Meta‐analysis and construction of simple‐to‐use nomograms for approximating testosterone levels gained from weight loss in obese men

Obesity‐induced hypogonadism, which manifests as erectile dysfunction and a lack of libido, is a less visible and under‐recognized obesity‐related disorder in men.


INTRODUCTION
Obesity, defined as a body mass index (BMI) ≥30 kg/m 2 , 1 has become a global health condition over several decades.In 2016, the prevalences of obesity in American men and women were 37.5% and 39.5%, and UK men and women were 29.3% and 31.3%.By 2031, these figures are projected to reach 43.6% and 44.4% of US men and women, and 36.9% of both sexes in the UK.Corresponding figures from 17 other European countries were not far behind. 2Obesity causes multiple health complications including type-2 diabetes, cardiovascular disease, hypertension, dyslipidaemia, osteoarthritis 3,4 and obstructive sleep apnea (OSA). 5By contrast, obesity-induced hypogonadism, which manifests as erectile dysfunction and a lack of libido, is a less visible and under-recognized obesity-related disorder in men.
As the global prevalence of obesity continues to increase, the number of men with obesity-induced hypogonadism has also increased, 6 with progressively more men being referred to endocrine clinics. 7though the standard approach for hypogonadism is testosterone replacement therapy (TRT), it is essential to offer help with weight loss as the initial treatment for obese men to disrupt the obesityhypogonadism cycle, 8 and at the same time improve or prevent other obesity-related health complications. 9Weight loss, by lifestyle management, or more effectively by low calorie diet (LCD) or bariatric surgery (BS) in obese men, has been shown to improve androgen levels.However, most studies of weight loss, including meta-analyses, 10 have focused primarily on total testosterone (TT), while information on bioactive (unbound) free testosterone (FT) has been sparsely documented.In recent years, BS for obese men has proliferated worldwide.
As a result, more studies, including randomized control trials (RCT) of large numbers of participants, have been published, providing the opportunity to examine more robustly the relationship between weight loss and changes to free testosterone, and determine factors influencing changes in testosterone induced by weight loss.In this study, we conducted a meta-analysis of the updated literature to assess the impact of weight loss on TT and in particular FT levels in obese men.
We also characterized the influences of age, BMI, sex-hormone binding globulin (SHBG), luteinizing hormone (LH), and testosterone levels at baseline on changes to testosterone levels induced by weight loss.
It was also possible to construct a set of nomograms to estimate the amount of testosterone gained from the extent of weight loss.This simple visual aid will be particularly useful in a clinical setting where consultation on the management of obesity-induced hypogonadism takes place, to help explain to the patient the potential improvement in gonadal function by weight loss.

METHODS
After the proposal of the concept of the study (TSH), two investigators (GK and TSH) together discussed and devised the strategy for conducting a literature search and data extraction, following guidelines from the Cochrane and PRISMA recommendations for conducting meta-analyses. 11,12

Search strategy and data extraction
Two investigators (GK and TSH) independently performed a literature search and extracted data from papers using MEDLINE, Google Scholar, and the Cochrane Library Central Register of Controlled Trials databases up to June 2023.Their results were then compared before creating a final database for analysis key terms used were: obesity, body mass index, BMI, testosterone, sex steroids, gonadotropins, gonadotrophins, luteinizing hormone, LH, fertility, male, men, weight loss, lifestyle modification, LCD, and BS.No filters for language or data were applied and the Boolean operators "AND" and "OR" were used to combine search terms.Relevant studies were manually searched from references within the identified papers.

Selection criteria
Studies examining the effect of weight loss on TT or FT in men were included irrespective of age, race, comorbidities, duration, or type of intervention.Those that fit the inclusion criteria were cohort studies or RCTs.Studies were excluded if they did not present numerical data for BMI or testosterone at baseline and end-point.

Risk of bias assessment
The quality of the reports was evaluated independently by two investigators (GK and TSH) using the risk of bias in non-randomized studies of interventions (ROBINS-I) tool for cohort studies 13 and risk of bias assessed using Cochrane Collaboration's tool for RCTs.

Data preparation
For consistency, all data were converted to système international d'unités (SI units).Reported standard errors (SE) were converted to standard deviations (SD) using the formula SD = SE × √ n, where n is the sample size.
After the literature search, data extraction and risk of bias assessment conducted independently by the two investigators (GK and TSH), their findings were compared and resolved where there was a disagreement.A final database was completed for analyses.Subsequently, the results generated were discussed in weekly "Lab meetings" between the pair of investigators at every stage up to the point of completion of the paper.

Statistical analysis
Meta-analyses were conducted using Review Manager (RevMan, Ver- were obtained via the DerSimonian and Laird method using a random effects model. 16Statistical significance threshold was accepted as P < 0.05, and heterogeneity of study results was assessed by the I 2 statistic. 17e influences of baseline characteristics of individuals on changes in testosterone levels induced by weight loss was assessed by multivariate regression, analyzing all predictive variables simultaneously to adjust for each other, using IBM SPSS Statistics, v28.0 (IBM Corp., Armonk, NY).

Nomogram construction
Data gained from weight loss.The regression equations were then translated into parallel-scale nomograms using R software (https://www.rproject.org/),presented for all ages and for two separate age groups using a cut-off of 40 years.

Baseline characteristics
From a TT of 1,419 articles initially identified on MEDLINE, Google Scholar, and the Cochrane Library Central Register of Controlled Trials databases, 1,015 titles were found to be not relevant or duplicate.The remaining 404 articles were screened and 288 were excluded because no original data were available.Further review of the remaining 118 full texts to check for eligibility against criteria showed that 74 articles were not eligible because there were incomplete data or they were duplicated from the same center.The remaining 44 studies comprising 1,774 participants met the search criteria (Figure 1).(Figure 4B).

F I G U R E 3 Continued
There was little evidence of bias due to confounding factors or in the selection of participants.Four studies were considered to suffer bias due to deviations from intended interventions as weight-loss treatment was relatively short (≤ 4 weeks), 18,34,52,56 and one LCD study with the addition of the glucagon-like peptide-1 agonist Exenatide), 33 and two LCD studies used very low-calorie ketogenic diet (35,36). 35,36,60 None of the studies had a bias in the measurement of outcomes, while there was insufficient information from any of the studies for assessing bias in the selection of the reported result (Figure 5).
The results of leave-one-out sensitivity analyses (Tables S1 and S2) showed that changes in TT induced by weight loss in obese men both in cohort studies and in RCTs were not substantially driven by any individ-ual study.This indicates that the results from this meta-analysis were statistically robust.

Nomogram construction
Regression analysis was conducted to estimate the amounts of TT and FT gained from weight loss, and used to construct a set of parallel-scale nomograms for men of all ages (Figure 6A) or for two different age groups, stratified at 40 years (Figure 6B,C).An estimated amount of TT or FT to be gained from weight loss could easily be inferred from nomograms by drawing a line (an isopleth) connecting an individual's current BMI to the desired target BMI.A similar set of nomograms was constructed for gain in free testosterone in relation to weight loss for all ages (Figure 6D) and for the two age groups (Figure 6E,F).Compared

DISCUSSION
In this meta-analysis of 2,159 datasets from 1,774 participants, including 1,317 datasets from 1,021 participants with measurement of free testosterone, we observed that weight loss by LCD, and more so by BS, significantly increased the levels of TT as well as FT.Multivariate regression analysis showed that greater amounts of TT and FT could be gained by weight loss in men relatively greater with higher baseline BMI, or lower levels of SHBG, TT and FT.On the other hand, gain in TT was relatively greater in older men and FT in younger men.The set of nomograms constructed from these data provide clinicians with an evidence-based and easy-to-use visual tool in a clinical setting, for approximating the amount of testosterone a patient could gain from a certain amount of weight loss.As far as we are aware, there are no such studies available in the existing literature.
Findings from this study and previous studies 10,61 support the benefit of weight loss leading to a gain in TT and FT levels, as well as improvement in many obesity-related conditions including type-2 diabetes, metabolic syndrome, 9 OSA 62 and osteoarthritis. 63Thus, changes in testosterone could serve as a barometer for an overall health status of a male individual.A meta-analysis conducted by Cignarelli et al. had shown that continuous positive airway pressure treatment for OSA in obese men did not alter their testosterone levels. 64This evidence indicates that testosterone and OSA, like other metabolic disorders, are among complications of obesity rather than causally related, and all of which could be improved by weight reduction.Our findings of an overall gain in TT and FT from weight loss are broadly in agreement with older and smaller meta-analyses of 24 papers comprising 479 participants (but only seven papers available at the time for analysis of free testosterone). 10These authors derived sues, thus is clinically more relevant than TT. 65,66This is because about 70% of TT is bound with high affinity to SHBG, and about 20%−30% is bound weakly to albumin.SHBG levels progressively increase with age but decline with increasing degree of adiposity, 67 while paradoxically, weight loss induces an increase in SHBG levels as observed in our Although weight loss has been shown to improve sexual function, 70 it remains unclear whether the concomitant gain in testosterone has a direct effect. 71This relationship is complicated by conditions such as diabetes, hypertension, and dyslipidaemia that frequently coexist with obesity.Such conditions are risk factors for microvascular disease and may also contribute to sexual and erectile dysfunction.Therefore, weight loss may not reverse erectile and sexual dysfunction in all individuals, even if testosterone levels are sufficiently regained, or when exogenous testosterone is given.
The relationships between obesity and its many secondary complications, including hypogonadism, are complex and may be bidirectional.
Previous studies have shown that obesity-induced hypogonadism appears to be worse in the presence of diabetes and metabolic syndrome, while hypogonadism itself is a risk for these conditions. 72,73pogonadotrophic hypogonadism in obese men is thought to be due to enhanced aromatase activity from excess adipose tissue, leading to greater conversion of androgens to estrogens, which in turn inhibit LH pulse amplitude through negative feedback to the anterior pituitary.
Consequently, the lack of LH signal to the testes leads to a reduction in testosterone production. 74Increased levels of circulating estrogens also promote gynaecomastia, a condition that occurs more frequently with higher BMI. 75The finding in this study of increased circulating LH We do recognize that weight loss may not always increase the levels of TT or FT to reference ranges.Thus, men who have not achieved sufficient testosterone through weight loss may benefit from TRT. 61 However, caution should be taken due to adverse effects associated with exogenous testosterone including prostate hypertrophy, which may increase the risk of prostate cancer, gynecomastia, or mood swings.
Amongst obese men, OSA is common and many are undiagnosed. 76T in the presence of OSA increases the risk of polycythemia, 77 which predisposes such individuals to thrombotic events. 78Weight loss, as the first line treatment, therefore serves an additional purpose in obese hypogonadal men who consider TRT in a safer manner by reducing aromatase activity and the risk of gynecomastia, as well as OSA and related polycythemia.
As expected for a meta-analysis, certain limitations were encountered in this study, and include differences in the duration and methods of intervention, namely LDC and BS.However, the two methods provide a wide range of weight losses, making it more suitable for regression analysis and development of nomograms.Details on compliance to LCD were not reported by most studies.The impact of weight loss on testosterone has been debatable because of conflicting findings between studies.A number of factors could contribute to these discrepancies, primarily the wide variation in study designs and smallsample sizes.For example, the duration of weight management varied from a few weeks to several months.The minimum time for full recovery of pituitary-gonadal function from weight loss is not known but results from studies of a few weeks would be unlikely to have a complete effect on the pituitary-gonadal axis.Studies have indicated that TT levels increased soon after weight loss and continued to increase up the end of the study period of 52 weeks, while free testosterone started to rise from 12 weeks of weight loss. 30The strengths of this study lie in its large number of studies and participants, comprising a wide range of age, BMI and TT and FT, as well as a wide range of weight loss.This has allowed the construction of age-stratified nomograms.

CONCLUSIONS
This meta-analysis of data, accumulated from a large number of participants and datasets, showed androgens, particularly free testosterone levels increased after weight loss among men.Those with higher baseline BMI, or lower levels of SHBG, TT and FT benefit the greatest testosterone gain, while younger men gained relatively greater amounts of free testosterone induced by weight loss.Nomograms constructed from large number of participants of a wide range BMI provide an evidence-based and simple-to-use tool for clinician in a clinical setting.
according to IBAN, International Bank Account Number (https://www.iban.com/country-codes);SD, standard deviation; BMI, body mass index; TT, total testosterone; FT, free testosterone; LH, luteinizing hormone; LCD, low calorie diet; BS, bariatric surgery.F I G U R E 1 Quality of reporting of meta-analyses (QUOROM) flow chart of literature search.F I G U R E 2 Distribution of percentage change in body mass index after intervention by low calorie diet (upper panel) and by bariatric surgery (lower panel).

sion 5 . 3 .
Copenhagen: The Nordic Cochrane Centre, the Cochrane Collaboration, 2014).The standardized mean difference (SMD, also known as Cohen's d) was used to determine the effect size on testosterone and SHBG to accommodate for a variety of ways by which they were measured.The SMD expresses the size of the intervention effect in each study relative to the variability observed in that particular study. 15Positive values indicated a gain of the variable with the intervention.The mean difference (MD) was used on the original scale of measurement to determine the effect size on TT and FT levels, as well as SHBG and LH.Pooled estimates of outcomes were used to develop regression equations based on the values of baseline BMI and target BMI (after weight loss from an intervention) to estimate the percentage and amounts of TT (nmol/L) and FT (pmol/L) F I G U R E 3 Changes in total testosterone (A), SHBG (B), free testosterone (C), and LH (D) in relation to weight loss in obese men in cohort studies and in RCTs.LH, luteinizing hormone; RCT, randomized control trials; SHBG, sex-hormone binding globulin.

F I G U R E 3
Continuedto older men (> 40 yr), younger men (≤ 40 yr) gained less TT but more FT for a given amount of weight loss.

F I G U R E 4
Changes in total testosterone (A) and free testosterone (B) in relation to weight loss in obese men in RCTs.RCT, randomized control trials.several univariate regression equations and found higher amount of TT testosterone gained in younger age.Our meta-analysis comprised more up-to-date publications, with a wider age-range, and substantially more papers (n = 42) as well as participants (n = 1,736).This has enabled us to examine the effects of weight reduction on TT as well as FT in RCTs, and further extend our work to the construction of evidence-based nomograms.In a busy clinic, nomograms are convenient both for the clinician and patient in the discussion of target BMI and the expected gain in testosterone from weight loss in a realistic manner.Of importance, our study also consisted of one of the largest collections of papers (n = 28) and participants (n = 1,021) on changes to FT levels induced by weight loss.FT is the unbound and the bioactive form of sex steroid hormone which has a pleiotropic action on body tis-

F I G U R E 4
Continuedand previous studies.68Changes in SHBG therefore play a key role in the availability of circulating FT after weight loss.Examination of the age stratified nomograms from our study revealed that compared to men > 40 years old, men aged ≤ 40 years gained lower amount of TT (Figure6B,C), but greater amount of free testosterone (Figure6E,F) for any given amount of weight reduction.This observation has not been well-documented in the current literature and may be explained by the influence of age on differential increases in SHBG induced by weight loss.This notion is supported by studies showing the increase in SHBG after weight loss was not as much for younger men as for older men.69This is confirmed by our observation of multivariate regression analysis of age on increases in SHBG after weight loss (adjusted for baseline BMI, TT and FT), showing a positive regression coefficient of 0.687 (i.e., relatively greater increases in SHBG induced by weight loss with increasing age).Therefore, despite a smaller gain in TT, younger men gain greater amount of unbound free testosterone after weight reduction.These observations emphasize the importance of assessing FT changes after weight loss, which is more clinically relevant than TT.

F I G U R E 5 F I G U R E 5
Risk of bias of cohort studies evaluated by ROBINS-I tool for low calorie diet (A) and bariatric studies (B), and RCTs evaluated by Cochrane Collaboration's tool (C).RCT, randomized control trials.Continued levels after weight loss is consistent with that from previous studies.10These observations indicate that gonadal function recovery in obese men is possible by relieving central inhibition induced by factors associated with excess adiposity, such as excess estrogens from increased aromatase activity.
Sieu Han created the study concept and design.Thang Sieu Han and Gie Ken-Dror reviewed the literature and performed data collection.Gie Ken-Dror performed data analysis under the guidance of F I G U R E 6 Nomograms for approximating a gain in total testosterone in obese men of all ages (A), ≤ 40 years (B) and > 40 years (C), and gain in free testosterone in obese men of all ages (D), ≤ 40 years (E) and > 40 years (F) after weight loss.F I G U R E 6 Continued Thang Sieu Han.Thang Sieu Han wrote the first draft of the manuscript.Christopher Henry Fry, David Fluck, and Thang Sieu Han edited subsequent versions.All authors checked, interpreted the results, and approved the final manuscript. 14 19

Influencing factors of gain in SHBG
In total, there were 2,159 datasets as some studies included several datasets from the same group of participants, but at different time points. The rlowest to highest) of baseline mean age, 21-68 years; BMI.26.2-71.2kg/m 2 ; TT, 7.0-20.2nmol/L;andFT,140-583pmol/L,wererecorded.Among all 44 papers, one used both LCD and BS treatments.Weight loss was controlled by LCD (n = 735 participants, 988 datasets) in 19 studies, and by BS in 26 studies (n = 1,039 participants, 1,171 datasets).Among LCD studies, 11 were cohort studies and eight RCT, and among BS papers, 20 were cohort studies and six RCT.Among all 44 papers, one used both LCD and BS treatments.There were 28 studies (11 controlled by LCD and 17 by BS) comprising 1,021 par-SHBG, TT and FT (negative regression coefficients).A greater gain of TT was observed in older age, while a greater gain in FT was seen in younger age.SHBG increased after weight loss among men with higher baseline BMI and TT, lower baseline SHBG and FT, and increasingly older age (Table2).TA B L E 2Multivariate regression analysis to assess individual factors influencing a gain in total and free testosterone after weight loss in obese men.Abbreviation: SHBG, sex hormone binding protein.