clinical attention given to the combined-arm result.¹ Extended follow-up at 13 years in the oestrogen-only arm produced a statistically significant reduction in breast cancer incidence: HR 0.79 (95% CI, 0.65 to 0.97), alongside a significant reduction in breast cancer mortality.¹ These numbers did not reshape prescribing culture the way the 2002 press release did.

The progestogen component matters. Medroxyprogesterone acetate, the synthetic progestogen used in WHI, has a different receptor binding profile and metabolic footprint from micronised progesterone, which is now available and widely used in transdermal and oral formulations in Europe.³ Observational data, including the large French E3N cohort study of 80,377 women, indicate that oestrogen combined with micronised progesterone carries a lower breast cancer risk than oestrogen combined with synthetic progestogens, though the absence of randomised data on this specific question limits the strength of that inference.³ The route of oestrogen delivery also affects venous thromboembolism risk: oral oestrogen is associated with a two- to three-fold increase in VTE risk, while transdermal oestrogen at standard doses does not appear to carry the same hepatic first-pass effect on clotting factors.³ The MHRA and NICE updated their guidance to reflect this distinction, but uptake of transdermal formulations in primary care has been inconsistent.⁴

Limitations of this evidence base are real and should not be discarded. The WHI remains the only large randomised trial of HRT with hard clinical endpoints. Observational studies are subject to healthy user bias, and the E3N data cannot substitute for a randomised comparison of formulation types. Absolute risk figures vary by individual baseline risk, and women with BRCA1 or BRCA2 mutations, a personal history of hormone receptor-positive breast cancer, or unexplained vaginal bleeding require separate risk assessment frameworks. The gap in the evidence is not the question of whether HRT causes harm at the population level of the WHI. It is the absence of a large modern randomised trial using transdermal oestrogen and micronised progesterone in women aged 50 to 59, with breast cancer and cardiovascular events as co-primary endpoints. That trial has not been done.