The clinical impact of androgens in endometrial cancer: cancer promoter or suppressor?

Abstract

Purpose

Androgens are associated with the risk of endometrial carcinoma (EC). However, their roles as prognostic factors are less clear. This study aims to investigate the prognostic impact of different androgens.

Methods

We conducted a retrospective cohort study of patients who had previous endometrial sampling that showed EC between August 19, 2003, and November 4, 2015. This study included 465 patients with an average follow-up of 37 months. We examined associations between endogenous testosterone levels (total, bioavailable and free testosterone) and the clinical outcomes in patients with EC. Serum levels of bioavailable testosterone were calculated using the total testosterone level and the sex hormone-binding globulin concentration (measured by chemiluminescent and radioimmunoassays, respectively).

Results

Higher circulating concentrations of total and bioavailable testosterone were associated with a higher risk of type I endometrial cancer (OR = 1.015, 95% CI: 1.008–1.022; OR = 1.395, 95% CI: 1.266–1.536, respectively), but the similar association was not found in type II patients (OR = 1.009, 95% CI: 0.998–1.020; OR = 1.208, 95% CI: 0.998–1.461, respectively). However, higher levels of testosterone in EC patients correlated with favorable clinicopathological features, especially bioavailable and free testosterone. Furthermore, Kaplan–Meier survival analysis indicated that elevated bioavailable and free testosterone levels were associated a longer overall survival (P = 0.048, P = 0.036, respectively), but this trend was not statistically significant after adjustment for other prognostic factors (P = 0.312, P = 0.272, respectively).

Conclusion

Elevated serum testosterone levels were associated with an increased risk of type I cancers. Although androgens showed certain prognostic values, they were not independent prognostic factors. Different androgenic indexes showed different clinical implications. The risk estimates could refer to total and/or bioavailable testosterone. Bioavailable and/or free indexes may be better markers for the probable prognostic effects.

Introduction

Endometrial carcinoma (EC) is the most common female genital tract cancer (1). It is well established that an excess-estrogen environment is associated with the risk of EC. Now, accumulating evidence supports the potential of androgens in the carcinogenesis of EC.

The preponderance of epidemiological studies has reported strong associations between elevated levels of testosterone and an increased risk of EC in postmenopausal women (2, 3, 4, 5, 6, 7). However, it is unclear if androgens influence the prognosis of EC patients. In addition, existing studies exploring circulating androgens and EC focused only on total hormone levels and/or its free fraction in plasma. Testosterone circulates in the bloodstream nonspecifically bound to albumin, specifically bound to sex hormone-binding globulin (SHBG), and only a small part dissolved freely. The SHBG-bound fraction is biologically inactive because of the high binding affinity of SHBG for testosterone. The free and nonspecifically bound plasma hormone levels are commonly identified as the bioavailable fraction and can potentially cross the cellular membranes exerting androgenic actions (8). It is important to determine which one is the best reliable index of testosterone among three androgenic indicators, including bioavailable testosterone, free testosterone, and total testosterone. In this study, the potential role of androgens as risk factors for EC was further evaluated based on the histology of EC (type I and type II). All of these were estimated using the three androgenic indexes to get the best reliable index of bioactive androgens in the circulation.

Materials and methods

Patient samples

A total of 465 patients with EC and 465 controls at Tianjin Medical University General Hospital were included in the study between August 19, 2003, and November 4, 2015. The patients were first diagnosed with histologically confirmed EC. None of them underwent hormonal therapy, radiotherapy, or chemotherapy prior to surgery. Tumor grade, stage (using the surgical staging criteria from the International Federation of Gynecology and Obstetrics, (FIGO)), lymph node status, myometrial invasion, cervical invasion, and ovarian invasion were retrieved by review of hospital notes and clinical databases. Women who went to Tianjin Medical University General Hospital for a routine examination were used as controls. None of them had a history of cancer, and they were time- and nationality-matched to the patients.

Blood collection and laboratory methods

Blood samples were collected early in the morning after 6 h of fasting and prior to surgery (for patients) or on the day of regular examination (for healthy subjects). The levels of total testosterone were measured using chemiluminescence reagents (Siemens Medical Solutions, USA). SHBG was quantified with the use of sandwich enzyme-linked immunosorbent assay (R&D Systems, USA) following the manufacturer’s protocols. The levels of bioavailable and free testosterones were calculated with a confirmed algorithm that used the total testosterone hormone level, SHBG level, albumin level, and association constants for the binding of testosterone to SHBG and albumin (8, 9). For the three androgenic indexes, quartiles 3 and 4 (Q3 and Q4) were labeled as ‘high level’ and Q1 and Q2 were designated as ‘low level’. All participants were informed about all aspects of this study and gave written informed consent. The Ethical Committee of Tianjin Medical University General Hospital approved this study, and the study was performed in a manner consistent with the Declaration of Helsinki (2013) (EthicaL No IRB2015-YX-009).

Statistical analysis

Data were statistically analyzed with SPSS version 22.0, MedCalc and GraphPad Prism 6. In clinical studies, the risk factors for EC were analyzed by conditional logistic regression. Kaplan–Meier survival plots were used to perform survival curves, and the log-rank test was used to compare significance. The chi-squared test was used to identify whether a significant difference existed between high-androgen and low-androgen groups. The two-tailed Student’s t test or analysis of variance for functional analysis determined statistical significance. The Cox regression model was used to determine independent prognostic factors. Multivariate models included only variables with P < 0.05 in univariate analysis. P value less than 0.05 was considered statistically significant.

Results

Demographic data

The details of the study population are displayed in Table 1. The significant differences between cases and control subjects in line with known EC risk factors were observed. EC patients had higher concentrations of androgens (total testosterone, free testosterone and bioavailable testosterone), especially total and bioavailable testosterone levels (P < 0.001, Table 1).

Table 1.

Basic characteristics of patients with EC and healthy subjects.

	Cases (n = 465)	Controls (n = 465)	P-value
Age (y)	57.98 ± 9.62	56.58 ± 10.23	0.032
BMI (kg/m2)	27.15 ± 4.91	24.80 ± 3.60	<0.001
Reproduction			<0.001
Never	44 (73.3%)	16 (26.7%)
Ever	421 (48.4%)	449 (51.6%)
Total testosterone (ng/dL)	50.30 ± 54.18	39.36 ± 17.62	<0.001
Free testosterone (ng/dL)	0.24 ± 2.05	0.08 ± 0.05	0.106
Bioavailable testosterone (ng/dL)	3.73 ± 9.18	1.93 ± 1.25	<0.001

BMI, body mass index.

Serum total/bioavailable testosterone levels were independent risk factors for type I EC

Of the 465 EC cases included in this study, 59 (12.7%) were non-endometrioid adenocarcinoma. We detected positive associations between type I EC risk and total/bioavailable testosterone levels in both unadjusted and risk factor-adjusted models. However, after adjustment for putative EC risk factors, the levels of total/bioavailable testosterone were not associated with the risk of type II EC (Table 2).

Table 2.

Odds ratio for EC risk using logistic regression.

	Type I (406 cases)						Type II (59 cases)
	Unadjusted			Adjusted*			Unadjusted			Adjusted*
	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value	OR	95% CI	P-value
Total testosterone	1.018	1.011–1.024	<0.001	1.015	1.008–1.022	<0.001	1.012	1.000–1.023	0.041	1.009	0.998–1.020	0.121
Free testosterone	1.000	-	1.000	-			1.000	-	1.000	-
Bioavailable testosterone	1.494	1.360–1.641	<0.001	1.395	1.266–1.536	<0.001	1.272	1.062–1.524	0.009	1.208	0.998–1.461	0.052

^*Adjusted for BMI, age and reproduction. OR, odds ratio; CI, confidence interval.

High levels of bioavailable and free testosterone were positively associated with favorable prognostic factors

Then, testosterone levels were further classified into two groups (high level and low level). The correlations between the levels of different testosterone indexes and clinicopathological parameters are summarized in Supplementary Table 1 (see section on Supplementary materials given at the end of the article). Both free and bioavailable testosterone levels correlated with those without deep myometrial, lymph node and cervical invasion (P < 0.05, Supplementary Table 1). They also showed positive correlation with early FIGO stages (P < 0.05, Supplementary Table 1).

This study also examined the association between total serum testosterone and clinicopathological characteristics, as shown in Supplementary Table 1. Lower total testosterone levels were observed only in patients with myometrial invasion (P = 0.002).

Bioavailable and free testosterone influenced patient outcomes

Follow-up data were available for 435 patients. During the follow-up, 30 deaths occurred (23 due to disease progression and 7 from other reasons). By November 2015, the average follow-up time was 37 months, ranging from 1 to 130 months. Generally, patients with high levels of androgens showed a longer overall survival (OS) than those with low levels of androgens (Fig. 1). A significant reduction in OS was observed in patients with low levels of bioavailable testosterone (112.790 ± 3.937 months vs 124.144 ± 3.059 months, P = 0.048) and free testosterone (112.444 ± 3.997 months vs 124.252 ± 3.033 months, P = 0.036) (Fig. 1A and B). Patients with a high total testosterone level also had a longer median OS (123.265 ± 3.256 months vs 112.883 ± 4.029 months). However, the difference between the two groups was not significant (P = 0.097) (Fig. 1C). The univariate analysis showed that high FIGO grade (G3), advanced stage (III + IV), positive lymph node status, cervical invasion, ovarian invasion and low level of bioavailable testosterone were significantly associated with progressive disease (Table 3). Furthermore, the Cox regression model found that only tumor grade (P = 0.051) and stage (P < 0.001) were independent prognostic indicators (Table 3).

Figure 1.

Correlations between levels of bioavailable testosterone (A)/free testosterone (B)/total testosterone (C) and overall survival were evaluated using the Kaplan–Meier survival curve.

Table 3.

Univariate and multivariate analyses of overall survival in EC.

Variables	Univariate	Multivariate analysis*		Multivariate analysis†
	P-value	P-value	HR (95% CI)	P-value	HR (95% CI)
Age
≤50 vs >50	0.637	Removed	——	Removed	——
BMI
<24 vs ≥24	0.295	Removed	——	Removed	——
Reproduction
Never vs ever	0.981	Removed	——	Removed	——
FIGO grade
G1 + G2 vs G3	0.002	0.051	2.710 (0.997–7.367)	0.051	2.710 (0.997–7.367)
FIGO stage
I + II vs III + IV	<0.001	<0.001	5.864 (2.367–14.531)	<0.001	5.864 (2.367–14.531)
Lymph node status
Negative vs positive	<0.001	0.565	——	0.565	——
Cervical invasion
No vs yes	0.009	0.104	——	0.104	——
Ovarian invasion
No vs yes	0.001	0.545	——	0.545	——
Bioavailable testosterone
Low vs high	0.048	0.312	——	Removed	——
Free testosterone
Low vs high	0.036	Removed	——	0.272	——
Serum total testosterone
Low vs high	0.097	Removed	——	Removed	——

HR, hazard ratio; CI, confidence interval.

^*The model was adjusted for FIGO grade, FIGO stage, lymph node status, cervical invasion, ovarian invasion and bioavailable testosterone.

^† The model was adjusted for FIGO grade, FIGO stage, lymph node status, cervical invasion, ovarian invasion and free testosterone.

Discussion

The importance of androgens in EC has been recognized since the last century. Increased levels of androgens were associated with an increased risk of EC. We have previously shown that androgens were risk factors for EC (2, 7). The data presented here highlight the importance of uncovering the role that androgens play in EC prognosis. To our knowledge, this is the first study to show the prognostic value of androgens in patients with EC.

In this study, we re-evaluated the association between androgens and EC risk. Previous reports of the potential role of androgens as risk factors for EC have not distinguished between endometrioid type I and non-endometrioid type II ECs (7). It has been well-known that an excess-estrogen environment is associated with a risk specifically of type I cancers. One of the mechanisms of estrogen production is the conversion of testosterone. In addition, the main source of estrogen production in women after menopause is peripheral aromatization in adipose tissues (10). In order to get better insight into the testosterone actions in EC, we separately assessed the association between testosterone and type I/type II cancers. We found that higher total and bioavailable testosterone levels had been more consistently associated with greater endometrioid type I EC risks. These positive associations attenuated after adjustment for EC risk factors (such as BMI), supporting the hypothesis that androgens influence EC risk via estrogenic pathways. Further support for this hypothesis comes from observational studies that the association of androgens with EC risk is weaker when the levels of serum estrogens are taken into account (3, 5). The higher level of testosterone could be converted to estrogen via peripheral or local aromatization.

The association between circulating testosterone concentration and type II EC risk has rarely been studied. Type II non-endometrioid adenocarcinomas are characterized by the absence of estrogen stimulation. In our study, we found statistically non-significant association between testosterone and type II cancer risk in the adjusted multivariable model. This finding should be interpreted cautiously as the number of type II cases was relatively small (n = 59 cases). However, this suggestive association is consistent with the well-established opinion that type II cancers are not hormone-dependent. Further larger and prospective studies are needed to examine the role of androgens in the development of type II cancers.

All of the above-mentioned results are not enough to reveal the effect of androgens in EC due to the lack of prognosis data. Therefore, we evaluated the associations between different levels of androgens and tumor parameters. Interestingly, a trend between higher circulating testosterone concentration and better EC prognosis was observed in our analysis. It is of note that the prognosis estimates after adjustment for other confounders did not reach the threshold of statistical significance, suggesting that the trend may be a consequence of the combined effect of oncogenic and tumor-suppressive roles. The protective effect of testosterone in EC may be largely attributed to its conversion to dihydrotestosterone (DHT), more potent than testosterone. Preclinical evidence indicates that DHT has an antiproliferative effect on EC cell growth that is regulated by the upregulation of the 17β-hydroxysteroid dehydrogenase type 2 expression (11). However, testosterone can be also catalyzed to the most potent estrogen, estradiol (10), which generally stimulates proliferation of EC cells. In addition, hyperandrogenism is often combined with several components, such as insulin resistance and visceral adiposity, which could contribute to the development of EC and partly counter the testosterone-mediated effects (2, 12, 13). Overall, current evidence is unsupportive to circulating testosterone level being a significant prognostic factor.

It is important to note that androgens cannot be considered equal in terms of action. Circulating testosterone in plasma is mostly bound to SHBG, and the SHBG-bound fraction is biologically inactive because of the high binding affinity of SHBG for testosterone (14, 15). Previous studies on circulating androgens and EC are mostly confined to total serum testosterone. A prospective study found a statistically significant inverse association between circulating SHBG concentration and EC risk (16). In our analysis, we found that bioavailable and free testosterone presented a better positive prognostic power, and risk estimates were more associated with bioavailable and total indexes. So, although free, bioavailable and total testosterone all represent the indexes of testosterone, different indexes may show different clinical significance.

A limitation of our analysis was that testosterone was measured by chemiluminescent immunoassay and not by the current gold standard mass spectrometry method (17); however, the association between increased testosterone levels and EC risk was consistent with the previous studies that measured androgens using liquid chromatography with tandem mass spectrometry (3, 4). In addition, the assessment of sex hormone levels was at a single point after their enrollment in this study and the circulating hormone concentrations may not reflect the local hormonal milieu of the endometrial tissue. A further potential limitation of our study is that we were unable to assess the influence of estrogens in mediating the relationship between testosterone and EC association due to the lack of related data.

In conclusion, our study supported probable causal associations between circulating testosterone and type I EC. However, its prognostic value in EC was limited. So, we concluded that testosterone was more prone to be a cancer promoter in EC.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

Funding

This study was funded by the National Natural Science Foundation of China (no. 82172626).

Ethical approval and consent to participate

The Ethical Committee of Tianjin Medical University General Hospital approved this study (No. IRB2015-YX-009). All methods were performed in accordance with the relevant guidelines and regulations. This study was performed in accordance with the Declaration of Helsinki. All participants were informed about all aspects of this study and gave written informed consent.