Hormonal Contraception and Cervical Cancer
Abstract & Commentary
Synopsis: An increased risk of cervical cancer is associated with long durations of use of hormonal contraceptives.
Source: Smith JS, et al. Lancet. 2003;361:1159-1167.
A team of epidemiologists performed a meta-analysis of 28 studies examining the relationship between invasive and in situ cervical cancer and hormonal contraception. The analysis included 12,531 women with cervical cancer in 4 cohort and 24 case-control studies. Hormonal contraceptive use was grouped into categories: short duration (less than 5 years), medium duration (5-9 years), and long duration (10 or more years). Overall, the relative risk of cervical cancer increased with increasing duration of use, from 1.1 (CI, 1.1-1.2) with short-duration use to 2.2 (1.9-2.4) with long-duration use. The risk declined gradually after discontinuation. When the analysis was confined to the 3000 cases that tested positively for human papillomavirus (HPV), the relative risks for short, medium, and long duration of use were: 0.9 (0.7-1.2), 1.3 (1.0-1.9), and 2.5 (1.6-3.9), respectively. The results were less marked in HPV-negative women; indeed, a slight increase did reach statistical significance. The relationship was observed for in situ and invasive cancer and both squamous cancer and adenocarcinoma.1
Comment by Leon Speroff, MD
Cancer of the uterine cervix is now believed to be significantly caused by persistent infection with certain types of the sexually transmitted HPV. The medical literature suggests that the use of hormonal contraceptives can influence whether HPV leads to cervical cancer. This meta-analysis supports this conclusion, finding an increasing risk with increasing duration of hormonal contraceptive use, persisting after adjustments for sexual partners, smoking, and use of barrier methods of contraception. The data largely reflect the use of combined estrogen-progestin oral contraceptives. Although the data with injectable contraceptives (largely progestin only) were limited, Smith and associates report a slight increase with this method, as well. All methods of hormonal contraception were lumped together; therefore, results according to specific types and doses are unavailable.
I have a slide that I have frequently used for more than 5 years. It reads: "Meta-Analysis is to Analysis like Meta-Physics is to Physics." The technique of meta-analysis dates back to the late 1970s when it was initiated as a method to bring together small, randomized trials in order to increase statistical power. In the 1980s and 1990s, meta-analysis was rapidly extended to case-control and cohort studies. Clinicians soon came to give great weight to conclusions from meta-analyses, believing that the sophistication of the method provided reliable conclusions.
The technique of meta-analysis contains an important element of subjectivity. A group of individuals makes judgments regarding published studies, decisions about how good or bad each study is, and whether the conclusions should be incorporated into their own decision-making. Meta-analysis cannot correct confounding problems and biases in the original studies, and the conclusion of a meta-analysis can be misleading. An excellent example is the 1996 meta-analysis concluding that induced abortion increased the risk of breast cancer.1 Subsequently it was discovered that case-control studies of this subject contained a major problem of recall bias (the healthy women in the control groups were reluctant to tell the truth about induced abortions). Appropriately designed studies (avoiding personal interviews) do not find an association between induced abortions and the risk of breast cancer.2,3
Cancer of the cervix is affected by many risk factors. Is it possible to lump 28 studies together and adequately control for confounding influences? Even Smith et al point out in their discussion that not a single one of the original studies adjusted for all confounding factors, and it would require uniform definitions and adjustments in each individual study to improve the reliability of the data. An excellent study from the Centers for Disease Control and Prevention (CDC) concluded that there is no increased risk of invasive cervical cancer in users of oral contraception, and an apparent increase in situ cancer is due to enhanced detection because of more frequent Pap smears.4 However, the World Health Organization Study found an increased risk of in situ cancer even when Pap smear frequency was adjusted.5 Very concerning is the rising incidence of adenocarcinoma of the cervix in young women over the last 20 years and impressive agreement among case-control studies that the risk of cervical adenocarcinoma increases with increasing duration of oral contraceptive use.
Fortunately, Pap smear surveillance is very effective for cervical cancer. The new methods incorporating HPV identification will be even better. This is another good argument against making hormonal contraception available over the counter. This meta-analysis of hormonal contraception and cervical cancer shouldn’t change clinical practice because evaluation for cervical cancer is already part of the routine care for these women. It makes sense to me to perform cervical cancer screening every 6 months in women using hormonal contraception for longer than 5 years, especially if they are at higher risk because of their sexual behavior.
Dr. Speroff is Professor of Obstetrics and Gynecology, Oregon Health Sciences University, Portland, Ore.
1. Brind J, et al. J Epidemiol Community Health. 1996; 50:481-496.
2. Mahue-Giangreco M, et al. Cancer Epidemiol Biomarkers Prev. 2003;12:209-214.
3. Erlandsson G, et al. Int J Cancer. 2003;103:676-679.
4. Irwin KL, et al. JAMA. 1988;259:59-64.
5. Ye Z, et al. Int J Epidemiol. 1995;24:19-26.