RESPONSE TO LETTER TO EDITOR

In Reply

We agree that the WHI’s repeated assertion that CEE+MPA increases the risk of breast cancer is not funny. Neither is their failure to follow protocol, their abandonment of basic statistical norms, and their repeated misrepresentation of their own data. It is not funny that these claims have scared millions of women away from Hormone Therapy (HT), the safest and most effective treatment for menopausal symptoms, prevention of bone loss and fractures, improvement in quality of life, and reduction of all-cause mortality when started within 10 years of menopause. The WHI’s persistent use of nominal statistics for their protocol-defined secondary endpoint of breast cancer, while ignoring protocol-mandated statistical adjustments,¹ has flown under the radar, which is why we explained in such detail what the WHI failed to do and why it matters.

This is the central point of our article: When appropriate statistical adjustments were made for covariates (differing distribution of baseline breast cancer risk factors across treatment groups)² and multiple comparisons (multiple statistical testing across time and across outcome categories),³ there was no increased risk of breast cancer among women randomized to CEE+MPA.

Drs. Chlebowski and Aragaki failed to respond to the major challenges raised by our paper, including:

1. Per the WHI protocol, breast cancer in the HT trials was a secondary outcome that required statistical adjustment for covariates and multiple outcomes. This is because participants were randomized for heart disease risk, not breast cancer risk. WHI investigators have ignored this a priori stipulation by re-branding breast cancer as the primary safety outcome and inappropriately analyzing breast cancer data as a primary outcome. That sleight of hand undermines the principles that maintain the integrity of clinical trials.

2. Whenever the WHI investigators have reported their findings with appropriate statistical adjustments, the results have failed to show a statistical difference in breast cancer risk between women randomized to CEE+MPA and those randomized to placebo.² Although the 2002 WHI paper generated headlines reporting an increased risk of breast cancer among CEE+MPA recipients,³ the authors actually reported that breast cancer “almost reached nominal statistical significance” with a 95% confidence interval (CI) of 1.00–1.59. It was clearly nonsignificant with the appropriate adjustment for multiple comparisons (95% CI= 0.83–1.92). In their 2003 paper, the first to claim a statistically significant increase in breast cancer risk for the use of CEE+MPA, when the most basic adjustment was made for sequential monitoring (for multiple unblinding of breast cancer data over time), the 95% confidence interval was no longer statistically significant (95% CI=0.97–1.59).⁴ Protocol-mandated adjustments for covariates and multiple outcomes that would further attenuate the findings were not reported.

3. In the randomized trial phase of the WHI, the increase in the hazard rate of breast cancer for women on CEE+MPA, compared to women on placebo, is attributable to the unexplained extremely low rate of incident breast cancer in the women with prior HT assigned to placebo, not to an increased breast cancer risk among those randomized to CEE+MPA.^2,5,6 Chlebowski and Aragaki now argue that this finding “disappears” over nearly 15 years of post-trial observational follow-up in the clinical trial participants. Yet because that assessment also lacked appropriate statistical adjustments, it compounds the analytic omissions in the clinical trial phase and does not contravene or undo the randomized clinical trial’s statistically nonsignificant findings of breast cancer risk.

For years, we and others have challenged the WHI’s failure to follow their protocol and basic statistical norms.^5,6,7 Although the WHI has consistently ignored these critiques, at worst these investigators claim a rare increase of 1 additional case of non-fatal breast cancer for every 1000 women on CEE+MPA for one year.⁸ Further, the WHI acknowledges that no increased risk occurs among women who had not taken HT prior to joining their study.^2,3 This, of course, is the situation for the vast majority of women who are started on HT. And yet in virtually all of their publications acknowledging HT’s many benefits, the WHI investigators keep inserting fear-generating warnings about breast cancer based on inappropriate analyses that lack protocol-mandated adjustment.^9,10 It is past time for the WHI to stop promoting fear and confusion about a treatment that has been shown to offer life-enhancing and life-saving benefits for countless menopausal women.