Higher Circulating Testosterone Linked to Higher CAD Risk in Men: Mendelian Randomization and Survival Analyses

Abstract

Context

Testosterone supplementation is increasingly widespread and has well-established beneficial effects on sexual function and metabolic health. However, there remains uncertainty regarding associated cardiovascular risks.

Objective

Human genetics studies demonstrated that Mendelian randomization approaches recapitulate the beneficial effects of testosterone therapy; here we apply this to cardiovascular disease.

Design

We performed a Mendelian randomization study to assess the causal effect of higher circulating testosterone on coronary artery disease (CAD). We also tested the phenotypic association between measured circulating testosterone and CAD in the cohort of men aged 40 to 69.

Patients or Other Participants

Testosterone genetic instrument data were derived from 425 097 European ancestry adults from the UK Biobank study and CAD from single nucleotide polymorphism-level summary statistics from 1 165 690 individuals in CARDIoGRAMplusC4D.

Main Outcome Measure(s)

CAD as defined in CARDIoGRAMplusC4D was the main outcome. In longitudinal analyses, CAD was defined according to medical records and self-report.

Results

We found that higher genetically predicted circulating testosterone conferred a higher risk of CAD in men [odds ratio (OR): 1.17, 95% confidence interval (CI) 1.07-1.27, P = 3.32 × 10⁻⁴]. There was no evidence of an effect in women (OR: 1.01, 95% CI 0.94-1.10, P = .73). The genetic association in men appeared to be mediated by higher blood pressure. In longitudinal observational analyses, a directionally opposite association was observed in men, likely arising due to confounding by type 2 diabetes and body mass index.

Conclusion

These data suggest that increased testosterone may increase the risk of cardiovascular disease and that this safety concern should be a focus in future clinical trials for testosterone supplementation.

Use of testosterone supplementation is increasing (1). In men, it is an approved treatment for hypogonadism, which typically manifests with fatigue and sexual dysfunction (2). Evidence from randomized controlled trials (RCTs) shows beneficial effects of testosterone supplementation on sexual function, lean mass, and muscle strength (3). Low circulating testosterone is also a risk factor for poor metabolic health (4), and there are increasing numbers of prescriptions for testosterone in “eugonadal” men (with normal circulating testosterone) (5). However, questions remain about the long-term impacts of testosterone levels on other clinical outcomes. There is also interest in the health impacts of testosterone supplementation in women, in whom its use is licensed for low sexual desire, and some hormone replacement therapies contain testosterone.

Coronary artery disease (CAD) is the leading cause of mortality and loss of disability adjusted life years globally (6). Observational studies have linked low measured testosterone with elevated CAD risk in men (7), suggesting that supplementation might have cardioprotective benefits (8, 9). RCTs have investigated the impact of testosterone supplementation on CAD risk in men with age-related testosterone decline (“functional hypogonadism”) (10, 11), as well as in men with pathological hypogonadism, caused by disorders of the hypothalamic-pituitary-testicular axis (12). RCTs have been underpowered and have reported conflicting findings; some reported no difference or decreases in cardiovascular events in hypogonadal men treated with testosterone (13) or improved CAD risk markers (14), while others reported increases in intermediate risk factors (11) and cardiovascular events (15), particularly in men with predisposing risk factors, including particular genetic profiles (16). Furthermore, cardiovascular warnings arose from trials where CAD was a secondary outcome (17). In women, some observational studies of testosterone replacement therapy have identified elevated CAD risk (18), while other studies reported no association (19). RCTs of hormone replacement therapies containing testosterone found no impact on cardiovascular events in women (20). However, few RCTs of testosterone supplementation in men or women measured cardiovascular disease endpoints as primary outcomes, and those that did were underpowered (21).

Mendelian randomization (MR) uses genetic variants as instrumental variables to infer causal links between exposures and disease (22). MR has been described as analogous to a large RCT, benefitting from long follow-up and large sample sizes by leveraging the data from very large observational studies. Previous MR evidence shows that higher circulating testosterone levels in men appear protective against type 2 diabetes but increase the risk of prostate cancer (4). Due to the previously reported distinct genetic architectures of testosterone between sexes (4) and the strong possibility of sex-specific effects of testosterone, we used sex-specific genetic instruments for circulating testosterone (4) as well as sex-specific data on the outcome, CAD, and conducted analyses in men and women separately. To distinguish between the effects of testosterone and SHBG, we use genetic instruments stratified by their effects on total testosterone, free testosterone, and SHBG (4). The use of sex- and hormone-specific genetic instruments, which have been validated with positive control outcomes (4), allows this study to evaluate cardiovascular safety in the absence of sufficiently powered RCT evidence.

Methods

Two-sample Mendelian randomization was used to assess the likely causal effects of testosterone and SHBG on CAD and also to identify possible mediators for any association.

Data

Genetic instruments for testosterone and SHBG in men and women were comprised of reported variants identified in previous sex-stratified genome-wide association (GWAS) analyses of circulating total testosterone, free testosterone, and SHBG in the UK Biobank (4). As previously reported, the arising GWAS signals were categorized by cluster analyses with the same hormone traits into 1 of 4 genetic instruments: 1 and 2—male or female testosterone-specific (variants associated with higher circulating total and free testosterone but not with SHBG) and 3 and 4—male or female SHBG (variants with primary effects on higher circulating SHBG and with secondary diverging effects on higher total and lower free testosterone) [Tables S1–S4 (23)]. Testosterone variants in men were weighted by their effects on free testosterone and in women by their effects on total testosterone, since these traits showed the most consistent associations in each cluster (4). Sex-specific summary level data for variant associations with CAD were obtained from a reported meta-analysis of the Coronary ARtery DIsease Genome wide Replication and Meta-analysis plus Coronary Artery Disease (CARDIoGRAMplusC4D) genetics consortium and the UK Biobank (24, 25). Lipid levels and blood pressure were considered as potential mediators in genetic association models. For these, we obtained summary-level, sex-specific variant associations with high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides from the Global Lipids Genetics Consortium (26) and for diastolic and systolic blood pressure measured in mmHg from a sex-specific GWAS in UK Biobank (27) [Table S9 (23)].

Phenotypic data on incident CAD in the UK Biobank men were derived from hospital electronic records, self-reported conditions and operations at baseline, and death records. CAD was defined using International Classification of Diseases, Tenth Revision codes I21-I23, I24.1, I25.2; International Classification of Diseases, Ninth Revision codes 410-412, 429.79; self-report of heart attack or myocardial infarction (UK Biobank code 1075 in field 20002; code 1 in field 6150); or self-report of coronary angioplasty or coronary artery bypass grafts (Office of Population Censuses and Surveys Classification System, version-4 (OPCS-4) codes K40.1-4, K41.14, K45.1-5, K49.1-2, K49.8-9, K50.2, K75.1-4, K75.8-9; UK Biobank codes 1070, 1095 in field 20004). This definition of CAD in the UK Biobank has been described elsewhere (28). Individuals who had CAD at or prior to recruitment were excluded from phenotypic analyses.

We assessed QRISK^®3 cardiovascular risk factors (29) as potential mediators or confounders in phenotypic association models. QRISK includes information on age, ethnicity, deprivation, systolic blood pressure (SBP), SBP variability, body mass index (BMI), total cholesterol to HDL cholesterol ratio, smoking status, family history of coronary heart disease, and medical history of various diseases. We identified data on these QRISK^®3 variables in the UK Biobank [Box 1. Table S13 (23)]. To assess variability in SBP, we calculated the SD of the first 2 SBP measures after recruitment in each individual. All other QRISK^®3 variables were based on the information recorded at baseline. Information on age of diagnosis of coronary heart disease in family members was unavailable in the UK Biobank, so all familial cases were assigned to “diagnosis before age 60.”

Box 1.

Variables used in QRISK3 algorithm

Age at study entry (baseline)

Ethnic origin (nine categories)

Deprivation (as measured by the Townsend score, where higher values indicate higher levels of material deprivation)

Systolic blood pressure

Body mass index

Total cholesterol: high-density lipoprotein cholesterol ratio

Smoking status [nonsmoker, former smoker, light smoker (1-9/day), moderate smoker (10-19/day), or heavy smoker (≥20/day)]

Family history of coronary heart disease in a first-degree relative aged less than 60 years

Diabetes (type 1, type 2, or no diabetes)

Treated hypertension (diagnosis of hypertension and treatment with at least 1 antihypertensive drug)

Rheumatoid arthritis (diagnosis of rheumatoid arthritis, Felty’s syndrome, Caplan’s syndrome, adult-onset Still’s disease, or inflammatory polyarthropathy not otherwise specified)

Atrial fibrillation (including atrial fibrillation, atrial flutter, and paroxysmal atrial fibrillation)

Chronic kidney disease (stage 4 or 5) and major chronic renal disease (including nephrotic syndrome, chronic glomerulonephritis, chronic pyelonephritis, renal dialysis, and renal transplant)

Measure of systolic blood pressure variability (SD of repeated measures)

Diagnosis of migraine (including classic migraine, atypical migraine, abdominal migraine, cluster headaches, basilar migraine, hemiplegic migraine, and migraine with or without aura)

Corticosteroid use (British National Formulary chapter 6.3.2 including oral or parenteral prednisolone, betamethasone, cortisone, depo-medrone, dexamethasone, deflazacort, efcortesol, hydrocortisone, methylprednisolone, or triamcinolone)

Systemic lupus erythematosus (including diagnosis of systemic lupus erythematosus, disseminated lupus erythematosus, or Libman-Sacks disease)

Second-generation “atypical” antipsychotic use (including amisulpride, aripiprazole, clozapine, lurasidone, olanzapine, paliperidone, quetiapine, risperidone, sertindole, or zotepine)

Diagnosis of severe mental illness (including psychosis, schizophrenia, or bipolar affective disease)

Diagnosis of erectile dysfunction or treatment for erectile dysfunction (British National Formulary chapter 7.4.5 including alprostadil, phosphodiesterase type 5 inhibitors, papaverine, or phentolamine)

Statistical Analysis

The genetic instruments for male testosterone, female testosterone, male SHBG, and female SHBG were used as the exposure in separate MR models. Where a variant was not present in the outcome GWAS CAD dataset, we identified a proxy with the highest r² value (all r² > 0.7) calculated using a random sub-sample of the UK Biobank white British sample, n = 20 000. Including proxies, the genetic instruments comprised 95 single nucleotide polymorphisms (SNPs) for male testosterone, 220 SNPs for female testosterone, 306 SNPs for male SHBG, and 322 SNPs for female SHBG [Tables S5–S8 (23)]. Genotypes at all variants were aligned to the hormone-increasing allele.

The inverse variance-weighted (IVW) MR model was selected as the primary model if the MR-Egger intercept indicated absence of horizontal pleiotropy (P_intercept≥.05). If horizontal pleiotropy was present (P_intercept < .05), we selected the MR-Egger model. We performed MR-PRESSO to identify any individual variants with outlier effects and reran models with these removed. We also ran leave-one-out analysis, where we iteratively removed each of the SNPs in our instrument from the MR model to ensure that the causal effect estimate was not being driven by the effects of just 1 SNP. In order to calculate the required sample size in a RCT to detect the odds ratio (OR) calculated in the MR in our study, we performed sample size calculations using the powerMediation package (version 0.3.4.). We assumed a baseline event rate of 7.3% in the male population, based on the incidence of CAD in the control group of the largest RCT of the impact of testosterone therapy on CAD risk (13). We also used a 2-sided α of .05, 90% power, and the OR calculated in our MR analyses.

Potential mediators were included as genetic covariates in separate sex-specific 2-sample multivariable MR (MVMR) models (30). Using genetic instruments for male testosterone and female testosterone, we obtained estimates for their associations with systolic blood pressure, diastolic blood pressure, and lipid levels (including LDL, HDL, and triglycerides), as well as their associations with CAD. GWAS summary statistics were harmonized across the exposure (testosterone), the potential mediators (blood pressure and lipids), and the outcome (CAD), ensuring alignment of effect alleles across datasets. We then applied MVMR to model the impact of testosterone on CAD while adjusting for the genetic associations of testosterone on lipid levels and blood pressure (30).

Cox proportional hazards models were used to test the phenotypic association between measured circulating testosterone and CAD in the cohort of men aged 40 to 69 at baseline in the UK Biobank (N = 228 989), with time since recruitment as the underlying time variable. Models compared men with measured testosterone <12 nmol/L to those with higher levels, as this threshold is typically used in clinical practice to indicate low testosterone (31). As covariates, QRISK^®3 variables were included altogether and individually in separate models.

We performed analyses using the MendelianRandomization (32), Survival (33), and MRPRESSO (34) packages in R version 4.3.2. software platform (R Development Core Team, Vienna, Austria).

Results

Sex-specific MR of Sex Hormones With CAD

We analyzed sex-specific genetic variant data on 425 097 individuals for sex hormone levels and 1 165 690 individuals for CAD, including 181 522 cases. In our initial IVW model, higher genetically predicted testosterone conferred higher risk of CAD in men [OR: 1.10, 95% confidence interval (CI) 1.01-1.19, I² = 41.2%, P = .03] [Table S10 (23)]. However, an outlier SNP (rs56196860) was detected by MR-PRESSO [Table S11 (23)]. rs56196860 is a missense variant in FKBP4, which is a reported candidate gene for androgen insensitivity syndrome (35). Leave-one-out analysis indicated this model underestimated the true effect of circulating testosterone on CAD due to the presence of this outlier SNP, as the β estimate in the IVW model was elevated once this SNP was excluded [Fig. S2 (36)]. After excluding this, higher genetically predicted testosterone conferred a clearer higher risk of CAD in men; each SD higher testosterone conferred a 17% higher risk of CAD (OR: 1.17, 95% CI 1.08-1.27, I² = 26.9, P = 3.32 × 10⁻⁴). This finding remained significant with a Bonferroni adjustment based on 4 tests (P < .0125). Similar findings were observed in sensitivity MR models (MR Egger and MR median) (Table 1). We used the powerMediation package (v0.3.4) in R to estimate the sample size needed to detect an OR of 1.17 in a RCT. Assuming 90% power, a 7.3% baseline incidence for CAD (which was the incidence in the control arm of a recent trial of testosterone supplementation (13)), and a 2-sided α of .05, the required sample size was 6439.

Table 1.

Results of Mendelian randomization of sex hormone levels to coronary artery disease

	Exposure	SNPs found	Proxies used	Total SNPs	Model	OR	95% CI	P	I2 (%)
Male	SHBG	245	61	306	IVW	0.79	(0.70-0.89)	1.22 × 10−4	63.6
					Egger	0.92	(0.77-1.10)	.38	97.8
					Intercept			.02
					Median	0.94	(0.82-1.09)	.40
	Testosterone	74	20	94	IVW	1.17	(1.07-1.27)	3.32 × 10−4	26.9
					Egger	1.24	(1.04-1.48)	.01	92.0
					Intercept			.43
					Median	1.22	(1.09-1.37)	<.001
Female	SHBG	259	63	322	IVW	0.72	(0.62-0.83)	6.2 × 10−6	76.9
					Egger	0.88	(0.70-1.11)	.28	96.2
					Intercept			.03
					Median	1.02	(0.84-1.24)	.85
	Testosterone	182	38	220	IVW	1.01	(0.94-1.10)	.73	43.3
					Egger	1.06	(0.92-1.22)	.41	95.7
					Intercept			.46
					Median	1.08	(0.96-1.22)	.19

Abbreviations: CI, confidence interval; IVW, inverse variance weighted; OR, odds ratio; SNP, single nucleotide polymorphism.

Regarding potential mediators, genetically predicted testosterone conferred higher diastolic blood pressure (β = .050, 95% CI 0.02-0.08, P = .001) and systolic blood pressure (β = .03, 95% CI 0.004-0.063, P = .024), measured in mmHg [Table S12 (23)]. Neither HDL cholesterol, LDL cholesterol, nor triglyceride levels showed significant associations with testosterone in MR analyses (HDL cholesterol: β = −.02, 95% CI −0.06-0.001, P = .23, LDL cholesterol: β = −.01, 95% CI −0.03-0.01, P = .33, triglycerides: β = −.06, 95% CI −0.03-0.02, P = .68), and so were not taken forward for MVMR models. The inclusion of diastolic blood pressure in a 2-sample MVMR model attenuated the genetic association between testosterone and CAD in men (OR: 1.09, 95% CI 0.88-1.35, P = .43).

By contrast, higher genetically predicted testosterone had no impact on CAD risk in women in primary (OR: 1.01, 95% CI 0.94-1.10, P = .73) or sensitivity models (Table 1). While there was moderate heterogeneity among individual variant estimates (I² = 43.3%), there was no evidence of directional pleiotropy (P_intercept  = .46), and results were consistent across sensitivity MR models (MR Egger and MR median).

In IVW models, higher genetically predicted SHBG levels appeared to confer lower risk of CAD in both men and women. However, there was evidence of directional pleiotropy (both P_intercept < .05), and the MR-Egger estimates were not statistically significant (Table 1). In addition, rs1799941, located within the SHBG gene and with a specific effect on SHBG levels, showed no association with CAD in men (OR: 0.99, P = .43) or women (OR: 1.004, P = .77). Taken together, we found no evidence for an effect of SHBG on CAD in men or women.

Phenotypic Analyses in UK Biobank

To assess the risk of incident CAD associated with low testosterone in men, we performed phenotypic analyses of a binary measured testosterone variable. Complete data on measured testosterone levels, all QRISK variables (29), and time-to-event data on CAD were available for 142 149 men in the UK Biobank [Table S13 (23)]. Descriptive statistics relating to relevant QRISK variables are presented in Table 2.

Table 2.

Descriptive statistics of testosterone-deficient and testosterone-sufficient men in UKBB with complete time to event and covariate data

Variable	Testosterone deficient (<12 nmol/L)	Testosterone sufficient (≥12 nmol/L)	P
n	80 059	67 463
Age at recruitment [mean (SD)]	57.37 (7.96)	56.67 (8.19)	<.001
CAD (%)	7.15	6.24	<.001
BMI [mean (SD)]	28.80 (4.41)	26.70 (3.69)	<.001
Type 2 diabetes (%)	16.0	8.1	<.001
Hypertension (%)	27.9	19.1	<.001
Corticosteroid use (%)	0.13	0.12	.70
Antipsychotic use (%)	0.06	0.05	.64
Rheumatoid arthritis (%)	1.76	1.58	.008
Atrial fibrillation (%)	12.21	11.25	<.001
Kidney disease (%)	4.30	2.56	<.001
Migraine (%)	1.46	1.41	.44
Systemic lupus erythematosus (%)	0.05	0.07	.12
Mental illness (%)	2.15	1.83	<.001
Erectile dysfunction (%)	0.33	0.36	.44
Townsend Deprivation Index [mean (SD)]	−1.38 (3.06)	−1.23 (3.15)	<.001
Cholesterol [mean (SD)]	5.48 (1.13)	5.59 (1.08)	<.001
HDL-c [mean (SD)]	1.24 (0.30)	1.34 (0.32)	<.001
Systolic blood pressure [mean (SD)]	142.52 (17.22)	139.84 (17.39)	<.001
Systolic blood pressure SD [mean (SD)]	5.38 (4.39)	5.25 (4.32)	<.001
Cholesterol:HDL [mean (SD)]	4.59 (1.18)	4.34 (1.12)	<.001
Family history of CAD (%)	40.2	38.4	<.001
Ever smoked (%)	66.2	65.1	<.001

Abbreviations: BMI, body mass index; CAD, coronary artery disease; HDL-c, high-density lipoprotein cholesterol; UKBB, UK Biobank.

During a median 15 years follow-up of these men (2 097 177 person-years), there were 9862 incident CAD cases. Before adjustment for QRISK covariates, measured testosterone <12 nmol/L at baseline conferred a 15% higher hazard of CAD [hazard ratio (HR) = 1.15, 95% CI 1.10-1.20, P = 8.6 × 10⁻¹²]. After adjusting for all QRISK variables, this association was fully attenuated (HR = 0.97, 95% CI 0.93-1.01, P = .11) [Table S14 (23)]. The QRISK variables with the largest impact on the phenotypic association were type 2 diabetes (54.3% attenuation) and BMI (48.1%) [Table S15 (23)]. Notably, inclusion of systolic blood pressure as a covariate slightly attenuated the association between low testosterone and CAD in men (adjusted HR = 1.11, 95% CI 1.06-1.15, P = 8.0 × 10⁻⁷).

Discussion

Principal Findings

Using genetic inference, we identified an apparent causal association between higher circulating testosterone levels and higher CAD risk in men, with no association in women. The genetic association in men was consistent across various sensitivity analyses and appeared to be mediated by the apparent effect of testosterone on higher blood pressure, which shows consistency with the effect of testosterone on blood pressure reported in RCTs of testosterone supplementation (10). This association between testosterone and CAD risk in men is consistent with evidence that shows increased coronary plaque volume in men treated with testosterone (11). Conversely, in longitudinal phenotypic analyses, high circulating testosterone levels at baseline appeared to be protective for CAD in men, but this was likely due to confounding by type 2 diabetes and BMI.

Comparison With Other Studies

Previous RCT evidence on the impact of testosterone supplementation on CAD is inconsistent and likely underpowered (11, 13-15), while observational studies of this topic benefit from larger sample sizes but carry limitations and the potential for selection bias (37). For example, a population-based matched cohort study gathered a sample size of over 30 000 men, composed of 10 311 men treated with testosterone replacement therapy and 28 029 controls, and found that longer-term exposure to testosterone supplementation was associated with reduced risks of mortality, cardiovascular events, and prostate cancer (37). Short durations of therapy increased the risk of mortality and cardiovascular events, however (37). The largest RCT to date, the TRAVERSE trial involving 5246 men with low testosterone levels, reported no harmful impact of testosterone supplementation on a composite of major cardiovascular events (HR = 0.96; 95% CI 0.78-1.17; P < .001 for noninferiority) (10). Larger trials to investigate the safety and efficacy of testosterone therapy should be encouraged. To detect the OR of 1.17 as found in our study would require an RCT of over 6400 men.

Previous MR analyses have produced differing findings. A previous MR study using a much smaller CARDIoGRAMplusC4D dataset of 547 261 individuals including 122 733 CAD cases reported a protective effect of testosterone on risk of CAD and atherosclerotic outcomes in men, apparently mediated by improved lipid profiles (38). However, that study used a sex-combined genetic instrument for testosterone, CAD, and sex-combined data for the mediators (38). Our previous work demonstrated the need for sex-stratified MR analyses using sex-specific variants for testosterone due to its complete sex-specific genetic regulation (genetic correction between sexes = 0.001) (4). This suggests there may be multiple pathways in the regulation of testosterone levels, which produce differing impacts on CAD. Another MR study reported a null association between total testosterone and CAD in men, but it used a much smaller outcome sample of 94 478 men including 14 315 cases, and an unrefined (noncluster derived) testosterone instrument (39). A third, older MR study found increased risks of thromboembolism, myocardial infarction, and heart failure in men, but it used a very limited genetic instrument for testosterone comprising variants from only 2 loci (JMJD1C and SHBG) (40). Variants in SHBG increase levels of total testosterone but without altering free or bioactive testosterone (4). Our current and previous findings (4) support the use of clustered instruments to distinguish between the effects of testosterone and SHBG.

Meaning of the Study

These results are notable as testosterone prescriptions have increased significantly, due to increased marketing, despite steady rates of pathological hypogonadism in men (1, 41). Testosterone replacement therapy for male hypogonadism requires the presence of characteristic signs and symptoms as well as low serum concentrations of total testosterone (<12 nmo/L) or free testosterone (42). These trends have raised concerns about the safety of testosterone supplementation in men, beyond the requirement to monitor for prostate cancer (42). In light of the uncertainty regarding cardiovascular risk associated with testosterone supplementation, Food and Drug Administration guidance issued in 2015 requires health care professionals in the United States to inform patients of this possible risk and that all prescription testosterone products are labeled accordingly (43). By contrast, there is no UK national guideline for testosterone prescribing in men. Our results support the need for more consistent warnings about the possible cardiovascular risks of testosterone supplementation while the RCT evidence is uncertain.

In women, testosterone use has surged, with a 10-fold increase in prescriptions from 2015 to 2022 (44). Testosterone products licensed for men are prescribed off-label to women at lower doses (45), driven by demand to treat low libido and claims of improved brain function and well-being. This is concerning given genetic evidence linking higher testosterone in women with higher risks of type 2 diabetes, endometrial and breast cancers, polycystic ovary syndrome, and gallstones (4, 38). While RCTs in women have identified adverse effects of testosterone on lipid profiles, the impact varied by delivery method (20). Our study found no effect of genetically predicted testosterone levels on CAD in women, but data on other cardiometabolic outcomes are needed, as well as RCTs featuring cardiovascular conditions as primary outcomes (45).

Strengths and Limitations

This study has several limitations. MR estimates the effects of lifelong differences in endogenous testosterone. These may differ from the effects of exogenous testosterone treatment, which might vary by duration, dosage, and age. Exogenous testosterone treatment may also exert biphasic effects, with short-term risks to health but longer-term benefits, which cannot be captured in an MR analysis (37). However, similar studies using this genetic instrument have confirmed the established association between testosterone and the risk of prostate cancer (4). Furthermore, MR models poorly estimate nonlinear or threshold effects, which are relevant to testosterone as most prescriptions are for men with hypogonadism. MR studies can produce unreliable findings due to the inclusion of SNPs with pleiotropic effects. We excluded 1 SNP (rs56196860), which we identified statistically as an outlier; this decision is supported by its purported role in causing androgen insensitivity syndrome (28), a condition where the functional link between circulating levels of testosterone and its bioaction is broken. While our sensitivity analysis diminished the likelihood of any major bias, we were unable to further validate the testosterone results in a biologically relevant manner as testosterone lacks a cis-SNP, unlike SHBG.

Strengths of this study include the use of clustered instruments to separate SHBG and testosterone effects. We leveraged the most recent GWAS of CAD to optimize the power of our 2-sample MR analysis and used sex-specific GWAS results for CAD to estimate sex-specific effects more precisely. Testosterone prescriptions have also been linked to a possible increased risk of stroke (43); however, we were unable to address this due to the lack of openly available sex-specific GWAS data on that outcome.

Conclusions

Sex-specific genetic causal modelling indicates that higher circulating testosterone increases the risk of CAD in men, mediated by the apparent effect of testosterone on higher blood pressure. While MR measures lifelong exposure to testosterone and not exposure to administered testosterone compounds, these findings support guidance to inform patients of the possible cardiovascular risks of testosterone supplementation, while the RCT evidence is uncertain.

Funding

E.J.M., F.R.D., N.J.W., K.K.O., and J.R.B.P. are supported by the UK Medical Research Council (unit programs: MC_UU_00006/1 and MC_UU_00006/2). This research was also supported by the National Institute for Health and Care Research Cambridge Biomedical Research Centre (NIHR203312). The BHF Cardiovascular Epidemiology Unit is supported by core funding from the British Heart Foundation (RG/18/13/33946: RG/F/23/110103), NIHR Cambridge Biomedical Research Centre (NIHR203312) [*], BHF Chair Award (CH/12/2/29428), and by Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and the Wellcome Trust. *The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

The sponsors had no role in the study design; collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit it for publication.

Disclosures

J.R.B.P. is an employee of Insmed Innovation UK and holds stock/stock options in Insmed, Inc. J.R.B.P. also receives research funding from GSK and has engaged in paid consultancy for WW International Inc., Ovartix Ltd., and Hertility Health.

Data Availability

This research was conducted using the UK Biobank Resource under application 9905. The data reported in this paper are available to other investigators by application directly to the UK Biobank. GWAS summary statistics for sex-stratified sex hormone levels are available from the GWAS Catalog. Sex-stratified GWAS summary statistics for CAD are publicly available at https://cvd.hugeamp.org/downloads.html#summary.

The genetic associations with the outcomes in the UK Biobank and CARDIoGRAMplusC4D consortium are provided in the supplementary data. Software code in R for implementing the Mendelian randomization analysis and survival analysis is available on GitHub at: https://github.com/emilymorbey/BMJ_FINAL_TESTOSTERONE/blob/main/Testosterone_cad_mr.r

For the purpose of Open Access, the author has applied a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising.

Ethics

The UK Biobank received ethical approval from the research ethics committee (REC reference for UK Biobank 11/NW/0382) and participants provided written informed consent. No ethics approval was acquired for the analysis using publicly available data (CARDIoGRAMplusC4D 1000 Genomes based genome wide association study).