A Systematic Review and Meta-Analysis of the Association Between Depot Medroxyprogesterone Acetate and Cerebral Meningioma.

The current study was conducted following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines ( Supplementary Table S1 ); the review protocol was not prepared and not registered. A search was performed in PubMed, Embase, and Web of Science for articles published through February 2026. For each search, terms included medroxyprogesterone acetate, depot medroxyprogesterone acetate, progestin, and meningioma. Results were limited to scientific journal articles, review articles, and letters written in English. Further inclusion criteria included the study being an observational study design (i.e., case-control or cohort; disproportionality analyses using adverse event reports were excluded); the study including human subjects of female sex only; and a study that assessed the association between dMPA and the incidence of new cerebral meningioma through estimation of relative measures of association including odds ratios (ORs), risk ratios (RRs), or incidence rate ratios (IRRs) in addition to the associated 95% confidence intervals (CIs). Once the final list of eligible articles was created, the titles and abstracts were reviewed by two of the authors (LR and RG) to determine which articles were to be included in the analytical sample for the current study. Reviewers were blinded to the journal, authors, and institution. There was a total of 45 disagreements between the reviewers; the reviewers discussed each of the studies, and a final decision was made. The full text of studies that were selected from initial review was then reviewed to determine final selection. For each selected study, using manual review, data was extracted on the study title, authors, year of publication, journal, study type (e.g., case-control, cohort), sample sizes, age range of participants, definition of dMPA exposure (i.e., source of exposure data, duration of exposure), confounders assessed, type of comparator group used, and relative measures of association. Type of comparator was defined as any non-dMPA exposed subject (which could include those who were taking a contraceptive other than dMPA), active comparator (i.e., a subject taking another contraceptive such as levonorgestrel or norethindrone), or non-active comparator (i.e., a subject taking no contraceptive). Exposure duration was categorized as any exposure regardless of duration, short-term duration (fewer than two years), or prolonged duration (two or more years). Of note, data from Griffin (2024) [ 14 ] were reanalyzed to make the exposure duration definition similar to studies that were subsequently published. Quality of the final selected studies was determined through use of the JBI critical appraisal checklist for case-control studies by two authors (LR and RG) [ 15 ]. The checklist assesses the quality of studies across ten questions assessing the reliability of exposure and outcome measurement, matching methodology (for matched case-control and cohort studies), and analytical methodology. For each question, an answer of Yes, No, Uncertain, or Not Applicable may be chosen. The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology was used to determine the quality of evidence for the association between dMPA and cerebral meningioma [ 16 , 17 ]. Potential publication bias was assessed through visual inspection of contoured funnel plots of the point estimates and inverse of the standard error (i.e., precision) of the natural log of the odds ratio. As a study could be included in multiple pooled analyses (e.g., twice if the study provided short-term and prolonged exposure associations), the funnel plot was conducted among any exposure, short-term exposure, and prolonged exposure associations. An Egger’s regression test was used for formal testing of symmetry, with p -values < 0.05 suggesting asymmetry and potential publication bias. Pooled analyses were conducted by each combination of type of comparator and duration of dMPA exposure. ORs were used as the measure of association for pooled associations across studies using both fixed and random effects analyses. If an OR was not reported for a given combination of exposure duration and comparator type, the study was excluded from that portion of the pooled analysis. Heterogeneity across studies was assessed utilizing an I 2 and Q statistic. If heterogeneity was significant ( p < 0.05) for a given duration category analysis, a random effects analysis utilizing the DerSimonian and Laird method [ 18 ] was used for pooled analysis; otherwise, a fixed effects pooled analysis utilizing the inverse variance method was used. In a post hoc sensitivity analysis, due to observed high heterogeneity, a leave-one-out analysis was performed using the weakest association to determine the effects of a study on the duration-specific effect and then separately for each combination of exposure type and comparator type. Within each duration category, pooled ORs were computed for each sub-group of comparator type and for the overall duration. A forest plot was used to present results of the pooled analyses. An absolute measure of effect for dMPA exposure was calculated using the non-malignant cerebral meningioma incidence rate reported among females aged 35–54 by [ 19 ] and the overall pooled OR calculated from the meta-analysis. In a planned sensitivity analysis, pooled associations for each duration category were conducted using (regardless of comparator) the weakest association and, separately, the strongest association from each study. Microsoft Excel (Version 2508 Build 16.0.19127.20570) was used for the literature review and determination of included studies, and SAS v9.4 was used for all meta-analyses.

Medroxyprogesterone acetate (MPA) is a synthetic form of progesterone called a progestin that is used for a variety of gynecological and obstetric conditions (e.g., amenorrhea, endometrial hyperplasia, endometriosis) and is, for the current study, most notably used as a contraceptive [ 1 ]. It acts by binding to progesterone receptors in the body and preventing the release of gonadotropin-releasing hormone (GnRH). This decrease in the levels of GnRH then inhibits follicular maturation and increases the thickness of the cervical mucus, the effects of which prevent ovulation and decrease sperm mobility through the cervix. There are two forms of MPA, oral and injectable. The former requires consistent daily intake while the latter is an intramuscular injection that forms a depot within the muscle that provides a slow release of MPA lasting up to three months; as a result, the injectable form of MPA is known as depot medroxyprogesterone acetate (dMPA). It is believed that progesterone receptors play an integral role in the formation of intracranial meningiomas, which are slow-growing, mostly benign tumors of the protective layer of the brain called the meninges [ 2 , 3 , 4 , 5 ]. Intracranial meningiomas are reportedly three times more prevalent among females than males [ 6 , 7 ] and have been observed to increase in size during periods of pregnancy when levels of progesterone are higher [ 8 ], particularly in healthy pregnancies [ 9 ]. Further, examination of meningiomas occurring during pregnancy revealed that all tumors expressed progesterone receptors while a lower expression frequency was observed among meningiomas that developed among women who were not pregnant [ 5 ]. A study examining progesterone receptor affinity among progestins observed that MPA has the fourth highest affinity behind progestins levonorgestrel, desogestrel, and nomegestrolacetate [ 10 ]. In 2024, a French study reported an over five-fold increased odds of cerebral meningioma among users of dMPA, an association that was driven by those who used dMPA for at least two years [ 11 ]. Since that time, a number of studies have been published that have examined the association between dMPA and cerebral meningioma among a variety of study populations; however, the reported associations range from weak to strong strengths of association due to differences in exposure duration, type of comparison group (e.g., active comparator, non-contraceptive user), study design, and prevalence of exposure. Recently, two review studies including the association between dMPA and cerebral meningioma have been published. The first of these studies provides a review with no meta-analysis, instead focusing on the implications of the South African population [ 12 ]. The second study [ 13 ] examined the association between cerebral meningioma and progestogens. While it did provide results of a meta-analysis, reporting a pooled odds ratio of 2.68 for dMPA, the study included only six dMPA-related studies, examined only any exposure to dMPA despite prior research finding that the association is particularly strong for prolonged dMPA exposure, and, in the pooled analysis, combined associations that included non-dMPA exposed individuals and non-dMPA users. Thus, the objective of this current systematic review and meta-analysis is to provide a description of the studies, assess the study quality, and address the limitations of the prior review studies by providing a pooled analysis of the association between dMPA and cerebral meningioma by duration of exposure (i.e., any, short-term, and prolonged) and type of comparator (i.e., non-exposed to dMPA, non-active, and active).

The literature search returned a total of 41 articles from Embase, 105 articles from PubMed, and 112 articles from Web of Science ( Figure 1 ). In addition, one article [ 20 ] not indexed in the above services found, from a Google Scholar search, that a prior article by two of the current authors [ 21 ] was included. After removing 62 duplicates, 197 unique articles were selected for review for final consideration. Initial review involving titles of articles resulted in the exclusion of 170 studies, the majority of which ( n = 88, 51.8%) were excluded due to not being a case-control or cohort study or case report ( n = 33, 19.4%); the remaining studies were excluded due to not being a study involving human subjects ( n = 19, 11.2%), not including MPA as an exposure ( n = 15, 8.8%), not including cerebral meningioma as an outcome ( n = 12, 7.1%), and not including females ( n = 3, 1.8%). After review of the full text of the remaining 27, a total of 17 were excluded due to not including dMPA exposure ( n = 12, 85.7%), not being a case-control or cohort study ( n = 2, 14.3%), utilizing a disproportionality analysis design ( n = 2, 14.3%), or examining only the association with meningioma grade ( n = 1, 7.1%). Of the ten selected studies [ 11 , 14 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ]—which included nine case-control studies and one cohort study—a multitude ( n = 8) were published between 2024 and 2026, though the first study to report an increased association with injection hormone exposure (particularly for prolonged use) was published in 2006 [ 22 ] ( Table 1 ). A total of 139,672 cerebral meningioma cases were included across the studies; the number of participants exposed to dMPA ranged from seven to 88,667. Eight of the ten [ 11 , 14 , 20 , 21 , 23 , 25 , 26 , 27 ] identified dMPA exposure through medical record review or through pharmacy claims data; one study defined dMPA exposure from use of HCPCS and ICD-10 CM codes [ 24 ]; and one study defined dMPA exposure through the participant’s self-reported use [ 22 ]. All but one study defined cerebral meningioma diagnosis through use of ICD-9 or ICD-10 CM codes, specifically 192.1, 225.2, C70, D32, and D42; a single study identified cerebral meningioma occurrence through histopathological diagnosis [ 20 ]. Nearly all studies received a Yes across all ten questions ( Table 2 ). The comparability of the cases and controls in the Wahyuhadi study [ 20 ] was uncertain as controls were patients who underwent a head CT; as a result, there is a potential for information bias if the dMPA exposure status of the controls is related to the medical condition that was the reason for seeking clinical care. In addition, there was no apparent adjustment for confounders in logistic models. The reliability of dMPA exposure measurement in Xiao [ 24 ] was uncertain as it was based on HCPCS and ICD-10 CM codes for an encounter for injectable contraceptive rather than through use of pharmaceutical claims or self-report in prior studies; however, any bias was likely non-differential between case and controls. Examining the contoured funnel plots, there is potential asymmetry for studies assessing short-term dMPA use with potential missing studies with stronger associations that are significant, resulting in a pooled short-term association that may be weaker than the true pooled association ( Figure 2 ); however, as the imputed missing studies would be towards the stronger, significant associations, it is likely that the exclusion is not due to publication bias based on statistical significance [ 28 ]. Further, results of Egger’s regression test suggest no statistical evidence of asymmetry with observed non-significant p -values for any exposure ( p = 0.2590), short-term exposure ( p = 0.6434), and prolonged exposure ( p = 0.9526). Including all associations across the studies, the pooled OR was 2.78 (2.20–3.52) ( Table 3 ). For any exposure to dMPA, the pooled association was OR 2.89 (95% CI 2.06–4.04), with the strongest association observed for comparisons to an active comparator (OR 4.43, 95% CI 1.99–5.54) ( Figure 3 ). By specific exposure duration, the strongest association was observed for prolonged dMPA exposure (pooled OR 3.49, 95% CI 2.35–5.18), particularly for comparisons to a non-exposed comparator (OR 3.84, 95% CI 1.98–7.46) with a weaker, yet still statistically significant, association compared to an active comparator (OR 2.52, 95% CI 1.28–4.93) ( Figure 4 ). The overall pooled association for short-term exposure was weakest among the three duration categories yet still statistically significant (OR 2.01, 95% CI 1.06–3.80) with no significant effects observed among the three comparison sub-group specific pooled associations. In sensitivity analyses by duration category and including (a) the weakest association from each study and (b) the strongest association from each study, pooled associations were similar to the above-reported associations for any exposure (weakest OR pooled: 2.95, 95% CI 1.92–4.51; strongest OR pooled: 3.37, 95% CI 2.15–5.29), short-term exposure (weakest OR pooled: 1.84, 95% CI 0.0.89–3.81; strongest OR pooled: 2.09, 95% CI 0.99–4.41), and prolonged exposure (weakest OR pooled: 3.41, 95% CI 2.13–5.46; strongest OR pooled: 3.69, 95% CI 2.38–5.74). There was significant heterogeneity for the global effect (I 2 = 92.2%, Q = 319.5, p < 0.0001); among studies that used any exposure to dMPA (I 2 = 91.3%, Q = 149.407, p < 0.0001); among studies assessing short-term exposure (I 2 = 94.8%, Q = 135.6, p < 0.0001); and among studies assessing prolonged exposure associations (I 2 = 81.4%, Q = 43.0, p < 0.0001). In leave-one-out analyses ( Figure A1 ), the association for any exposure remained relatively stable with a low association of OR 2.52 (95% CI 1.66–3.82) estimated odds ratios based on random-effects estimation after excluding Wahyuhadi [ 20 ] and a high association when excluding Wigertz [ 22 ] (OR 3.19, 95% CI 2.02–5.06). For both short-term exposure and prolonged exposure, exclusion of Tettamanti [ 24 ] resulted in no heterogeneity (short-term: I 2 = 4.7%, Q = 4.2, p = 0.3799; prolonged: I 2 = 0.0%, Q = 3.9, p = 0.5701) and the weakest associations (short-term: OR 1.36, 95% CI 1.16–1.60; prolonged: OR 2.50 (95% CI 2.13–2.93)). The strongest association for short-term exposure was observed when excluding Reynolds [ 25 ] (OR 2.06, 95% CI 0.91–4.68) and for prolonged exposure when excluding Griffin (2024) [ 14 ] (OR 3.95, 95% CI 2.63–5.92). When performing the leave-one-out analysis by combinations of exposure duration and comparator type, associations remained similar in strength for the any exposure categorization ( Figure A2 ). Among short-term exposure associations ( Figure A3 ), the pooled comparison to a non-active comparator was lowest when excluding Griffin (2025) [ 21 ] (OR 1.31, 95% CI 0.89–1.93, I 2 = 0.0%, Q = 0.7, p = 0.4057), and pooled comparisons to non-exposed subjects were lowest when excluding Tettamani [ 26 ] (OR 1.34, 95% CI 1.16–1.55, I 2 = 0.0%, Q = 0.5, p = 0.4645). Among prolonged exposure comparisons ( Figure A4 ), there was no heterogeneity among comparisons to active or non-active comparator subjects; the significant heterogeneity for comparisons to non-exposed subjects was removed when excluding either Griffin 2024 [ 14 ], which resulted in the strongest pooled association (OR 5.75, 9% CI 4.66–7.09, I 2 = 0.0%, Q = 1.9 p = 0.3874) or Tettamani [ 26 ], which resulted in the weakest pooled association (OR 2.81, 95% CI 1.83–4.32, I 2 = 34.5%, Q = 3.1, p = 0.2171). GRADE assessments for reviews based on observational research begin with a low grade of evidence. Based on the funnel plots and Egger’s regression test, there is no concern for publication bias ( Table A1 ). Further, as all studies were observational and mostly based on registry or claims data, there is no concern noted for risk of bias due to selection or information bias. No concern was noted for imprecision as a majority of associations were statistically significant and were based on studies using large population registry or medical claims data. The heterogeneity was driven by select studies, the exclusion of which had limited effect on the inference of the pooled effect; as a result, no concern of inconsistency was noted. Regarding residual confounding, as observational studies do not have a controlled, randomized aspect to the design, there is potential for residual confounding in the included studies; however, per Balshem [ 16 ] and Prasad [ 17 ], concerns for risk of bias are not noted if the confounding effect is towards the null. For the included studies, there is no reason to believe that residual confounding in the included studies, a vast majority of which are based on registry or claims data, would be differential by cerebral meningioma status or depot medroxyprogesterone acetate exposure status. As a result, the non-differential confounding bias results in reported associations being biased towards the null, and accounting for the confounding would result in stronger associations; thus, no concern for risk of bias is present. That said, due to the difference in the source populations and the significant, high heterogeneity, a very serious concern of indirectness was noted, downgrading the GRADE by two ( Table 3 ). From the pooled analysis, the GRADE evidence quality was upgraded due to a large, pooled effect size (i.e., >OR 2.00) and presence of a dose–response gradient (prolonged association stronger than short-term association). This yielded a final quality of evidence rating of moderate, suggesting that the true estimate is near the reported pool observations, though a true effect different than the reported pooled association cannot be ruled out. Based on the overall pooled OR of 2.78 (2.20–3.52) and an incidence of 12.5 per 100,000 persons from the prior literature [ 19 ], the estimated risk of cerebral meningioma among those with dMPA exposure is 34.8 (95% CI 27.5–44.0) per 100,000, a risk difference of 22.3 (95% CI 15.0–31.5) per 100,000.

The results of this meta-analysis confirm associations from the previous literature. Specifically, the current study reported an over two-fold association between exposure to dMPA and cerebral meningioma. This association is strongest for prolonged use (i.e., at least two years of continuous use) rather than short-term use (i.e., fewer than two years), though it is present at a significantly increased effect for any exposure and short-term exposure. Further, the observed pooled association is persistent whether the comparison group comprises those with either no exposure to hormonal contraceptives or users of non-dMPA hormonal contraception. This study adds to the published literature an assessment of the quality of evidence in addition to potential biases such as publication bias of studies published from inception through February 2026; in addition, the study is the first to provide a meta-analysis by dMPA exposure duration and comparator type, providing a more thorough examination of pooled associations of dMPA and cerebral meningioma. It is worth noting that over two-thirds of the 32 associations included in the current analysis were statistically significant. Of the nine associations that were not significant, three were from the same study that was the earliest performed and had the second smallest sample size [ 22 ] behind Wahyuhadi [ 20 ]. Further, of the studies examining prolonged dMPA use, only two reported non-significant associations. Both were based on smaller case-control studies [ 21 , 22 ] but reported strengths of association similar to the other studies. Though the physiological mechanism underlying the association between dMPA and cerebral meningiomas is not fully described, there are potential mechanisms that have been discussed. Progestin-associated cerebral meningiomas have been reported to occur more often among the skull base [ 29 ], a location that has been associated with higher rates of mutations to the PIK3CA gene [ 30 ]. The PIK3CA gene is believed to be involved in the apoptosis of meningeal cells, and the mutations result in an activation of the PI3K/AKT/mTOR pathway. This then results in a cascade of events that lead to decreased apoptosis [ 31 ], formation of tumors [ 32 ], and tumor progression [ 33 ].This proposed mechanism could potentially explain the observed pattern in the current meta-analysis that prolonged dMPA exposures are most strongly associated with cerebral meningioma as the longer duration of exposure potentially results in an increased chance of PIK3CA mutation due to longer exposure time. These results should be viewed in light of the limitations of the current search methodology, mainly that only three services (PubMed, Embase, Web of Science) were reviewed; thus, it is possible that studies were excluded from the review if published in a journal not indexed in one of the three services. However, research has suggested that as few as two services is sufficient to include related literature on a given topic [ 34 ] while another suggests the three services used in the current analysis allow for appropriate coverage of literature on a topic [ 35 ]. In addition, the review was limited to observational research studies; however, in the search, no studies based on clinical trials were identified. As a final limitation, there was high heterogeneity in the current meta-analysis that was particularly driven—as evidenced in the leave-one-out-analysis—by Tettamanti [ 26 ] and Griffin (2025) [ 21 ]. Despite the high heterogeneity, there was minimal change in the strength and inferences of the associations during the leave-one-out analysis, lending credence to the robustness of the pooled associations despite the high heterogeneity. It is also important to place these studies in the context of their strengths and limitations. First, nearly all studies utilized pharmaceutical claims or medical record reviews to determine dMPA exposure. While this type of medication exposure ascertainment can lead to misclassification bias if the subject does not truly take the medication, this bias is likely not present for dMPA, which is a one-time injection given every three months. Thus, the claims and medical record data accurately present whether the person was exposed. That said, it is possible that the injection is not reported in the claims or medical record data, resulting in the person being incorrectly misclassified as unexposed; however, this would result in a bias towards the null as it is unlikely to be differential between cerebral meningioma case and controls and, if it were to occur, would result in an overestimate of cerebral meningioma in the unexposed group, decreasing the strength of association. As another limitation, since the data for most of the studies were derived from a hospital or medical claims data, controls may not be representative of the exposure prevalence of the general population. The effects of this limitation, however, are likely minimal as the claims data encompasses inpatient and outpatient claims across multiple medical facilities. Finally, the prevalence of dMPA exposure was low among cases in multiple studies, which could affect statistical power and could be due to the exposure prevalence of the studies not being representative of the general population; however, there is no reason to suspect the non-representativeness to be differential by case or non-case status. The results of this meta-analysis have important clinical implications; however, prior to discussion of the clinical impact, one must take into consideration the potential causation of associations derived from observational studies. Specifically, one must consider strength of association, consistency of association, dose–response relationship, temporality of exposure, and specificity of association. The first three are described within the GRADE methodology. In the current analysis, dMPA exposure was deemed a moderate quality due in part to the strong association, the consistency of the association, and the dose–response relationship. The latter two criteria are described by Bradford Hill [ 36 ] and often used to aid in epidemiological research in the determination of potential causality of an association. The studies included in this analysis all meet the temporality criteria as the use of data from prior cohorts and administrative data allowed researchers to determine the timing of dMPA exposure relative to the diagnosis of cerebral meningioma. Regarding specificity, null effects have been reported for dMPA exposure association and spinal meningioma and oral MPA exposure for cerebral meningioma or spinal meningioma [ 20 ], suggesting that the exposure increases the risk only for cerebral meningioma and for dMPA. Taken together, the associations reported in the current analysis provide evidence to support a potential causal association between dMPA and cerebral meningioma; however, causation is not a certainty given the limitations of observational research regarding confounding and selection bias, and, further, more rigorous research is needed.

Despite the stated limitations, the selected studies provide evidence that dMPA is associated with the diagnosis of cerebral meningioma. The source of the study populations of the studies has been varied—including random selection of the source general population, cancer patients treated at a medical center, and enrollees in a public or private insurance—yet the reported associations remain consistent across the studies. Further, the studies have observed that the association is specific to dMPA (reported associations for oral MPA have been null), specific to cerebral meningioma (reported associations for spinal meningiomas have been null), and increase in strength with longer duration of use, particularly an over three-fold increased association for prolonged dMPA use. The results of the current analysis and review provide evidence that the association between dMPA and cerebral meningioma could be causative; however, further research is warranted on the topic of injectable MPA and the association with meningioma. In particular, it would be of interest to examine the association between dMPA and meningioma WHO grade as the literature on that topic is lacking as of the current date. In addition, research is needed on the progression of dMPA-related meningiomas (in order to determine, for example, the proportion of time surgical treatment is needed) as well as long-term cognitive and quality-of-life outcomes of females with MPA-related meningiomas. Further, while keeping in mind the discussion of the concept of causation in epidemiology, the results of this meta-analysis suggest that, first and foremost, clinicians should discuss with their patients the potential risk of meningioma associated with dMPA. The American College of Obstetricians and Gynecologists recommends a shared decision-making approach in which clinicians provide clear explanations of the evidence of the risks associated with contraception use, allowing the patient to make an informed decision on whether to use the contraceptive [ 37 ]. Part of that discussion should include the recent FDA requirement that dMPA should include a warning regarding the risk of meningioma [ 38 ]. A recent review article of dMPA and cerebral meningioma provided further suggestions, including advising against the use of dMPA for more than two years and for women who use dMPA to monitor for signs of cerebral meningioma including headaches, vision changes, or seizures [ 12 ]. Research has reported that progestin-related meningiomas decrease in size following discontinuation of non-MPA progestins including cyproterone acetate, chlormadinone acetate, and nomegestrol acetate [ 39 , 40 , 41 , 42 , 43 ]. Finally, there are other contraceptives within the progestogen class such as levonorgestrel or desogestrel that have been reported to not be associated with a risk of cerebral meningioma [ 44 , 45 ]. Clinicians should discuss the use of other progestin contraceptives if they are available to the patient; however, caution should be used in the selection of an alternative progestin as research has reported an increased risk of meningioma with the use of cyproterone acetate [ 40 ], chlormadinone acetate [ 46 ], and nomegestrol acetate [ 47 ].