Depot-Medroxyprogesterone Acetate Use Is Associated with Decreased Risk of Ovarian Cancer: The Mounting Evidence of a Protective Role of Progestins.

We included 7,977 women with ovarian cancer and 11,820 control women in our analysis ( Table 1 ). Ninety women with ovarian cancer (1.1%) and 252 control women (2.1%) had ever used DMPA ( Table 1 ). The prevalence of DMPA use in control women ranged from 0.8% to 3.5% across the seven studies. Overall, ever-use of DMPA was associated with a 35% reduction in risk of ovarian cancer (OR = 0.65, 95% CI 0.50–0.85). This inverse association was observed in all seven OCAC studies ( Figure 1 ). When restricting to women who were pre-menopausal and ≤45 at the time DMPA use was approved for contraception (56 users/2,660 cases and 197 users/4,884 control women), a similar association was observed; women who used DMPA had a 42% reduction in risk of ovarian cancer compared to never users (OR=0.58, 95% CI 0.42–0.81). DMPA use was inversely associated with all histotypes except low-grade serous where the odds ratio was 1.0 ( Figure 2 ). Likely due to small numbers, only the association with high-grade serous was statistically significant (OR=0.60, 95% CI 0.41–0.88). The results across histotypes were not statistically significantly different from each other (p>0.05 for all comparisons). As the duration of DMPA use increased, the magnitude of the inverse association with ovarian cancer became stronger (p for trend<0.001; Table 2 ). Compared to never users, women who used DMPA for 3–9 months (1–3 injections) had an 8% reduced risk of ovarian cancer (OR= 0.92, 95% CI 0.54–1.57; Table 2 ). Women who used DMPA for 12–18 months (4–6 injections) had a 44% reduced risk (OR=0.56, 95% CI 0.29–1.07), and those who had used DMPA for 21+ months had a 56% reduced risk (OR=0.44, 95% CI 0.26–0.77) ( Table 2 ). In the same model, women who used COCs for 12–59 months had a 24% risk reduction compared to never users (OR=0.76, 95% CI 0.68–0.84); and women with two births had a 35% risk reduction compared to nulliparous women (OR=0.65, 95% CI 0.57–0.74) ( Table 2 ). From the 4,702 records found through the PubMed search, 4,696 were removed for the following reasons: duplicates (n=34), DMPA use for different purposes (e.g., stimulating drugs for infertility; n=2,181), non-ovarian cancer outcomes (e.g., other cancer types; n=1,879), not original research (i.e., review articles; n=320), studies in animals (n=162), no English full texts (n=74), in vitro studies (n=45), or no full text available (n=1). The summary of the remaining six studies which met the a priori review criteria is presented in Table 3 . Of the six studies identified, four specifically studied the association between DMPA use and ovarian cancer risk; the other two did not distinguish between DMPA and NET-EN use. A cohort study of 5,003 Black women in the U.S. who used DMPA for an average of ~18 months and were followed for an average of 8.6 years observed only one ovarian cancer case compared to an expected 1.16 cases (SIR=0.86, 95% CI 0.1–4.6) ( 9 ). Two hospital-based case-control studies, one carried out by the World Health Organization (WHO) in Mexico and Thailand (22 DMPA users/224 invasive and borderline ovarian cancer cases) ( 10 ) and a second conducted in Thailand (59 DMPA users/330 invasive ovarian cancer cases) ( 11 ), observed inverse associations between DMPA use and invasive ovarian cancer risk (OR= 0.81, 95% CI 0.36–1.8 and OR=0.52, 95% CI 0.33–0.88, respectively). In contrast, a population-linkage study carried out in Denmark observed a six-fold increase in ovarian cancer risk among DMPA users (n=3 cases) (RR=6.56, 95% CI 2.11–20.40) ( 12 ). Of the two studies that did not specify the type of injectable contraceptive used, one found an inverse association and the other found a positive association. Urban and colleagues, in their hospital-based case-control study in South Africa, found that ever use of injectable progestin-only contraceptives with no use of oral contraceptives (n=10 cases), was associated with reduced risk of invasive ovarian cancer compared to never use of either (OR=0.35, 95% CI 0.17–0.71) ( 23 ). A cohort study in Shanghai, China observed a positive association between injectable contraceptive use and ovarian cancer risk (HR=1.33, 95% CI 0.58–3.04) ( 24 ).

All studies included in this analysis obtained institutional ethics committee approval and followed recognized ethnical guidelines including the Declaration of Helsinki, the Belmont Report, and/or the U.S. Common Rule. All participants provided written informed consent. We used data from seven case-control studies participating in the OCAC that had DMPA use information available - six in the U.S. and one in Australia: the Diseases of the Ovary and their Evaluation Study (DOV) ( 13 ), the Hawaii Ovarian Cancer Study (HAW) ( 14 ), the Hormones and Ovarian Cancer Prediction Study (HOP) ( 15 ), the North Carolina Ovarian Cancer Study (NCO) ( 16 ), the New England Case Control Study (NEC) ( 17 ), the University of Southern California Study of Lifestyle and Women’s Health (USC) ( 18 , 19 ), and the Australian Ovarian Cancer Study (AUS) ( 20 ). Data were self-reported and collected by in-person or telephone interviews using structured questionnaires. Data were sent to the OCAC data-coordinating center (Duke University) for central harmonization ( 21 ). The OCAC Epidemiology Working Group created a codebook which was distributed to each study site. The Working Group reviewed the data that were received, carried out logic checks and queried sites to resolve inconsistencies. Cases were women with invasive epithelial ovarian cancer (hereafter referred to as ovarian cancer). Controls were women without a personal history of ovarian cancer who had at least one intact ovary. Ever/never use of DMPA was available for all seven studies, and duration of DMPA use was available from four of these studies (DOV, NCO, USC and AUS). Duration of DMPA use was provided by each study in three-month intervals (i.e., 3, 6, 9, etc.). The association between DMPA and ovarian cancer risk overall and by histotype was modeled using conditional logistic regression. All models were conditioned on age at diagnosis for cases/reference age for controls (<40, five-year age groups to 74, 75+ years), race/ethnicity (Non-Hispanic White, Hispanic White, Black, Asian, other) and OCAC study site (n=7). Education level (less than high school, high school, some college, college graduate), COC use (never use, <1 year, 1–4 years, 5–9 years, 10+ years of use), parity (0, 1, 2, 3+ births), and breastfeeding (never, <12 months, 12–23 months, 24+ months) were considered important confounders a priori and were included in all models. The inclusion of duration of COC use and parity allowed us to compare the effects of these hormonal exposures to DMPA use on ovarian cancer risk. The impact of additional potential confounders, including a personal history of endometriosis (yes/no), first-degree family history of ovarian cancer (yes/no), body mass index (<25, 25-<30, 30+ kg/m 2 ), incomplete pregnancies (0, 1, 2+), and smoking (never, current, former), on the association between DMPA use and ovarian cancer risk were considered. None of these exposures changed the beta-coefficients for the DMPA use-ovarian cancer risk association by >10%; they were not included in the final models. Of the total of 20,446 participants, only women with complete data were included; 649 (3.2%) had missing data on at least one variable in the final model and were excluded. The variables with the most missing values were education level (1.7%), breastfeeding (1.0%), COC use (0.9%), and race/ethnicity (0.3%). Based on age at diagnosis and year of birth, many women who participated in the study would have been post-menopausal at the time DMPA was approved for contraception in the U.S. (1992) or Australia (1994). Although it was possible for women in the U.S. to receive DMPA for contraception at least as early as 1967 as some care providers received approval to use it as an investigational new drug ( 9 ), we conducted a sensitivity analysis which excluded women who were post-menopausal or >45 years of age in 1992 for the U.S. studies and 1994 for the Australian study to ensure that they could have used DMPA for contraception. We also carried out multinomial logistic regression to determine whether the results by histotype were statistically different from each other. Statistical significance was defined as pࣘ0.05 using 2-sided tests. Data were analyzed using R studio 1.1.463. A systematic search in PubMed was conducted to look for original research that reported the association between use of injectable DMPA as a contraception method and ovarian cancer risk. The search terms used were “ovarian cancer”, “depot-medroxyprogesterone acetate”, “progestin”, “contraceptive” and their synonyms. A manual search of the references provided in the selected articles was also conducted. Studies had to be published in English before August 1 st , 2019 and be available in full text. Studies on DMPA use for other purposes (such as cancer treatment), review studies, in vitro studies, and studies conducted in animals were excluded. After duplicates were removed, publications were screened based on titles, abstracts, and full texts, using Rayyan ( 22 ) as the analysis system. For articles that met the inclusion criteria, we extracted information about authors, year of publication, study design, time and place of recruitment, number and characteristics of participants, and the findings related to the objectives of this review.

In the pooled OCAC data, we observed a 35% reduced risk of ovarian cancer among women who had used DMPA; this decreased risk was consistently observed in all seven OCAC studies included. We also observed a dose-response relationship with greater risk reduction associated with longer duration of DMPA use when we pooled data across the four of the OCAC studies with this information. The results were broadly consistent across histotype, robust to sensitivity analysis restricted to women who were pre-menopausal and <46 years of age at the time DMPA was approved for contraception in the U.S. and Australia, and not likely to be confounded given that we were able to take COC use, parity, breastfeeding, education level, race/ethnicity, age, and other factors into account in the analysis. Six previous studies on DMPA use and ovarian cancer were identified through our systematic review; four of these studies found inverse associations with effect estimates ranging from 0.35 to 0.86 ( 9 – 11 , 23 ). Of the two studies that did not find an inverse association, one also did not observe the expected inverse association with combined oral contraceptive (COC) use or tubal ligation ( 24 ); systematic bias may have been introduced by only measuring contraceptive exposures at baseline in this 12 year follow-up study ( 24 ). . Further, these two studies also had the lowest prevalence of DMPA use among cases (six users out of 174 cases ( 24 ) and three users out of 1,249 cases ( 12 )), making these results particularly susceptible to uncontrolled confounding. Overall, based on the results of our pooled analysis and systematic review, the preponderance of evidence indicates an inverse association between DMPA use and ovarian cancer risk. Our results are also consistent with studies that examined the association between ovarian cancer and progestin-releasing IUDs, another progestin-only long-acting contraceptive ( 12 , 25 , 26 ). The prevalence of progestin-releasing IUD use has increased significantly, but because these types of IUDs were not marketed until 2000 in the U.S. and Australia ( 27 ), there are limited data on their association with ovarian cancer risk. However, the three studies that have examined the relationship between progestin-releasing IUD use and ovarian cancer found an inverse association (RR=0.84, 95% CI 0.53–1.35 ( 12 ),RR=0.53, 95% CI 0.32–0.88 ( 25 ), and SIR=0.59, 95% CI 0.47–0.73 ( 26 )), however the last of these studies did not adjust for potential confounders. In the literature, prevalence of progestin-only pill use ranged from 2.1% to 9.3% among controls in the four case-control studies that have reported this information; the prevalence of exclusive use of progestin-only pills ranged from 0.2%−1.8%. Based on these four studies, the association between progestin-only pill use and ovarian cancer risk is equivocal ( 28 – 31 ). COC use and parity are two other exposures that are associated with higher progestin/progesterone exposure compared to normal cycling. The average serum progestin concentration produced by DMPA is 5.1-fold higher than that produced by a COC with the same duration of use ( 32 – 34 ), whereas the progesterone exposure during pregnancy is approximately equivalent to the progestin dose in DMPA. In our data, the OR for 12–59 months of COC use was 0.76 compared to 0.56 for 10–18 months of DMPA use ( Table 2 ) which is suggestive of a stronger magnitude of effect for DMPA use. The DMPA OR is similar to that observed with parity (two births OR=0.65; Table 2 ). A DMPA injection is associated with residual anovulatory effects that last as long as nine months after the injection which may explain the stronger inverse association with ovarian cancer compared to COCs ( 35 – 37 ). Also, the high levels of estrogen and other hormones during pregnancy may affect the magnitude of the inverse association between parity and ovarian cancer risk. The DMPA duration data are sparse making firm conclusions impossible, but it does appear that DMPA is at least as protective as COCs and parity. The mechanisms underlying the protective effect of progestins are unclear. A simple explanation of blocking ovulation is possible, but if this were the only mechanism, the effect of all such exposures would be the same, which is not what has been observed. It is also possible that there is a direct hormonal effect ( 3 ). These direct effects may be due to the induction of cell death, thus preventing oncogenic transformation of precursor cancer cells. Indeed, it was found that progesterone induces necroptotic cell deaths in transformation-related protein 53 (Trp53) mouse oviduct epithelium and in immortalized human p53-defective fallopian tube epithelial cells ( 38 ). Possibly more importantly, in a randomized study of women undergoing risk-reducing bilateral salpingo-oophorectomy, women who received levonorgestrel had statistically significantly lower cell proliferation (Ki-67) in the ovarian surface epithelium (OSE) compared to women who received placebo (p=0.011) ( 39 ), but OSE is not considered to be the cell of origin for most ovarian cancers. There was some suggestion in this study that progestins cleared genetically abnormal cells in the fallopian tube fimbriae based on karyometic results ( 39 ). In addition, several in vitro studies have demonstrated that progestins induce apoptosis in human ovarian carcinoma cell lines ( 40 – 42 ). A major strength of this study is the large sample size which included 90 women with ovarian cancer who used DMPA. The results of this study are widely generalizable as we included participants in many locations in the U.S. and Australia. This allowed us to explore histotype-specific associations as well as the overall association by duration on a subset of studies. It is possible there is misclassification of duration of DMPA use as a woman may find it difficult to remember precisely how many shots she received. However, there is little reason to expect this misclassification to be differential as DMPA use is not a stigmatized exposure, thus the impact of any non-differential recall error would be toward the null. It is possible that participants may have misreported NET-EN use as DMPA use. However, DMPA is used every three months while NET-EN is used every two months. We had duration of use data from four OCAC studies and the durations were in multiples of three suggesting that this is not NET-EN use. Overall, the results from this study of DMPA use build on the body of evidence of a protective role for progestins/progesterone for ovarian cancer. Further investigation is warranted to understand the mechanism underlying the decreased risk. This is particularly relevant given the increasingly popularity of progestin-releasing IUDs.

Combined oral contraceptives (COCs), which include both an estrogen and progestin component, have been shown to reduce risk of invasive epithelial ovarian cancer by approximately 20% per five years of COC use ( 1 , 2 ). It has long been hypothesized that progestins in particular are associated with reduced risk of ovarian cancer ( 3 ). However, the relationship between progestin-only contraceptives and ovarian cancer risk is not well understood. Injectable progestin-only contraceptives, depot-medroxyprogesterone acetate (DMPA) and norethisterone oenanthate (NET-EN), are long-acting ( 4 ). Both have been available in some countries since the 1960s for contraception, but not until 1992 in the U.S. ( 5 – 7 ) and 1994 in Australia ( 8 ). Previous studies of the association between DMPA use and ovarian cancer risk have been small and the results were equivocal ( 9 – 12 ). Understanding the relationship between DMPA use and ovarian cancer may provide a broader understanding of the role of progestins in ovarian cancer risk, has significant value given its wide-spread use in some populations, and may be informative with respect to the potential for progestin-releasing intrauterine devices (IUDs) as chemopreventive for the disease. We used data from seven studies from the international Ovarian Cancer Association Consortium (OCAC) to conduct a pooled analysis examining the association between use of DMPA and ovarian cancer risk. The number of women with ovarian cancer using DMPA in this study is greater than in previous studies, thus our analysis presented here is the largest to date. The association between ovarian cancer risk with duration of DMPA use was examined and compared to the association with duration of COC use and number of births. We also conducted a systematic review to identify previously published studies that have examined the relationship between DMPA use and ovarian cancer risk to put the results of our analysis in context with the existing literature.