Associations between serum sex steroid hormone metabolites and gastric cancer and precancerous lesions in men: A 11.8-year prospective study

Selection and description of participants

A population-based upper gastrointestinal cancer screening cohort was established from 2007 to 2012 in Linzhou, China. Given the substantial differences in sex hormone levels between men and women, and the fact that women experience significant hormonal fluctuations before and after menopause while men’s hormone levels decline gradually with age and are unaffected by reproductive cycles, we selected men as study participants. This approach ensures hormonal stability and is consistent with the design of previous related studies. After a baseline questionnaire interview and physical examination, a total of 558 men were enrolled and 10 mL of fasting blood was collected from each participant who had fasted for at least 12 h prior to the physical examination. Serum samples were immediately separated, aliquoted, and stored frozen at -70 ℃ until they were prepared for the laboratory analysis.

The endoscopic examination and biopsy procedures were described in detail in a previous study.^[27] Specifically, the participants received a local anesthetic (5 mL of 1% lidocaine solution) orally 5 min before endoscope insertion, followed by a complete visual examination of the esophagus and stomach. Then, the stomach was sprayed with 0.2% indigo carmine, which stains the gastric fundic gland mucosa light red and imparts dull yellow to the pyloric gland mucosa. In areas of abnormal gastric mucosa, the dye demonstrates abnormal deposition, resulting in accentuated staining. Next, all areas with accentuated staining > 5 mm in diameter were biopsied, with the number of specimens collected (range 1–3) depending on lesion dimensions (5–19 mm, 1 biopsy; 20–39 mm, 2 biopsies; and ≥40 mm, 3 biopsies). Biopsy specimens underwent standard processing including 10% buffered formalin fixation, paraffin embedding, sectioning at 5-μm, and hematoxylin-eosin (HE) staining. All sides were independently evaluated by two well-trained pathologists, with discordant cases reviewed jointly to reach diagnostic consensus. Finally, the histological criteria strictly adhered to the 2020 edition of Chinese Technical Guidelines for Screening, Early Diagnosis, and Treatment of Upper Gastrointestinal Cancers.^[28]

We excluded men who were diagnosed as severe atypical hyperplasia or carcinoma of the esophagus at baseline (n = 53) (Figure 1). Among the 505 eligible men, there were 32 participants diagnosed with GC (n = 19) or high-grade lesions (n = 13), 146 diagnosed with IM, and 317 normal participants. We randomly matched the IM cases with normal participants by age and village with a ratio of 1:2, and finally included 32 high-grade lesions and GC cases, 146 IM cases, and 292 normal participants in the cross-sectional analytical dataset (n = 470). From 2007 to 2021, the village health workers checked vital status and the occurrence of incident cancers and ascertained causes of deaths for all participants by monthly home visits, supplemented by quarterly crosschecks of the data in the Linzhou Cancer and Death Registries. New GC cases and all causes of death were reviewed by a panel of professional experts from the Cancer Hospital, Chinese Academy of Medical Sciences in Beijing, China. Diagnostic materials used in these expert reviews included medical records, pathology and cytology slides, biochemical results, X-rays, ultrasound, endoscopy, and surgical reports. The International Classification of Diseases for Oncology, Third Revision code C16.0 was used to define CGC, and NCGC was defined with topography code C16.1–C16.9.^[29] Primary causes of death were classified by the International Classification of Diseases-10.^[30] By December 2021, a total of 32 new GC cases were diagnosed in the prospective analytical dataset (n = 438, including 146 IM and 292 normal participants).

Figure 1

Flowchart of the study for diagnostic analysis and prospective analysis. IM, intestinal metaplasia; Gastric cancer includes non-cardia gastric cancer and cardia gastric cancer.

Data collection

All participants underwent a physical examination at local hospitals and completed a standardized questionnaire according to the national upper gastrointestinal cancer screening guideline.^[29] The questionnaire included items on demographics (age, sex and residence location), height, weight, smoking status (regular cigarette or pipe use for at least 6 months), alcohol consumption (any alcohol consumption in the past 12 months), history of upper gastrointestinal disease (i.e., ulcer, esophagitis, gastroenteritis, and hepatitis), and family history of cancer (at least one first-degree relative with cancer).

Laboratory analysis

H. pylori infection was measured by immunoblot assay at the laboratory of Linzhou Cancer Hospital. SHBG was tested with electrochemical luminescence immunoassay at the laboratory of the Cancer Hospital, Chinese Academy of Medical Sciences. Sex steroid hormone metabolites were assessed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) system at the Integrated Chinese and Western Medicine Laboratory of Dalian Medical University. A total of 20 sex steroid hormone metabolites were quantified, including testosterone, 16 α-hydroxytestosterone, androstenedione, 17α-hydroxyprogesterone, 17α-hydroxypregnenolone, progesterone, estrone, epitestosterone, DHEA, pregnenolone, dihydrotestosterone, etiocholanolone, 17-epiestriol, estradiol, 2-methoxyestrone, 2-methoxyestradiol, 16-epiestriol, androsterone, 11-oxoetiocholanolone, and 6β-hydroxytestosterone. Quality control (QC) samples were prepared by evenly mixing aliquots of randomly assigned samples and were inserted into the analytical sequence. A total of 35 QC samples were evenly distributed on each solid phase extraction plate and analyzed by LC-MS/MS. The merits of the analytical performance, including reproducibility, i.e., coefficients of variance (CVs) of the concentration determinations in these QC samples, and sensitivity of the LC-MS/MS method (limits of detection (LODs), limits of quantification (LOQs)), were reported in the Supplementary Materials. In these QC samples, CVs were less than 46% (range = 1.14%–45.98%), LODs were less than 5 μg/L (range = 0.25–5 μg/L), and LOQs were less than 5 μg/L (range = 0.25–5 μg/L).

Statistics

Baseline characteristics and the serum sex steroid hormone metabolites levels of participants were presented as median or percentage between normal and case groups. The Wilcoxon-Mann-Whitney test and chi-squared test were used to compare the statistical differences between groups for continuous and categorical variables, respectively. As the continuous sex hormone values were right-skewed, these values were logarithmically transformed to follow a normal distribution. In addition, sex steroid hormone metabolites were categorized into quartiles, based on the distributions among the normal participants at baseline. Tests for linear trend were performed on the basis of the quartiles.

In the cross-sectional analysis, conditional logistic regression, based on the age-matched IM-normal pairs, was used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for associations between sex steroid hormone metabolites and IM risk. Binary logistic regression was used to evaluate ORs and 95% CIs for associations between sex steroid hormone metabolites and high-grade lesions or GC risk. In the prospective analysis, Cox proportional hazards regression was used to evaluate the hazard ratios (HRs) and 95% CIs for associations between sex steroid hormone metabolites and GC risk. The proportional hazards assumption was assessed using Schoenfeld’ residuals and Kolmogorov-type supremum analysis, with evidence of equiproportionality being detected. Follow-up time in years was used as the underlying time metric and was calculated from baseline to the date of diagnosis of GC, the date of death, or the end of the follow-up period (December 31, 2021), whichever occurred first. Multivariable models were adjusted for age, body mass index (BMI), smoking status, alcohol consumption, history of upper gastrointestinal disease, family history of cancer, and H. pylori CagA infection. Additionally, participants were stratified by the H. pylori CagA infection status for the sensitivity analysis to test the effects of H. pylori infection on associations between sex steroid hormone metabolites and GC and precancerous lesions.

In addition to individual sex hormones, we also calculated the following parameters using the aforementioned analytical approaches: parent estrogens (the sum of estrone and estradiol), testosterone: parent estrogens ratio, testosterone: estradiol ratio, and androstenedione: estrone ratio. Statistical significance was defined as P < 0.05, and all P values were two-tailed. Analyses were performed with SAS, version 9.4 (SAS Institute, Inc., Cary, NC) and R 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria).

Baseline characteristics and levels of serum sex hormone metabolites

Demographic characteristics, distributions of risk factors, and levels of serum sex hormone metabolites for participants with different baseline pathological diagnoses are summarized in Table 1. A total of 470 participants were enrolled in the cross-sectional analysis. Compared to normal participants, those with IM were more likely to be infected with H. pylori (65.1% vs. 52.1%, P = 0.01). High-grade lesions or GC patients were older than normal participants (73.5 years vs. 68.0 years, P < 0.01), and generally exhibited higher median concentrations of sex hormones compared to both IM patients and normal participants.

Cross-sectional analysis results

Figure 2 presents the associations between serum sex steroid hormone metabolites and the risk of IM and high-grade lesions or GC in the cross-sectional analysis. We found that higher concentrations of androstenedione were associated with an increased risk of high-grade lesions or GC (OR_continuous= 2.45, 95% CI: 1.01–5.95), and a similar association was observed between androstenedione and IM (OR_Q2 vs. _Q1 = 2.07, 95% CI: 1.12–3.82). Meanwhile, higher concentrations of 17α-hydroxypregnenolone (OR_continuous= 2.35, 95% CI: 1.13–4.88) and estrone (OR_continuous = 5.36, 95% CI: 1.28–22.52) were associated with an increased risk of high-grade lesions or GC, with consistent results when analyzing quartiles of 17α-hydroxypregnenolone (OR_Q4 vs. _Q1 = 3.49, 95% CI: 1.24–9.82, P_trend = 0.01) and estrone (OR_Q4 vs. _Q1 = 4.22, 95% CI: 1.27–13.98, P _trend = 0.01). Additionally, higher concentrations of progesterone (OR_continuous = 2.68, 95% CI: 1.09–6.61) were also associated with an increased risk of high-grade lesions or GC in participants.

Prospective analysis results

In the prospective cohort of 438 participants diagnosed with either normal or IM at baseline, we found positive associations between several serum sex steroid hormone metabolites and GC risk during a median follow-up of 11.8 years (IQR: 9.7–13.7 years). Specifically, statistically significant associations were observed between higher levels of SHBG (HR_continuous = 2.57, 95% CI: 1.04–6.36), epitestosterone (HR_continuous = 2.10, 95% CI: 1.07–4.15), and pregnenolone (HR_continuous= 1.30, 95% CI: 1.03–1.63) and incident GC risk (Figure 3). The results remained consistent when analyzing the quartiles of SHBG (HR_Q4 vs. _Q1 = 5.04, 95% CI: 1.23–20.63, P_trend = 0.01), epitestosterone (HR_Q4 vs. _Q1 = 3.82, 95% CI: 1.02–14.30, P_trend = 0.04), and pregnenolone (HR_Q4 vs. _Q1 = 4.70, 95% CI: 1.47–15.02, P_trend < 0.01).

Sensitivity analysis

To further explore the potential modifying effect of H. pylori infection, we stratified the analysis by H. pylori CagA serostatus. Consistent associations between sex steroid hormone metabolites and the risk of GC or its precursors were observed in both the overall dataset and the stratified subgroups. Among H. pylori CagA-positive participants, higher levels of estrone were significantly associated with high-grade lesions or GC risk (Figure 4–5). Meanwhile, in H. pylori CagA-negative participants, increased levels of androstenedione, 17α-hydroxypregnenolone, and progesterone were associated with the risk of high-grade lesions or GC, and both SHBG and pregnenolone exhibited positive associations with GC risk (Supplementary Figure S1–2). However, no statistically significant associations were observed between epitestosterone levels and the risk of GC and precancerous lesions in subgroup analyses, possibly reflecting limited statistical power due to small sample sizes within each stratum.