Sex and Gender Differences in Risk, Pathophysiology and Complications of Type 2 Diabetes Mellitus

A. Body mass index (BMI)

Important physiological and pathophysiological sex differences of anthropometric, metabolic, and endocrine parameters are summarized in Figure 3. A short overview of the most interesting risk factors and markers are presented in Table 1.

Across the age range, European men are usually diagnosed with diabetes at an earlier age (Figure 2B) and at lower BMI than women, with the most prominent sex difference being at younger age (13). In Sweden, time trends revealed that the male predominance in 1940, with a male to female ratio up to 1.4 in the ages 10–55 years, increased and expanded over time especially in the age group 45–65 years reaching a ratio of 2 (14). Men were diagnosed 3–4 years earlier and at a BMI 1–3 kg/m² lower. This trend was partly explained by an increase of automation and decrease of physical work particularly in men. Diabetic women, on the other hand, are more obese than diabetic men in most studies and show a stronger association between increase of BMI and diabetes risk, despite similar curvilinear associations between increasing BMI and diabetes risk in both sexes (15). Sex differences in body composition and fat deposition clearly contribute to sex-dimorphic diabetes risk (16). BMI overestimates body fat mass in men, who generally have more fat-free muscle compared with women.

B. Body fat distribution

During puberty, increased accumulation of gluteo-femoral fat promoted by estrogen results in a “gynoid shape” of premenopausal women (Figures 3 and 4). Males feature a greater trunk and visceral fat (VAT), upper extremity mass, and liver fat compared with females with same age and BMI (16, 17). Nonetheless, men and women with similar degree of insulin resistance show comparable intraabdominal and liver fat (18). In an Asian population, women with normal waist circumference (WCR) and BMI were diagnosed with visceral obesity by computer tomography. This even showed greater cardiometabolic risk in women, in terms of glucose and lipid abnormalities compared with males (19). However, VAT and age were independent predictors of greater cardiometabolic risk in males, whereas the VAT to subcutaneous fat (SAT) ratio independently predicted higher risk in females.

In general, men not only featured larger amounts of VAT for any degree of total body fat but also higher levels of fatty acids (FAs) turnover with higher rates of lipolysis and lipogenesis in VAT compared with women (20). Women, instead, have higher rates of FA uptake in leg fat tissue and lower rates of release in gluteal and femoral regions. Also females expressed higher lipogenetic rates from SAT compared with males. Increased leg adiposity was found to be associated with a decreased cardiometabolic risk especially in women, whereas higher trunk adiposity is generally related to clustering of cardiometabolic risk factors in cross-sectional population-based studies (21, 22). Aging and in particular menopause transition, with loss of estrogen production, is associated with changes in body shape and a preferential increase of abdominal fat in women shifting to the android “visceral adiposity” (23).

In line with this, women have a more prominent increase of WCR with increasing age than men. The relationship between WCR and intraabdominal fat mass is stronger for intraabdominal SAT in younger women than men; but in menopausal women, the associations become more similar to the male patterns in cross-sectional analysis testing for sex and age differences (24). In British elder subjects waist was the best predictor of diabetes in women, whereas in males the predictive value of BMI and waist were comparable (25). These results are confirmed by data from various other cohorts from different countries (26, 27) and further expanded by trajectories of anthropometric parameters. In pooled analysis of 2 prospective population-based cohort studies, German women who gained 1 cm of their WCR had an increased risk for incident diabetes of 31% per year, compared with 28% if they gained 1-kg body weight (28). In men, the corresponding increase of risk for incident diabetes was 29% and 34%.

C. Brown adipose tissue (BAT)

Sex differences are described regarding mass and activity of BAT in adults (Figure 3), which was recently supposed to impact whole-body energy metabolism, insulin resistance, and obesity-related T2DM. Women have much higher prevalence and activity of BAT, which was related independently and inversely to age in both sexes, but only to BMI in males and only to VAT in females, in a large population-based study (29). In mice, expression of factors involved in BAT activity, like fibroblast growth factor families, was positively regulated by the presence of ovaries and estrogens (30). BAT transplantation reversed obesity, increased adiponectin, and reduced insulin resistance and liver steatosis in leptin-deficient animals (31). Therefore, overall higher impact of BAT could also contribute to lower diabetes risk in women.

D. Metabolic syndrome (MetS)

Clustering of traditional metabolic risk factors associated with insulin resistance, often termed MetS, disregard risk factors like age, sex, family history, SES, and lifestyle. Recent analysis of National Health and Nutrition Examination Survey data show comparable prevalence in both sexes with greatest increase in young women (32). Diabetes appears to diminish the in general more favorable cluster of risk factors of females compared with males, leading to greater differences in central adiposity and risk factors related to coagulation and inflammation between diabetes and nondiabetes in women rather than in men (33). Clustering of risk factors varies between sexes and ethnicities, but abdominal obesity and increased WCR as surrogate markers seem to be the dominant factors in women (34–36). Overall, adjustment of risk factors for MetS appears to have greater impact on women's CVD risk. There are several diabetes risk scores, including sex in risk calculations together with various other risk parameters, but only a few include WCR or social factors like social deprivation (37, 38).

As recently shown in a collaborative analysis of 10 large cohort studies, women appear to feature more often the metabolically healthy obese phenotype with normoglycemia and without dyslipidemia and hypertension (7%–28%) compared with males (2%–19%) (39). As demonstrated by a recent metaanalysis of prospective cohort studies even obese men and women with normal cardiometabolic clustering had a 4-fold higher relative risk of developing T2DM, although this risk was only half of that of metabolically unhealthy obese patients regardless of sex differences in the progression toward T2DM (40).

E. Adipokines

Sexual dimorphism is also evident in the expression and the predictive value of some fat-related biomarkers (16, 41, 42). Leptin is important in the regulation of satiety, food intake, and energy expenditure. It also influences the insulin glucose axis as well as peripheral insulin resistance (43). Similarly, adiponectin has manifold effects on lipid and glucose metabolism and increases insulin sensitivity in target organs. Dysregulation of adiponectin action is relevant in the development of T2DM (44).

Women show an up-regulation of expression of adiponectin and its receptor in abdominal adipose tissue, possibly contributing to their lower cardiometabolic risk. In general, metaanalyses have shown that women have higher leptin and adiponectin levels than men of comparable age and BMI, which may be related to their sexual hormones (41, 45). In several longitudinal studies, increased plasma leptin, which mirrors body fat mass and is strongly associated to SAT, relates to increased diabetes risk in males (37). On the other hand, an inverse correlation between plasma adiponectin levels and insulin sensitivity is seen in obese and diabetic subjects, which tends to be somewhat more pronounced in women (45–47). In addition, androgens may decrease adiponectin secretion. However, it is still unclear whether hypoadiponectinemia is a cause or a consequence of insulin resistance or hyperinsulinemia (37).

F. New biomarkers

There are also a number of new risk factors reported with sexual dimorphism, such as the hepatokine fetuin A, which was shown to be related to T2DM onset only in women in the Rancho Bernardo Study (42). In the Prevention of Renal and Vascular Endstage Disease Study, copeptin, the C-terminal portion of the precursor of vasopressin and reliable marker of arginine vasopressin secretion, was shown to be associated with the risk of future diabetes in women but not in men (48). Inclusion of copeptin in risk models based on traditional risk factors was of additive value in predicting diabetes in women. This may point to a closer link between arginine vasopressin stress adaptation system and pathogenesis of T2DM in women. The development and validation of new risk scores with sex-specific weighting of risk factors could be a promising tool for future prediction models.

Another novel biomarker is proneurotensin, the precursor molecule of neurotensin, which is peripherally released from the endocrine-like N-cells of the small intestine after fat intake (49). It acts as neurotransmitter in the central nervous system but behaves as a hormone in the periphery, stimulating pancreatic and biliary secretion, inhibiting gastric motility, and facilitating FA translocation. Fasting proneurotensin plasma levels are usually lower in women than men but predict incident diabetes and CVD as well as total and cardiovascular mortality in women and, however, not in men (50). Each standard deviation increase of baseline proneurotensin was associated with an increased risk of 41% for new-onset diabetes in women during the follow-up of 13 years.

In a cross-sectional population-specific study, low 25(OH) vitamin D3 was found in middle-aged Caucasians independently associated with T2DM in women but not in men (51). A significant interaction between sex and vitamin D was found before sex-stratified analysis. The probability of having a newly diagnosed or known diabetes more than doubled in women with levels below a cut-off of 15 ng/mL. In men, seasonally adjusted values only marginally predicted T2DM. In a previous metaanalysis, an inverse association between vitamin D and diabetes was confirmed in both men and women (52). However, some differences between the studies could be explained by ethnicity and age as sex hormones, particularly 17α-estradiol, may influence these associations with variations over time (53). In fact, in the Korean population low levels were related to increased diabetes prevalence in younger women and older men over 50 years. Vitamin D may also directly stimulate the expression of the insulin receptor, thereby improving glucose transport in human cells (54).

In a large community-based prospective cohort study, increased liver enzymes (alanine aminotransferase, aspartate aminotransferase, and γ-glutamyl transferase [GGT]) preceded the incidence of T2DM in both sexes (55). The strongest association with incident T2DM was seen for GGT. This could be explained by the fact that GGT is more closely related to fatty liver, oxidative stress, and thus to insulin resistance compared with the other enzymes (55). The independent association between liver enzymes and diabetes risk was continuously extending in the normal range, hence it remained significant by use of sex-specific quartiles and showed no significant sex interaction overall. Women usually have lower levels and lower liver fat than men, of comparable BMI and age, and appear to be protected by estrogen at premenopausal age (56). Although overall males have a higher prevalence of increased liver fat, a marked rise is described in elder women (57). In a historical cohort of the Brisighella Heart Study, the fatty liver index, including liver enzymes, triglycerides, WCR, and BMI, was even a better predictor of the MetS in women than in men (58). Furthermore, a sex-specific association between liver transaminase levels and insulin sensitivity was described (59). Alanine aminotransferase independently predicted muscle glucose uptake measured by hyperinsulinemic euglycemic clamp in females only, whereas in males, fasting insulin and leptin were stronger predictors of insulin resistance.

Additionally, low sex hormone-binding globulin (SHBG) levels may indicate diabetes-risk potentially mediated via SHBG gene polymorphisms (60, 61). In general, women tend to have higher SHBG levels than men and low SHBG concentrations may be associated with even higher diabetes risk in women compared with men. In the Diabetes Prevention Program, SHBG and SHBG-single nucleotide polymorphisms did not predict incident diabetes in any sex, but diabetes incidence was directly associated with estradiol and estron and inversely with testosterone in men (62). Although not directly evaluated in this study, the association between circulating estrogen and diabetes risk could be attributed to systemic estrogen resistance in men (63). However, in this study, sex steroids did not relate to diabetes risk in women (62). The authors conclude that, although SHBG may be able to predict diabetes in unselected populations, in high-risk groups, elevated glucose and weight are more potent indicators of development of diabetes. However, in a large population-based sample, an independent inverse relationship was proven between SHBG and MetS, as well as incident T2DM, especially among postmenopausal women (64).

G. Imbalance of sex hormones

Cardiometabolic similarities were described among women with androgen excess and men with androgen deficiency. The balanced proportion between estrogens and androgens plays an important role in maintenance of energy metabolism, body composition, and sexual function. Also the bidirectional modulation of glucose and lipid homeostasis by sex hormones and their receptor activation in central and peripheral targets in both sexes are influenced by estrogens and androgens (16, 17, 65–69). In women, higher levels of androgens lead to increased body weight and VAT; this is also seen in female to male transsexuals (70). Overall, relatively higher testosterone levels in women and lower levels in men relate to incident diabetes (71).

The polycystic ovary syndrome (PCOS) describes a female-specific state of androgen excess and hyperinsulinemia related to obesity, T2DM, and higher cardiometabolic risk (72, 73). An influence of genetic aspects is supported by higher prevalence of metabolic disorders in both male and female. First-degree relatives of women with PCOS and impaired glucose metabolism (IGM) among men is strongly mediated by obesity (74). In addition, a sex difference in the parental metabolic phenotype was reported referring to fathers, which feature a higher risk of fasting dysglycemia and evidence for pancreatic β-cell secretory defects, when compared with mothers of women with PCOS. Nevertheless, only maternal heritability exerted a significant impact on the prevalence of fasting dysglycemia in these women (75).

Obese or diabetic males feature a 2- to 4-fold higher rate of late-onset hypogonadism with low testosterone levels and higher prevalence of erectile dysfunction (76, 77). Overweight/obese males showed accelerated aromatization of androgens to estrogens, inhibiting gonadotropin secretion by activation of estrogen receptors (ERs) of the hypothalamus that promote hypogonadism (78). Aromatization of testosterone to 17β-estradiol impacts energy homeostasis. A higher testosterone-estrogen ratio can promote visceral obesity in males, but androgen deficiency itself associates with increased VAT. Whether testosterone deficiency itself causes metabolic derangement or testosterone levels are decreased due to aging, changes of body composition, or illness (reverse causality) is not yet fully understood and needs further clarification (79, 80). However, testosterone replacement therapy can improve insulin sensitivity and hyperglycemia in hypogonadal diabetic males (81).

H. Prediabetes

The prevalence of prediabetic categories differ between sexes (Figure 2B) giving rise to clinical implications: men more often develop impaired fasting glucose (IFG), whereas women more often show IGT (Figure 2B). IFG is characterized by increased hepatic glucose output and impaired early insulin secretion, whereas IGT is primarily due to peripheral insulin resistance (82). IGT may better predict progression to diabetes and mortality risk relates more strongly to an increased cardiovascular risk. This fact may explain why World Health Organization criteria, including IGT status may be superior to other definitions of MetS in prediction of diabetes and CVD in women (83, 84). It further highlights the importance of performing oral glucose tolerance tests to screen for IGT, especially in women.

I. Gestational Diabetes Mellitus (GDM)

GDM is a heterogeneous entity were mostly insulin-resistant overweight/obese women are affected. It serves as an independent and strong female risk factor for eventual progression of T2DM (85). Nonetheless, normal weight women may also be susceptible to gestational diabetes (GDM) due to genetic traits, along with physiologically increasing insulin resistance during the course of pregnancy. Although intervention strategies might be an effective approach to reduce progression to T2DM, women with a history of GDM face a more than 70% higher incidence than prediabetic women do (86). Throughout literature, GDM is associated with several adverse pregnancy outcomes affecting not only mothers but also their offspring in a sex-specific way (87, 88). Recent studies report that pregnant women carrying a male fetus have higher risk for developing GDM (see section V.D) (89–91).

A. Socioeconomic status

SES, assessed by educational level, position, and income, is inversely associated with prevalence of obesity and T2DM in developed countries. Steeper gradients among women can be observed in a national population health survey in Canada (94). This study found persisting associations between low education and income and self-reported diabetes after controlling for obesity and physical activity in women. Consistently, a population-based European survey, the Kooperative Gesundheitsforschung in der Region Augsburg (KORA) study (95) found stronger associations between SES indicators, abdominal obesity, and physical activity in women. Additionally, a strong inverse association between occupation and newly detected diabetes was presented only in women (95). On the other hand, low SES, evaluated by occupation, relates to risk of IGT in men, independent of other confounders. Confirmed by a metaanalysis of case-control and cohort studies low SES is an important risk factor for T2DM in both sexes worldwide (96).

Furthermore, the application of a validated diabetes risk prediction algorithm in a nationally representative cross-sectional survey in Canada showed that among the individual level SES variables, such as lower household income and food insecurity, predicted a higher diabetes risk in women but not in men (97). On the other hand, a strong protective effect was found only for women living in ethnically dense areas, which is an area-level indicator of SES used by the Canadian Marginalization Index for ethnic concentration. In a longitudinal population-based study, childhood SES, assessed from fathers' occupation or education, was a robust predictor of incident diabetes, especially among women, which had a cumulative risk effect for both childhood SES and adult BMI (98). Higher levels of physical inactivity, energy intake, smoking, and stress factors enhance neuroendocrine perturbations in women compared with men with low SES. Conclusively, studies claim that women appear more sensitive to socio-contextual predictors, such as education, income, and occupation, for future diabetes risk development. This may be the cause due to multiple environmental and behavioral mechanisms; however, more studies are definitely needed to clarify this complex issue.

B. Psychosocial stress

Females appear to be more vulnerable to the adverse effects of cardiometabolic impact of psychosocial stress, occupational stress, and sleep disturbances as well as partly by unhealthy behavior (Table 1) (99–105). However, all together results are controversial (Table 1) (106–108). Greater amounts of unpaid housework and responsibilities in the family may contribute to feelings of conflicting demands and sustained stress levels in females, even in matched highly educated groups of employees (109, 110). Discrimination and gender roles may further increase the environmental psychosocial stress, as well as the stress responses especially in women. Besides, large geographical differences in countries, depending on culture and gender equality, were observed. The so-called “allostatic load,” ie, the imbalance between the ability to adapt to environmental demands and overexposure to environmental stress, increase the risk of cardiometabolic diseases via neuroendocrine, autonomic, and immune mediators (see section V.P) (111).

C. Sleep deprivation and work stress

Discrimination and posttraumatic stress disorders exert greater negative impact on sleep health in women rather than in men. In a sex-specific metaanalysis of epidemiological studies, women at all ages were shown to be at a 40% higher risk for suffering from insomnia (112). In turn, sleep loss, short-sleep duration, and impaired quality of sleep correlated with obesity and even more strongly with IGM related to insulin resistance (113). In the Nurses' Health study, a close link between less than 5 hours of sleep and incidence of hypertension was found in younger women (114), and sleep deprivation also exerted more detrimental cardiovascular effects among women in the Whitehall II cohort. This was, however, attenuated after correcting for other cardiovascular risk factors and depression in the prospective analysis (115). In a metaanalysis, both short sleep (<5 h) and difficulties initiating or maintaining sleep were associated with higher diabetes risk. However, comparable effect estimates were observed in both sexes after stratification by sex (116). In one smaller prospective study looking for sex differences as primary outcome, sleep deprivation led to increased food and fat intake; however, males were more susceptible to weight gain based on greater daily caloric intake, especially during night (106). The results of a metaanalysis of observational studies, with subgroup analysis by sex, presented that shift work was associated with greater diabetes risk in men (107). Generally, diverse results exist regarding sex and gender differences, referring to the impact of shift work, work stress, and coping (Table 1). Unfavorable effects of testosterone secretion, due to changes of the circadian timing system, were suggested as one explanation. In population-based cohort and occupational cohort studies, job strain overall implicated a higher diabetes risk in women, especially in those perceiving a combination of lack of control and high job demand, as well as low emotional support (100–102). Only 20% of the stress induced diabetes could be attributed to obesity and biological risk factors, thus a mediating effect of employment grade and low SES was suggested. In a population-based study from Germany, men and women with job strain had a 45% higher risk to develop T2DM, independent of traditional risk factors and without relevant sex differences (117).

Controversial results of sex and gender differences in the work-stress-diabetes risk interrelationship may be explained by differences in the interindividual shift work tolerance, in the selection of the occupational groups and the specific definitions of job strain in studies as well as differences in opportunities for recovering from work stress between men and women. In addition to rotating shift work, the individual chronotype and work related circadian misalignment can modulate the diabetes risk in the workforce, as shown for women by the Nurses Health Study 2 (103, 104). T2DM incidence increased in early chronotypes with night shift work and parallel duration of shift work exposure, whereas late chronotypes featured the greatest risk, working daytime schedules. A longitudinal cohort study from Brazil confirmed a sex-specific association between night work exposure and T2DM incidence. After adjustment of confounders, including obesity, much stronger and earlier effects in women were shown (105). At Korean population level, the evening chronotype was more common in younger women and linked to metabolic diseases with sex dimorphism, independent of lifestyle and sleep duration in subgroup analysis. It related to a 3-fold higher risk of T2DM in men and to a 2-fold higher risk of MetS in women (108). Additionally, late chronotype was associated with lower lean mass in males as opposed to females who demonstrated a higher (visceral) fat mass. Another study found that circadian misalignment increased insulin resistance, diabetes risk, and inflammatory parameters, independent of sleep loss with significant sex-by-group interaction. However, the study was underpowered for additional analysis of sex differences (118). Overall, these studies suggest that misalignment between the circadian clock and social rhythms and between sex-dependent biological factors such as body composition and gender-dependent social timing impact pathogenesis of diabetes in men and women.

However, more prospective longitudinal studies are needed to further study these complex sex-dimorphic associations between sleep, work stress, and diabetes. These could help to implement sex-specific prevention programs in specific groups of (shift) workers.

A. Lifestyle

There are consistent sex differences in health behavior, nutrition, and physical activity, closely associated with risk of T2DM. According to health survey research data stratified by sex, women are overall more inactive but put more effort in healthy nutrition by consuming more fruits and vegetables and less meat (119–122). A prospective, but possibly underpowered, cohort study testing a priori for sex interactions states that women tend to consume more sugar, although high glycemic index diets seem to increase abdominal fat, particularly in sedentary women in contrast to sedentary men (123, 124). In a metaanalysis including predominantly women from observational studies, high glycemic index diets related to increased risk of T2DM (124). As derived from British household surveys, the density of fast food outlets associated with body weight more strongly among women possibly reflecting their greater responsibility in the family setting or maybe lower control of appetite; thus, hinting at a stronger association between body weight and environmental factors. However, the availability of low-priced meals was associated with obesity in both sexes (125).

B. Sugar-sweetened beverages (SSBs)

The rapid economic development and the simultaneously rising consumption of fast food provokes a higher consumption of SSBs, which contributes to the epidemic of T2DM independent of adiposity. In a metaanalysis of prospective cohort studies, men and women consuming SSBs in the highest quantile had 26% excess risk of developing T2DM compared with those in the lowest quantile (126). Half of the effects could be mediated by weight gain in women. Large female and male cohorts evidenced a gene-environmental interaction showing that greater consumption of SSBs is linked to higher genetic predisposition to obesity risk in both sexes (127). In addition, a relationship was found between consumption of more than 2 drinks per day with incident coronary heart disease (CHD) (35% greater risk) in women, followed up for 24 years (128). In a prospective cohort study with separate analysis for men and women, only women showed an increased risk of incident T2DM over 10 years, with a doubled risk seen in women with daily consumption of soft drinks compared with nonconsumers (129). By sex pooling data from national dietary surveys, the model-estimated global burden of diabetes associated with SSB consumption in 2010 revealed 133 000 deaths per year from diabetes and 4.5% of diabetes-related disability-adjusted life years with small differences between men and women. Generally, only a slightly higher number of deaths, however, a minimally lower proportion of deaths attributable to SSBs, was found for diabetes in women compared with men (130). However, there were large regional differences, with high mortality, related to SSB consumption in elder men in Latin America and the Caribbean. Low mortality was observed in younger women in Western Europe and Australia.

C. Alcohol

Moderate alcohol consumption was shown to be associated with a lower risk of T2DM in several observational studies. A systematic review and dose-response metaanalysis of observational studies indicated that relative to current nondrinkers and never drinkers, risk reduction was found in all levels of alcohol intake below 63 g/d with increasing risk above that threshold revealing a significant sex interaction (131). Sex- and ethnicity-stratified analysis demonstrated that risk reduction was specific to women only and non-Asian population. A possible explanation for the sex dimorphism could be that men more frequently have worse drinking behavior with heavy episodic drinking or that alcohol exerts sex-dimorphic effects on glucose metabolism. Indeed, another metaanalysis based on intervention studies showed that moderate alcohol consumption improved glycated haemoglobin in both sexes but tended to improve insulin sensitivity in women only (132). Cross-sectional analysis from the Nurses Health Study indicated that frequent alcohol intake is independently related to higher endogenous estradiol levels and that estradiol alone, or combined with SHBG, influenced the protective association between alcohol consumption and diabetes risk in postmenopausal women (133). Further research is warranted to clarify sex-specific dose-response relationships between alcohol drinking and T2DM risk and the exact underlying mechanisms.

D. Smoking

On the basis of a metaanalysis of cohort studies with subgroup analysis by sex, both active and passive smoking is related to higher risk of developing T2DM in both men and women without known prominent sex differences (134). In a prospective European case-cohort study stratified by sex, overall effects tended to be slightly stronger in men compared with women, although adjustment for confounding factors like obesity, physical activity, or educational level attenuated the association in men but strengthened it in women (135). Based on a sex-specific analysis of a recent metaanalysis, it was estimated that if the association was causal 11.7% of T2DM cases in men and 2.4% in women were attributable to current smoking worldwide (136). However, smoking behavior substantially changed between men and women. In the past decade, it particularly increased in young women, potentially contributing to higher smoking-related diabetes incidence in females in the future (137). In addition, a recent metaanalysis showed that the relative risk of myocardial infarction (MI), an important and frequent complication in diabetic subjects, conferred by smoking appears to be 25% higher in women than in men (138).

A. Developmental origins of health and disease

In the vulnerable phase of pregnancy, many environmental factors have strong influence on fetal development in a sex-specific way (139). Plenty of conditions occurring in pregnancy, such as over- or undernutrition, hyperglycemia, and acute stress situations, for example, are known to influence the phenotype of the progeny via epigenetic effects without affecting the genetic coding directly (140). These epigenetic changes involve DNA methylation, histone modifications, or micro-RNAs. They have the potential to activate or inactivate genes and their subsequent products in manifold ways and can even modify evolution of future generations in a transgenerational and sex-specific mode (140, 141).

B. Fetal programming/epigenetics in animals

Studies on epigenetic effects on diabetes risk in humans are scarce; thus, we mostly rely on studies on rodents. There is evidence that sex-specific intragenerational transmission of glucose tolerance and fat distribution, from one generation to subsequent ones, is caused by maternal undernutrition or hyperglycemia in an epigenetic manner (Table 2) (142–144). In mice, IGT was transferred through both parental lineages, whereas obesity only advances through the maternal line. Reduction in birth weight only appears in transmission through the paternal lineage from F1 to F2 generation (142). In contrast, in another study, a substantial increase of birth weight in F2 was found through impaired glucose tolerant paternal lines (143). Nonetheless, both mothers and fathers transfer an increased danger for IGT to F2 generations, which is especially pronounced through paternal lines (Table 2) (143). Maternal high-fat diet (HFD) in mice was reported to cause sex differences in glucose metabolism in offspring (145). Male offspring in the HFD group had increased oxidative stress, decreased insulin secretion, islet area, and insulin content compared with female mice. Female mice, with mothers on controlled diets demonstrated lower estradiol levels compared with male offspring. The authors pointed out that the sex difference may be explained by higher oxidative stress in male β-cells, which related to decreased estradiol levels, potentially leading to a loss of protection of the β-cells.

Earlier studies in overfed mice in gestation and lactation periods demonstrated IGT in male and female offspring of obese mice. Similar differences were shown with lower pancreatic insulin content in male offspring of obese dams (146). Furthermore, in the obese descendant group insulin levels were higher than in the control animal group, independent of sex.

Most recently high susceptibility to obesity and diabetes was demonstrated in a sex- and parent of origin- specific mode in murine progeny (147). Sperm and oocytes of HFD mice were isolated and transferred into healthy foster mothers. In F1 generation, female offspring of HFD parents showed higher adiposity, with nearly similar effects of maternal and paternal gametes. Both male and female F1 offspring exhibited insulin resistance and higher glucose, inherited predominantly through maternal gametes (147).

Maternal HFD plus high sucrose but without obesity during pregnancy exerted sexual dimorphic effects on the regulation of the hypothalamic transcriptome of the offspring (148), showing higher female vulnerability to metabolic disturbances (148). Female maternal HFD offspring presented lower insulin sensitivity and fasting hyperglycemia compared with controlled littermates after birth (148). In males, paraventricular hypothalamic gene expression was down-regulated, potentially indicating an important adaptation to maintain glucose homeostasis in male offspring (148). Additionally, after insulin-induced hypoglycemia, Crh mRNA expression was up-regulated in female offspring only, also demonstrating sexual dimorphism in stress response (148).

Fetal programming was shown to influence hypothalamic neurocircuit formation through central insulin signaling (149). Furthermore, effects of maternal HFD during lactation was recently reported causing obesity and IGT in the offspring through impairment of proopiomelanocortin (POMC) and agouti-related peptide neurons projections to hypothalamic target areas (150). Central hypothalamic insulin signaling interfered with negative feedback to the hypothalamus-pituitary-adrenal (HPA) axis in stressed male mice (151), which potentially explains high comorbidity rates of mental and metabolic disorders (see section V.P). However, sex-specific differences are not yet well documented and need further investigation.

C. Fetal programming/epigenetics in humans

Human males and females born in times of low nutritional resources (famines) are more vulnerable regarding diabetes risk in adulthood with greater risk seen in men. This was experienced for 2 of 3 famines in Austria, which was in between and after the 2 world wars (152). Thus, even in humans, male offspring appear to be more vulnerable in intrauterine life and early postnatal period. However, other studies did not specify any sexual dimorphisms after perinatal undernutrition. Additionally, increased postprandial glucose levels, as well as hyperinsulinemia, were detected in adults affected by food limitation during their pregnancies (153, 154). During Dutch famine, sex-dimorphic alterations of epigenetic profiles were seen in offspring exposed to malnutrition (Table 2) (155). Such variations may explain sex differences in fetal development, at birth, and in later life. Furthermore, the Dutch famine study population showed higher BMI and dyslipidemia only in women, which were exposed to food limitation (156). Further nutrition-dependent sex dimorphisms are detailed in Table 2 (157).

A Scottish human population with and without diabetes, reported a high degree of variation between female and male first generation offspring throughout transgenerational transmission of unfavorable cardiometabolic traits (158). Female offspring, whose mothers had diabetes, were more often affected by MetS, higher glucose levels, and body fat content, rather than female offspring of fathers with diabetes, or no parent diseased at all. Furthermore, lower high-density lipoprotein (HDL)-cholesterol was seen in female offspring, if both parents were affected with diabetes. Both sexes had higher blood pressure (BP), when the mother had diabetes as opposed to the fathers. However, all offspring of diabetic parents have higher risk of increased waist, BMI, and body fat content compared with offspring of parents with no diabetes history. In the light of these sex differences, conveyed by parental T2DM, it is no surprise that predictors of MetS differ in a sex-specific way as well. The most eminent predictors in females were diabetic mother, BMI, and age, whereas in males, they were BMI and body fat. A Japanese study revealed that the body weight of the offspring and parental history of T2DM are determinants of future risk of T2DM in offspring, in a sex-specific way (159). Maternal history of T2DM was related to higher T2DM risk in normal-weight subjects, and history of T2DM in the father was associated with higher T2DM risk in overweight subjects, without showing impact of sex of the offspring.

Regarding smoking, fetal exposure to parental smoking associated independently with risk of T2DM in adult daughters only (160).

D. Fetal sex and risk for GDM

Recently, fetal sex was found to be relevant in pregnancies, for defining the risk of developing GDM and subsequent risk of developing T2DM after pregnancy (89–91). Women carrying a boy in their first pregnancy have a 3%–4% higher risk of GDM and a 7% higher risk when carrying a boy in their second pregnancy. Carrying a girl implies to maintain normal glucose tolerance (NGT) in the first pregnancy. Interestingly, women having GDM in the first pregnancy had 6%–7% higher risk developing T2DM over a median follow-up time of 5 years when carrying a girl (90, 91). In a previous study, it was found that compared with women carrying a female fetus, women with a male fetus had decreased β-cell capacity and higher postprandial glucose levels during glucose challenge (89). Because the underlying mechanism is unclear, the authors speculated different pathophysiological causes for these sex differences in β-cell function in mothers. This might be related to actions of the Y chromosome on sex-specific variations in placenta-derived hormones, because placental lactogen and prolactin or other proteins are involved in β-cell mass expansion (89). Interestingly, especially the placenta shows many sex-specific alterations, also in regard to epigenetic mechanisms, which might truly have huge impact on complications in and after pregnancy. These were recently reviewed in detail elsewhere (161, 162). Moreover, the hypothesis behind higher risk of T2DM in women, after carrying a girl in a GDM pregnancy, could include already existing poorer β-cell capacity of these mothers compared with mothers with a male fetus. The male fetus causes a decrease in β-cell function only in pregnancy, which resolves after delivery and restitutes former capacity. This theory was not proven against actual measures of β-cell function and thus remains speculative. However, it seems that GDM mothers with female fetus have lower β-cell capacity, which lead to higher risk of T2DM and earlier onset over time.

In a secondary analysis, another research group demonstrated that women carrying a female fetus were less insulin resistant when fasting in an early stage of pregnancy but not in a late one (163). As shown in a few previous studies, female newborns had higher insulin resistance compared with male offspring. Interestingly, an earlier study speculated that the usual lower birth weight in girls, compared with boys, might be due to higher insulin resistance in female fetuses during pregnancy. The female fetus does not react to insulin and its trophic actions in the same way as the male fetus does (164). Based on studies reporting higher insulin resistance throughout childhood in females, the authors concluded a genetic background of their Gender Insulin Hypothesis. So far, explanations for these differences are elusive; however, we are now aware of already existing sex-specific differences at the very beginning of life and potentially responsible for many differences in health and disease, in men and women later on. Altogether, these new results highlight the impact of fetal sex on maternal glucose metabolism. The constant interaction between fetus and mother, with potential future negative impact affecting the health of both, clearly demonstrates a health determinant neglect not only in the field of glucose metabolism but also in the field of GDM and fetal sex.

E. Neonatal fat distribution

Already at birth females have more SAT and a more centralized pattern of SAT, assessed by skinfold thickness (165). Remarkably, sex-specific differences in SAT accumulation of neonates were related to their insulin levels (166). The associations between cord blood insulin and different SAT locations were more pronounced in areas typically described as preferential fat storage location, specific for each sex. Overall, these associations were higher in male neonates. Among the 15 measured body sites, neck and upper abdomen were mostly affected by insulin levels. In female neonates instead, this association was only observed with SAT thickness on the hip. Based on these observations, sex-specific body shape and lipid accumulation could already be determined in utero, especially in insulin sensitive locations varying by fetal sex. Furthermore, these observed differences in SAT might be allegeable with smaller insulin effects on SAT, due to above described higher insulin resistance in female neonates.

F. Small or large for gestational age neonates

Birthweights under and over the normal limits are associated with metabolic disease as reported throughout literature (167). Sex differences were found in several studies with controversial results, regarding the risk of T2DM for subjects born with low birth weight (LBW) or high birth weight. In a recent Swedish register study (168) investigating nearly 760 000 individuals, high birth weight was related to increased risk of T2DM and obesity. Males already had a higher risk for T2DM in lower weight categories, among high birth weight group. This risk was exaggerated in the highest birth weight categories in men compared with women. In a Danish register, with more than 220 000 men and women, LBW and high birth weight were reported to result in a higher risk for development of T2DM in women (169). Women in the high birth weight group had a higher risk for T2DM compared with men. In the LBW group, only women were affected by the higher T2DM risk. A small observational study found the opposite in very LBW offspring, with male sex being an independent risk factor for hyperglycemia (170). In this study, men with very LBW had higher levels of glucose, lower levels of insulin and reduced β-cell function compared with women with comparable insulin resistance.

These studies demonstrate a strong relation of birth weight and T2DM risk in both sex cohorts. Register studies revealed associations of LBW and T2DM in women and a higher risk for men categorized as large for gestational age (168, 169). Nonetheless, studies are controversial and further research is needed to fully understand sex-specific associations of LBW and high birth weight with glucose metabolism.

Additionally, both over- and undernutrition as well as hyperglycemia in pregnancy are associated with increased risk for cardiorenal disease in the offspring. Especially in this vulnerable phase, the kidney is prone to a number of mechanistic changes driven by epigenetic alterations. These can lead to renal dysfunction, glomerular hypertrophy, diminished vasodilative renal vessel function, changes in renin-angiotensin system expression followed by hypertension in adult life, usually more often affecting men. These aspects were summarized in a recent review, also highlighting sex dimorphism in kidney disease (171).

A study reporting on induced fetal lung maturation, using glucocorticoid bethamethasone, identified increased insulin concentrations in adult female offspring only (172). A higher reactivity of HPA axis, after prenatally prescribed glucocorticoids, was reported in female offspring between 6 and 11 years of age (173). Therefore, higher sensitivity of HPA axis may already exist in this very early period of life in female offspring. A more responsive HPA axis was also described in elder women compared with men (174). However, in both sexes, higher HPA axis activity was clearly related to lower birth weights with no sex difference (174). These results confirmed previous findings in animal studies, which showed higher HPA reactivity in female offspring after exposure to glucocorticoids or stress in the mother (175, 176). Interestingly, excess HPA axis reactivity was identified in some studies to be involved in pathogenesis of psychiatric disorders, as depression, as well as cardiometabolic disease (177, 178). Further studies are needed to clarify the relationship between birth weight, pregnancy-related stress and HPA axis activity of offspring and the impact of programming adult diseases in young men and women.

G. Endocrine disruptors

Exposure to endocrine disruptors is supposed to contribute to higher risk of obesity and T2DM in humans (179). A dose-dependent positive relationship was noticed between urinary phthalate metabolites and parameters of glucose metabolism (fasting glucose and insulin resistance) in both sexes (180). However, pathophysiological mechanisms behind disrupting actions are mostly based on surrogate markers and not well understood at the moment. Endocrine disruptors are already able to act as hormones in low but persistent dosages, mostly mimicking estrogen properties. They either activate or inactivate cellular receptors, cell responses, and other targets and can cause higher insulin resistance and hyperinsulinemia. Furthermore, persistent exposure to small dosages appears to be related to mitochondrial dysfunction due to intracellular gluthathione depletion. This associates to inflammation and ectopic fat, potentially leading to T2DM. Candidate obesogens cover a wide range of compounds, including bisphenol A (BPA) and phthalates (179). Endocrine disruptors mostly travel with lipids and accumulate in adipocytes causing reactive changes in adipokine levels, which happen in a sex- and disruptor-specific way.

In the Canadian Maternal-Infant Research on Environmental Chemicals Study, newborns showed significant sex differences in leptin and adiponectin levels, which were associated to maternal in utero BPA exposures (181). Female offspring had higher leptin levels than males, whereas adiponectin did not differ but was inversely related to BPA in males. On the other hand, high leptin levels in males were especially seen in moderate to high exposure when compared with mono-(3-carboxypropyl). In another study investigating mother-child pairs, late in utero BPA exposure related to increased plasma leptin levels in boys, whereas early exposure was linked to higher plasma adiponectin in girls at the age of 9 years (182). Sex differences in key metabolism-related hormones, referring to BPA and phthalate exposures in utero and childhood, were corroborated by another study showing additional results regarding exposure time and differentiating by pubertal status next to sex in adolescent offspring (183). In utero monoethyl phthalate was associated with higher leptin levels in girls and decreased insulin secretion in pubescent boys. Monobenzyl phthalates related to lower leptin levels in girls only. This study also investigated peripubertal BPA and phthalate exposure, which was related to higher leptin in boys. The combination of mono-n-butyl-phthalates and monoisobutyl phthalates associated with C-peptids in boys. In addition, only pubertal boys had lower IGF-1 concentrations, which were also related to various urinary phthalate metabolites. In girls, another compound, di-2-ethylhexyl phthalate, was associated with increased IGF-1 levels in prepuberty. Pubertal girls were affected by higher glucose levels during fasting, which was associated with various phthalate metabolites.

Therefore, overall sex-specific associations between BPA and leptin were found in boys in both studies but with different exposure times and developmental periods (182, 183). These discrepancies are supposed to be caused by different populations either from urban or rural areas with varying BPA exposures. Although underlying mechanisms of endocrine disruption through chemicals in humans are still unclear, the peroxisome proliferator-activated receptors (PPARs) and reactive oxygen species are yet suspected to play a major role (183). In animal models, PPAR expression is altered through endocrine disruptors, and thus subsequent alterations in metabolic parameters result. Of note, PPAR expression is sex-dependent (184). Hence, if endocrine disruptors have the potential to change metabolism in humans via alterations in PPAR expression, sex-specific differences will have a causal explanation.

Further studies need to give wider insight into the uncharted interactions of genes and environment in animals and humans. Considering their transfer and phenotypic transmission throughout the offspring generations. At present, in human surveys, the underlying mechanisms are limited and sex-specific investigations are urgently needed for epigenetic effects on limited nutrient supplies, as well as food overload, environmental factors, and hyperglycemia.

H. Genetic predisposition

Even autosomes display divergent expression patterns in gene regulation either mediated through hormones or directly modulated by sex chromosomes. Sexually dimorphic gene expression across organs vary, between 14% in brain and 70% in liver, with mostly small effects (185). Additionally, sex differences were recently also described in the transcriptome of human placental cells, demonstrating an influence of fetal sex on placental gene expression in a cell-type dependent manner (186). Thus, placental function appears to be affected differently for male or female offspring, potentially framing sex differences in metabolic, immunological, and inflammatory responses (186). In a recent study presenting genome-wide association metaanalysis, the genetic background of body fat accumulation and its relation with cardiometabolic traits was investigated (187). Sex-specific differences were found in 20 out of 49 genetic loci, which are involved in regulatory functions of adipose and insulin biology. According to the waist to hip ratio, stronger effects were seen in women in almost all associations of these loci. In particular, these associations were related to higher WCRs in women and to lower hip circumferences in men.

Next generation sequencing in obese diabetic ZSF1 rats, featuring a model of T2DM, revealed 103 genes showing sex differences in genes expressed in the liver (188). Genes involved in lipid metabolism and glycolysis were associated with female-specific genes, whereas hepatic metabolism, detoxification, and secretion were associated with male-specific genes. These results suggest a huge variety of genes showing sex dimorphism in hepatic genes, which might influence drug pharmacokinetics in a sex-specific way. This calls for a further urgent clarification of its impact on clinical outcome in humans.

Another way of looking at genome-wide association study results is the so called pathway-based approach, which links SNPs considered mutually involved in a pathway. This multilevel approach might give further understanding of mechanism behind genes and their relevance in diseases. Heterogeneity between males and females was identified in gene sets associated with T2DM in humans (189), 5 pathways were identified in the male group and 13 in the female group, with only 3 pathways overlapping. In the male group, the gene TCFL7L2 largely contributes to the significance of these pathways, whereas in women, no such effect was reported.

I. Gonosomes

The number of X chromosomes within cells contributes to sex differences in adiposity (190). Accelerated weight gain on HFD, incident hepatic steatosis, and hyperinsulinemia of XX animals mainly depended on the amount the X chromosomes. Higher expression of a subset of genes on the X chromosome, which escaped inactivation, is seen in adipose and liver tissue of XX animals compared with XY animals (female cellular mosaicism). GH expression might be stimulated through the number of X chromosomes and X inactivation and is involved in impaired energy metabolism (191). Furthermore, one locus near dual specificity phosphatase 9 on the X chromosome is directly associated with higher T2DM risk (192).

A couple of sex differences in genes associated with T2DM were investigated so far (Table 2). Most of these identified genes, conveying sex differences in diabetes risk, increased in one sex without showing any effect in the other. In some genes, the ability of reducing T2DM risk was observed to differ between men and women as well. Furthermore, differences in leptin, lipid, and glucose metabolism were demonstrated, which are based on genetic sex differences (Table 2). In T2DM, cardiovascular mortality was lower in men compared with women carrying a leucine repeat in carnosinase gene.

Sex-specific differences in the ANKK1 (rs1800497) polymorphism of dopamine receptor D2 in humans were reported with increased risk for T2DM in women, which could not be found in men (193). Furthermore, another polymorphism in women was found to be associated with elevated first-phase insulin secretion (193). However, sex differences in regulatory mechanisms of genes could be the key element in explaining sexual dimorphism, leading to variations in phenotype through gene-environment interactions (194). Although a large number of T2DM risk genes were found so far, the effect size of single risk alleles or the predictive accuracy of combined genetic risk scores of incident T2DM is rather low up to now (195). A small improvement in risk prediction could be achieved by the use of a genetic risk model, even in a sex-specific very high-risk group like women with GDM. The additive value, in addition to traditional anthropometric and metabolic clinical parameters, in prediction of diabetes risk appears to be small. However, one reason among many others could be the heterogeneity of subjects with T2DM next to the polygenetic nature of the disease, missing heritability or that important biological factors including sex, age and ethnicity were neglected in many studies. Sex had a small but significant impact on the genetic T2DM risk model, which disappeared in the combined genetic and lipid metabolites risk model in the Framingham offspring cohort (196). Still, also in this study, genetic markers of diabetes risk only modestly improved the predictive accuracy of future occurrence of T2DM, based only on traditional clinical risk factors. Nevertheless, improved and maybe sex-specific genetic and epigenetic risk models could help to develop personalized medicine in the future.

A comprehensive overview of genes associated with T2DM or involved in metabolic function, relating to complications or mortality featuring sex dimorphism in T2DM so far, are shown in Table 2.

J. Glucose tolerance

IFG is more prevalent in men and IGT more frequent in women, regardless of their ethnicity. These metabolic conditions, together with T2DM, are related to glucose handling by the whole body, which is evaluated by the glycemic levels at fasting and in postprandial conditions. The 2 major processes responsible for the maintenance of normal glucose levels are insulin resistance/sensitivity and insulin secretion. In general, NGT is maintained if increased insulin secretion is able to compensate the reduction of insulin sensitivity, which occurs for instance, with increasing BMI or age (197). Figure 5 shows this phenomenon of the classic hyperbolic paradigm, ie, the nonlinear inverse relationship between the 2 processes. As long as a subject is able to balance the 2 processes, he/she remains on the “normal” curve. When the balance insulin sensitivity/secretion and their interplay fail, because 1 or both deteriorate, the subject moves in the lower part of the graph in the bad “zone,” featuring prediabetes/IGM and eventually reaching a status of overt T2DM (197, 198). We have applied these concept to a large European population to evaluate possible sex differences in the metabolic parameters across various glucose tolerance categories (199).

K. Insulin sensitivity and secretion

Nonobese subjects exhibit glucometabolic sex differences only in the NGT group, where women are more sensitive than men (Figure 5) (200). When glucose tolerance deteriorates toward IGM, insulin sensitivity in women is reduced more than in men. An increased secretion is observed in both sexes, which was enough to maintain the subjects close to the appropriate compensation curve, derived from the normal-weight healthy control subjects. In T2DM, the impairment of insulin sensitivity and insulin secretion is substantial and similar in both sexes. These evidences are confirmed also when introducing age and BMI as covariates into the whole population, ie, insulin sensitivity decreased with increasing BMI at the same rate for both men and women in all glucose tolerance categories, whereas insulin secretion increased with BMI at a faster rate in men, which better compensated the increasing insulin resistance. This trend may partly explain why, in general, women show better insulin sensitivity and overall more favorable cardio-metabolic risk profiles, than males if normoglycemic. This may be associated with sex hormones and their receptors, different body fat distribution and related biomarkers, such as higher adiponectin, which are discussed earlier. Estrogen demonstrated protective effects from β-cell apoptosis (201), stimulates β-cell secretion (202) and improves insulin sensitivity (203) with antidiabetic effects primarily described to ERα (68, 204). Estradiol administration increases insulin-mediated glucose disposal in early postmenopausal women but such benefit is not seen in late postmenopausal women (205).

With age-induced deterioration of glucose tolerance, sex differences are reduced, leading to more pronounced aggravation of the metabolic profile in women than in men (206, 207). This is mediated by greater adiposity and insulin resistance (33). This aggravation, therefore, may be related to the fact that women have to gain more weight to develop T2DM (208). Nonetheless, the rate of onset of T2DM is greater for men than for women in different ethnic groups (209) at lower BMI levels. Another reflection of the above conclusions is the importance of performing, when possible, a “full” test with insulin measurements, for instance, an oral glucose test (210), in order to better characterize the single metabolic parameters in both healthy subjects and patients, especially women.

L. Incretin hormones

Differences in insulin secretion could be partly ascribed to the effect exerted by the incretin hormones during an oral test on pancreatic β-cells. To the best of our knowledge, there is no recent study that thoroughly investigates the role of sex on the incretin effect. In an early review, it has been reported that the magnitude of the effect of glucagon-like peptide-1 (GLP-1) does not greatly depend on patient characteristics, such as age and sex (211). More recently, GLP-1 response was evaluated in a large cohort of subjects of both sexes, with different degrees of BMI and ranging from NGT to T2DM. Women with IGM or T2DM, independently of age or obesity, had 25% lower GLP-1 response to an oral challenge than those with NGT with consequently lower GLP-1 concentration (212). Because higher GLP-1 responses are associated with better insulin secretion, the low incretin levels could partly explain the reduction of β-cell function observed in women when becoming IGM or T2DM.

M. Gastric emptying and glucose absorption

In studies carried out with an oral test, “postprandial” glucose metabolism is influenced by gastric emptying and intestine glucose absorption. Gastric emptying of carbohydrate containing meals has been demonstrated to positively correlate with postprandial glucose levels (213), and it has been studied in healthy control subjects and in diabetic patients. In both cases, gastric emptying was found to be slower in women than in men, although the etiology of this sex difference remains to be clarified (214). Recently, attention has been devoted to gut glucose absorption, because it could contribute to explain why in general women exhibit higher 2-hour glucose levels during an oral test. However, only few studies have been performed. In general, in lean subjects, meal appearance in the peripheral circulation was found higher in women than in men, irrespective of age (215). More recent investigations showed no differences in the amount of absorbed glucose when adjusted for age, height, and/or fat-free mass (212, 216) but only proved a prolonged gut glucose absorption in women, probably due to a slower gastric emptying. These factors could then contribute to higher glycemic levels during the oral test and thus to the evidenced higher prevalence of IGT in women.

N. Ectopic fat

Impaired lipid metabolism with excess release of FA from adipose tissue leads to overspill of circulating lipids and thus to ectopic nonadipose lipid accumulation in the liver hepatocellular lipids (HCL), muscle intramyocellular lipids (IMCL), and myocard myocardial lipids (MYCL) with detrimental consequences of lipotoxicity. Of these facts potential sex differences may emerge (56).

Sex-dimorphic results were seen in IMCL, depending on total body fat content and lifestyle. Among younger normal-weight sedentary volunteers (217) and also in obese and diabetic subjects (218), females had higher IMCL in leg muscle (219), higher lipid area density, and greater number of IMCL droplets than males. Women also tend to oxidize more fat than glucose after endurance exercise training (220). Women presented to have double the amount of IMCL as men and to experience a net reduction in IMCL during prolonged exercise. Women at high risk of T2DM, with previous insulin-treated GDM, showed an IMCL increase of up to 60%, whereas insulin sensitivity and ATP-synthesis were only slightly decreased at that stage (221, 222).

Men more often developed fatty liver disease than females, due to more pronounced insulin resistance, different effects of gonadal hormones and environmental aspects (Figures 2 and 3). Estrogen is protective in both sexes, and estradiol treatment improved insulin resistance and HCL levels in animals (223). Nevertheless, the balance of sex hormones appears important as inappropriate low testosterone, or high estrogen concentrations in males, may also relate to fatty liver disease and metabolic abnormalities. In normotolerant young women, with previous GDM, liver fat was twice as high as in women with normal pregnancy, considering insulin resistance and body fat (222). Also later the manifestation of overt diabetes (224) was predicted.

Increased MYCL is one of the many potential pathophysiologic mechanisms implicated in “diabetic cardiomyopathy” (225). Healthy women have lower MYCL and pericardial fat than men (226). Male sex and glycemia independently predict heart steatosis. Glucose tolerance deteriorates the increase of MYCL, which was more pronounced in women, abrogating sex differences in T2DM. This potentially attenuates cardiac protection in diabetic women. Consistently, MYCL increased in diabetic women, whereas stroke volume decreased and heart rate increased; this was already initiated in insulin-resistant nondiabetic women (227). Also nondiabetic women, with previous GDM, had no cardiac steatosis or dysfunction, although myocardial wall thickness related to glycemia and BP. Thus, increased MYCL does not appear to be an early marker in the pathogenesis of diabetes but a consequence of chronic hyperglycemia (228).

O. Energy imbalance

Sex dimorphism, in substrate and energy partitioning (Figure 3), has been clearly described in details in recent reviews (229, 230). At rest, women have greater storage of free FAs than men, but during exercise and conditions of sustained increased demand, women were shown to exert higher oxidation of lipids in relation to carbohydrates. Males instead rely relatively more on glucose and protein metabolism (Figure 3) (66). During times of food deprivation, females reduce energy expenditure with consequent loss of fat stores contrary to males. This can explain greater harm in males in periods of undernutrition.

Estrogen can decrease food intake directly by effects in the brain. Tight interacting of leptin, insulin, neuropeptide Y (NPY), and ghrelin seem to play a vital role (231). Moreover, estrogen can exert direct effects on fat tissue by enhancing proliferation of preadipocytes, especially in females (232), and by up-regulating sc α2A-adrenergic receptors promoting SAT accumulation, notably in premenopausal women (233). Polymorphisms in the ERα gene were found to relate to increased abdominal fat in young women (234). Both female and male ERαKO mice were shown to develop adiposity, which may be primarily due to reduced energy expenditure (235).

Women display less reduction of insulin sensitivity with increasing body fat and lower resting energy expenditure, which declines more rapidly with ageing compared with men. Menopause is associated with decrease of total energy expenditure due to the loss of estrogen effects on the hypothalamus via increased release of orexigenic hormones NPY, agouti-related peptides, and a decrease in lean body mass (67). In postmenopausal women, the most important source of estrogens is fatty tissue where adrenal androgens are converted into 17β-estradiol and estrone catalyzed by aromatase. A higher conversion rate in obese and/or older women is probably due to the higher number of adipocytes (236). Females may exert a higher capacity for adipocyte enlargement and adipose tissue plasticity may play a role in obesity-related metabolic abnormalities and ectopic fat deposition (237). The sum of all these effects could contribute to the higher rate of the metabolically healthy obese phenotype in women.

Testosterone has shown to inhibit lipoprotein lipase and uptake of triglycerides by adipose tissue (238), to increase lipolysis via β-adrenergic receptors (238), and to promote increase of fat-free mass and muscle insulin sensitivity by increasing mitochondrial capacity (78). It is noteworthy that estrogen deficiency but not testosterone deficiency per se is responsible for the increase in body fat mass in males, but androgen deficiency induced a decrease in lean body mass, muscle size, and strength (239).

Recently, also the composition of gut microbiota has shown to be associated with energy balance, the risk of insulin resistance and diabetes, potentially via changes of the FA metabolism or via release of gut-derived hormones (56). Preliminary results have pointed out potential sex-specific gut signatures related to obesity and diabetes with more unfavorable proinflammatory gut microbiota components in males (240). However, up to now, only very few studies have reported results, stratified by sex. Future research would demand bigger studies to clarify whether sex differences also play a relevant role in microbiome-dependent hormone regulation and microbiome-related metabolic diseases, like T2DM, as shown for type 1 diabetes (241).

P. HPA axis activity and stress model

The hypothalamus regulates food intake and energy homeostasis and coordinates important metabolic and endocrine processes, as well as the autonomous nervous system. Females have higher estrogen-related POMC, an anorexigenic prohormone synthesized by the hypothalamus and pituitary gland, which results in important derivates like ACTH and α-melanocyte-stimulating hormone, which influence appetite and energy expenditure (242). POMC neurons express insulin and leptin receptors. Glucocorticosteroids play an important role in adipogenesis, lipolysis in fatty tissue, and hepatic insulin resistance, potential sex differences regarding basal or stress-induced HPA axis activity have been observed (243). 11β-hydroxysteroid dehydrogenase type 1 converts cortisone in cortisol, thereby increasing local glucocorticoid levels in liver and adipose tissue, particularly in VAT (244). Some studies have shown higher sensitivity to ACTH and cortisol in women and lower ACTH and cortisol levels in younger women including lower circadian variations. However, there was no difference observed between postmenopausal women and men of comparable age (245). Across the cycle, estrogens may increase ACTH release and corticosterone levels, related to estrogen levels, whereas androgens exerted an inhibitory effect on the HPA axis stress response (245). Especially women with abdominal obesity feature increased cortisol release to acute stress challenges. Interestingly, the adrenal response to ACTH was higher in healthy young women as well as in postmenopausal ones, compared with men. However, these sex differences disappeared in patients with overt T2DM (246). Therefore, diabetic men had higher peak cortisol levels compared with controls, but the clinical implications of this finding are so far unclear. Further studies present an association in women between the stress level and the activation of the vascular endothelial nitric oxide (NO) system, as well as with the carotid intima media thickness, suggesting a close link to stress induced cardiovascular dysfunction in women (99).

Insulin resistance/hyperinsulinemia influence neuroendocrine pathways and relate to sex hormone imbalance. HPA axis and sex hormones appear to mutually interact by impairing stress responses, thereby enhancing visceral obesity phenotype and related metabolic abnormalities (111, 247). Acute stress leads to increased glucocorticoid release from the adrenal glands combined with sympathetic arousal, activation of the renin-angiotensin system, release of inflammatory cytokines, and changes of the immune system (247). A vicious cycle is then established: chronic stress results in disturbed energy homeostasis and feeding behavior (248), leading to VAT accumulation and insulin resistance, which is further aggravated by sympathetic overactivity. Testosterone deficiency aggravates the MetS in men by bringing forward obesity and hyperinsulinemia, which in turn suppresses testicular androgen production (“vicious cycle”) (also discussed in II.G). In obese women, androgen excess, related to increased HPA activity, implicates unfavorable cardiometabolic effects.

Ghrelin and the orexigenic anxiolytic NPY increases with stress contributing to “emotional eating,” which appears to play a greater role in females, mediating the stress-obesity-diabetes relationship. Stress potentiates craving for energy-dense food and impacts food reward. Women appear more vulnerable to stress-related weight gain (249). In addition, women were shown to have lower ability to suppress hunger (250). In general, women appear to exert higher HPA sensitivity and therefore tend to be at higher risk, than males, of diseases associated with HPA dysfunction. HPA dysfunction relates to eating pathologies and psychiatric disorders, like major depression and anxiety disorders. These are much more frequently diagnosed in obese or diabetic women than men. Depression may disrupt HPA axis, thereby increasing levels of cortisol, catecholamines and inflammatory cytokines, potentially even in a sex-specific manner (251, 252). Additionally, depression can unfavorably influence lifestyle, dietary intake, and physical activity further aggravating diabetes and obesity. As discussed above, it has been speculated that sleep loss or sleep disturbances, related to rotating shift work, and desynchronized circadian rhythm may induce glucose intolerance by stimulating similar neuroendocrine pathways (113, 253). Interacting circuits of the circadian clock system and metabolic elements involve central pacemaker neurons of the suprachiasmatic nucleus regulating sleep-wake cycle, feeding schedule and behavioral rhythm. Besides the peripheral clocks at the level of liver, muscle, fat, and pancreas are also regulated (254). A possible link between circadian clocks, regulation of glucose metabolism, and melatonin signaling in the pancreas was indicated by genetic studies (255). Melatonin is a hormone that causes chronobiotic peaking at midnight. Reduction of its release during shift work appears to contribute to deterioration of insulin resistance and glucose intolerance (256). Furthermore, lower melatonin levels were described in depressed patients and were found in elder women compared with elder men, independent of light exposure (257). Additionally, circadian misalignment is associated with alterations in cortisol and leptin secretion in relation to wake-sleep and feeding-fasting cycles causing further aggravating dysregulation of glucose and energy homeostasis (258, 259).

Q. Hypothalamic melanocortin system

Sex-different behavior is influenced by a sex-dimorphic hypothalamic melanocortin system (66, 260). POMC peptides are involved in the regulation of energy and glucose metabolism. Male mice showed decreased anorexigenic neuropeptide POMC genes and protein expression, which related to increased energy intake (260). In these POMC neurons, leptin receptors in the hypothalamic arcuate nucleus display important sexual differences (261). Glucose metabolism is influenced to a greater extent by leptin receptors in POMC in males, whereas lipid distribution is more intensely influenced in females (261). Furthermore, female rats were more sensitive to the anorectic effects of centrally injected leptin rather than male rats, which on the contrary were more sensitive to centrally injected insulin (262, 263). After ovariectomy, female rats exhibited higher central insulin and lower leptin sensitivity, which could be restored by estrogen administration (262). Therefore, the mechanism proposed behind decreased female central insulin sensitivity is a modulation caused by brain estrogen signaling, potentially mediated by estrogen effects on ERα in various regions of the hypothalamus. These effects are also centrally involved in the regulation of body weight. Interestingly, estrogen effects within the brain with consequent alterations of leptin and insulin sensitivity could also be verified in male animals after estrogen administration (262). This is in agreement with findings in juvenile male and female rats with low levels of gonadal hormones. These featured similar anorexigenic effects of intranasal insulin with decrease in meal frequency and total food intake together with comparable changes in hypothalamic gene expression (262, 264). These anorexigenic central nervous insulin effects shall be interesting for potential therapeutic approaches to treat childhood obesity in both sexes.

Central insulin action is impaired by binge drinking to a similar extent in male and female rats (265). This may be potentially caused by decreased insulin receptor and protein kinase B phosphorylation in the hypothalamus, such as increased inflammation and expression of down-regulating elements of the insulin signaling cascade (265). Glucose tolerance was more disrupted in female rats after binge drinking than in male rats. The authors demonstrated that hepatic and adipose insulin effects were disturbed as a secondary effect to hypothalamic insulin resistance (265). Moreover, it was found that the inactivation of the brain insulin receptor caused increased food intake in female mice (266). However, both sexes showed an increase of body fat, insulin resistance, leptin, and triglyceride levels as well as impaired reproduction (266).

In genetically programed insulin-resistant obese mice with increased appetite and reduced physical activity, targeted restoration of Pomc function only within 5- hydroxytryptamine 2c receptor containing cells induces sex differences in energy balance (267). Although food intake and insulin sensitivity improved in both sexes, female mice remained physically inactive with low total energy expenditure and higher fat mass (267). This may also have implications for antiobesity treatment in humans, because new drugs target these key regulators of energy balance, restoring impaired POMC neuron function.

Fundamental sex differences in central insulin signaling were corroborated in humans (16, 262, 267, 268). Also in humans, both insulin and leptin play important roles in regulation of energy metabolism and body weight via central and peripheral effects. Intranasal insulin decreased fat mass and reduced feelings of hunger and leptin levels in men only, confirming higher sensitivity to catabolic insulin action of male brains (16, 262, 269). Additionally, intranasal insulin application suppressed systemic lipolysis and reduced circulating free FA concentrations in healthy men and women, independent of blood insulin or leptin concentrations, presumably mediated by central nervous system insulin signaling (268).

Based on all these studies, new treatment options may be available in future, which target the central regulation of energy metabolism and glucose homeostasis. Translating these evidences into human terms, elicits that consideration of possible sexual dimorphism is compulsory. Generally, more studies in humans are currently needed regarding this important field of research.

A. Risk factors

Table 3 gives an overview of sex differences regarding major diabetic complications and frequent comorbidities. It is hypothesized that diabetes attenuates the general female biological advantage by protecting against cardiovascular complications across all ages (206, 207). Reproductive factors, differences in experience, and presentation of symptoms or psychosocial stress may also play a role in the more unfavorable situation of diabetic women. Additionally, comorbidities, including depression, greater “cardiometabolic load,” and inflammation, are already present in an early phase of T2DM, as well as later diagnosis in women probably contributes to persisting discrepancies in cardiovascular complications between men and women. A global study showed that there were sex differences among various risk factors for MI; besides diabetes, also hypertension, low physical activity, and high alcohol intake were stronger predictors for MI in women rather than in men (270).

B. Coronary heart disease

In newly diagnosed diabetic subjects without clinical CVD, increased atherosclerosis and higher intima media thickness was found in carotid arteries (271). In general, male sex, besides age, was stronger associated with carotid plaques than female sex. However, subgroup analysis revealed that as opposed to male groups, carotid atherosclerosis was more prevalent in newly diagnosed diabetic women than in nondiabetic female controls. This data confirms that diabetes, even at an early stage, attenuates the protective effect of female sex and increases the risk for CVD in females to a greater extent than in males.

The presence of MetS and visceral adiposity predicted development of CVD largely independent of diabetes in men but primarily via IGM in women (272). Diabetes is also a greater risk factor for nonfatal incident CVD in women than men (273, 274). The cause of this excess risk of diabetic women is still not completely understood, but many biological and environmental factors play a role. Physiologically speaking female sex hormones have beneficial effects on cardiovascular wall properties, which are reviewed in detail elsewhere (275, 276). However, diabetes impairs endothelial response in women more dramatically than in males modifying the beneficial hemodynamic estrogen effects by complex interactions between insulin and estrogen signaling. Hyperglycemia adversely modifies the balance of expression and activity of ERs (277). Also, an increase in oxidative stress and endothelin-1 can be observed. This further promotes vasoconstriction and platelet aggregation (278) and decreases endothelium-dependent relaxation and NO production via impaired insulin signaling (277). In diabetic rat models, female mesenteric arteries show greater abnormal vascular responses demonstrated by greater impairment of acetycholin-induced relaxation and an increased expression of endothelial nitric oxide synthase. Thus, a relative reduction of endothelium-derived relaxing factors with greater reliance on NO can be expressed. These modulations of vascular reactivity in diabetic animals may predispose female arteries to vascular injury (279).

Hyperglycemia inhibits antiproliferative effects of estrogen on vascular smooth muscle cells. These are mediated through selective ERα activation under normoglycemic conditions. Hence, beneficial effects are counterbalanced by simultaneous ERβ activation, leading to loss of protective estrogen effects (280). This causes a proinflammatory environment accelerating atherosclerotic processes and CVD particularly in diabetic women. However, this might also serve as a potential therapeutic pathway. In addition, mutations of ERα relate to insulin resistance and early atherosclerosis in young males (281, 282).

An indisputable effect of sex hormones on endothelin-1 expression was reported in humans with androgens increasing and estrogens decreasing plasma levels, which was also verified in male to female and female to male transsexuals. On the other hand, sex differences in endothelin receptor density and higher binding capacity in men were reported (283). Furthermore, an increased receptor mediated vasoconstrictor tone was recently found in middle aged males, which might be an additional contributor to lower cardiovascular risk in normoglycemic women (284). However, in diabetic cells endothelin-1 expression is increased (285). Endothelin-1-mediated vasoconstriction in mesenteric artery rings was higher in female diabetic mice, whereas no sex differences were seen in newly manifested animals (286). However, diabetes induced cardiac remodeling in mice was increased in females at an earlier stage than in males. This was measured by prohypertrophic and prooxidant gene expression as well as cardiomyocyte size (287).

C. Coagulation

Further aspects, which could partly explain higher cardiovascular risk for women with T2DM compared with diabetic men, were described in a cross-sectional study, which documented in vitro sex differences in fibrin structure function (288). It is worth noting that diabetic women had a higher prothrombotic fibrin profile, with denser fibrin clots and prolonged fibrinolysis, which was shown in increased maximum absorbance, clot formation, and lysis time. These results were not changed after adjusting for higher plasminogen activator inhibitor 1 and fibrinogen levels, which were found in women. A further aggravation was described with increasing BMI and WCR, as well as lower HDL-cholesterol levels in diabetic women only. In men only, worse glycemic control was related to an increased atherothrombotic risk. Of note, increased fibrinolysis was only found in men at higher age, which could possibly be explained by decreasing PAI-1 levels with increasing age. In both sexes renal impairment, low glomerular filtration rate, smoking, and peripheral arterial occlusive disease, measured by ankle brachial index, were associated with increased atherothrombotic risk in T2DM, whereas for ischemic heart disease, a significant relation was found in men, with a trend shown in women. Sex-specific findings implicate a higher atherothrombotic risk for women with T2DM, which potentially could have clinical consequences. An individualized therapy strategy might be needed with more aggressive antithrombotic therapy in diabetic women and other higher risk groups, but a more cautious approach in older men, to minimize bleeding risk due to increased fibrinolysis. However, longitudinal and in vitro studies need to be undertaken to confirm these findings.

Another study investigating arteriogenesis in patients with stenotic or occluded coronary vessels, recently found that female sex and T2DM were independent risk factors for poor collateral vessel development in chronic total occlusion, in a collective of patients with severe coronary artery stenosis (289). Estrogen decline after menopause might be an underlying cause for women having worse outcome in this study. This hypothesis is corroborated by animal studies showing that angiogenesis was negatively influenced by estrogen loss in oophorectomized animals (290), whereas estrogen therapy induced collateral and microvascular remodeling in a similar animal model (291).

D. Cardiac energy supply

Although little is known about the glucose metabolism in the human heart, the energy supply through glucose in the diabetic heart is disturbed (292). Sex seems to influence glucose utilization in obese and diabetic subjects substantially (293, 294). In general, male hearts use more glucose, whereas female hearts prefer lipids (295). A recent study reports sex differences in intramyocardial glucose metabolism, assessed by positron emission tomography scans, with reduced intramyocardial glucose kinetics (ie, fractional glucose uptake and metabolism) in diabetic women compared with men. However, the reduction of glycogen synthesis, glycolysis, and glucose oxidation was more pronounced within the group of men, with decreasing levels from normal-weight to obese and T2DM group (296). The authors highlight that these sex differences have to be considered before starting metabolically active drugs, because higher plasma glucose levels were also associated with increased glucose kinetics, in diabetic patients, and this might have clinical relevance especially during ischemia. They suppose that mildly elevated plasma glucose levels in T2DM patients may improve glucose handling by the human heart and thus increase myocardial glucose metabolism.

E. Mortality

Presence of MetS was associated with a 2-fold higher risk of CVD and 1.5-fold increase of mortality, with consistently higher risk for women than men, including all-cause mortality (297). Diabetes is associated with an almost 2-fold increased risk of death (298). The sex ratio of risk of fatal CHD or stroke showed a greater risk in women, which did not improve over time (273, 274). An Italian register study found that diabetes increased the risk of first ever ischemic stroke by more than 50% in both men and women (299). Interestingly, this risk decreased with advancing age in both sexes but at a faster rate among men. Therefore, postmenopausal women at the age of 55–74 years were at higher risk for ischemic stroke than men of comparable age. Furthermore, the recurrence of stroke within the next years was increased by T2DM in women independent of age, whereas in diabetic men such enhancement was only observed in those younger than 70 years.

A. Functional limitations

In most clinical studies, women with T2DM show a higher burden of risk factors and comorbidities as well as more cognitive and physical functional limitations than males do (335, 336). Quantification of diabetes comorbidity risk across life, using nation-wide big claim data, further revealed age-related sex differences (335). More comorbidities in males were observed up to the age of 60 years, whereas more comorbidities in females thereafter. Additionally, worse glycemic control, higher BP and BMI, in combination with lower exercise levels directly mediated the association between sex and functional limitations in a large community-based T2DM cohort (336). Higher rates of depression, lower perceived family support and self-efficacy in women served as indirect mediators in this gender-functional limitations link. Overall, impaired cognitive function appears to affect self-care abilities to a greater extent in women.

In another population-based sample of older people, T2DM also exerted greater impact on the risk of geriatric conditions in women than in men (337). A greater proportion of diabetic women had cognitive impairment, depression, and falls than did diabetic men. Diabetes doubled the risk of these geriatric conditions among women, whereas no associations were found in diabetic men.

B. Mental disorders

Anxiety and eating disorders as well as depression are more common in diabetic subjects, especially in women (Table 3) (338). In addition, mental disorders adversely affect glycemic control, adherence to therapy, and development of complications. Furthermore, presence of depression doubles mortality risk in diabetic subjects (339). However, compared with nondiabetic subjects, the increase of the risk for depression seems even more pronounced in diabetic men. Nevertheless, diabetic men seem to live more effectively with their disease, showing lower prevalence of depression and anxiety, more active problem-oriented, and solving approach strategies, a better health-related quality of life and positive wellbeing (340).

C. Sexual function and reproduction

Diabetic males and females are prone to suffer from sexual dysfunctions and reproductive problems, which appear to be underestimated in clinical practice. Prepregnancy care in diabetic women of child-bearing age is still unsatisfying, due to consistently increasing rates of serious perinatal complications, including excess malformation rates and higher mortality (Table 3) (341). More than half of the investigated sexually active obese diabetic women and men complained about sexual dysfunction, which is much higher than in healthy lean subjects (342, 343). Sexual dysfunctions were strongly related to psychosocial factors and depressed mood in women but more to physiological factors, like cardiorespiratory fitness, glycemic control, and MetS in males. These problems may be regarded as markers for cardiovascular risk (344). Without it being clinically relevant the presence of erectile dysfunction can indicate doubling of new onset CVD events during the next decade, in middle-aged diabetic men (345). Impaired penile microcirculation may serve as an early marker of endothelial dysfunction, indicating higher cardiovascular risk. Additionally, males with erectile dysfunction commonly suffer from lower quality of life and depression.