Low testosterone at age 31 associates with maternal obesity and higher body mass index from childhood until age 46: A birth cohort study

Low testosterone (T) levels in men associate with increased risks of obesity, type 2 diabetes, metabolic syndrome, and cardiovascular diseases. However, most studies are cross‐sectional with follow‐up‐time < 10 years, and data on early growth are limited.


INTRODUCTION
Several studies have shown that androgen deficiency (hypogonadism) is associated with obesity in men. 1,23][4] Further, regulation of LH pulse amplitude by adipocyte-secreted hormones, such as leptin, is disrupted in obese men.In addition, a decreased level of sex hormonebinding globulin, which is linked to low T, has been associated with obesity, especially increased visceral adiposity. 3,5Low T also associates with a higher risk of abdominal obesity, 6 metabolic syndrome, 7 type 2 diabetes, 8,9 and cardiovascular diseases. 81][12] In the newborn, BMI increases from birth, reaching a maximum at age 7−9 months (adiposity peak, AP), after which it decreases until age 5−7 years.The second rise in BMI, the adiposity rebound (AR), occurs after this BMI decrease and lasts until adulthood. 11,12Early AR has been associated with obesity and worse metabolic profile in adulthood both in men and women, 11,12 as well as with a diagnosis of polycystic ovary syndrome (PCOS) in women. 13In previous studies, maternal obesity has also been shown to increase the risk of offspring's obesity, [14][15][16][17] but limited data exist on the association between maternal obesity and reproductive function in male offspring.
2][3][4] Physiological, gradual T decline has been reported to occur, especially after the age of 40 18,19 ; however, more recently, the onset of T decline has been recognized to happen even earlier in life. 4,20[23][24][25][26] In this large population-based birth cohort study, our aim was to evaluate the association of serum T levels in males at age 31 with prenatal factors, birth weight (BW), and the development of BMI from birth until late adulthood.

Study population
The study population was derived from a general population data set, the Northern Finland Birth Cohort 1966 (NFBC1966), recruited at gestational week 24 from the two northernmost provinces of Finland.The cohort included a total of 12,058 live births (of them, 6169 males), covering 96% of all births in this area. 27,28The study was

Definition of the study populations
A serum level of T < 12.1 nmol/L at age 31 was defined as a cutoff for low T levels, according to the recommendations of the European Association of Urology, the International Society for the Study of the Aging Male, and International Society for Sexual Medicine. 21,24,26The use of T preparations (n = 0), statin medications (n = 0), and opioids (n = 8) was evaluated from a self-reported list of medications at age 31.The results did not change after excluding opioid users (data not shown).0][31] Men with primary and compensated hypogonadism were excluded from the main analyses due to their small number and different etiology.This exclusion did not change the results (data not shown).A comparison of the main outcomes between the four groups is also shown in Table S1.The final study population included 2693 men, who were divided into two groups: men with low T (n = 132, 4.9% of the whole study population) and control men with normal T (n = 2561, 95.1% of the whole study population).

Growth data
Weight and height from infancy to adolescence were collected from individual records reported by child health and welfare nurses, and later by school nurses, as part of the national child health program which is free for all children in Finland and covers almost 100% of the children.Growth data were available for 80% of the study population, some of the data were missing during the process of digitalization of the records from child health and welfare nurses.Twins' growth patterns differ from single-born babies, therefore, they were excluded from the analyses (n = 66).
BMI and the timing of AP and AR were derived from fitted growth curves, as described previously. 32Previous literature has shown that BMI is an acceptable proxy for the estimation of fat mass in children. 33sed on the natural changes observed in childhood BMI, the data were split into two age windows: infancy (2 weeks−1.Height and weight at age 14 were measured and reported by the parents and BMI was calculated.According to the criteria of the World Health Organization (WHO) and Center for Disease Control and Prevention, BMI groups at the age of AR and 14 were stratified into underweight (< 5th percentile [pc]), normal weight (5th−85th pc), preobese, that is, overweight (85th−95th pc), and obese (> 95th pc). 34I groups at ages 31 and 46 and maternal BMI were stratified according to the criteria of WHO: underweight (< 18.5 kg/m 2 ), normal weight (18.5−24.9kg/m 2 ), pre-obese, that is, overweight (25.0−29.9kg/m 2 ), and obese (> 30.0 kg/m 2 ).35  self-reported 27.20 p = 0.987).

Laboratory methods
At age 31, the serum levels of total T were measured using Agilent triple quadrupole 6410 LC/MS equipment with an electrospray ionization source, operating in positive-ion mode (Agilent Technologies).
Multiple reaction monitoring was used to quantify testosterone by ficient of variations (CVs) of the method were 5.3%, 1.6%, and 1.2% for testosterone at 0.6, 6.6, and 27.7 nmol/L, respectively.The inter-assay CVs were 5.3%, 4.2%, and 1.0% for the respective concentrations.LH at age 31 was determined using an immunochemiluminometric method (Advia Centaur; Siemens Healthcare Diagnostics).

Covariates
Maternal out separately in infancy and adulthood, as described previously. 22e outlier values of T were excluded using the first quartile cutoff, −1.5 × IQR (interquartile range), for the lower limit and the third quartile cutoff, +1.5 × IQR, for the upper limit. 39Using these TA B L E 1 Baseline characteristics of the study population.

Adiposity peak and adiposity rebound
There were no differences between the case and the control groups regarding the age of AP ( cases and controls with early AR (Figure 3).Age at AR and BMI at AR correlated significantly with serum T levels (r = 0.103, p < 0.001, r = 0.049, p = 0.026) and the associations were linear (Figure S1B,C).
After adjustments, the association between age at AR and T levels remained significant (p < 0.001), but the significance was lost between BMI at AR and T levels (p = 0.211).

BMI and BMI change from childhood and adolescence to late adulthood
Prevalence of pre-obesity and obesity at ages 14 and 31 and of obesity at age 46 were higher in the case group (Figure S2).Men with low T and with early AR had the highest BMI at ages 14, 31, and 46 (Figure 3) and the greatest weight gain between ages 14 and 31 (Table 3).Cases with normal/late AR compared to control men with normal/late AR also had higher BMI at ages 31 and 46 but not at age 14 (Figure 3).BMI increased more between the age of AR and age 14 and between ages 14 and 31 in the case group, but the BMI increase between ages 31 and 46 was greater in the control group (Table 2).At ages 31 and 46, WHR was greater in the case group compared to controls (Table 2).The results did not change after adjustments (Table S2).

DISCUSSION
The In previous studies, low BW, 40,41 SGA, 42 and prematurity 43 have been shown to associate with decreased serum T levels in adulthood, although opposite results have also been published. 43,44In this study, BW, gestational age, or prematurity did not associate with low T. As a novel finding, maternal obesity was associated with low T serum levels in adult men, independently of adulthood abdominal obesity, but the association vanished after adjusting for BMI in childhood.However, the inverted J-shaped association between continuous T level and maternal BMI remained significant.Developmental programming occurs in utero resulting in increased susceptibility to health problems also in adulthood. 45,46Epigenetic modification of gene expression is a likely mechanism connecting the early development of excess adiposity with the future risk of an adverse metabolic and hormonal phenotype. 47deed, these changes have long-lasting effects, not only by increasing the risk of obesity but also the risk of metabolic and endocrine disorders. 48,495][16][17] This association was also seen in our study population.Of note, the association between maternal prepregnancy BMI and low T lost its significance after adjustment with BMI in childhood, suggesting that the effect of maternal obesity might be mediated through elevated childhood BMI.In line with the present finding, extreme weight loss after bariatric surgery has been shown to reduce the risk of offspring obesity compared to the siblings born before bariatric surgery. 50This result might suggest that the expression of obesity in offspring can be modulated by the mother's weight loss, but the nature of the association with hormonal parameters later in life remains unclear.One putative mediator of this association is leptin, the level of which has been shown to be affected by maternal obesity and nutrition during pregnancy.Leptin concentrations are higher in the circulation of obese mothers as well as in the cord blood, 51,52 and children exposed to higher leptin levels prenatally have higher BW. 52,53Paradoxically, higher leptin levels have been associated with slower weight gain in infancy and in early childhood, 54,55 but in later childhood and adolescence, a positive correlation between cord leptin levels and adiposity has been detected. 56reover, increased leptin levels and leptin resistance are associated with obesity later in life, 3,51 and high leptin levels have been shown [3][4] Younger age and higher BMI at the time of AR onward were associated with low T at age 31.Notably, men with early AR and low T presented the highest BMI from AR onward and the greatest WHR at ages 31 and 46, suggesting a more severe phenotype for those with early AR.This finding suggests a strong positive interaction between early AR and the development of obesity with increased susceptibility to low T levels in adulthood.The association between early AR and low T remained significant after adjustment for abdominal obesity at age 31, suggesting also an independent role of AR regarding the development of hypoandrogenemia.Two previous studies in men have shown that higher BMI at prepubertal age and "rising to high" BMI growth trajectories are associated with lower T levels in adulthood. 43,44Moreover, early AR has also been linked to adverse metabolic outcomes and obesity in men and women, 11,12,[57][58][59] and recently with PCOS in women, independently of adulthood obesity. 13However, studies describing the association between early AR and T levels in adult men have not previously been reported.
Interestingly, we detected lower weight gain between ages 31 and 46 among men with low T.This novel finding is perhaps surprising.
One explanation could be the loss of skeletal muscle mass in late adulthood in men with hypogonadism, as shown in a previous study concerning men with functional hypogonadotropic hypogonadism. 60re studies are needed to confirm this observation and the etiology behind it.
The strengths of our study are that it involves a large, populationbased cohort and incorporates longitudinal data from birth until age 46.Our study population was also ethnically homogeneous.All measurements-except for BMI at age 14-were performed by trained professionals at every step.Additionally, we were able to evaluate the use of testosterone supplementations, statins, and opioids, albeit self-reported.We were also able to define secondary and primary hypogonadism by taking LH into account.Another novel contribution of the present work was that in addition to analyzing specific childhood and adolescent BMIs, childhood growth trajectory data were also assessed, which have been shown to be a valuable tool to evaluate the BMI and metabolic risks in adulthood.
This study had some limitations.First, only a single T sample for each subject was available for analysis. 21,26However, to minimize potential bias, all blood tests were drawn in the morning when the physiological higher T levels occur. 61Another limitation is that, despite high participation rates at all data collection points, full childhood growth data were not available for all participants.Further, we were not able to take paternal obesity into account.Also, the study results should also be tested in other ethnicities.Lastly, adiposity, lean body mass, T levels in childhood, or timing of pubertal onset were not available in the cohort.

CONCLUSIONS
In conclusion, maternal obesity and early AR were associated with low T serum levels in men in early adulthood.The BMI growth trajectories appeared to deviate at the time of AR onward and weight gain was faster in men with low T until age 31.Men with both early AR and low T were at the highest risk of developing obesity in adulthood, suggesting a more severe phenotype in this group of men.This finding might suggest that early AR timing may be the first sign of a developmental process leading toward T deficiency in adulthood.

Finland
or in Helsinki metropolitan area participated in a clinical examination, including anthropometric measurements and blood samples for hormonal analyses.BMI was calculated as the ratio of weight (kg) and height squared (m 2 ).Waist and hip circumference were measured and waist-hip-ratio (WHR) was calculated at the 31-and 46-year clinical examinations.The data on current medication were verified by asking the question during the clinical examination.
5 years) and childhood (1.5−13 years).There were, on average, seven measurements during infancy and 16 during childhood for each child.Calculation of AP (1 month−1.5 years) and AR (1.5−13 years) was done only for the children having at least three measurements during childhood (AP n = 1584, AR n = 2008).We set an age cutoff value for early AR by dividing the study population into quartiles according to the age at AR.The early AR quartile limit value was ≤5.2 years in the present male population, in line with previously published data.11,12As late timing of AR has not been associated with adverse metabolic outcomes, the three upper quartiles (children with AR > 5.2 years) were pooled together to represent men with normal/late AR.

46 F I G U R E 2
present prospective, population-based cohort study reports for the first time the association of serum T levels in adulthood with prenatal factors, AR timing, and BMI development from birth until age Body mass index trajectories in men with low testosterone and controls.95% confidence intervals are shown.Men with low testosterone present early deviation in BMI trajectories from adiposity rebound onward until late adulthood.Abbreviations: BMI, body mass index; T, testosterone.in men.Maternal obesity was associated with an increased risk of low T levels in adulthood.Further, the BMI trajectories already started to deviate during childhood in men with low T, showing earlier AR and higher BMI from childhood until age 46.
36ctors (pre-pregnancy BMI, weight gain during pregnancy, maternal age, smoking at the end of pregnancy) and BW as well as gestational age were included in the confounding factors as they all associate with childhood growth.36Full-term,moderately preterm, and very preterm birth, as well as small for gestational age (SGA), 37,38priate for gestational age (AGA), and large for gestational age (LGA) were defined according to the literature.37,38Information on the mothers was collected by the local midwifes in the antenatal clinics using a questionnaire.Smoking, educational level, and alcohol consumption at age 31 were classified based on the questionnaire.According to the literature and the study population's characteristics, maternal parameters (mothers' pre-pregnancy BMI and smoking at the end of pregnancy) as well as BW and gestational age, BMI at AR, and parameters at age 31, that is, alcohol consumption, education, smoking, and WHR (which depicts abdominal obesity and body fat distribution), were selected for the adjustment models when appropriate.Differences in the continuous anthropometric parameters were analyzed via an independent t-test, a Mann-Whitney U-test, a one-way analysis of variance, or the Kruskal-Wallis test, as appropriate.To assess differences in the categorical parameters, a chi-square test was used.For these tests, results were reported as means or medians, groups and to perform regression analyses.Longitudinal BMI modeling was created with R-studio version 4.0.3. Modelng was carried

Table 2
AR age compared to controls with normal/late AR, but not between Body mass index at the age of adiposity rebound (AR) and ages 14, 31, and 46 in men with low testosterone at age 31 (cases) and controls.The study population was divided into four groups according to the occurrence of early or normal/late AR.Adiposity rebound was considered to be early when occurring before age 5.2 years.Body mass index changes between ages AR, 14, 31, and 46; waist-hip-ratios at ages 31 and 46 and hormonal parameters at age 31 in men with low testosterone at age 31 (cases) and controls with early versus normal/late adiposity rebound.The results are reported as mean ± standard deviation (SD) and the comparison of differences between groups were performed using Mann−Whitney U test or Kruskal−Wallis test.AR was considered early when occurring before age 5.2 y. p-values <0.05 are bolded.Abbreviations: AR, adiposity rebound; BMI, body mass index; n, number; WHR, waist-hip-ratio; y, year.a Numbers may vary in different analysis due to some missing data.b Case versus control in the early AR or normal/late AR groups.c Case + early AR versus case + normal/late AR. d Control + early AR versus control + normal/late AR.