The WHI-Wayne State Fracas
By Leon Speroff, MD
Women’s Health Initiative (WHI) investigators from Wayne State University performed an analysis using the WHI database and presented the results at the 2004 annual meeting of the American Society of Reproductive Medicine (ASRM).1,2 An oral presentation entitled Adverse Cardiovascular Disease Outcomes Are Reduced in Women with a History of Oral Contraceptive Use was selected as the prize paper by the ASRM-affiliated organization, the Society of Reproductive Endocrinology and Infertility. This analysis concluded that previous users of oral contraceptives (OCs) had a reduced risk of various cardiovascular outcomes, including hypercholesterolemia, angina, myocardial infarction, transient ischemic cerebrovascular attacks, and peripheral vascular disease. Another presentation indicated that previous users of OCs had reduced cancer risks (endometrial, ovarian, and breast).
Two months later, the presentation on the effects of oral contraception on cardiovascular disease drew a critical response from the director of the WHI, Barbara Alving, MD, who is also the acting director of the National Heart, Lung, and Blood Institute. Alving issued a statement revealing that the work from Wayne State had been reviewed by the WHI and that flaws were discovered in both the design and interpretation.3 Indeed, the WHI review indicated no evidence that OC use is linked to a lower risk of cardiovascular disease. Appropriately the WHI pointed out that the Wayne State analysis was performed post hoc, using the information provided by the WHI participants based on their recall upon entry to the study, a study that was not designed to examine a relationship between OCs and cardiovascular disease. Alving emphasized that the original presentations were not reviewed by the WHI or NIH, and that subsequent review by the WHI statisticians could not find a relationship between OC use and cardiovascular disease, and that the statisticians further doubt the validity of the presentation on cancer risks.
Unfortunately, Alving then stated that "there is a large and reputable body of higher scientific evidence linking current OC use to future increases in risk of stroke and heart attack, especially in older women and in smokers." She further stated that research indicates an increased risk of breast cancer in women who have recently used OCs. Alving acknowledged a decreased risk of ovarian cancer in OC users, but stated that the decreased risk of endometrial cancer is only a slight one.
In response to the objections of the WHI, Wayne State University issued a press release, authored by 2 of the co-authors of the original ASRM presentations.4 The press release essentially agreed with the position of the WHI, pointing out that analysis of the WHI baseline data is "exploratory" at best. But the press release goes further, essentially retracting the original conclusions, stating that "analyses of these data don’t support inferences of either cardiovascular disease benefit or risk when additional account is taken of the complex relationships between the ages of participating women and OC use patterns and cardiovascular disease."
This is an embarrassing moment, to be sure, but also another example of the inappropriate dissemination of information by the WHI to the public. It is not good science and it is not good for the public welfare to generate conclusions from data created for a different purpose and in a design not directed for the publicized end point. The WHI criticized its own investigators for this action, but this whole scenario is not that different from previous WHI conclusions that were applied to all postmenopausal women when they were derived from a small and special group of older women. Furthermore, the WHI compounds this unpleasant story by making statements that are flat out wrong.
There is no evidence of an increase in risk of cardiovascular disease among past users of oral contraception.5-7 In the Nurses’ Health Study, the Royal College of General Practitioners’ Study, and the Oxford Family Planning Association Study, long-term past use of OCs was not associated with an increase in overall mortality.8-10 Part of the concern for a possible lingering effect of OC use was based on a presumed adverse impact on the atherosclerotic process, which would then be added to the effect of aging and, thus, would be manifested later in life. Instead, the findings have been consistent with the contention that cardiovascular disease due to oral contraception is secondary to acute effects, specifically estrogen-induced thrombosis, a dose-related event.
Almost all myocardial infarctions and strokes in OC users occur in users of high-dose products, or users with cardiovascular risk factors older than the age of 35. In the Oxford Family Planning Association cohort, cardiac deaths occurred only in women who smoked 15 or more cigarettes per day.10 OCs containing less than 50 µg ethinyl estradiol do not increase the risk of myocardial infarction or stroke in healthy, nonsmoking women, regardless of age. The effect of smoking in women younger than age 35 is, as we have long recognized, not detectable in the absence of hypertension. After age 35, the subtle presence of hypertension makes analysis difficult, but studies indicate that increasing age and smoking by themselves have little impact on the risk of stroke in low-dose OC users. The studies do indicate that hypertension should be a major concern, especially in regards to the risk of stroke.
Increased Risk of VT
All low-dose OCs, regardless of progestin type, have an increased risk of venous thromboembolism, concentrated in the first 1-2 years of use. The risk increases with increasing age and body weight. The actual risk of venous thrombosis with low-dose OCs is lower in the new studies compared with previous reports. Some have argued that this is due to preferential prescribing and the healthy user effect. However, it is also logical that the lower risk reflects better screening of patients and lower estrogen doses.
The importance of good patient screening cannot be overemphasized. The occurrence of arterial thrombosis is essentially limited to older women who smoke or have cardiovascular risk factors, especially hypertension. The impact of good screening is evident in the repeated failure to detect an increase in mortality due to myocardial infarction or stroke in healthy, nonsmoking women.10-12 Although the risk of venous thromboembolism is slightly increased, the actual incidence is still relatively rare, and the mortality rate is about 1% (probably less with OCs, because most deaths from thromboembolism are associated with trauma, surgery, or a major illness). The minimal risk of venous thrombosis associated with OC use does not justify the cost of routine screening for coagulation deficiencies.
OCs and Cancer
I want to emphasize that the use of OCs for at least 12 months reduces the risk of developing endometrial cancer by 50%, with the greatest protective effect gained by use for more than 3 years (this is hardly a "slight" reduction).13-18 This protection persists for 20 or more years after discontinuation (the actual length of duration of protection is unknown) and is greatest in women at highest risk: nulliparous and low parity women.18,19
How about Alving’s definitive statement that there is an increased risk of breast cancer in recent users of OCs? I suspect that the statement refers to the collaborative group reanalysis of data from 54 studies in 26 countries, a total of 53,297 women with breast cancer and 100,239 without breast cancer.20,21 The relative risk analyzed by duration of use was barely elevated and not statistically significant (even when long-term use, virtually continuous, was analyzed). Women who had begun use as teenagers had about a 20% statistically significant increased relative risk. In other words, recent users who began use before age 20 had a higher relative risk compared with recent users who began at later ages. The evidence was strong for a relationship with time since last use, an elevated risk being significant for current users and in women who had stopped use 1-4 years before (recent use). Even though the data indicated that young women who begin use before age 20 have higher relative risks of breast cancer during current use and in the 5 years after stopping, this is a time period when breast cancer is very rare; and, thus, there would be little impact on the actual number of breast cancers. It is possible that early and recent use of OCs also affects the growth of a preexisting malignancy, explaining the limitation of the finding to current and recent use and the increase in localized disease. The largest case-control study thus far included 4575 American women with breast cancer, and most importantly, the women were 35 to 64 years old.22 The risk of breast cancer was not increased in current users or past users of oral contraception. There was no adverse effect of increasing duration of use or higher doses of estrogen, with no differences in current or recent users. Initiation at a younger age had no impact, and there was no increase in risk in women with a family history of breast cancer. This large American study had consistently negative results.
It’s disappointing that the NIH and the WHI can’t get things right.
1. Victory R, et al. Fertil Steril. 2004;82:S52-S53.
2. Victory R, et al. Fertil Steril. 2004;82:S104-S105.
3. Alving B. Statement on Oral Contraceptive Study: www.nhlbi.nih.gov/new/press/04-12-15.htm 2004.
4. Hendrix S, Prentice R. Cardiovascular Disease. Press Statement from Wayne State University and the Women’s Health Initiative. www.whi.org/news/press_oral_contraceptives.php 2004.
5. Croft P, et al. Br Med J. 1989;298:165-168.
6. Rosenberg L, et al. Am J Epidemiol. 1990;131: 1009-1016.
7. Stampfer MJ, et al. Am J Obstet Gynecol. 1990;163: 285-291.
8. Colditz GA, and the Nurses’ Health Study Reasearch Group. Ann Intern Med. 1994;120:821-826.
9. Beral V, et al. Br Med J. 1999;318:96-100.
10. Vessey M, et al. Lancet. 2003;362:185-191.
11. Jick H, et al. Lancet. 1995;348:1589-1593.
12. Petitti DB, et al. Stroke. 1997;28:280-283.
13. The Cancer and Steroid Hormone Study of the CDC and NICHD. JAMA. 1987;257:796.
14. Schlesselman JJ. Contraception. 1991;43:557.
15. Vessey MP, Painter R. Br J Cancer. 1995;71:1340.
16. Schlesselman JJ. Hum Reprod. 1997;12:1851-1863.
17. Salazar-Martinez E, et al. Cancer Res. 1999;59: 3658-3662.
18. Weiderpass E, et al. Cancer Causes Control. 1999;10: 277-284.
19. Jick SS, et al. Obstet Gynecol. 1993;82:931-935.
20. Collaborative Group on Hormonal Factors in Breast Cancer. Lancet. 1996;347:1713-1727.
21. Collaborative Group on Hormonal Factors in Breast Cancer. Contraception. 1996;54:1S-06S.
22. Marchbanks PA, et al. N Engl J Med. 2002;346: 2025-2032.
Leon Speroff, MD, Professor of Obstetrics and Gynecology, Oregon Health and Science University, Portland, and Editor for OB/GYN Clinical Alert.