Do testosterone and sex hormone-binding globulin affect cancer risk? A Mendelian randomization and bioinformatics study

Abstract

Using Mendelian Randomization (MR) and large-scale Genome-Wide Association Study (GWAS) data, this study aimed to investigate the potential causative relationship between testosterone and sex hormone-binding globulin (SHBG) levels and the onset of several cancers, including pathway enrichment analyses of single nucleotide polymorphisms (SNPs) associated with cancer allowed for a comprehensive bioinformatics approach, which offered a deeper biological understanding of these relationships. The results indicated that increased testosterone levels in women were associated with a higher risk of breast and cervical cancers but a lower risk of ovarian cancer. Conversely, increased testosterone was linked to lower stomach cancer risk for men, whereas high SHBG levels were related to decreased risks of breast and prostate cancers. The corresponding genes of the identified SNPs, as revealed by pathway enrichment analysis, were involved in significant metabolic and proliferative pathways. These findings emphasize the need for further research into the biological mechanisms behind these associations, paving the way for potential targeted interventions in preventing and treating these cancers.

Keywords:

• Sex hormone binding globulin
• testosterone
• cancer
• mendelian randomization
• bioinformatical analysis

1 Introduction

Cancer is the leading cause of death worldwide [1,2]. The etiology of cancer consists of various factors, including smoking, a poor diet, and lack of physical exercise, which have been widely known. However, the correlation between sex hormones, especially testosterone and SHBG, and cancer remains to be discovered.

Previous studies on the impact of gender on several cancers have shown that females have a lower risk and overall better prognosis of colon cancer, skin cancer, head and neck cancer, esophageal cancer, lung cancer, and liver cancer [2–4], while the risk of thyroid cancer is lower in men [3]. In addition to sex hormone levels, many risk factors are caused by gender differences (such as differences in cooking oil smoke caused by sex [4]). However, the causal relationship between sex hormone levels and cancer is still uncertain.

Testosterone is often referred to as a male hormone, in part because males have about ten times higher concentrations of testosterone compared to women, although women are more sensitive to testosterone [5]. Produced by the Leydig cells in the testes, testosterone directly interacts with the androgen receptors in reproductive organs, muscle, bones, brain, and skin, as well as in the prostate gland, and testosterone plays other important roles in health and disease. Currently, testosterone therapy is approved primarily for the treatment of delayed male puberty, low production of testosterone (whether due to failure of the testes, pituitary or hypothalamus function) [6] and certain inoperable female breast cancers [7]. Observational studies have shown that plasma testosterone levels are associated with a 30%–80% increased risk of early death after cancer in men and women from the general population [8]. In a case-control study with 61 breast cancer patients and 200 healthy women, elevated testosterone levels correlated with breast cancer risk [9]. In contrast, another study with 169 ovarian cancer cases and 410 controls found no overall association, but higher androgen levels increased the risk of non-serous ovarian cancer [10]. Therefore, testosterone has different risks for different types of cancer.

Sex hormone-binding globulin (SHBG) is a homodimer glycoprotein with a molecular weight of about 90 kDa [11]. It has been proved to bind with other sex steroid hormones and act as a transport protein [12], and is, therefore, an important regulator of their bioactivity [13]. The distribution of SHBG-binding testosterone is different in men and women; the SHBG level in women is twice as high as that in men, thus reducing the body’s exposure to androgens and estrogens [14]. Researchers have found an association between sex hormones and cancers in several studies. Zoё Hyde and his team found that the higher the testosterone level in the body, the greater the risk of prostate and lung cancer [15]. Niki Dimou et al. discovered that the SHBG level was negatively correlated with breast cancer risk [16]. Moreover, Katherine Ruth observed a positive correlation between female testosterone levels and the risk of endometrial cancer [17]. In addition, according to the Women’s Health Initiative Association, the group with the highest SHBG level was more than twice as likely to develop colon cancer as the group with the lowest SHBG level [18].

However, there are no related reports about total testosterone and SHBG attainment in pan-cancer analysis. To further explore the relationship between testosterone and SHBG levels in vivo and other cancers, we conducted a Mendelian randomization analysis (MR) study. A new epidemiological research method that evaluates the causal relationship between exposure and outcome variables using data from genome-wide association studies (GWAS) [19,20]. Because it uses genetic variation as a tool variable (IV), it is less susceptible to confounding variables and reverses causality than other methods [21]. We conduct bioinformatics, including enrichment analysis, to fully comprehend the causal relationship between exposure and outcome variables.

2 Method

2.1. Statistical analysis

To determine MR estimates of testosterone and SHBG attainment for lung, breast, ovarian and other cancers. Estimates of genetic causality were obtained by applying fixed effects IVW analysis, which was done to determine the overall effect of SHBG concentration on each outcome. In the MR analysis, the statistical significance threshold was p < 0.05. For each single nucleotide polymorphism (SNP), the ratio of the causal effect of exposure on the outcome was estimated as the ratio of the effect of the SNP on the outcome to the effect of the SNP on the exposure. In IVW, the overall estimate was derived from an IVW randomized analysis of the ratio estimates for all variables in a set of instrumental variables. Considering that IVW methods can be affected by directional pleiotropy (genetic variation affecting the outcome through pathways other than exposure), we used MR-Egger and weighted median methods to check the robustness of IVW method estimates.

We evaluated the relationship between single nucleotide polymorphism and the results to obtain the odds ratio (OR) and average difference.

MR-Egger can detect and correct for bias due to directional pleiotropy because it allows for a non-zero intercept and within the internal (instrument strength independent of direct effects) assumptions to provide consistent estimates of causal effects (genetic associations associated with exposure independent of the direct effect of genetic variation on the outcome). The median of the ratio instrumental variable estimates was examined using a weighted median approach, which complements the IVW analysis. The MR-Egger method was used to account for the potential pleiotropy of genetic variants that may affect the results through pathways other than exposure and thus may bias the analysis. The MR pleiotropy residual sums and outliers (MR-PRESSO) method was also used to test for horizontal pleiotropy. The application of this method allows the identification of abnormal genetic variants. In addition, traits associated with SHBG-related SNPs were searched on the V2 phenom scanner website to determine whether the selected instrumental variables were associated with any confounding factors in the potential association between SHBG levels and cancer. Therefore, IVW analysis should be performed after the removal of genetic variants in the sensitivity analysis.

Because of limitations in the observational design, genetic variants have been proposed as potential instrumental variables (IVs), usually SNPs, to mimic the effects of variable environmental exposure on disease susceptibility, known as MR [22]. MR examines the causal effects of exposure on health outcomes (disease occurrence and development) in ways in which germline genetic variation is used as a tool for exposure (e.g. environmental factors, biological characteristics, or drug pathways). If the association between exposure and outcome is statistically significant and can be fully explained by the association of the two genetic variants, exposures are determined to be causal.

To investigate potential mechanisms from testosterone and SHBG to cancer, first, we selected the GWAS results corresponding to our study’s cancer pair of interest by searching the public database. Then we selected the GWAS results corresponding to testosterone and SHBG, followed by a 2SMR analysis; we selected single nucleotide polymorphisms in cancers related to lung, breast, ovarian, liver and others as instrumental variables, testosterone and SHBG as exposure factors, and gender as a confounder. After adjusting the parameters, by using MR-base, we performed a simple two-sample MR analysis. Finally, testosterone and SHBG levels were associated with the risk of developing breast, ovarian, liver, prostate, colorectal and leukemia with some causality, and the effect of gender on the results was greatly discarded.

Our research used MR-base (http://www.mrbase.org/) for data analysis and three R packages (R version 4.20), MRInstruments, TwoSampleMR, and googleAuthR, for statistical analysis.

2.2. Enrichment analysis

Gene ontology (GO) analysis [23], including biological processes (BP), molecular functions (MF) and cellular components (CC), is a commonly used method for large-scale functional enrichment research. Kyoto Gene and Genome Encyclopedia (KEGG) [24] is a database that stores information about genomes, biological pathways, diseases, and drugs. For functional annotation analysis and information on differentially expressed senescence-associated genes, R package clusterProfiler [25] and KEGG pathway enrichment analysis were used as tools, respectively.

3 Result

3.1. GWAS summary data on various cancers

In our research, the Genome-Wide Association Study (GWAS), a genome-wide association study of the primary exposure characteristics, was obtained from the UK Biobank. Lung cancer and C64 kidney cancer data were obtained from GWAS MD Anderson Cancer Center (MDACC), GWAS Cancer Research Institute (ICR), GWAS National Cancer Institute (NCI), and GWAS International Agency for Research on Cancer (IARC). Breast cancer data were obtained from the Breast Cancer Consortium (BCAC) results. Ovarian cancer data were obtained from the Ovarian Cancer Association Consortium (OCAC) and the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA). Prostate cancer data were obtained from the OncoArray project, UK GWAS, and others. Leukemia data were obtained from England’s National Health Service Blood and Transplant (NHSBT). Oral cancer data were obtained from the International Head and Neck Cancer Epidemiology Consortium (INHANCE), the European cohort study (EPIC), and the cases and case controls from the UK Case series (HN5000) [17,26–28] (Table 1). Additionally, our MR studies identified and extracted exposure-associated SNP information at the genome-wide significance level (p = 5 × 10⁻⁸) (Table 2).

Table 1. Details of studies included in Mendelian randomization analyses.

Trait	Number of cases	Number of controls	Sample size	Population	PubMed ID	Year
Lung cancer overall	11,348	15,861	27,209	European	24880342	2014
Adenocarcinoma	3442	14,894	18,336	European	24880342	2014
Squamous cell carcinoma	3275	15,038	18,313	European	24880342	2014
Breast cancer	122,977	105,974	228,951	European	29059683	2017
Ovarian cancer	25,509	40,941	66,450	European	28346442	2017
Liver cancer	168	372,016	372,184	European		2021
C64 Malignant neoplasm of kidney	1114	461,896	463,010	European		2018
Esophageal cancer	740	372,016	372,756	European		2021
Malignant melanoma	2677	334,482	337,159	European		2017
Prostatic cancer	79,148	61,106	140,254	European	29892016	2018
Colorectal cancer	5657	372,016	377,673	European		2021
Leukemia			3,301	European	29875488	2018
Cervical cancer	563	198,523	199,086	European		2021
Thyroid cancer	649	431	1080	European	23894154	2013
Oral Cancer	1,223	2,928	4,151	European	27749845	2016
Stomach cancer	633	170,006		European		2021

Table 2. Details of studies of confounders and mediators.

Trait	Population	First author	Consortium	Sample size	Number of variants	Year	PubMed ID
Total testosterone	Female	Rebecca Richmond	NA	199,569	16,132,861	2020	NA
	Male	Rebecca Richmond	NA	199,569	16,132,861	2020	NA
SHBP	Female	Rebecca Richmond	NA	214,989	12,321,875	2020	NA
	Male	Rebecca Richmond	NA	185,221	12,321,875	2020	NA

SHBG: sex hormone binding globulin.

3.2. MR estimates for multi‑polymorphism scores

3.2.1. Total testosterone

The results of the MR estimates for circulating total testosterone concentration and each type of cancer risk are presented in Table 3. In the random-effects, IVW models, higher genetically predicted circulating total testosterone concentration was causally associated with an elevated risk of women’s breast cancer (OR 1.14, 95% CI 1.06-1.23; p = 0.01) (Supplementary Figure 1) and cervical cancer (OR 1.01, 95% CI 1.01-1.01; p = 0.01) (Supplementary Figure 2). However, the risk of women who have ovarian cancer (OR 0.89,95% CI 0.81–0.99; p = 0.03) (Supplementary Figure 3) and the risk of men who have stomach cancer (OR 0.65,95% CI 0.48–0.89; p = 0.01) (Supplementary Figure 4) are reduced. We have not found a causal relationship between circulating total testosterone concentration and lung cancer, liver cancer, C64 malignant neoplasm of the kidney, esophageal cancer, malignant melanoma, colorectal cancer, leukemia, thyroid cancer and oral cancer.

Table 3. Mendelian Randomized estimation of the relationship between total testosterone and SHBG and the histological types of pan-cancer.

(1) females
Outcome	IVW method		MR-Egger		Weighted median method
	OR (95% CI)	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value
Total testosterone
Lung cancer overall	0.98(0.85-1.13)	0.82	1.03(0.79-1.34)	0.83	1.20(0.97-1.47)	0.09
Adenocarcinoma	0.92(0.74-1.15)	0.48	0.86(0.56-1.31)	0.48	1.17(0.82-1.67)	0.38
Squamous cell carcinoma	1.02(0.81-1.29)	0.88	0.96(0.62-1.49)	0.86	1.29(0.93-1.78)	0.13
Breast cancer	1.14(1.06-1.23)	0.01	1.25(1.08-1.44)	0.01	1.13(1.04-1.23)	0.01
Ovarian cancer	0.89(0.81-0.99)	0.03	0.90(0.75-1.09)	0.30	0.92(0.76-1.10)	0.34
Liver cancer	1.00(0.99-1.01)	0.61	1.00(0.99-1.01)	0.57	1.00(0.99-1.01)	0.84
C64 Malignant neoplasm of kidney	1.00(0.99-1.01)	0.61	1.00(0.99-1.01)	0.64	1.00(0.99-1.01)	0.99
Esophageal cancer	1.00(0.99-1.01)	0.60	1.00(0.99-1.01)	0.66	1.00(0.99-1.01)	0.37
Malignant melanoma	1.00(0.99-1.01)	0.90	1.00(0.99-1.01)	0.64	1.00(0.99-1.01)	0.48
Colorectal cancer	1.00(0.99-1.01)	0.16	1.00(0.99-1.01)	0.72	1.00(0.99-1.01)	0.59
Leukemia	1.16(0.97-1.39)	0.10	1.00(0.71-1.41)	0.99	0.95(0.73-1.25)	0.74
Cervical cancer	1.01(1.01-1.01)	0.01	1.00(0.99-1.01)	0.26	1.00(0.99-1.01)	0.14
Thyroid cancer	0.79(0.29-2.11)	0.63	0.08(0.01-1.66)	0.11	0.78(0.01-1.66)	0.72
Oral Cancer	1.23(0.78-1.93)	0.37	1.31(0.41-4.17)	0.65	1.78(0.94-3.36)	0.08
Stomach cancer	0.89(0.60-1.32)	0.56	0.66(0.32-1.35)	0.26	0.69(0.35-1.40)	0.31
SHBG
Lung cancer overall	1.05(0.93-1.17)	0.44	0.97(0.78-1.19)	0.74	1.00(0.85-1.17)	0.98
Adenocarcinoma	1.05(0.90-1.22)	0.57	0.99(0.75-1.31)	0.95	0.90(0.71-1.14)	0.37
Squamous cell carcinoma	1.08(0.91-1.27)	0.39	0.90(0.67-1.21)	0.49	0.94(0.74-1.19)	0.58
Breast cancer	0.98(0.92-1.06)	0.65	0.95(0.84-1.08)	0.43	0.97(0.91-1.04)	0.46
Ovarian cancer	1.02(0.95-1.10)	0.58	1.00(0.87-1.15)	0.99	1.02(0.91-1.15)	0.72
Liver cancer	1.00(0.99-1.01)	0.52	1.00(0.99-1.01)	0.81	1.00(0.99-1.01)	0.35
C64 Malignant neoplasm of kidney	1.00(0.99-1.01)	0.16	1.00(0.99-1.01)	0.67	1.00(0.99-1.01)	0.14
Esophageal cancer	1.00(0.99-1.01)	0.68	1.00(0.99-1.01)	0.49	1.00(0.99-1.01)	0.86
Malignant melanoma	1.00(0.99-1.01)	0.44	1.00(0.99-1.01)	0.44	1.00(0.99-1.01)	0.56
Colorectal cancer	1.00(0.99-1.01)	0.72	1.00(0.99-1.01)	0.78	1.00(0.99-1.01)	0.84
Leukemia	1.01(0.89-1.14)	0.93	0.95 (0.76-1.19)	0.67	1.05(0.85-1.30)	0.65
Cervical cancer	1.00(0.99-1.01)	0.45	1.00(0.99-1.01)	0.58	1.00(0.99-1.01)	0.99
Thyroid cancer	1.07(0.53-2.18)	0.85	1.60(0.32-8.10)	0.57	1.09(0.39-3.00)	0.88
Oral Cancer	0.94(0.72-1.24)	0.67	1.50(0.90-2.52)	0.12	1.31(0.86-2.01)	0.20
Stomach cancer	0.78(0.59-1.03)	0.08	0.95(0.57-1.56)	0.83	0.60(0.37-0.99)	0.04
(2) males
Outcome	IVW method		MR-Egger		Weighted median method
	OR (95% CI)	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value
Total testosterone
Lung cancer overall	1.02(0.89-1.16)	0.82	1.13(0.88-1.44)	0.34	0.96(0.80-1.15)	0.67
Adenocarcinoma	1.15(0.97-1.37)	0.12	1.27(0.92-1.75)	0.15	1.08(0.83-1.40)	0.57
Squamous cell carcinoma	0.95(0.78-1.16)	0.63	1.17(0.81-1.68)	0.40	1.12(0.85-1.47)	0.42
Breast cancer	0.94(0.88-1.00)	0.07	0.84(0.75-0.95)	0.01	0.90(0.84-0.97)	0.01
Liver cancer	1.00(0.99-1.01)	0.26	1.00(0.99-1.01)	0.76	1.00(0.99-1.01)	0.35
C64 Malignant neoplasm of kidney	1.00(0.99-1.01)	0.72	1.00(0.99-1.01)	0.58	1.00(0.99-1.01)	0.67
Esophageal cancer	1.00(0.99-1.01)	0.87	1.00(0.99-1.01)	0.85	1.00(0.99-1.01)	0.81
Malignant melanoma	1.00(0.99-1.01)	0.93	1.00(0.99-1.01)	0.75	1.00(0.99-1.01)	0.68
Prostatic cancer	0.99(0.91-1.08)	0.81	1.03(0.88-1.20)	0.70	1.07(0.98-1.16)	0.12
Colorectal cancer	1.00(0.99-1.01)	0.61	1.00(0.99-1.01)	0.47	1.00(0.99-1.01)	0.79
Leukemia	1.12(0.98-1.27)	0.09	0.97(0.78-1.22)	0.82	0.95(0.77-1.18)	0.64
Thyroid cancer	1.58(0.57-4.37)	0.37	1.12(0.13-9.91)	0.92	0.60(0.19-1.89)	0.38
Oral Cancer	0.99(0.74-1.31)	0.92	1.10(0.66-1.84)	0.72	1.41(0.84-2.37)	0.20
Stomach cancer	0.65(0.48-0.89)	0.01	0.86(0.50-1.48)	0.59	0.70(0.42-1.18)	0.18
SHBG
Lung cancer overall	0.99(0.90-1.09)	0.87	1.08(0.90-1.29)	0.39	0.95(0.82-1.10)	0.47
Adenocarcinoma	1.02(0.89-1.18)	0.76	1.19(0.93-1.53)	0.17	0.93(0.75-1.15)	0.50
Squamous cell carcinoma	0.94(0.81-1.09)	0.42	0.87(0.67-1.15)	0.33	0.78(0.64-0.96)	0.02
Breast cancer	0.94(0.89-0.99)	0.04	0.93(0.84-1.04)	0.22	0.94(0.90-0.99)	0.02
Liver cancer	1.00(0.99-1.01)	0.83	1.00(0.99-1.01)	0.39	1.00(0.99-1.01)	0.32
C64 Malignant neoplasm of kidney	1.00(0.99-1.01)	0.59	1.00(0.99-1.01)	0.38	1.00(0.99-1.01)	0.10
Esophageal cancer	1.00(0.99-1.01)	0.54	1.00(0.99-1.01)	0.88	1.00(0.99-1.01)	0.89
Malignant melanoma	1.00(0.99-1.01)	0.72	1.00(0.99-1.01)	0.78	1.00(0.99-1.01)	0.33
Prostatic cancer	0.93(0.86-0.99)	0.04	1.00(0.87-1.14)	0.96	0.92(0.86-0.98)	0.01
Colorectal cancer	1.00(0.99-1.01)	0.09	1.00(0.99-1.01)	0.89	1.00(0.99-1.01)	0.50
Leukemia	1.07(0.96-1.19)	0.22	0.98(0.81-1.20)	0.86	0.96(0.80-1.16)	0.67
Thyroid cancer	0.91(0.44-1.91)	0.81	0.81(0.19-3.47)	0.77	0.73(0.31-1.69)	0.46
Oral Cancer	1.08(0.85-1.38)	0.51	1.64(1.05-2.56)	0.03	1.30(0.88-1.93)	0.19
Stomach cancer	0.81(0.63-1.04)	0.09	0.72(0.46-1.12)	0.14	0.73(0.48-1.09)	0.12

SHBG: sex hormone binding globulin; OR: odds ratio; CI: confidence intervals; SD: standard deviation; IVW: Inverse-variance weighted; P-value for heterogeneity in inverse-variance weighted analysis.

a: ORs per 1-SD increment; b: ORs per 1-unit increment (log scale); c: ORs per 1-SD increment; d: ORs per 1-SD increment; e ORs per 1-SD increment; f ORs per 1-unit increment (log scale).

3.2.2. SHBG

In Table 3, the IVW models show men with high SHBG concentration were associated with reduced breast cancer risk (OR 0.94, 95% CI, 0.89-1.00; p = 0.04) (Supplementary Figure 1) and prostate cancer risk (OR 0.93,95% CI 0.86-0.99; p = 0.04) (Supplementary Figure 5), we have not found a causal association between SHBG and other cancers.

3.3. Functional enrichment analysis

In our study, SHBG may be associated with the development of prostate and breast cancer in men, while total testosterone is associated with the development of breast cancer in women. Based on the above studies, we collected overlapping SNPs between exposure factors and outcome variables and selected SNPs with p < 0.05 for gene annotation. Therefore, we performed GO and KEGG pathway enrichment analyses to evaluate further the function and relationships of corresponding genes of the identified SNPs. The results of GO enrichment analysis showed that the corresponding genes of SHBG and prostate cancer in males are mainly enriched in the cytoskeleton, scaffold protein binding, and embryonic placenta development (Figures 1A-C). Corresponding genes of the identified SNPs of SHBG in men and breast cancer are mainly enriched in the telomerase catalytic core complex, death receptor binding, and apoptotic signaling pathway (Figures 2A–C). The corresponding genes of the identified SNPs between total testosterone and breast cancer in females are mainly enriched in the death-inducing signaling complex, the gastrointestinal peptide receptor activity, and the regulation of the primary metabolic process (Figures 3A–C). Moreover, pathway enrichment analysis with KEGG pathway enrichment showed that the differentially expressed SNPs between SHBG and male prostate cancer are mainly concentrated in human papillomavirus infection, salmonella infection and human cytomegalovirus infection (Figure 1D). The differentially expressed SNPs between total and breast cancer in women were mainly concentrated in Human cytomegalovirus infection, human T cell leukemia virus 1 Infection, and cAMP signaling pathway (Figure 3D). These enrichment analysis results may reveal the potential occurrence and development mechanisms of SHBG and total testosterone and related cancer.

Figure 1. Correlation analysis between SHBG and prostate cancer enrichment pathway in men.

A–C. Bar charts show the results of the GO enrichment analyses for SHBG and prostate cancer in males, cellular component (A), molecular function (B), and biological process (C), respectively. A darker blue indicates a lower P value.

D. Bubble chart shows the results of the GO enrichment analysis for SHBG and prostate cancer in males. The size of the circle represents the amount of enrichment, and the larger the circle, the more enrichment data. A darker blue indicates a smaller p-value.

Figure 2. Correlation analysis between SHBG and breast cancer enrichment pathway in men.

A-C Bar charts showing GO enrichment analyses for SHBG and breast cancer in males, including Cellular Component (A), Molecular Function (B), and Biological Process (C), respectively.

Figure 3. Correlation analysis of total and breast cancer enrichment pathway in women.

A–C. Bar charts showing the results of the GO enrichment analyses for total and breast cancer in females, including Cellular Component (A), Molecular Function (B), and Biological Process (C), respectively.

D. Bubble chart showing results of KEGG enrichment analysis of total and breast cancer in females. The size of the circle represents the amount of enrichment, and the larger the circle, the more enrichment data. A darker blue indicates a smaller p-value.

3.4. Sensitivity analysis

Increasing the number of SNP instruments enhances trait prediction, thereby boosting the statistical power of Mendelian Randomization (MR) analyses. However, this augmentation may also introduce elevated levels of heterogeneity in the effects of genetic instruments. For total cholesterol levels in females, it is noteworthy that heterogeneity is observed in lung adenocarcinoma, lung squamous cell carcinoma, breast cancer, malignant neoplasm of the kidney (C64), colorectal cancer, and oral cancer (Pheterogeneity < 0.05). Conversely, other causal estimates exhibit no heterogeneity or directional pleiotropy (Pheterogeneity > 0.05; PMR-Egger intercept > 0.05), as presented in Table 4. In the case of total cholesterol levels in males, heterogeneity is observed except for lung cancer overall, lung adenocarcinoma, lung squamous cell carcinoma, breast cancer, liver cancer, prostatic cancer, and colorectal cancer (Pheterogeneity < 0.05). Furthermore, breast cancer demonstrates both heterogeneity and diversity (Pheterogeneity < 0.05; PMR-Egger intercept < 0.05), while other causal estimates do not exhibit heterogeneity or directional pleiotropy (Pheterogeneity > 0.05; PMR-Egger intercept > 0.05), as outlined in Table 4.

Table 4. MR-Egger pleiotropy test of the associations between total testosterone and SHBG and the histological types of pan-cancer.

Outcome	Heterogeneity test	MR-Egger
	p-value	intercept	p-value
(1) females
Total testosterone
Lung cancer overall	0.05	−0.01	0.70
Adenocarcinoma	0.02	0.01	0.70
Squamous cell carcinoma	0.01	0.01	0.76
Breast cancer	0.01	−0.01	0.16
Liver cancer	0.08	0.01	0.43
C64 Malignant neoplasm of kidney	0.01	0.01	0.73
Esophageal cancer	0.19	−0.01	0.29
Malignant melanoma	0.08	0.01	0.64
Colorectal cancer	0.03	0.01	0.66
Leukemia	0.91	−0.01	0.94
Thyroid cancer	0.14	0.07	0.12
Oral Cancer	0.03	−0.01	0.91
Stomach cancer	0.89	0.01	0.33
SHBG
Lung cancer overall	0.01	−0.01	0.33
Adenocarcinoma	0.03	−0.01	0.47
Squamous cell carcinoma	0.01	0.01	0.19
Breast cancer	0.01	0.01	0.51
Liver cancer	0.05	−0.01	0.35
C64 Malignant neoplasm of kidney	0.38	0.01	0.67
Esophageal cancer	0.16	0.01	0.12
Malignant melanoma	0.01	0.01	0.74
Colorectal cancer	0.01	0.01	0.22
Leukemia	0.41	0.01	0.23
Thyroid cancer	0.01	0.01	0.72
Oral Cancer	0.52	−0.01	0.62
Stomach cancer	0.97	−0.01	0.23
Outcome	Heterogeneity test	MR-Egger
	p-value	intercept	p-value
(2) males
Total testosterone
Lung cancer overall	0.01	0.01	0.36
Adenocarcinoma	0.03	0.01	0.65
Squamous cell carcinoma	0.01	0.01	0.15
Breast cancer	0.01	0.01	0.04
Liver cancer	0.01	−0.01	0.94
C64 Malignant neoplasm of kidney	0.19	−0.01	0.75
Esophageal cancer	0.46	0.01	0.77
Malignant melanoma	0.06	−0.01	0.67
Prostatic cancer	0.01	−0.01	0.53
Colorectal cancer	0.01	−0.01	0.92
Leukemia	0.14	−0.01	0.50
Thyroid cancer	0.05	−0.01	0.59
Oral Cancer	0.57	−0.02	0.04
Stomach cancer	0.67	−0.01	0.35
SHBG
Lung cancer overall	0.01	−0.01	0.26
Adenocarcinoma	0.01	−0.01	0.15
Squamous cell carcinoma	0.01	0.01	0.52
Breast cancer	0.01	0.01	0.86
Liver cancer	0.01	0.01	0.27
C64 Malignant neoplasm of kidney	0.06	−0.01	0.19
Esophageal cancer	0.09	−0.01	0.73
Malignant melanoma	0.14	−0.01	0.92
Prostatic cancer	0.01	−0.01	0.19
Colorectal cancer	0.01	0.01	0.35
Leukemia	0.55	0.01	0.50
Thyroid cancer	0.01	0.01	0.85
Oral Cancer	0.28	−0.02	0.03
Stomach cancer	0.92	0.01	0.52

SHBG: sex hormone binding globulin

Regarding SHBG levels in females, heterogeneity is evident except for lung cancer, lung adenocarcinoma, lung squamous cell carcinoma, breast cancer, malignant melanoma, colorectal cancer, and thyroid cancer (Pheterogeneity < 0.05). Conversely, other causal estimates in this context do not display heterogeneity or directional pleiotropy (Pheterogeneity > 0.05; PMR-Egger intercept > 0.05), as reported in Table 4. In contrast, for SHBG levels in males, heterogeneity is observed except for lung cancer, lung adenocarcinoma, lung squamous cell carcinoma, breast cancer, liver cancer, prostatic cancer, colorectal cancer, and thyroid cancer (Pheterogeneity < 0.05). Like the female cohort, other male causal estimates do not show heterogeneity or directional pleiotropy (Pheterogeneity > 0.05; PMR-Egger intercept > 0.05), as detailed in Table 4.

3.5. Assessment of MR assumptions

Since the SNPs included in this study were selected at the genome-wide significance threshold of p < 5 × 10⁻⁸, the first MR assumption was satisfied. Statistical tests and sensitivity analyses were conducted to evaluate the second assumption’s potential violation. Sensitivity analyses were conducted using MR-Egger regression to test for global pleiotropic effect in each type of cancer, which suggested weak evidence for the existence of horizontal pleiotropy in each MR analysis since all the intercepts were not statistically significant (p > 0.05). As for the third MR assumption, we found no evidence in all published GWASs that included testosterone and SHBG-associated SNPs were genome-wide significantly associated with any other phenotype except for testosterone and SHBG, suggesting no violation of the assumption. Therefore, no evidence was suggested that our genetic instruments of testosterone and SHBG-associated SNPs were genome-wide significantly associated with any other phenotypes, which supports that the third MR assumption was not likely to be violated in our study.

4 Discussion

Main findings and comparisons with the literature from this large Mendelian randomized study revealed a causal association of testosterone and SHBG levels with risk for breast, ovarian, cervical, stomach and prostatic cancers. Furthermore, we performed SNP enrichment analyses for exposure factors and outcome variables with p < 0.05, exploring potential mechanisms between SHBG and breast and prostate cancer in men and between total and breast cancer in women. This is the first study with sufficient sample size under Mendel’s random hypothesis and demonstrates a lack of causal relationship between total testosterone and SHBG concentrations and the risks of lung cancer, liver cancer, kidney cancer, esophageal cancer, malignant melanoma, colorectal cancer, leukemia, thyroid cancer and oral cancer.

4.1. Breast cancer

In our findings, genetic prediction indicates that a higher concentration of circulating total testosterone increases the risk of breast cancer in women, but for men, no correlation was observed. Meanwhile, high concentrations of SHBG were associated with a risk reduction of breast cancer in men, an association we did not observe in women. The association of testosterone with overall BC and ER + BC risk has been reported before using MR methods [29]. A large prospective study observed an increased risk of breast cancer among women with relatively high circulating levels of testosterone, while those with relatively high circulating levels of SHBG had reduced risk [30]. A negative association between SHBG levels and breast cancer was observed in a meta-analysis of prospective studies. Similarly, similar MR reports have been made that testosterone increases the risk of breast cancer [31,32], consistent with our study. The possible reason for this phenomenon is that after testosterone is combined with SHBG, testosterone can be used to reduce the part that can be converted into estradiol, reduce the combination of estradiol and ER, induce the transcription of growth-positive genes and reduce the expression of cell growth negative regulators, thereby inhibiting the proliferation of breast cancer cells [33].

4.2. Ovarian cancer

Results from our IVW approach showed that women with high total testosterone concentrations had a reduced risk of ovarian cancer, and SHBG was not causally associated with ovarian cancer. Prior observational studies suggest not observing associations between circulating androgen, SHBG levels and risk of ovarian cancer [34,35]. However, the relationship between testosterone, SHBG and ovarian cancer may be related to its histological subtypes, especially in postmenopausal women [10]. Three prospective studies evaluated the risk of circulating testosterone by subtype, and the results showed that higher concentrations of estrogen and androgen were associated with the increased risk of non-serous infiltrating epithelial ovarian cancer subtype [36–38]. In a recent pooled analysis of cohort studies, higher levels of testosterone were related to an increased risk of ovarian cancer [39]. Previous epidemiologic studies have not consistently shown associations between circulating androgens and overall ovarian cancer risk. Laboratory data indicate that androgen-related signal transduction is associated with ovarian cancer by increasing cell proliferation and reducing cell apoptosis rate [40]. Therefore, testosterone may play an important role in the onset and progression of ovarian cancer or subgroups.

4.3. Cervical cancer

Our MR results indicate a positive correlation between circulating testosterone concentration and cervical cancer. The relationship between testosterone and SHBG and cervical cancer has been reported. In a large cohort study, Rinaldi S found a significant positive correlation between testosterone levels in premenopausal and postmenopausal women and cervical cancer [41]. In the 1-year serum testosterone level detected after surgery for cervical cancer, it was found that the testosterone level decreased significantly [42]. This is consistent with our research results. There is a certain correlation between testosterone levels and cervical cancer. Interestingly, the reported SHBG concentrations between patients with cervical cancer and the control group were not the same [43–45]. Therefore, more research is needed to support our results.

4.4. Stomach cancer

In previous studies, testosterone has shown different roles in gastrointestinal tumors. In a large prospective cohort by Mc Menamin C, the results showed that testosterone was almost unrelated to gastrointestinal cancers such as gastric, esophageal, and colorectal cancers [46]. Our results also support this study. Testosterone is not associated with esophageal and colorectal cancer, but our study found that high concentrations of testosterone increase the risk of gastric cancer in men. No studies have investigated the impact of circulating sex hormone levels on the risk of gastric cancer. It is worth noting that basic research has shown that testosterone inhibits the occurrence and development of gastric cancer in male rats. Previously, a Mendel analysis by Chang J showed similar results [47]. The role of testosterone in the digestive tract is currently unclear.

4.5. Prostatic cancer

In our MR analysis, testosterone was not associated with prostate cancer risk, while high concentrations of SHBG were associated with a causal relationship with a reduced risk of prostate cancer. In a prospective, nested case-control study, high levels of circulating testosterone and low levels of SHBG were found to be associated with an increased risk of prostate cancer [48]. In a blood-based MR analysis by Watts EL, SHBG was negatively correlated with overall and early-onset prostate cancer. We support the association between SHBG and prostate cancer risk [49]. However, there is still some other evidence that contradicts us [50–52], and further research is needed to confirm that.

4.6. Other cancer types

As for other cancers, our research also yields some conclusions. The relationship between lung cancer, testosterone, and SHBG has rarely been explored. However, it has been previously elucidated that androgen receptors and androgen-dependent gene expression exist in the lung. It seems to be feasible to study testosterone and SHBG on lung squamous cell carcinoma. Our results by the IVW method showed that no connection was found between total testosterone concentration and lung cancer. There was also no association between SHBG concentrations and lung cancer. However, in the Weighted median method, we found an association between genetically predicted circulating SHBG concentrations and Squamous cell carcinoma risk in men (OR 0.78, 95% CI 0.63–0.96; p = 0.02). There is still a lack of mechanisms by which testosterone and SHBG levels affect lung cancer, and their specific mechanisms in the progression of lung cancer still require further research.

4.7. Mechanism exploration

In our study, we performed enrichment analyses of SHBG with prostate and breast cancer in men and total testosterone with breast cancer in women to further explore potential mechanisms. Consistent with our conclusions and enrichment analysis, Fachen et al. [48], in a Mendelian randomized study, found an association of exposure factors such as SHBG, telomere length, and HDL-cholesterol with breast cancer. Meanwhile, in contrast to our findings, blood-based analyses by Eleanor L. Watts et al. [53] showed a negative correlation between SHBG and overall prostate cancer. However, there is still a lack of in vivo and in vitro experiments related to the above conclusions, which may be the mechanism for further research in the future.

4.8. Strengths and limitations of this study

Our studies have some advantages. First, this study used a series of MR analyses to evaluate the effect of circulating total testosterone and SHBG on the risk of as many as fourteen cancers, including lung, breast, and ovarian, covering common malignant diseases of six major human systems. Second, others rarely have established a causal association between elevated circulating total testosterone levels and cancer risk. At the same time, to clarify the possible mechanism, we also conducted KEGG and GO enrichment analysis, which may become a hot topic for further research. Our findings suggest that testosterone and SHBG supplementation holds promise for cancer prevention and even as a cancer treatment. Finally, we adopt a more effective method, MR, rather than observational research methods for causal inference, which can effectively avoid confounding factors [54] as well as tell the "upstream" factor and the "downstream" consequence apart . Besides, the resulting phenotype of MR is highly heritable, and using a large number of IVs, multiple statistical methods, and heterogeneity testing methods can provide more reliable causal effect estimation. We validated a good genetic variant of IV in the UK Biobank, able to minimize polymorphism of IV.

There are some limitations of our study. Firstly, we need to perform bidirectional MR to avoid horizontal pleiotropy. Moreover, the ability of our study to detect causal associations may be limited by the number of IVs, so the validity of IVs was examined as rigorously as possible. Besides, our study lacks subgroup data stratified by age and is mainly based on European ancestry participants, so it isn’t easy to compare differences in causal effects between subgroups or figure out relationships varied by race.

5. Conclusion

In conclusion, our study shows that testosterone and SHBG are strongly associated with reduced risk of several cancers, suggesting that modulating testosterone and SHBG concentrations may help to inhibit cancer development and progression. In addition, testosterone and SHBG supplementation might be recommended as a strategy for primary cancer prevention based on further mechanism explorations.

Authors’ contributions

Xiwen Liu, Lixuan Lin, Qi Cai and Caichen Li designed the article. Haoxiang Xu participated in editing the code. Ruiqi Zeng and Mingtong Zhang performed data analysis and made the graphs. Xinyi Qiu, Shiqi Chen, Xizhe Zhang and Linchong Huang collated the data results. Xiwen Liu, Lixuan Lin and Qi Cai wrote the first draft of the manuscript. Xiwen Liu, Caichen Li and Haoxiang Xu revised the primary version. Wenhua Liang and Jianxing He revised the final version of the manuscript. All authors contributed to revisions of the manuscript. LXW is the guarantor. All authors approved the final manuscript.

Acknowledgements

We acknowledge the First Affiliated Hospital of Guangzhou Medical University and State Key Laboratory of Respiratory Disease & National Clinical Research Center for Respiratory Disease for providing access to necessary resources for conducting this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Funding

This study was supported by the China National Science Foundation (Grant No. 82022048 & 81871893), the Key Project of Guangzhou Scientific Research Project (grant No. 201804020030).