By Jeffrey T. Jensen, MD, MPH

Dr. Jensen is Leon Speroff Professor and Vice Chair for Research, Department of Obstetrics and Gynecology, Oregon Health & Science University, Portland.

OB/GYN Clinical Alert’s editor, Jeffrey T. Jensen, MD, MPH, is a consultant for, on the Advisory Boards of, and receives grant/research support from HRA Pharma, Bayer Healthcare, Merck, Agile Pharm, Population Council, AbbVie, Evofem and ContraMed; and is a consultant for Teva Pharmaceuticals and MicroChips.
 

Synopsis: A large case-control study using the Danish National database found a slight increase in the risk of glioma in ever-users of hormonal contraception, which increased with duration of use. Users of progestin-only methods were at higher risk. However, these findings should not influence clinical practice or perception of contraceptive safety.

Source: Andersen L, et al. Hormonal contraceptive use and risk of glioma among younger women: A nationwide case-control study. Br J Clin Pharmacol 2014; Oct 26. doi: 10.1111/bcp.12535 [Epub ahead of print].

The authors used the extensive Danish population-based administrative and health registries to perform a case-control study evaluating the association of current and past use of hormonal contraception and glioma. Cases included all women in the Danish Cancer registry database between the ages of 15 to 49 years with a first-time diagnosis of a histologically verified glioma between 2000 and 2009. Each case was age-matched to eight population controls identified in the Danish National Patient Register. Hormonal contraceptive use, categorized according to type and duration, was inferred using prescription data from a pharmacy database. The categories of contraceptive use were combined estrogen-progestogen, progestogen-only, or mixed use.

Only use of oral products and the levonorgestrel intrauterine system (LNG-IUS) were evaluated, as few women reported use of the vaginal ring, implant, injection, or patch. Duration was categorized as < 1 year, 1 to < 5 years, or > or = 5 years. The authors used conditional logistic regression to compute odds ratios (ORs) with 95% confidence intervals (CIs) for glioma associated with hormonal contraceptive use, adjusting for potential confounders (age, years of schooling, and history of allergy or asthma). A total of 317 cases were identified and matched to 2126 controls. Ever-use of a hormonal contraceptive was associated with an elevated risk of glioma (OR, 1.5; 95% CI, 1.2-2.0), and the OR increased slightly with longer duration of use (> 5 years; OR, 1.9; 95% CI, 1.2-2.9) and then disappeared with past use (OR, 1.2; 95% CI, 0.8-2.0). Of great interest, the increase in risk appeared confined to users of progestogen-only methods. For progestogen-only tablets, the odds were significantly increased (OR, 3.3; 95% CI, 1.6-6.9); the point estimate was only increased for the LNG-IUS (OR, 2.4; 95% CI, 0.8-6.8).

The overall increase in risk associated with combined methods (OR, 1.4; 95% CI, 1.0-1.8) was not statistically significant, but the odds ratio became significant with use of > or = 5 years (OR, 1.7; 95% CI, 1.1-2.8). In contrast, the odds of glioma associated with > or = 5 years use of a progestogen-only method was 4.1 (95% CI, 0.8-20.8), but, again, the confidence interval included 1.0. The authors conclude that long-term use of hormonal contraceptives was associated with an increased risk of glioma, and that further studies are needed to evaluate this potential risk.

Commentary

This paper received a burst of attention from the media that quickly fizzled. But I suspect trial lawyers are already preparing Internet websites to serve the “interest” of families of young women diagnosed with one of these rare tumors. At first review, several features of this manuscript should draw suspicion. It is highly unlikely that the British Journal of Pharmacology is on your monthly reading list or that it was the preferred journal for publication of this paper. This strongly suggests failed peer review in more appropriate journals. This is not necessarily a black mark, as many papers make the rounds. But most authors prefer publication in a relevant journal. In this case, that would include the numerous publications devoted to contraception, gynecology, reproductive endocrinology, neurology, neurosurgery, or epidemiology. This is important, as these journals maintain lists of reviewers with sufficient expertise to evaluate manuscripts and place results in context with what is known about a topic.

The published results of this paper warrant critical appraisal. First, although glioma is a devastating diagnosis, these tumors are extremely rare. Data from the Central Brain Tumor Registry of the United States (1998-2002) puts the average annual age-adjusted incidence rate of primary malignant brain and central nervous system tumors among adults aged 20 years as 9.0 per 100,000 person-years. For glioblastoma, the female rate was 2.4/100,000 and the median age of diagnosis was 64 years.1Although the incidence has been slowly increasing, this is thought to be related to more common use of diagnostic imaging (MRI, CT). Suffice to say, the disease is rare, particularly in young women. On the other hand, if a young woman develops a brain tumor, it is likely to be a glioma.

Although brain tumors are rare in young women, hormonal contraceptive use is common. This makes evaluation of risk difficult. A case-control design is appropriate for a rare disease, but we must be careful to not over-interpret the data. The Bradford Hill criteria for epidemiologic studies provide guidance:2 1) Strength of association: Risk estimates should be moderate to large. Most of the odds ratios in the present study are < 2, suggesting a weak association. Of note, the association with progestogen pills was OR 3.3 (95% CI, 1.6-6.9). 2) Dose response: The biologic effect should increase with greater exposure. In this study, increasing duration of use did increase the OR, but the associations remained weak except for progestogen-only pills. 3) Biologic plausibility: There should be a biologic mechanism to mediate the effect. There is evidence to suggest that estrogens may influence the development of gliomas;3 the incidence is lower in premenopausal females compared to males, some gliomas and glioblastomas express estrogen receptors (ER) and aromatase, and experiments in rodents have shown that estradiol and certain SERMs inhibit proliferation of gliomas. Progesterone receptor (PR) is expressed in some gliomas, and progesterone is believed to be endogenously produced in the central nervous system and responsible for normal development. Could the increased PR exposure of the progestogen-only pill, in association with ovarian suppression and low estrogen levels, be a mechanism for an increase in glioma risk? Possibly, but the evidence is weak. Progesterone is known to be neuroprotective following stroke or head trauma, and is associated with schwannomas and meningiomas. But, in contrast to these tumors, treatment with mifepristone does not influence the growth of gliomas.4 In other words, progesterone signaling seems to be related to growth of myelin and meninges, and not to neurons and other brain connective tissue. 4)Consistency of effect: A recent meta-analysis of 11 case-control studies of exogenous hormones in relation to glioma in women found a lower risk of glioma in women who were ever-users of exogenous hormones. Although the Danish study was limited to younger women, there is certainly no hint from the meta-analysis to support an increase in risk.

Should you counsel young women about the risk of glioma with hormonal contraception? In my opinion, the study does not provide sufficient information to warrant a change in practice. But this study will be out on the Internet, so it makes sense to be aware of the issue. Note that the progestogen-only pill used in Denmark (Cerazette®, a desogestrel pill) is not marketed in the United States. This pill inhibits ovulation in most cycles. Even if we consider a three-fold increase in risk to be real, the attributable risk of glioma from use of a progestogen-only pill would be 4.9/100,000. Given that the maternal mortality risk is about 1/10,000, this risk should be put into perspective.

References

  1. Schwartzbaum JA, et al. Nat Clin Pract Neurol 2006;2:494-503.
  2. Schulz KF, Grimes DA. The Lancet Handbook of Essential Concepts 
    in Clinical Research
    . Elsevier; 2006.
  3. Kabat GC, et al. Cancer Epidemiol Biomarkers Prev 2010;19:
    2421-2427.
  4. Ramaswamy R, et al. Br J Neurosurg 2012;26:336-339.