LAUSR

Offspring: A Call for More Conclusive Study B irth defects remain the leading cause of infant mortality in many geographic regions, including the United States where they account for approximately 20% of deaths (1). Given that most birth defects have a relatively low incidence, they can be exceedingly difficult to study and often require investigation at the population level and over time to ensure sufficiently powered epidemiologic investigations. Thesemethodological challenges can delay the identification of emerging teratogens, especially in the context of rapid changes in the health status of reproductive-aged populations, such as the growing percentage of young adults who are overweight or obese. Wensink and colleagues (2) deserve commendation for addressing a critical data gap regarding paternal exposures and birth defects. Their well-designed, population-based registry cohort study comprising more than 1.2 million births in Denmark between 1997 and 2016 identified an association between paternal metformin exposure during the period of spermatogenesis and genital–urinary malformations in male infants, as reflected in an elevated adjusted odds ratio with a relatively narrow 95% CI (adjusted odds ratio, 1.40 [95% CI, 1.08 to 1.82]). The authors found no evidence of an association for paternal use of metformin outside the sensitive window of spermatogenesis or among unexposed siblings. Moreover, no association was seen for other diabetic pharmacologic agents used by either partner. Of note, the study did not examine maternal use of diabetes drugs because it excluded offspring of mothers with diabetes. These interesting and alarming findings are further strengthened by extensive sensitivity analyses, including attention to the well-known clustering (repetition) of adverse pregnancy outcomes among other analytic permutations. Moreover, reliance on registry data prompted the authors to empirically assess a possible unmeasured confounder that may account for the findings and concluded it was unlikely in light of the large adjusted odds ratio of 2.7 or greater required to attenuate findings. Another key finding is the reversal in the secondary sex ratio as reflected in a female excess (51%) associated with paternal metformin use rather than the typical male excess of live births. Are these findings plausible? Available evidence supports a longstanding association between maternal diabetes and birth defects, and diabetes-related sequalae for male reproductive health, possibly through hypogonadism and impaired spermatogenesis. Serum lipids have been associated with diminished semen quality (3), which in turn has been associated with genital–urinary malformations in male offspring (4). Bicephalic sperm also have been associated with an altered secondary sex ratio favoring an excess of female births (5). These data suggest that altered testosterone levels may be an underlying mechanism raising concern about the antiandrogenic activity of oral diabetes pharmacologic agents, including metformin (6). Given the prevalence of metformin use as first-line therapy for type 2 diabetes, corroboration of these findings is urgently needed. Meanwhile, clinical guidance is needed to help couples planning pregnancy weigh the risks and benefits of paternal metformin use relative to other medications. Important in this guidance will be communicating that the adverse relationship was specific to metformin during the period of spermatogenesis. These novel findings should prompt targeted basic and causal inference research to help delineate underlying mechanisms and to differentiate drug effect from alterations in metabolic state or other confounders. A key limitation of this study noted by the authors is the lack of data on men's adherence to prescribed medications and glycemic control. Both are important considerations when assessing the cause of genital–urinary malformations relative to pharmacologic agents. Continued efforts to promote healthy lifestyles are also important and reported to be associatedwith a lower incidence of diabetes and greater weight loss relative to medical management in cohorts (7). With overweight and obesity rates remaining high for many populations, it is not surprising to find metformin use being reported by reproductive-aged couples (8), including those trying for pregnancy in relatively small preconception cohorts (9). If corroborated, metformin would not be the first medication to pose iatrogenic risk or be recognized as a human teratogen. However, its attributable risk could be large given its prevalent use. Wensink and colleagues' work supports a global call for more conclusive study of the potential risks for paternal use of diabetes drugs to offspring. The efficacy and safety of pharmacologic agents for future generations deserve nothing less.