Metabolic Effects of Testosterone Replacement Therapy in Patients with Type 2 Diabetes Mellitus or Metabolic Syndrome: A Meta-Analysis

Background Testosterone replacement therapy (TRT) is commonly used for the treatment of hypogonadism in men, which is often associated with type 2 diabetes mellitus (T2DM) and metabolic syndrome (Mets). Recent compiling evidence shows that TRT has beneficial metabolic effects on these patients. Objective A meta-analysis has been conducted to evaluate the effects of TRT on cardiovascular metabolic factors. Methods We conducted a systemic search on PubMed, Embase, Cochrane Library, Wanfang, and CNKI and selected randomized controlled trials (RCTs) to include. The efficacy of TRT on glycemia, insulin sensitivity, lipid profile, and body weight was meta-analyzed by Review Manager. Results A total of 18 RCTs, containing 1415 patients (767 in TRT and 648 in control), were enrolled for the meta-analysis. The results showed that TRT could reduce HbA1c (MD = −0.67, 95% CI −1.35, −0.19, and P=0.006) and improve HOMA-IR (homeostatic model assessment of insulin resistance) (SMD = −1.94, 95% CI −2.65, −1.23, and P < 0.0001). TRT could also decrease low-density lipoprotein (SMD = −0.50, 95% CI −0.82, −0.90, and P=0.002) and triglycerides (MD = −0.64, 95% CI −0.91, −0.36, and P < 0.0001). In addition, TRT could reduce body weight by 3.91 kg (MD = −3.91, 95% CI −4.14, −3.69, and P < 0.00001) and waist circumference by 2.8 cm (MD −2.80, 95% CI −4.38, −1.21 and P=0.0005). Erectile dysfunction (measured by IIEF-5) did not improve, while aging-related symptoms (measured by AMS scores) significantly improved. Conclusions TRT improves glycemic control, insulin sensitivity, and lipid parameters in hypogonadism patients with T2DM and MetS, partially through reducing central obesity.


Introduction
Male hypogonadism is defined as insufficient testosterone due to variable pathology in any part of the hypothalamuspituitary-testes axis [1]. It presents with primary symptoms of decreased libido, erectile dysfunction, and infertility. e incidence of male hypogonadism increases with age [2], with the prevalence in men aged 30-79 years at 3.1-7.0%, which increases markedly to 18.4% among men over 70 years [3]. Hypogonadism has a profound negative impact on both the physical health and the life satisfaction for middle-aged and elderly men.
Previous studies have reported a complex relationship between low testosterone levels and deteriorating metabolic status, such as hyperglycemia, obesity, and poor lipid profile [3]. e incidence of androgen deficiency in male patients with metabolic syndrome (MetS) and type 2 diabetes mellitus (T2DM) is significantly higher than that in the normal population [4]. Studies have found that the prevalence of hypogonadism in T2DM is 46% (n � 333) [5] in Poland and 33.1% (n � 112) in China [6]. MetS is a group of clinical syndromes consisting of obesity, hypertension, hyperglycemia, dyslipidemia, and other metabolic disorders. It is a major risk factor for T2DM and cardiovascular diseases.
Considering the close interlinked relationship between diabetes, obesity, and hypogonadism [6,7], we believe that further investigation in the causality among these factors is worthwhile.
Testosterone replacement therapy (TRT) is the primary treatment for male hypogonadism [8]. It has been confirmed that TRTcan improve the symptoms of hypogonadism [1,8]; however, the metabolic effects of this treatment on male hypogonadism remain controversial. Evidence supporting TRT improvement of glucose control, lipid profile, and weight control is still insufficient. For this reason, screening for testosterone deficiency and TRT for men with T2DM and MetS is not routinely recommended in some countries [9]. In recent years, compiling evidence consistently shows that TRT can improve metabolic factors, a compelling issue that leads us to conduct this meta-analysis.

Materials and Methods
Following the reporting recommendations made by the PRISMA (Preferred Reporting Items for Systemic Reviews and Meta-Analyses) statement, we met all 27 items stated therein in our study.

Inclusion Criteria.
e inclusion criteria are fully published, randomized, and controlled clinical trials; aiming to evaluate the metabolic effects of TRT on patients with MetS and/or T2DM; language in English or Chinese.

Exclusion Criteria.
e exclusion criteria are (1) case reports and reviews, (2) language in non-Chinese or non-English, (3) republished articles, (4) literature studies in which data were missing or unavailable, and (5) literature studies that did not provide standard deviation or quartile spacing.
Groups: the TRT group was defined as patients treated with testosterone supplementation. e control group was defined as a blank or receiving placebo.

Search Strategy.
e search strategy was designed by an expert on public health ( Figure 1). Databases, including PubMed, Embase, Cochrane Library, CNKI, and Wanfang, were searched in order to identify qualified trials published up to Dec 2019. Potentially eligible studies on the reference lists were searched by hand. Meta-analysis was conducted by two independent reviewers (Li and Zhao). Discrepancies were resolved by discussion or submitted to a third party.

Data Extraction.
Two investigators, Li and Zhao, extracted the relevant data independently using a standardized form. e data included demographic information, diagnosis of T2DM and/or MetS, number of participants, baseline testosterone levels, therapeutic regimen, and treatment duration. Differing opinions were resolved by consulting a third party.

Quality
Assessment. Methodological quality evaluation was performed using a quality evaluation tool recommended by the Cochrane Reviewers' Handbook ( Figure 2).

Primary Outcome.
e primary outcomes following TRT were improvement of glycemia, lipid parameters (HDLc, LDLc, and triglyceride), body weight, and waist circumference.

Secondary Outcome.
e secondary outcome upon TRT was improvement of sexual function, as assessed by the International Index of Erectile Function-5 (IIEF-5) scores, as well as symptoms of senescence, as assessed by the Aging Males Symptoms (AMS) scores. In addition, we also evaluated the possible adverse effects of TRT, including blood pressure, prostate-specific antigen (PSA), hemoglobin, and hematocrit.

Data Synthesis and Statistical Analysis.
e two investigators, Li and Zhao, independently analyzed the extracted data. Review Manager (RevMan5.3) software was used to generate the meta-analysis of the clinical efficacy from various studies. e heterogeneity of the included literature studies was firstly analyzed by the I 2 test. e fixed-effect model was adopted if I 2 was <50%. Otherwise, a randomeffect model would be adopted. A funnel plot was used to analyze publication bias. In clinical studies, if a standard deviation (SD) was not provided, it was calculated using the following formula: the required correlation coefficient Corr was directly obtained or obtained through the included studies [10][11][12]:

Ethical Approval.
is meta-analysis review does not require approval from an ethics committee.
A total of 15 studies (676 patients in TRT and 588 patients in control) were enrolled in order to clarify the effect of TRT on fasting blood glucose (FBG). e findings revealed that TRT significantly changed FBG by −0.86 mmol/l (95% CI −1.15, −0.56, and P < 0.001) (Supplementary Figure S3). We next conducted subgroup and sensitivity analysis according to the duration of TRT. e results indicated that the studies by Dhindsa et al. [17], Khripun et al. [15], Heufelder et al. [26], and Yang [23] were the sources of heterogeneity (Supplementary Figure S4).

e Effects of TRT on
Twelve studies (564 patients in TRT and 511 patients in control) were included to clarify the effect of TRTon LDLc. e data showed a significant reduction in LDLc (SMD � −0.50, 95% CI −0.82, −0.19, and P � 0.002) ( Figure 5).  Changes in the lipid profile following TRT are summarized in Supplementary Table S2.

e Effects of TRT on Body Composition (Weight, BMI, and WC).
Seven studies (343 patients in TRT and 299 in control) were included for the evaluation of TRT's effect on body weight. We found that the inter-trial heterogeneity was not significant (Cochrane Q-test P � 0.68, I 2 � 0%), and so we adopted a fixed-effect model for combining the data. Testosterone therapy resulted in an average weight loss of 3.91 kg (MD � −3.91, 95% CI −4.14, −3.69, and P < 0.00001) (Supplementary Figure S8).
We also included a total of 14 studies (620 patients in TRT and 540 in control) to investigate the change in BMI after TRT. e results showed that TRT significantly reduced BMI (MD � −0.81, 95% CI −1.21, −0.42, and P < 0.0001) ( Figure 6).
Additionally, we evaluated the change in waist circumstance after TRT by including 14 studies (656 patients in TRT and 539 patients in control). We found that TRT significantly reduced waist circumstance by 2.8 cm (MD � −2.8, 95% CI −4.38, −1.21, and P � 0.0005) (Supplementary Figure S9). e changes in weight, BMI, and waist circumstance are summarized in Supplementary Table S3.

e Effects of TRT on Erectile Function and
We included a total of 6 studies (362 patients in TRT and 325 patients in control) for the evaluation of TRT's effects on AMS scores. We found that testosterone therapy significantly reduces the AMS score (MD � −4.65, 95% CI −8.76, −0.54, and P < 0.0001) (Supplementary Figure S11), indicating that TRT improves aging-related symptoms.

4.7.
e Safety of Testosterone Replacement erapy. Changes in systolic blood pressure (SBP) following TRT were evaluated through the inclusion of 12 studies (465 patients in TRT and 393 controls). We found that TRT did not significantly influence SBP (MD � −0.31, 95% CI −0.89, 0.28, and P � 0.91, data not shown).
Furthermore, changes in diastolic blood pressure (DBP) after TRT were evaluated through the inclusion of 13 studies (478 patients in TRT and 419 controls), which revealed no significant difference after TRT (MD � −0.67, 95% CI −3.21, 1.88, and P � 0.61, data not shown).
We then evaluated the effect of TRT on PSA by examining 3 studies (122 patients in TRT and 91 in control). Because the inter-trial heterogeneity was not significant (Cochrane Q-test P � 0.14, I 2 � 0%), we adopted a fixedeffect model for combining the data. e results showed that TRT did not increase PSA (MD � −0.12, 95% CI −0.06, 0.30, and P � 0.20 ) (Supplementary Figure S12).
Changes in the blood pressure, PSA, and hemoglobin after TRT are summarized in Supplementary Table S4.

Risk of Publication Bias.
Publication bias was evaluated by examining the change of HbA1c after TRT. e funnel plot showed that the distribution was symmetrical with slight publication bias (Figure 7).

Discussion
We employed rigorous inclusion criteria in this metaanalysis, ultimately including 18 studies. We then set out to systemically investigate the effects of TRT on HbA1c, LDLc, body weight, waist circumstance, blood pressure, and hemoglobin. Our analysis concluded that TRT had favorable metabolic effects on glycemia control, lipid profile, and weight loss.
Our findings indicated that testosterone supplementation could improve glycemia control. is is based on our observation that TRT reduced HbA1c by 0.67%, fasting blood glucose by 0.86 mmol/L, and fasting insulin and insulin resistance index (HOMA-IR) by 1.23. ere was significant heterogeneity in the meta-analysis of HbA1c, and sensitivity analysis revealed that the heterogeneity was primarily derived from differences in treatment duration. Subgroup analysis according to treatment duration showed that HbA1c did not improve in patients who underwent TRT for 6∼12 months, possibly due to the small sample size. Another subgroup analysis, stratified by baseline HbA1C, showed that patients with higher baseline HbA1c values would have a greater reduction in HbA1c. Increasing insulin sensitivity by testosterone may be explained by various mechanisms: (1) testosterone can upregulate the expression of the insulin receptor, insulin receptor substrate 1, and GLUT4 [28,29] and (2) testosterone inhibits lipoprotein lipase activity and thus reduces triglyceride flowing into adipocytes. Visceral adipocytes express more androgen receptors than subcutaneous adipocytes [14]; (3) testosterone increases antiinflammatory cytokine IL-10 and decreases proinflammatory cytokines, such as IL-1b, IL-6, and TNF-a. Suppression of the inflammatory state may improve insulin sensitivity [30].  We also found that TRT significantly decreased TG and LDLc levels. Considering their atherosclerogenic effects [31], decreasing TG and LDLc may have promising protective effects on cardiovascular status. Despite this, the role of TRT on HDLc remains controversial. Some studies have shown that TRT may increase HDLc levels [32], while others have not [33]. Our results supported that TRT would not significantly influence HDLc levels. NonRCTstudies concluded that testosterone replacement therapy can reduce LDLc and increase HDLc [34]. ese results were not fully shown in our research, because it is a nonRCT study, and improvement in dyslipidemia would be achieved after a long period of TRT.
Furthermore, our analysis indicated that TRT resulted in a remarkable reduction in weight, BMI, and waist circumference, reflecting that testosterone therapy would improve abdominal (central) adiposity. On average, TRT leads to a weight loss of 3.91 kg, equal to 4-5% of body weight. is weight loss could partially explain the beneficial effect of TRTon glucose metabolism [32]. Long-term TRT (8 years) can reduce body weight by about 10% in patients with hypogonadism and prediabetes and can successfully prevent the progression to T2DM [35]. e lower reduction of body weight in our meta-analysis, 4-5% as opposed to 10%, is possibly due to the short duration of TRT (3-24 months). Longer treatment periods could result in lower body weights and may thus bring a further reduction in cardiovascular risks. A low level of testosterone is associated with obesity. Obesity may reduce testosterone levels by conversion of testosterone to estrogen in adipose tissue. On the other hand, testosterone deficiency may slow down the metabolism of triglycerides and increase the accumulation of adipose [36,37]. e supplementation of testosterone leads to weight loss by increasing metabolic function and energy utilization [36,38,39].
TRT may partially improve hypogonadism symptoms. e reduction of the AMS score (denoting improvement in aging-related symptoms) in this meta-analysis is consistent with the concept that TRT may improve the androgen deficiency associated with the symptoms of aging [9]. To our disappointment, TRT did not significantly improve the IIEF-5 score (erectile function) in this meta-analysis. Of all 5 studies included, only one by Gianatti showed a negative effect on erectile function resulting from TRT, which influenced the final result. On the one hand, multiple factors are involved in erectile dysfunction [40]. It is possible that neurological and microvascular complications associated with aging and diabetes, rather than low testosterone levels, may be the dominant pathology for erectile dysfunction [12]. ese patients would have poor response to TRT. Too short period of observation in RCT studies may be another reason for no improvement in the IIEF-5 score. erefore, the effect of TRT on erectile function is still uncertain, and long-term RCT studies are needed.
No significant change was observed in systolic and diastolic blood pressure after TRT [10]. One RCT revealed that TRT over a long period (60 months) might lower blood pressure. A recent nonRCT study has confirmed that longterm TRT can result in a significant improvement in arterial stiffness and blood pressure control [41]. Unfortunately, due to few randomized controlled trials, the comprehensive effect of TRT on the vascular wall was not evaluated in our meta-analysis. e controversy of TRT on CVD is still existing [42]. In recent years, many evidence showed that normalized testosterone is a protective factor for cardiovascular disease [34]. It is found that TRT can reduce the incidence of CVD by improving glycemic and lipid metabolism. A study, lasting for 8 years and focusing on testosterone treatment on CVD, found that TRT reduced not only the incidence of CVD but also all-cause mortality and CVD-induced mortality [41].
TRT can consistently increase hemoglobin and hematocrit, as indicated by our study. Bone marrow in the elderly is more sensitive to testosterone than that in younger men [1]. It has been shown that, during 6∼12 months of TRT, hemoglobin increases and then remains within the normal range [43].
Some studies have brought forth concerns regarding TRT-related prostate hyperplasia and cancer [44]. Our meta-analysis revealed that the physiological dosage of testosterone does not result in an increase in the PSA biomarker. Guidelines were developed to assess the risk of prostate cancer before or during TRT [44,45]. However, at least three studies found that TRT may reduce the incidence of prostate cancer (1.08% in the TRT group vs. 7.35-9.6% in the control group) [34,46], possibly by reducing the risk factors, such as antiinflammatory effect and weight loss [47]. e final effects of TRT on prostate cancer should be clarified by future high-quality and large-scale RCTs.
A study found that 97.2% of patients with testosterone deficiency may have multiple comorbidities, including dyslipidemia, hypertension, obesity, type 2 diabetes, and chronic obstructive pulmonary disease (COPD) [39]. TRT can improve all these comorbidities by its beneficial effect on metabolism. TRT may improve COPD by improving respiratory muscle function and by strengthening exercise capacity [48][49][50][51]. More studies are needed to further evaluate the effect of TRT on COPD. Some limitations should be addressed. First, most of the participants in this study have moderate diabetes. us, the effect of TRT on patients with poor glycemic control cannot be assessed herein. Second, because most of the patients included are obese, it is, therefore, prudent to extrapolate our conclusion to the nonobese population with T2DM and MetS. ird, the analysis included different routes of administration and dosages of testosterone therapy, which may have influenced the effects of TRT on metabolic factors. Fourth, when PSA and hemoglobin were investigated, only 3 studies were included, yielding results that may potentially be unreliable. Finally, hard indicators of cardiovascular diseases, such as myocardial infarction, heart failure, and cerebral infarction, were not fully evaluated. Studies that include a large sample size and long-term follow-up are needed to demonstrate the favorable effects of TRT on the cardiovascular system.

Conclusion
TRT can improve multiple cardiovascular risk factors, including blood glucose control, insulin sensitivity, dyslipidemia, and central obesity. Short-term TRT is safe; however, more high-quality RCTs are needed to fully clarify the effects of long-term TRT on cardiovascular disease.

Data Availability
e data used to support the findings of the study are available from the corresponding author upon request.

Ethical Approval
All the included studies are published and need not to be approved by the ethical committee

Conflicts of Interest
All authors declare that they have no conflicts of interest.

Authors' Contributions
Li Shu-ying and Zhao Ya-ling contributed equally to this work.