In this large cohort of Brazilian adults who were born and came of pubertal age before the economic and epidemiologic transition in Brazil, menarche onset less than 11 years was associated with higher risk of diabetes. This association remained after adjustment for potential confounders, including race, maternal education, parental diabetes, birth weight, relative leg length–a marker of early-life nutritional status –and BMI at age 20 years.
Our study provides estimates of average age at menarche in a sample of Brazilian women participating in a large, free-living occupational cohort, born between 1934-1975 and raised during a period when the average gross domestic product of Brazil increased by 81% . The earlier age at menarche observed with younger participants reflects the secular trend toward earlier age at menarche in Brazil . While this trend may have been influenced by increases in childhood fatness, our study, corroborating evidence from others [20, 22], suggests that the association between age at menarche and adult diabetes is not accounted for by early-life BMI. Our study adds an important piece of evidence to the literature base on this topic, showing that early menarche was associated with diabetes after full multivariable adjustment-including BMI at age 20 years-in Brazilian adults born and raised in an era when childhood stunting was more common (26.7%) than overweight (8.6%) .
Our findings largely align with literature on this topic from Asian cohorts that had low rates of childhood overweight/obesity and lived through rapid epidemiologic and nutritional transition after coming of pubertal age. Conway et al. found that in 69,385 middle-aged adult Chinese women from the Shanghai Women’s Health Study, a 1-year increment in menarcheal age was associated with a 5% lower risk of diabetes; this association was attenuated upon adjustment for BMI measured when participants were 40-70 years . In the Singapore Chinese Health Study, menarche at or before 12 years of age (compared to 13-14 years) was associated with an 18% increased risk of diabetes even after adjustment for BMI self-reported when participants were aged 45-74 years . In a study of 3,304 post-menopausal women from Fujian, China, age at menarche did not appear to be associated with diabetes, but the earliest category of menarche age in this analysis was 9-14 years thus potentially masking an association between diabetes and early menarche as defined by our study (<11 years) .
Studies from countries with longer-standing economic prosperity have also reported an association between earlier menarche and increased risk of diabetes [19–23]. Similar to findings from more recently transitioned populations, evidence from Western societies suggests the menarcheal age-diabetes association is independent of confounding by BMI. Findings from our study indicate the association between early age at menarche and diabetes is not explained by BMI at 20 years of age (the earliest measure of BMI in our cohort). These results are consistent with those from a British birth cohort , the EPIC-InterAct study , and the ARIC study , which found that BMI measured pre-menarche, at age 20, and at age 25, respectively, did not explain the age at menarche-diabetes association. Furthermore, of the studies mentioned herein, ours was the first to find that age at menarche was associated with diabetes independent of relative leg length, indicating these markers represent different aspects of early-life growth and development.
The mechanisms underlying the association between age at menarche and type 2 diabetes are likely myriad and interrelated. One extensively studied pathway is the triggering of puberty by adipocytes and related hormones. Accretion of fat (and therefore increased leptin) has been shown to promote hypothalamic gonadotropin releasing hormone pulse generator activity [34, 35], and, thereby, attenuate gonadal feedback suppression of luteinizing hormone secretion  and augment aromatase activity in the ovarian granulosa cells . However, our findings, which were largely independent of BMI at 20 years of age, provide evidence that some independent direct effect, acting through additional biological pathways, may be at play.
Earlier menarche was associated with higher levels of insulin like growth factor (IGF)-I, androstenedione, dehydroepiandrosterone sulfate (DHEAS), leptin, and fasting insulin, and with lower levels of IGF binding protein-I, and sex hormone binding globulin (SHBG) at age 8 years in a study of 329 girls . IGF-I, androstenedione, and DHEAS remained associated with earlier menarche after adjustment for BMI and height, suggesting independent functional roles of these hormones, which have been associated with low birth weight and early catch up weight gain, in regulating puberty timing in girls .
Insulin-resistance induced hyperinsulinemia manifesting early in life may be an important pathologic perturbation contributing to the observed association between earlier age at menarche and higher diabetes risk. Randomized controlled trials in low-birth weight, precocious-puberty girls have shown that administration of metformin, an insulin-sensitizing medication commonly used to treat type 2 diabetes, delayed age at menarche and improved post-menarcheal (up to 15 years of age) insulin resistance, inflammation, liver fat content, and other pernicious metabolic parameters [39–42]. Further research from cell, animal, and human experiments is needed to shed etiologic light on the interplay of adiposity, insulin resistance, inflammation, the IGF axis, SHBG, and sex hormones, among other biologic intermediaries in the association of earlier menarche with diabetes.
A novel finding from our study was the association between earlier menarche and elevated alanine aminotransferase, triglycerides, and high sensitivity C-reactive protein. This is the first evidence, to our knowledge, linking menarcheal age with markers of liver dysfunction and inflammation. One hypothesis is that these associations are driven by an underlying association between age at menarche and non-alcoholic steatohepatitis–otherwise known as the hepatic manifestation of the metabolic syndrome . While studies finding that early maturing women had higher risk of the metabolic syndrome and its components provide indirect support for this hypothesis [44–47], observational and experimental research is needed to directly examine the potential link of pubertal timing with liver steatosis and inflammation.
This current study took advantage of a rich database in a unique population to study the role of age at menarche in relation to adult diabetes and other cardiometabolic disease risk factors. The fact the association was little changed after adjustment for participant age at enrollment, and factors related to menarcheal age and type 2 diabetes that may vary across birth cohorts, such as prevalence of paternal diabetes, education, birth weight, and early adulthood BMI, makes it very unlikely that the association between age at menarche and diabetes is due merely to a correlation between secular trends of increasing diabetes prevalence and decreasing age at menarche in Brazil.
An important limitation to consider when interpreting the results of this analysis was the self-recall of menarcheal age many years after the event; thus, misclassification was inevitable. However, adulthood retrospective reports of age at menarche are highly correlated (r = 0.79) with original adolescent reports . One might also consider the cross-sectional nature of the data collection a limitation. Yet, in our cohort, all diabetes diagnoses occurred after menarcheal onset, and were therefore incident in relation to age at menarche. We cannot rule out the potential for residual confounding by factors related to both age at menarche and type 2 diabetes. For example, ELSA-Brasil does not have childhood anthropometry (e.g., BMI) or physiologic measures. Childhood fatness and related hormones may cause early puberty and diabetes. To address this concern, we adjusted for BMI (based on self-reported weight) at age 20 years which, based on evidence  that BMI tracks well through life, may proxy pre-pubertal BMI.
Understanding how factors along the life course impact type 2 diabetes and cardiometabolic disease risk, and how such factors can be cost-effectively and sustainably modified, are among the transcendent public health challenges of our time. These challenges are especially critical for population health in low- and middle-income countries, including those in South America, in which changing demographics, in addition to behavioral and environmental risks, are producing rapid increases in diabetes, and where the alternative to risk factor prevention–the widespread use of costly drug- and device-related interventions–is neither practical nor cost-effective. As such, beyond its potential for extending quality years of life, investigation into the primordial prevention of type 2 diabetes and other cardiometabolic diseases is critically important to the future economic viability of health care globally.