Prediagnostic circulating levels of sex hormones and survival in esophageal adenocarcinoma

Abstract Sex hormonal differences may contribute to the strong male predominance in esophageal adenocarcinoma (EAC), but whether sex hormone levels influence survival in EAC is unstudied. Our study aimed to assess associations between prediagnostic sex hormone levels and survival in EAC. In a population‐based cohort study, 244 male EAC patients from the Janus Serum Bank Cohort in Norway were followed up through 2018. Associations between prediagnostic serum levels of 12 sex hormone measures and disease‐specific mortality were assessed using multivariable Cox regression, providing hazard ratios (HR) with 95% confidence intervals (CI) adjusted for age, calendar year, body mass index, tobacco smoking, physical activity and surgical resection. Higher levels of sex hormone‐binding globulin (SHBG) indicated decreased disease‐specific mortality (HR 0.68, 95% CI 0.44‐1.07, highest vs lowest tertile). In stratified analyses by surgery, such associations remained in nonoperated patients (HR 0.58, 95% CI 0.35‐0.96, highest vs lowest tertile), but not in operated patients. Higher levels of follicle‐stimulating hormone (FSH) were associated with increased disease‐specific mortality in an exposure‐response pattern; HRs for the middle and highest tertiles vs the lowest tertile were 1.35 (95% CI 0.89‐2.05) and 1.61 (95% CI 1.06‐2.43), respectively. No clear associations were observed with serum levels of dehydroepiandrosterone sulfate, luteinizing hormone, prolactin, testosterone, 17‐OH‐progesterone, progesterone, estradiol, androstenedione, testosterone:estradiol ratio or free testosterone index. These findings suggest that higher endogenous levels of SHBG and lower levels of FSH may increase the survival in EAC. The other 10 examined sex hormone measures may not influence the survival.

We did not include female patients because of the low incidence of EAC and the great menstrual variations in sex hormone levels in women. The male EAC patients were followed up from the date of diagnosis until death, emigration or the end of the study (31 December 2018), whichever occurred first.

| Hormone level exposures
The participants donated serum samples when they entered the Janus Serum Bank Cohort. To assess the validity of the archived samples, repeated stability experiments have confirmed that relevant serum components are stable after long-term storage. [19][20][21][22] We evaluated the following 12 sex hormone measures in serum samples: sex hormone-binding globulin (SHBG); the nine steroid sex hormones dehydroepiandrosterone sulfate, follicle-stimulating hormone (FSH), luteinizing hormone, prolactin, testosterone, 17-OH-progesterone, progesterone, estradiol and androstenedione; and finally calculated free testosterone index (testosterone × 10/SHBG) and testosterone:estradiol ratio. These hormone measures cover key points in the biosynthesis of sex hormones, 7,11,23 and are at detectable levels in serum in men with the available analytic methods.
These analyses were conducted at the Hormone Laboratory, Oslo University Hospital. Testosterone, 17-OH-progesterone and androstenedione were analyzed using liquid chromatography-mass spectrometry, and the remaining hormones were analyzed using immunoassays, all according to standard laboratory protocols. 24

| Mortality outcomes
The main outcome was disease-specific mortality after diagnosis of EAC, defined by esophageal cancer as the underlying cause of death, through 31 December 2017 (the latest date for which data on causes of death were available). The secondary outcome was all-cause mortality after EAC diagnosis until end of the entire study period, that is, 31 December 2018.

| Covariates
The covariates considered were six factors that might influence the survival in EAC: age, calendar year at diagnosis, body mass index (BMI), tobacco smoking, physical activity and curatively intended

What's new?
Esophageal adenocarcinoma (EAC) occurs more frequently in men than women. Whether this pattern is linked to differences in sex hormone levels and whether such differences impact EAC survival remain unclear. In this study of male EAC patients in Norway, analyses of prediagnostic sex hormone measures uncovered associations between reduced disease-specific mortality and increased sex hormonebinding globulin and decreased follicle-stimulating hormone levels. The associations were detected only in patients who had not undergone surgery. Ten other sex hormone measures also analyzed had no influence on survival. Additional investigation is needed to better understand relationships between sex hormone levels and EAC survival. esophagectomy. Data on age, calendar year and surgery were retrieved from the Cancer Registry of Norway, whereas information about BMI, smoking and physical activity came from questionnaires completed at baseline, that is, when the participants were recruited to the Janus Serum Bank Cohort. The assessment of the questionnaire data has been described in detail elsewhere. 15,25

| Statistical Analysis
Pairwise Spearman correlation analysis was used to assess correlations between the 12 sex hormone measures. We assessed associations between levels of each sex hormone measure and mortality using multivariable Cox regression, providing hazard ratio (HR) with 95% confidence interval (CI), adjusted for covariates. The sex hormone measures were treated as categorical variables based on the cut-off values of tertiles (three equal-sized groups), except for progesterone, estradiol and androstenedione, where over one third of the values were below the limit of detection. Levels of these three sex hormones were instead categorized into three groups, that is, one group of values below the limit of detection and the other two groups of approximately equal sizes for the detectable values. Two Cox models were derived for each sex hormone, one basic model with adjustment for age and calendar year at diagnosis (both as continuous variables) and a full model with further adjustment for BMI (continuous), tobacco smoking (never or ever), physical activity (high/medium or low/inactive) and esophagectomy (yes or no). We also assessed possible exposure-response relations by treating each sex hormone measure as a discrete variable (ie, values as 1, 2 or 3) in the Cox regression. Because of the differences in survival between patients who had undergone surgery and those who had not, we further stratified the analyses by esophagectomy for the main outcome, that is, disease-specific mortality. An experienced biostatistician (FM) performed all analyses according to a predefined protocol, using the statistical software package SAS 9.4 for Windows (SAS Institute Inc., Cary, NC). All statistical tests were two sided.

| Sex hormone levels
The distribution of serum levels of each sex hormone is presented in Supplementary

| Sex hormone levels and disease-specific mortality
Patients with higher SHBG levels had decreased risk of disease-specific mortality in EAC (fully adjusted HR 0.68, 95% CI 0.44-1.07, highest vs lowest tertile) ( Table 2). In analyses stratified by surgery, higher SHBG levels were associated with decreased disease-specific mortality in nonoperated patients (adjusted HR 0.58, 95% CI 0.35-0.96, highest vs lowest tertile), while no such decrease was found in operated patients (Table 3).
Patients with higher FSH levels had increased disease-specific mortality in an exposure-response pattern (  (Table 3).
Higher luteinizing hormone levels were associated with increased point estimates of disease-specific mortality (adjusted HR 1.38, 95% CI 0.90-2.10, highest vs lowest tertile), but the risk estimates were not statistically significant (Tables 2 and 3).
No clear associations were observed between serum levels of dehydroepiandrosterone sulfate, prolactin, testosterone, 17-OH-progesterone, progesterone, estradiol, androstenedione, testosterone: estradiol ratio, or free testosterone index, and the risk of diseasespecific mortality in EAC patients (Tables 2 and 3). Hazard ratio with 95% confidence interval, adjusted for age (continuous), calendar year at diagnosis (continuous), body mass index (continuous), tobacco smoking (never or ever), physical activity (high/medium, or low/inactive) and surgical treatment (yes or no).
T A B L E 3 Associations between prediagnostic sex hormone levels and disease-specific mortality in esophageal adenocarcinoma, stratified by surgical treatment (N = 217) a

| Sex hormone levels and all-cause mortality
Associations between levels of each sex hormone and all-cause mortality are shown in Table 2 in the stratified analyses in the operated patients.
The observed associations between higher FSH levels and an increased disease-specific mortality in EAC in our study are biologically plausible, and might thus not be a chance finding. Our recent Mendelian randomization analysis using data from international consortia of genome-wide association studies suggested that higher genetically predicted levels of FSH increase the risk of EAC and Barrett's esophagus in both men and women. 28 Other investigations smoking (never or ever) and physical activity (high/medium, or low/inactive).