Maternal Diabetes and Infant Sex Ratio

Abstract

Purpose of review:

Evolutionary hypotheses on the ratio of males to females at birth posit that women terminate pregnancies with low likelihood of surviving and producing grandchildren. Thus, females are preferred to males under unfavorable conditions. Much of this literature has focused on catastrophic disruptions that induce maternal stress and result in fewer males. Diabetes may similarly affect the sex ratio.

Recent findings:

A male bias at birth among infants born to women with GDM is widely recognized; mild hyperglycemia experienced early in pregnancy may signal favorable conditions and warrant investment in males. There are sparse data on women with pregestational diabetes, but some evidence for a female bias born to those with type 1 diabetes and severe hyperglycemia (i.e., requiring insulin). Disease-related maternal stress in these women may led to the selective termination of male fetuses.

Summary:

Further examination of pregestational diabetes stands to contribute to scientific understanding of the sex ratio.

Introduction

Scientific inquiry into the human secondary sex ratio, or the ratio of males to females at birth, dates back to at least 1662 (1) when John Graunt observed an excess of male over female births, and noted spatial and temporal variation in the secondary sex ratio. There has been a great deal of inquiry into the subject since that time, with scholarly contributions from the fields of demography, medicine, sociology, and economics, among others. Contemporary scholars continue to document a male excess at birth, and suggest that the secondary sex ratio oscillations are due to both endogenous (i.e., natural variation) and exogenous effects (i.e., population-level shocks) (2). Accumulating evidence points to dips in the secondary sex ratio (i.e., fewer liveborn males) following exposure to exogenous effects, or major population-level disruptions that negatively affect maternal condition. As such, it has recently been suggested that the secondary sex ratio may be a useful indicator of public health (3, 4).

Over the last decade, scientific understanding of events antecedent to birth has increased and sophisticated analyses testing evolutionary hypotheses pertaining to the secondary sex ratio have been published. This review summaries the literature in this area overall, focusing on the relationship between maternal diabetes and sex ratio. Maternal diabetes has increased in recent decades and affects the maternal condition (5, 6). Please note this review takes an evolutionary perspective and thus focuses on environmental and biological factors at the population-level (e.g., catastrophic events, disease states), as opposed to the many cultural, social and societal factors influencing preferences for and selection of one sex over the other (7–11).

The Secondary Sex Ratio

It has long been recognized that males outnumber females at birth (1), with the ratio varying, very slightly, around 1.04 (12). Across societies and time periods, males are more likely to die than their female counterparts of the same age (13), and male infants in particular are more likely to die than any other sex/age group through reproductive age (14). The literature hypothesizes that natural selection would maintain biological mechanisms whereby women would terminate pregnancies requiring too great a maternal investment relative to the offspring’s ability to survive to reproductive age and produce grandchildren (15). The frailty of males, especially male infants, suggests that such a mechanism would affect males more than females (14), since from an evolutionary perspective, males are the riskier investment (13, 14). Approximately 73% of conceptions do not survive the first 6 weeks gestation (16), including approximately 22% of implantations that end before pregnancy is clinically detectable (17). Amon Avalos et al. reported that, among gestations surviving to 5 weeks, the overall cumulative risk of miscarriage through 20 weeks is 11–22%, with substantial variation in estimates of miscarriage risk prior to 14 weeks (18). It is certainly plausible that evolutionary based mechanisms are driving some portion of these fetal losses, and that they are sex-specific and based upon the maternal condition to maximize future reproductive success.

Unexpected, catastrophic disruptions at the population level, or exogenous shocks (2), have been linked to dips in the secondary sex ratio (i.e., fewer liveborn males) in numerous settings and periods. These extreme population level stressors are perceived to be an environmental threat to the offspring’s’ survival and frailer male fetuses are thus selectively aborted in order to maximize future reproductive success (i.e., yield of grandchildren). The literature documents dips in the secondary sex ratio associated with: self-assessed early pregnancy stress (19), anxiety disorders (20), maternal occupational stress (21), a stressful ‘lifestyle’ as identified by commute time ≥ 90 min (22), severe pre-conceptional life events (23), terrorist attacks (24–27), floods and heavy fog (28), earthquakes in Japan (29), Iran (30), and Chile (31), economic downturns (32, 33), and mass layoffs (34). The unexpected outcome of the 2016 U.S. presidential election, hypothesized to be a societal stressor for liberal-leaning regions of Canada, preceded a reduction in the secondary sex ratio, the lowest occurring 4 months after the election (35). Conversely, and less commonly explored in the literature, are data suggesting increases in the secondary sex ratio (i.e., more liveborn males) following periods of public elation [i.e., the birth of Prince William (36), the 2010 FIFA World Cup (37), and the Years of the Dragon (38), which are considered more propitious for the birth of a child].

Purported Mechanisms

There are two popular hypotheses outlining potential mechanisms underlying variation in the secondary sex ratio from exogenous effects. The first, by James (3, 39–41), suggests that male conceptions and births are favored overall at the population level but particularly under circumstances of good parental and societal health, whereas female biased conceptions and births indicative of poor parental and societal health. James posits that the hormone levels of one or both parents around the time of conception serve as the indicator of health, but acknowledges that the mechanisms by which parental hormones exert their effects on the sex ratio at this early stage are unknown (3). James points out that maternal androgens are increased among stressed women, and that high androgen levels in pregnancy have been associated with fetal growth restriction and spontaneous abortion (40). Experimental evidence in pregnant ewes (i.e., administration of testosterone having similar effects) is also cited (40). Overall, James suggests that more males are conceived, miscarried, and survive to birth (41), with variation in these attributable to parental hormones at conception. James maintains that high levels of maternal testosterone and/or estrogen play a role in fetal loss (3, 40, 41), and may result from a variety of stressors, including maternal disease, and thus potentially, maternal diabetes.

Catalano and colleagues provide an alternative hypothesis, which favors the idea of selection in utero driving variation in the secondary sex ratio. They used interrupted time-series analyses to examine the secondary sex ratio following the September 11 attacks [in California (26) and New York City (27)] and reported a prominent dip in December 2001 (26) and January 2002 (27), i.e., among those in utero at the time of the attacks. Notably, the secondary sex ratio did not differ from expected values in the 9, 10 or 11 months following the attacks (26). These findings and others’ (42) support their hypothesis that, with exposure to an acute stressor, frail male fetuses are ‘culled’ after conception, leading to dips in the secondary sex ratio.

Additional evidence comes from analyses of population-level stressors and the sex ratio of subsequent fetal deaths (26) and very low birthweight (26), low birth weight (43), and preterm births (43), indicators of decreased survival and thus low likelihood of future reproductive success. Following the September 11^th attacks, the fetal death sex ratio increased (i.e., more male fetal deaths) (44), and the sex ratio among very low birthweight births decreased (i.e., fewer very low birthweight males were born), further support for the idea that ‘weak’ males had been selectively culled (26). Similarly, in the months following the November 2015 Paris attacks, the secondary sex ratio fell, and more preterm births were observed among males, but not females (45). Analyses of the rate of small for gestational age (SGA) among 4.9 million male term births in California also found that SGA was less frequent following labor market contractions (46).

Overall, this work suggests that exogenous stressors alter the secondary sex ratios primarily by affecting which fetuses survive the course of gestation, not impacting the sex ratio at conception. Catalano proposes a maternal mechanism that ranks gestations by expected yield of grandchildren and then applies a variable threshold below which a woman would spontaneously terminate the pregnancy (14, 47–49). They posit that the threshold for termination would be dependent on both the woman’s environment, a proxy for the offspring’s future survival, and the sex of the fetus. Therefore, exposure to poor environments would result in the culling of males after conception, particularly weak or frail males, because they are less likely than females and strong males to survive and achieve reproductive success (41, 46, 50, 51). Using individual-level life-history data from historical church records in Finland (i.e., ~8,000 males born 1790–1870), Bruckner et al. observed decreases in male infant mortality risk following declines in the secondary sex ratio: weak males had already been culled in utero from the cohorts favoring fit males (i.e., those resulting in lower secondary sex ratios); interestingly, the males in those cohorts were subsequently slightly more likely to survive infancy (52).

Gestational human chorionic gonadotropin (hCG) has been proposed as one potential means by which offspring ‘hardiness’, i.e., potential for survival and future reproductive success, could be detected by the mother (14, 47). Low levels of hCG predict spontaneous abortion, and male gestations yield endemically lower hCG levels than females from the 4^th week onward (53). In a sample of almost 2 million conceived in California in 2001–2007, Catalano et al. examined prenatal screening test data from the birth registry and found that women carrying males in conception cohorts subjected to contracting economies exhibited higher mid-pregnancy hCG levels (54). Catalano et al. also examined 1.56 million gestations conceived in California in 2002–2007 and found elevated median mid-pregnancy hCG levels among women carrying males in the conception cohorts with the fewest males, supporting the idea of a raised threshold for termination with exposure to ambient maternal stress (i.e., termination of male conceptions which would have warranted maternal investment under more favorable conditions) (14, 47).

The Primary Sex Ratio

While scholars agree that variation in the secondary sex ratio is driven, at least in part, by the preferential loss of males in utero, debate pertaining to the primary sex ratio, or the sex ratio at conception, is on-going (26, 41, 42, 55). In 2015, Orzack et al. (55) assembled data from multiple sources in an effort to construct a comprehensive examination of the primary sex ratio and the trajectory of the sex ratio over the course of gestation. The authors used amniocentesis, chorionic villus sampling (CVS), and induced abortion data to define fetal sex, in addition to inferring fetal sex from 3- to 6-day old embryos derived from assisted reproductive technology (ART) procedures and U.S. census records of fetal deaths and livebirths. Their results suggest that the sex ratio at conception is 1:1 (i.e., unbiased) but then varies over the course of gestation. A female-bias develops within the first 1–2 weeks following conception due to higher male mortality, but the proportion of males then increases during the rest of first trimester and into the second as females are preferentially lost (55). There is a leveling off and then the proportion of males decreases as they are preferentially lost in the third trimester (55). Overall, the female losses exceed male losses, resulting a secondary sex ratio that is biased towards males (55).

A decade before the Orzack et al. publication (55), Boklage similarly suggested that the primary sex ratio was unbiased (42). Reports of 1:1 ratios for X versus Y chromosomes in sperm, or a small excess of X-bearing sperm over Y-bearing, were provided as indirect evidence against the primary sex ratio being biased towards males. It was proposed that bias in the secondary sex ratio arises from the preferential loss of females between fertilization and clinical recognition of pregnancy (i.e., during embryogenesis). Boklage posited that population stressors impact the secondary sex ratio by affecting embryogenesis, at least in part via epigenetic pathways (e.g., genomic imprinting), thereby creating an excess of males during the transition from embryogenesis to clinically recognized pregnancy (42).

Maternal Diabetes, Hyperglycemia, and the Secondary Sex Ratio

In 2012, we estimated and compared the secondary sex ratio across categories of maternal hyperglycemia among 288,009 mother-infant pairs delivering at Kaiser Permanente Northern California in 1996–2008. We observed that women with pregestational diabetes delivered the fewest males [sex ratio (SR)= 1.01], followed by those with normoglycemic pregnancies (SR= 1.05), and that women with gestational diabetes mellitus (GDM) delivered the most males (SR= 1.07) (56). Although the adjusted odds ratio estimates did not attain statistical significance, these results are consistent with the hypothesis that women spontaneously abort fetuses who were least likely to survive to reproductive age and produce grandchildren (15). From the standpoint of future reproductive success, pregestational diabetes is unfavorable, deficient metabolic regulation resulting in hypoglycemia and/or hyperglycemia associated with oxidative and metabolic stress. As the theory goes, women with pregestational diabetes would optimize their future reproductive success by ‘culling’ week males and/or investing in the maintenance of fit female fetuses, and thus ultimately exhibit a lower secondary sex ratio (i.e., female-biased).

In the Kaiser Permanente Northern California data (56), further examination of the secondary sex ratio in those with pregestational diabetes revealed effect modification by type. Women with pregestational type 1 diabetes exhibited a female-bias sex ratio (SR= .87), while those with pregestational type 2 displayed a male-bias sex ratio (SR= 1.05) which was identical to that observed among women with normoglycemic pregnancies (56). All women in this study were members of a large group practice, prepaid health plan, and thus received monitoring and treatment for their disease, but it is reasonable to expect that those with type 1 diabetes entered their pregnancies with a higher burden of disease and were exposed to more stress, overall, than women with pregestational type 2 diabetes.

There are few studies comparing the secondary sex ratio between women with pregestational type 1 versus type 2 diabetes, though differences in the causes of pregnancy loss have been reported by diabetes type (57). Similar to the Kaiser Permanente Northern California data, Rjasanowski et al. (58) reported that more female than male offspring were born to women with severe hyperglycemia (i.e., requiring insulin) (SR= 0.45). In a study by James (59), women with type 1 diabetes exhibited a female-bias (SR= 0.92), and those with type 2 diabetes a male-bias (SR= 1.39). Interestingly, in Catalano et al.’s examination of conception cohorts in California, an increase in the percentage of mothers with severe hyperglycemia was significantly associated with an increase in median hCG levels among male gestations (i.e., ‘survivors’) in the cohort (14). This finding suggests that the threshold for spontaneous termination proposed by Catalano et al. may be elevated in pregnancies ‘stressed’ by severe hyperglycemia, and that these women thus select only the most fit males for maternal investment to term.

Unlike pregestational diabetes, GDM is most commonly diagnosed between 24 to 28 weeks gestation (60). The condition is characterized by the increased availability of metabolic substrates which results in fetal overnutrition (56). Importantly, this excess availability of metabolic substrates increases as the pregnancy progresses. Thus, early in the pregnancy, when the majority of fetal losses occur, those who go on to develop GDM are likely to exhibit increased availability of metabolic substrates, but perhaps not at a level high enough to constitute overt disease and induce systemic stress (56). This low grade excess of metabolic substrates may in fact signal increased maternal ability to invest in and sustain ‘weak’ male fetuses and male fetuses overall, thereby resulting in a higher secondary sex ratio (i.e., more boys) in this group (56).

A 2015 meta-analysis (61) of pooled data from more than 2.4 million women reported a statistically significant increased risk of GDM among those who went on to deliver a male infant, leading the authors to conclude that the sex of the fetus may impact maternal glucose metabolism during pregnancy. The authors had previously examined data from a Canadian population-based administrative database and found that carrying a male fetus was associated with impaired maternal beta cell function (62), and thus suggested beta cell function as the potential mechanism through which fetal sex may lead to GDM. However, there remains the possibility for another upstream variable or variables associated with both male fetal sex bias (i.e., at or soon after conception) and the subsequent development of GDM. Indeed, the authors of the meta-analysis recognized that their findings could also be explained by spontaneous termination of female fetuses due to impaired pre-pregnancy beta cell function (61). Currently, the mechanisms underlying beta cell dysfunction in GDM pregnancy are largely unknown (63).

Retnakaran et al. also provided evidence that, among women with GDM, those carrying a girl have a slightly higher future risk for type 2 diabetes (64, 65). If those carrying a girl are also more likely to have ‘culled’ a weak male or male fetus overall, perhaps due to more severe hyperglycemia at conception and in early pregnancy, then the theory of Catalano et al. (14, 47–49) would still hold. Across the spectrum of GDM severity, those with more severe disease or unrecognized type 2 diabetes (56) could have experienced some degree of metabolic dysregulation early in their pregnancy, thereby resulting in a female-bias in this sub-group of GDM women (e.g., those who progress to type 2 diabetes within 3 years postpartum).

Another Retnakaran et al. study (66) reported that twin pregnancies had increased risk of GDM, but affected twin pregnancies had a lower risk of postpartum progression to type 2 diabetes as compared to GDM women carrying singletons. They concluded that the impact of carrying twins on maternal glucose metabolism supplanted that of fetal sex (66); indeed, placental adaptations needed to support multiple fetuses are likely involved (63). Interestingly, in the Retnakaran et al. study (66), which examined more than 13,000 twin pregnancies, male/male twins had a reduced risk of GDM and male/female twins a slightly increased risk as compared to female/female twins, though neither estimate attained statistical significance.

As proposed by Catalano et al., the maternal mechanism that ranks gestations by expected yield of grandchildren (and then applies a variable threshold below which a woman would spontaneously terminate the pregnancy) would include twins and be ordered, by preference: 1) female/female twins, 2) female singleton, 3) male singleton, 4) male/male twins (49). Maternal stress raising the criterion for spontaneous termination on such a scale would cull male twins before male singletons, and cull female singletons before female twins. Indeed, in their examination of 17 years of birth data from the State of California, Catalano et al. found support for this hypothesis (49). They later also reported that twins among male Norwegian infants fell below expected, while twins among females rose above expected levels, among those gestation in July 2011 when 77 Norwegians were murdered (48). Unfortunately, biologically induced increases in maternal hyperglycemia to maintain and support multiple fetuses complicates examination of twin sex ratio theory in the context of maternal diabetes.

Glucose’s Effects on Early Development

Evidence from animal models suggests that the secondary sex ratio deviations attributed to maternal diabetes may be related to glucose’s impact on early embryonic development. In bovine embryos, brief exposure to glucose after fertilization skews development toward males (67). Time-lapse video recordings of bovine embryo development in vitro suggest that, in the presence of glucose, male embryos cleave earlier, whereas females cleave are earlier in the absence of glucose (67). In the presence of glucose, more male than female embryos reach the morula and blastocyst stage, but there is no sex difference in this progression in the absence of glucose (67). In vitro exposure of bovine blastocysts to excess glucose-containing medium thus appears to selectively block females at the morula to blastocyst transition (56, 67–69). By impairing female bovine embryonic development, high glucose levels may increase the sex ratio (i.e., male-bias), and metabolites generated by increased pentose phosphate pathway activity among females have been implicated in their higher rate of early developmental failure (70).

Environmental disturbances to the in vitro embryo culture can influence gene expression patterns regulated by X-chromosome dosage, thereby differentially effecting male and female embryos (67). In bovine embryos, the X-chromosome carries the gene glucose-6-phosphate dehydrogenase (G6PDH), which encodes the enzyme responsible for the rate-limiting step of the pentose phosphate pathway. Female intolerance to excess glucose may reflect the presence of two active X-chromosomes in female blastocysts, presumably through to the morula stage, when random inactivation of one X-chromosome is initiated (71, 72). Over activation of the pentose phosphate pathway may thus be responsible for the higher failure rate of female versus male embryos in the presence of glucose (71). The synthesis of interferon-tau (IFNT), a biomarker of embryo viability that is natural higher in female bovine blastocysts, is hypothesized to be associated with the pentose phosphate pathway, as inhibiting this pathway was observed to eliminate sex differences in IFNT (71). Conditions, such as excess glucose, which compromise development may reduce IFNT production, thereby disrupting, “an embryo’s normal dialogue with the maternal reproductive tract during the preimplantation period” (71).

Acute Maternal Metabolic Disruption and the Sex Ratio: Ramadan as a Natural Experiment

Only about half of all pregnancies in the U.S. are intended (73), making it difficult to conduct generalizable, population-based studies examining maternal exposures at the time of or soon after conception and the secondary sex ratio. However, studies examining the Islamic holy month of Ramadan allow for exploration of the impact of an acute and specifically timed maternal metabolic stress on the secondary sex ratio. Ramadan follows a lunar calendar, and its month-long duration overlaps with more than 75% of pregnancies at a variety of stages (74). Almond and Mazumder (74) examined natality data from Michigan, which has a large Muslim population, and reported a reduction in the secondary sex ratio in mothers who were likely to have fasted in the month prior to conception and at conception. It is important to note that pregnant women may be exempt from fasting and may thus be less likely to fast than those women who have not yet recognized their pregnancy; this could explain the lack of significant findings for Ramadan exposure later in gestation (74). Among Indonesian Muslims exposed to Ramadan in utero, a lower proportion of males has also been reported, most likely but not definitively due to increased male death prior to birth (75).

Conclusions and Future Research

The male bias among infants born to women with GDM is now widely recognized. The male bias in GDM may also be consistent with the hypotheses of Catalano et al. (14, 26, 27, 47–49). Specifically, mild hyperglycemia at and soon after conception in these women may signal favorable conditions and warrant maternal investment in male fetuses, including frail males who would have otherwise been culled. Although data on women with pregestational diabetes are sparse, it seems there is a female bias among infants born to women with type 1 diabetes and severe hyperglycemia (i.e., who require insulin). Perhaps a consequence of disease-related maternal stress, these women may be more likely to cull male fetuses. Women with type 2 diabetes appear to display a secondary sex ratio comparable to normoglycemic women or slightly biased towards males, potentially reflecting a less severe state of maternal disease at and soon after conception.

Pregestational diabetes is linked to an increased risk of adverse pregnancy outcomes, but the prevalence of pregestational diabetes has increased substantially in recent decades (6). This is most likely due to the increasing incidence of diabetes among women of reproductive age, as well as improvements in preconception and prenatal care (6). Women with recognized diabetes are encouraged to obtain preconception counseling and achieve adequate glycemic control prior to pregnancy to reduce the risk of adverse outcomes. With the proliferation of continuous glucose monitoring devices (CGM) in recent years, detailed data on maternal glucose values around conception and over the course of pregnancy will soon be available to contribute to this field. However, existing data, such as the prenatal screening test data from the State of California previously analyzed by Catalano et al. (54), may offer more clues. Such data could be used to examine hCG levels among women with severe hyperglycemia over time. If more of these women have been achieving and maintaining pregnancy, as the data suggest (6), women with severe hyperglycemia carrying males in more contemporary conception cohorts would be expected to exhibit lower median hCG levels than those carrying males in earlier conception cohorts, and secondary sex ratio deviations (i.e., the female bias) would similarly dampen over time. Improvements in disease management have undoubtably improved the maternal condition among women with severe hyperglycemia (i.e., requiring insulin). Further examination of this group in particular stands to contribute to scientific understanding of the human sex ratio.