Neutral effect of exenatide on serum testosterone in men with type 2 diabetes mellitus: A prospective cohort

Endogenous testosterone increases with weight loss from diet, exercise, and bariatric surgery. However, little is known about testosterone levels after weight loss from medication.


| INTRODUC TI ON
Male hypogonadism has become a common diagnosis, and testosterone therapy is often prescribed as a first line option without full consideration of adverse effects and alternative strategies to treat underlying causes. Potential complications of testosterone therapy may include polycythemia, 1 worsening of obstructive sleep apnea, 2 and cardiovascular disease. 3 An increased risk of stroke has been reported in several studies, particularly within the first year of testosterone therapy (3; 4). Cardiovascular risks or benefits of testosterone therapy are not proven and large randomized controlled trials with this focus are needed. 5 Men with diabetes mellitus have an increased risk of cardiovascular disease and hypogonadism, 6 raising concern about the safety of testosterone therapy in this population.
Weight loss from diet, exercise, and bariatric surgery consistently increases serum testosterone in men with functional hypogonadism, 7 and may represent a preferred initial strategy over exogenous testosterone in select patients. 6 In addition to diabetes mellitus, obesity and metabolic syndrome are associated with functional hypogonadism. 8 Weight loss in patients with these disorders is a fundamental component in disease management, improving both disease progression and other quality-of-life factors, to include gonadal function. Many diabetes medications also affect weight with a predictable impact on testosterone levels. For example, pioglitazone promotes weight gain and appears to lower testosterone; 50 eugonadal men taking pioglitazone for 6 months gained an average of +1.9 kg and their testosterone levels significantly decreased. 9 In contrast, metformin facilitates weight loss and is associated with increased testosterone. There was a study of 35 eugonadal men with metabolic syndrome who took metformin for 4 months. They lost an average of −2.3 kg, serum total testosterone increased on average by +44 ng/ dL, and free testosterone increased on average by +4.6 pg/mL. 8 Glucagon-like peptide-1 receptor agonists (GLP-1 RA) typically result in a −2 to −3 kg weight loss in people over 3 to 12 months. 10 However, there is little information about how GLP-1 RA therapy affects testosterone. One study 11 showed a decrease in diurnal variations of serum testosterone in men taking GLP-1 RA therapy over a 24-hour period; this trial did not have a sufficient duration to affect weight. There was a retrospective study of 30 men with obesity, diabetes, post-puberty hypogonadism, and erectile dysfunction who had been making lifestyle changes as well as taking testosterone and metformin for a year. 12 Of this group, 16 men added liraglutide for an additional year. Their total testosterone values increased from 466.1 ± 63.6 ng/dL to 481.7 ± 57.3 ng/dL but their free testosterone did not significantly change. Their weight decreased from 99 ± 7.6 kg to 93.7 ± 6.3 kg, and HbA1c values improved from 8.3 ± 0.3% to 7.3 ± 0.3%. They also reported improvement in erectile function. A more recent study 13 showed exenatide therapy for 3 months had beneficial effects on serum testosterone.
It was unclear if weight loss from GLP-1 RA therapy stimulates testosterone production or if decreases in testosterone diurnal variations significantly impact testosterone levels. The aim of our study was to uncover the 6-month biochemical effects of GLP-1 RA therapy on serum testosterone. with consideration of the rising rates of T2DM in the U.S. population. The 2020 National Diabetes Statistics Report highlighted the continued increased incidence rates of T2DM in children and young adults. 14 Obesity has become more prevalent in younger males over time as well. Males younger than 18 years of age were not included in this study, due to differences in normal testosterone ranges for pediatric patients.

| MATERIAL AND ME THODS
Eligible participants met with a member of the study staff to obtain written informed consent and schedule fasting baseline laboratory studies. They provided Health Insurance Portability and Accountability Act (HIPAA) authorization to allow study staff to collect study-related data from the electronic medical record.
Participants' names and identifiers were recorded in a master key associated with a unique identifier to connect data to the participants.
Each participant received at least 6 months of GLP-1 RA therapy to allow sufficient time for GLP-1 RA to reach full therapeutic effects and for patients to reach their nadir weight. Patients in the study were followed for up to 12 months. Study staff collected the following information during screening and after at least 6 months of GLP-1 RA therapy: height (baseline only), weight, HbA1c, total testosterone, calculated free testosterone, sex hormone binding globulin (SHBG), albumin, current medications, and other medical conditions. The calculation for free testosterone was the Vermeulen equation. Participants' data were stored on a limited access, password-protected computer network.
Power analysis estimated means and standard deviations for total testosterone, free testosterone, and SHBG based on data in a prior study about testosterone before and after 6 months of weight loss from lifestyle changes. 15 In that study, participants lost on average −16.3 kg. SHBG went from 27.6 ± 11.9 nmol/L to 32.6 ± 12.9 nmol/L; the pooled standard deviation was 12.4 nmol/L, and effect size was 0.403 so the power analysis was that the number of participants per group would need to be 51 paired. Free testosterone went from 185 ± 66 pmol/L to 212 ± 84 nmol/L; pooled standard deviation was 75 pmol/L, and effect size was 1.7 so the power analysis was that the number of participants per group would need to be 5 paired. Total testosterone went from 12 ± 4 nmol/L to 14 ± 4 nmol/L; the pooled standard deviation was 4 nmol/L, and effect size was 0.5 so the power analysis was that the number of participants per group would need to be 34 paired. SPSS Sample Power 3.0 (IBM) estimated the required sample size for a power of 80% with a level of confidence of 95%. A correlation coefficient of <−0.50 was considered a good index of an inverse relationship between body weight and serum testosterone. According to the method of Kraemer and Thiemann, 16 detecting a correlation coefficient of <−0.50 required 51 participants.
Primary outcome variables were total testosterone, SHBG, and calculated free testosterone. Paired t-tests compared these variables at baseline and after at least 6 months of GLP-1 RA therapy.  Table 1 shows participant baseline characterizes. All participants started exenatide extended-release (ER) 2 mg weekly due to formulary restrictions. One participant switched to dulaglutide 0.75 mg weekly after 3 months, and the remainder continued exenatide ER 2 mg weekly throughout the study.

| RE SULTS
Over the study period, participants on average lost −2.27 kg  Table 2).
Subgroup analysis did show a significant increase in total testosterone for men who had an HbA1c improvement ≥1% compared to those with HbA1c change <1% (p = 0.0413). HbA1c improvement of ≥1% occurred in men with higher baseline HbA1c (mean value of 8.96% versus 7.7%). Total testosterone increased from 351.6 ± 123.9 ng/dL to 394.4 ± 136.5 ng/dL for men with HbA1c improvement ≥1%. In contrast, total testosterone decreased from the baseline of 331.8 ± 128.6 ng/dL to 316.1 ± 126.2 ng/dL for men with a HbA1c change <1% (Figure 1). There was no significant change in free testosterone from baseline regardless of A1c change   There were large variations in testosterone values among participants. Notably, there was no significant relationship between testosterone levels and weight loss. The strongest, but not significant, relationship was between improvements in total testosterone and improvements in HbA1c.

Figure 4 depicts bar graphs comparing men who had HbA1c
improvements ≥1% to those with HbA1c change <1%, breaking it down by change in total testosterone, free testosterone, SHBG, and weight. This analysis also showed that change in testosterone was not related to the small weight loss in this study. The bar graphs did show a significant change for total testosterone with HbA1c improvement ≥1% compared to HbA1c change <1%; it did not show this relationship for free testosterone. The change in SHBG accounted for the differences between total and free testosterone; the increase in SHBG correlated with the increase in total testosterone while the free testosterone did not change.
Initially, subgroup analysis by t-test appeared to show a difference for participants who were older or younger than 60 years of age; older participants appeared to have greater increases in testosterone. However, the confirmatory Wilcoxon test did not show statistical significance ( Figure 1).
Interestingly, subgroup analysis showed a statistically significant difference for men whose baseline testosterone was <320 ng/dL compared to ≥320 ng/dl (p = 0.0172). We selected this threshold because one study of 3369 participants 17 showed that symptoms of sexual dysfunction were more likely to be present when total testosterone values were less than 320 ng/dL. In our study, total testosterone increased from 238. 5 (Figure 1).
Similarly, the change in free testosterone was significantly different between men whose baseline testosterone was <320 ng/dL compared to ≥320 ng/dL (p = 0.00807). Free testosterone increased from 55.2 ± 12.8 pg/mL to 57.2 ± 17.6 pg/mL for men with baseline total testosterone values <320 ng/dL.
Free testosterone started at 74.7 ± 16.3 pg/mL and decreased to 64.2 ± 17.7 pg/mL for men with higher baseline total testosterone values (Figure 2).

| DISCUSS ION
Our study mirrors the results of a trial where 18 healthy men received GLP-1 RA therapy over an 8-hour infusion and a placebo infusion on separate days. 18 Participants in this placebo-controlled crossover study ate less after GLP-1 RA therapy than after placebo and had no significant change in testosterone levels. A similar study of 9 healthy men undergoing a GLP-1 RA infusion during a euglycemic clamp showed no change in free/total testosterone or luteinizing hormone (LH). 11 However, there was a decline in free/ total testosterone following an oral glucose tolerance test (OGTT).
A larger study 19 of 57 men showed a similar decline in testosterone and LH following an OGTT. In a separate OGTT study, this effect on testosterone and LH was present regardless of glucose tolerance. 20 Whether this acute effect has any long-term consequences on testosterone production remains unknown.
Our study contrasts with a trial of 176 men with T2DM who received metformin and either exenatide or a sulfonylurea for 3 months. 13 These participants received exenatide 5 μg twice daily for 4 weeks and then increased to 10 μg twice daily thereafter.
Investigators encouraged participants to monitor blood sugars, diet, and exercise on a monthly basis. The HbA1c for both groups improved by about 1.7%. Those receiving the GLP-1 RA lost more weight (−5.7 kg) than those receiving a sulfonylurea (−0.75 kg).
Additionally, total testosterone increased more in those receiving GLP-1 RA therapy (+121.72 ng/dL) than those receiving sulfonylurea (+34.67 ng/dL) but free testosterone change did not substantially differ between groups. SHBG increased more in those receiving GLP-1 RA therapy (+10.36 nmol/L) compared to those receiving sulfonylurea (+0.33 nmol/L).
SHBG is a transport molecule for testosterone; lower levels of SHBG lead to lower total testosterone levels. 21 Moreover, their subgroup analysis showed that participants who lost >5% of their body weight had the most significant increases in serum testosterone. This is consistent with a meta-analysis on weight loss interventions that found people who lost the most weight had the greatest increase in serum testosterone. 7 Weight loss is a relative-class effect of GLP-1 RA therapy.
However, the amount of weight loss with each GLP-1 RA is variable. It is minimal with albiglutide, moderate with exenatide, and robust with semaglutide. 27 Using an alternative GLP-1 RA to exenatide ER could show greater weight loss benefits, such as liraglutide 1.8 mg daily with weight losses of −3.57 kg versus −2.68 kg 29 or, even more so, semaglutide 1.0 mg weekly with weight losses of −5.6 kg versus −1.9 kg. 30 The weight loss appears to nadir 24-30 weeks and remains relatively stable thereafter for GLP-1 RA therapy.
There were significant testosterone increases in a recent study of 30 men with obesity who received either liraglutide 3 mg dailythe treatment dose for obesity-or testosterone replacement therapy for 16 weeks. 31 They found significantly more weight loss with GLP-1 RA therapy (−7.9 kg) than with testosterone replacement therapy (−0.9 kg) (p < 0.001). Total testosterone significantly increased with liraglutide (+75 ± 101 ng/dL) and with testosterone replacement therapy (+170 ± 208 ng/dL); the testosterone increase in both groups was not significantly different from each other (p = 0.239). Sexual function improved with both therapies.
They concluded that the GLP-1 RA therapy showed an overall health benefit for men with obesity and low total testosterone levels. This is an important finding in a time when many men are drawn to testosterone replacement therapy, believing it is the single best therapy for their symptoms.
Our study showed significant testosterone increases when HbA1c improvement ≥1% compared to HbA1c change <1%.The mechanism for diabetes improvement resulting in gonadal function improvement is not fully understood. In vitro studies showed that insulin stimulates the secretion of gonadotropin-releasing factors, gonadotropins, and testosterone. Hypogonadism may result because the hypothalamic-pituitary-gonadal axis does not fully respond in an insulin resistance state. Insulin resistance can lead to lower total testosterone and more frequent diagnosis of hypogonadism in men with obesity. 32 Additionally, insulin resistance and hyperinsulinemia can inhibit hepatic SHBG secretion in vitro and in vivo; SHBG influences the bioavailability of testosterone. 33 Furthermore, hyperglycemia can increase inflammation. There was a study of mice with type 1 diabetes who had progressive testicular dysfunction and a 30% reduction in testis weight compared to control mice. This effect was thought to be caused by inhibitory effects of chronic hyperglycemia mediated by testicular inflammatory cytokines. 34 It is possible that reversal of hyperglycemia improves inflammation and allows testosterone to increase. Free testosterone is the biologically active form and more consistently correlates with hypogonadal symptoms. 35 Obesity and T2DM are risk factors for hypogonadism associated with low free testosterone 36 so consideration should be given to measurement or calculation of free testosterone in this population.
Our study does have several limitations. The most significant was that as an observational study, there is no true control group.
To the best of our ability, we had patients act as their own control by assessing testosterone levels before and after GLP-1 RA therapy.
Participants were at liberty to make choices about diet, exercise, and sleep that could potentially impact their testosterone levels. Furthermore, targeted subgroup evaluation was not pre-specified in the original study protocol power analysis calculations. Therefore, our conclusions within the subgroup analysis, regarding patients with baseline total testosterone levels <320 ng/dL and HbA1c improvement ≥1%, are interesting, but would require further evaluation using appropriately powered subgroups to evaluate for significant changes.
Despite the limitation, similar small subgroup analysis has been seen in previously published data. 13 Moreover, the subgroup statistics has shed light onto potential future research focuses. In conclusion, this study showed that men taking GLP-1 RA therapy for T2DM lost on average of 2.3 kg in 6 months and had a 0.7% improvement in HbA1c without significant changes in serum testosterone levels. Furthermore, it appears there is greater potential for increasing total testosterone in men starting at lower values and in those with more robust glucose responses to GLP-1 RA therapy.

CO N FLI C T O F I NTE R E S T
The authors take public responsibility for the content of this manuscript and adhered to ethical standards. They declare that there are no conflicts of interest.

D I SCL A I M ER
The