Background
Aging is associated with a 1% decline in testosterone levels in males, though the causes remain unclear [
1]. Testosterone deficiency (TD) refers to a low level of serum testosterone and may induce a series of clinical symptoms [
2]. Androgen deficiency may lead to dysfunctions of the skeletal, reproductive, and cardiovascular systems. Patients with TD also seem to be at higher risk of sustaining fractures
3. An epidemiological study [
3] of 50,613 patients with prostate cancer who survived for at least five years reported a higher incidence of fractures in patients who received androgen-deprivation therapy (ADT) than in patients who did not (19.4% versus 12.6%,
p < 0.001).
Given the association between TD and fracture revealed by the observational studies mentioned above, it is believed that androgen supplementation therapy can prevent osteoporosis and increase bone mass. However, several randomized controlled trials (RCTs) failed to demonstrate that testosterone supplementation increases bone density in patients with TD [
4‐
6]. Furthermore, clinicians have also expressed concern about other associated risks of prescribing testosterone to middle–aged or aging patients with TD, especially the risk of cardiovascular and prostatic events [
7‐
11]. Whether testosterone supplementation increases the risk of cardiovascular events remains a focus of debate. Two large cohort studies [
9,
10] reported that testosterone therapy increases the risk of myocardial infarction. One RCT that enrolled 209 patients [
11] also reported that the application of testosterone gel was associated with an increased risk of cardiovascular events. However, in another RCT [
8], the authors found that the use of testosterone did not increase the risk of carotid artery intima-media thickness or coronary artery calcium in 308 men 60 years or older with low or low-normal testosterone levels.
There is also uncertainty among clinicians about whether testosterone supplementation in aging males is protective against other risks, such as all-cause mortality and prostate cancer. Although several systematic reviews [
7,
12‐
15] on this topic have been published, they did not fully address the above questions [
7,
12,
13]. While one review [
13] investigated the effect of testosterone replacement on patients’ quality of life, it did not investigate the effect of testosterone replacement on bone mineral density (BMD), cardiovascular disease, and all-cause mortality. Three reviews [
14‐
16] evaluated the efficacy of testosterone therapy in males with late-onset hypogonadism (LOH) and found that testosterone increased BMD. However, these reviews were either out of date or they omitted relevant studies; several RCTs reported no effect of testosterone on BMD after these reviews [
17,
18] were published. Given these conflicting results, an update of the evidence regarding the impact of testosterone supplementation on BMD is required. Two systematic reviews investigated the risk of cardiovascular events after testosterone therapy, but the findings were inconsistent. One review [
8] found that testosterone therapy increases the risk of cardiovascular events in aging males, while the other review [
16] simply made reference to the controversy surrounding this issue. Given that the evidence to date is both conflicting and insufficient, this systematic review aims to evaluate the effect of testosterone supplementation on BMD and its potential risks (fracture, falling, all-cause mortality, cardiovascular disease, and prostate events) in middle-aged or aging males with TD.
Methods
Materials and methods
We registered our protocol in PROPERO (CRD42018109738). The systematic review and meta-analysis (study level) were conducted in alignment with the Cochrane Handbook of Interventional Reviews and reported in accordance with the PRISMA standard.
Inclusion and exclusion criteria
Aging male adults (aged ≥40 years old) with a diagnosis of TD were included in this review. Because of the lack of a uniform definition of TD, we accepted any criteria used in the included studies to define TD. We only included studies involving patients with TD who were not interested in fertility and who were determined to have well-controlled obstructive sleep apnoea syndrome (OSAS). Any RCT in which testosterone therapy was used alone or in combination with other therapies (such as calcium or vitamin D) were included without restrictions regarding treatment dosage, frequency, and duration. Testosterone therapy might have included oral capsules, gels, patches, injections, pellets, sublingual testosterone. The comparator was placebo. The exclusion criteria were i) studies including patients with prostatic cancer who had received castration therapy (including endocrine therapy or testectomy) or androgen therapy; ii) studies including patients with testicular cancer; iii) studies including patients with primary hypogonadism induced by pituitary disease or pituitary surgery; iv) studies including patients with secondary hypogonadism (e.g., Paltauf’s dwarfism, pituitary tumour, acromegalia, or Cushing’s syndrome); and v) studies including patients who received other medications that influence androgen levels (e.g., finasteride, sildenafil).
Our primary outcome was total BMD. Secondary outcomes included lumbar spine BMD, total hip BMD, or other BMDs, the incidence rates of hip fracture, falling, total fracture, vertebral or non-vertebral fracture, all-cause mortality, and cardiovascular events (defined as myocardial infarction, angina, coronary artery disease, hypertension, stroke, or other definitions used in the original studies), as well as quality of life, total cost, sexual function, adverse events, prostate-specific antigen (PSA) level, and prostate events, such as prostate cancer or prostatitis.
Searching and study screening
We conducted electronic searches in MEDLINE, Cochrane Library, EMBASE and PubMed on 9 December 2019. The search strategy was developed by an information specialist and is presented in Additional file
1. There was no limitation on language, document type, and publication status. We also hand searched the references of relevant systematic reviews to identify additional RCTs for inclusion. Two reviewers screened the search results. Disagreements were resolved by discussion with assistance from a third party if necessary.
Data extraction and synthesis
Data from each study were extracted independently by two separate reviewers using a standardized data extraction form. Any disagreements were resolved by discussion with the assistance from a third party if necessary.
We synthesized data using a fixed-effect method for all analyses. An I2 estimate greater than or equal to 50% accompanied by a statistically significant χ2 statistic was interpreted as evidence of a substantial level of heterogeneity. Where substantial heterogeneity was found, we explored potential sources that may have caused this heterogeneity. If we could not definitively locate the sources of heterogeneity, we synthesized the data using a random-effects model. We summarised all dichotomous outcome data using risk ratios (RRs) and all continuous outcome data using mean differences (MDs) and calculated their respective 95% confidence intervals (CIs).
Risk of bias assessment
We made the risk of bias judgments based on the methods endorsed by The Cochrane Collaboration, which included the following domains: patient allocation, blinding, selective reporting, attrition of study participants, and any other detected sources of bias [
19].
Additional analysis
We assessed the quality of the body of evidence for the primary and secondary outcomes based on the GRADE approach [
20]. To test the robustness of the results of the synthesis, we conducted a trial sequential analysis (TSA) [
21] for the primary outcomes. The required information size (RIS) was calculated based on the empirical mean difference and variance with a two-sided alpha of 0.05 and a beta of 0.20 [
21].
Discussion
This review included 5067 participants with TD. Evidence showed that compared with placebo, testosterone supplementation did not i) increase total BMD, vertebral, hip and femoral BMD; ii) decrease the risk of falling or fracture; or iii) increase the risk of cardiovascular events, all-cause mortality or prostatic events, such as PSA increase or prostatitis; however, testosterone supplementation was associated with improved quality of life and sexual function. Nonetheless, the above findings may be influenced by the presence of attrition bias and selective reporting in individual RCTs. Furthermore, the small total sample size and the unexplained heterogeneity between studies also impacted the quality of the body of evidence, especially for long-term outcomes and the risk of cardiovascular events. In terms of sexual function and quality of life, the indirect approach used to interpret the results of the screening tools somewhat reduces our level of confidence in these findings. All the included studies used surrogate outcome measurements, namely the mean difference in the scores of each scale, to reflect improvement in these two outcomes. However, clinicians must also consider whether the differences in the scores between the two compared groups are clinically significant.
Testosterone receptors are widely distributed in bone tissues. When combined with these receptors, testosterone facilitates skeletal growth and development, for instance by stimulating the proliferation of preosteoblasts and the differentiation of osteoblasts (non-dependent oestrogen conversion) and by promoting the maturation and ossification of cartilage cells and deposits of calcium on bone [
72]. Theoretically, testosterone supplementation can improve bone health in patients with TD. However, the current meta-analysis failed to demonstrate this effect, a finding that is consistent with previous systematic reviews [
15,
73‐
75]. Contrary to our findings, a guideline published in 2010 [
76] stated that although testosterone had no effect on vertebral, hip and femoral BMD, it was associated with an increase in lumbar BMD. A possible reason for this inconsistent finding is that this guideline focused on patients with osteoporosis, while we included only a very small proportion of participants with osteoporosis. Nonetheless, even with our negative finding, there are several reasons why caution must be exercised in concluding that testosterone does not affect BMD. First, the finding that testosterone supplementation did not improve BMD in the short-term (< 2 years) may due to inadequate duration of treatment. It is well known that the effect of testosterone on BMD is only evident after more than 2 years of use. However, only 156 participants from two studies used testosterone for > 2 years, and the sample size is too small to detect a significant difference between the groups. Second, although all participants were androgen deficient, most did not have any abnormality in bone mass density or any evidence of osteoporosis at baseline; therefore, the change in BMD before and after testosterone supplementation may be nonsignificant. We also did not find any difference in the risk of fall or fracture between the testosterone supplementation and the placebo groups, though this may also be due to inadequate treatment duration and the small sample size.
Several studies [
7,
9,
11] indicated that testosterone increases the risk of cardiovascular events. However, we did not find this effect in our meta-analysis, possibly because only a small proportion (5.45%) of participants in our review had a history of CAD at baseline, while several studies [
9‐
11] included a larger number of patients with a history of CAD. Furthermore, the age range of participants also differs between our review and the above studies, with the latter including participants older than 60 years compared with our inclusion of participants over 40 years of age. One cohort study [
10] indicated that a history of CAD and an age greater than 65 years were risk factors for cardiovascular events in patients treated with testosterone.
With regard to all-cause mortality, our review found that testosterone did not decrease the risk of all-cause mortality in patients with TD. We concluded that this negative result was due to inadequate sample size, as there was an obvious trend towards a reduction in the rate of all-cause mortality in the testosterone supplementation group; however, the 95% confidence interval was too wide to detect a significant difference. While this result is consistent with another review [
72] in which testosterone supplementation was found not to increase the incidence of severe adverse events, including mortality, it is contrary to the finding reported in a cohort study of a positive association between testosterone supplementation and all-cause mortality [
10]. These inconsistent findings may in part be explained by the variations in the baseline characteristics of the participants, especially the differences in ages and associated comorbidities. More RCTs are clearly needed to better identify the effect of testosterone supplementation on mortality.
Three studies [
12,
13,
77] found that testosterone therapy increases sexual function in patients with a low testosterone level, which is consistent with our finding. We also found improved quality of life in the testosterone therapy group. However, one study did not find this favourable effect [
76]. In that study [
76], the included participants had complex comorbidities, which may explain this difference. In our meta-analysis fewer participants had comorbidities, and thus, a significant improvement in quality of life may have been more evident.
Persistent concerns revolve around whether testosterone supplementation increases the risk of prostate cancer or BPH. Consistent with other observational studies [
78‐
80], this review did not find an association between testosterone supplementation and prostate cancer. Interestingly, subnormal testosterone levels have been reported to be associated with high-grade prostate cancer [
81].
This systematic review has some strengths. First, the search strategy was developed by a professional information specialist. In addition, we searched both electronic databases and hand searched the references of relevant systematic reviews. This approach allowed us to collect as many relevant RCTs as possible. Second, the study screening and data extraction process were conducted by two researchers independently to minimize bias.
The systematic review also has some limitations. For instance, the long-term data for primary or secondary outcomes were insufficient to detect a clear difference between the groups. Furthermore, significant heterogeneity between populations was identified, such as the definition of TD or LOH and differences in the presence of comorbidities at baseline. Despite the presence of significant heterogeneity, we were unable to determine whether the variations in the effect of testosterone supplementation across subgroup populations were due to insufficient data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.