Association between Helicobacter pylori infection and mortality risk in prostate cancer patients receiving androgen deprivation therapy: A real‐world evidence study

Abstract Purpose Helicobacter pylori (H. pylori) is a major risk factor for gastric cancer and may affect androgen activity in men. The association between H. pylori and androgen deprivation therapy (ADT) in patients with prostate cancer (PCa) remains unclear. Methods This retrospective cohort study linked National Health Insurance (NHI) data to Taiwan Cancer Registry (TCR) and Taiwan Death Registry (TDR) between 1995 and 2016. PCa patients who received ADT were classified into H. pylori infection and non‐H. pylori infection groups. The outcomes were overall mortality, prostate cancer‐specific mortality, and castration‐resistant prostate cancer (CRPC). Propensity score matching was adopted for the primary analysis and inverse probability of treatment weighting (IPTW) was used for the sensitivity analysis. Results Of the 62,014 selected PCa patients, 23,701 received ADT, of whom 3516 had H. pylori infections and 20,185 did not. After matching, there were 3022 patients in the H. pylori infection group and 6044 patients in the non‐H. pylori infection group. The mean follow‐up period for the matched cohort was 4.8 years. Compared to the non‐H. pylori group, the H. pylori group was significantly associated with decreased risks of all‐cause mortality (hazard ratio [HR] 0.90; 95% confidence interval [CI] 0.84–0.96) and prostate cancer‐specific mortality (HR 0.88; 95% CI 0.81–0.95) in the matched analysis. Conclusions H. pylori infection was associated with a reduced risk of mortality in PCa patients receiving ADT.


| INTRODUCTION
Helicobacter pylori (H. pylori) is a gram-negative curved bacillus that colonizes the stomach and that was first described in the 1980s. 1 The gastric colonization of H. pylori is associated with gastric cancer. 2 H. pylori is also associated with extra-gastrointestinal diseases including cardiovascular disease, autoimmune disorders, skin disorders, and platelet disorders. 3,4 These extra-gastrointestinal phenomena of H. pylori may occur due to the direct effects of the bacterium, the activation of inflammatory processes resulting from cytokine release, or the mimicry between bacterial and host antigens. 5,6 The prostate may also be affected by H. pylori as extra-gastrointestinal infection. 7 Eastern Asian populations have higher incidences of gastric cancer and lower incidences of prostate cancer (PCa) than Western countries. 8,9,10 In addition, Eastern Asian populations have lower androgen-related parameters, such as testicular weight, than Caucasians. The effects of H. pylori may be associated with the above observations. In the elderly, H. pylori could modulate androgen production. 11 It has further been reported that H. pylori is associated with lower androgen activity in men. 12 Lower androgen activity may, in turn, have some beneficial effects on PCa and androgen deprivation therapy (ADT).
Prior to the present study, however, a comprehensive analysis of the relationship between H. pylori infection and ADT for PCa had not been undertaken. We therefore conducted the first large-scale real-world evidence study utilizing data from the National Health Insurance program of Taiwan in order to determine whether H. pylori is associated with the outcomes of patients receiving ADT for PCa.

| Data source
This retrospective cohort study linked National Health Insurance (NHI) data to Taiwan Cancer Registry (TCR) and Taiwan Death Registry (TDR) data using unique and de-identified civil identification numbers. The NHI data were taken from a national health and welfare administrative database for Taiwan's NHI program since 1995. The NHI data were collected by the National Health Informatics Project (NHIP) and managed by the Health and Welfare Data Science Center (HWDC). We analyzed the NHI, TCR, and TDR data from 1995 to 2016. Data for gender, birth date, medications, and diagnostic codes (the International Classification of Diseases, Ninth or Tenth Revision, Clinical Modification; ICD-9-CM or ICD-10-CM) were retrieved for the analyses performed in this study.

| Study population
The patients with PCa in the TCR between March 1995 and December 2016 were identified using ICD-9-CM, ICD-10-CM, or ICD-O-3 (International Classification of Diseases for Oncology, Third Edition) diagnostic codes (Table S1). The patients with PCa who received subsequent ADT therapy were then identified. The ADT categories are detailed in the Supplementary Appendix. The use of ADT therapy was detected using the Taiwan NHI reimbursement codes, which can be referred to using the anatomic therapeutic chemical (ATC) classification system 13 (Table S2). The date of receiving ADT therapy for PCa was defined as the index date. Patients who were less than 40 years old; had a cancer other than PCa; or received a bilateral orchiectomy, chemotherapy, abiraterone acetate, or enzalutamide were excluded. Finally, patients who had a history of peptic ulcer disease without subsequent eradication were excluded because the status of H. pylori infection for these patients was unknown (Figure 1).

| Definition of exposure
The exposure of interest in this study was H. pylori infection. The H. pylori infection was defined as having a diagnosis of peptic ulcer and receiving subsequent eradication therapy before the index date. In Taiwan, H. pylori eradication consisting of treatment with a proton pump inhibitor (PPI) or histamine receptor-2 blocker (H2 receptor blocker), clarithromycin or metronidazole, and amoxicillin or tetracycline with or without bismuth for a duration of 7-14 days is the standard treatment for H. pylori Conclusions: H. pylori infection was associated with a reduced risk of mortality in PCa patients receiving ADT.

K E Y W O R D S
androgen deprivation therapy, Helicobacter pylori, mortality, prostate cancer (Table S2). Patients were divided into two groups according to the status of H. pylori infection. Patients with H. pylori infection were further divided into two subgroups according to the timing between the initial diagnosis of peptic ulcer and the eradication therapy. Patients who received H. pylori eradication therapy within 1 year of the diagnosis date of peptic ulcer were classified as the early eradication subgroup, while the other patients were classified as the late eradication subgroup ( Figure 1). 14

| Study outcomes
This study included three outcomes: all-cause mortality, PCa-specific mortality, and castration-resistant prostate cancer (CRPC). The survival status, date of death, and cause of death of patients were verified in the TDR. PCa patients diagnosed with CRPC were defined as PCa patients who received chemotherapy, abiraterone acetate, or enzalutamide. The drugs used to treat patients were also identified using the Taiwan NHI reimbursement codes (Table S3). The end dates for follow-up were extracted from the survival status in the TDR. Patients were followed until the date of death, event occurrence, or 31 December 2016, whichever came first.

| Covariates
The covariates in this study consisted of potential confounders including age, urbanization level, monthly income, comorbidities, history of events, Charlson Comorbidity Index (CCI) score, PCa stage at diagnosis, disease extent, and medications. The demographics and socioeconomic information were included in the Registry for Beneficiaries, which is a sub-database of the NHI database. Comorbidities at the index date were identified based on having two outpatient diagnoses or a single inpatient diagnosis in the previous year. The history of a given event (e.g., cerebrovascular disease) was identified based on a single inpatient diagnosis before the index date. The ICD-9-CM and ICD-10-CM diagnostic codes for the detection of diseases are listed in Table S1. The PCa stage at diagnosis was extracted from the TCR, while medication use in the year preceding the index date was also extracted.

| Statistical analysis
To reduce confounding due to potential selection bias, a propensity score matching (PSM) method was employed as the primary analysis in this study. The propensity score was the predicted probability of being in the H. pylori group given the values of covariates using logistic regression. Each patient in the H. pylori group was matched with two counterparts in the non-H. pylori group. The quality of matching was checked using the absolute value of the standardized difference (STD). As to the fatal time to event outcomes, the risks between the two groups were compared using the Cox proportional hazard model with hazard ratio (HR). The incidences of CRPC for the two groups were compared using the Fine and Gray subdistribution F I G U R E 1 Patient selection hazard model 15 , which considered all-cause mortality as a competing risk. PSM and the same survival analyses were also performed to compare the early eradication subgroup with the late eradication subgroup. Subgroup analysis was performed to determine whether the effects of H. pylori infection on all-cause death and PCa-specific death were consistent among the different levels of the nine prespecified subgroup variables listed in the Supplementary Appendix.
Sensitivity analysis was performed to assess the robustness of the primary analysis. Propensity score matching causes some amount of data loss that would reduce the statistical power. Therefore, we used inverse probability of treatment weighting (IPTW) based on the propensity scores to estimate the average treatment effect of H. pylori infection without loss of patients in a simulated population. A two-sided p value < 0.05 was considered to be statistically significant and no adjustment of multiple testing (multiplicity) was made in this study. All statistical analyses were performed using SAS version 9.4 (SAS Institute). The demographic characteristics of the study subjects are listed in Table 1. Before matching, the patients in the H. pylori group had higher prevalence of hypertension, diabetes, coronary heart disease, hyperlipidemia, chronic obstructive pulmonary disease, chronic kidney disease, chronic liver disease, cerebrovascular disease, and higher CCI scores. The H. pylori group also had more distant metastases. In addition, more patients in the H. pylori group were prescribed non-steroidal anti-inflammatory drugs, clopidogrel, Cox-2 inhibitors, oral hypoglycemic agents, and angiotensin-converting enzyme inhibitor/angiotensin II receptor blockers. After matching, however, there were no differences between the two groups.

| All-cause mortality and prostate cancer-specific mortality
The clinical outcomes of the PCa patients who received ADT in the H. pylori infection group and the non-H. pylori group are shown in Table 2. Compared to the non-H. pylori group, the H. pylori group was significantly associated with decreased risks of all-cause mortality (HR 0.90; 95% CI 0.84-0.96) and PCa-specific mortality (HR 0.88; 95% CI 0.81-0.95). The decreased risks of all-cause mortality and PCa-specific mortality in the H. pylori infection group were still observed in the sensitivity analysis by IPTW (Table S4). However, the risk of CRPC was not significantly different between the two groups ( Table 2). The unadjusted cumulative event rates of all-cause mortality and PCa-specific mortality and the cumulative incidence rates of CRPC are depicted in Figure 2.

| Subgroup analysis according to H. pylori eradication status
The pre-specified subgroup analysis results of all-cause mortality and PCa-specific mortality are shown in Figure 3. The observed decreased mortality risk from H. pylori infection was more apparent in patients who were younger than 75 years old, without hypertension, without diabetes, and who had locoregional PCa (p < 0.05) ( Figure 3A). The observed decreased PCa-specific mortality risk was more obvious in patients who did not have diabetes and locoregional PCa (p < 0.05) ( Figure 3B).
We further compared outcomes in patients with H. pylori infection between the early eradication and late eradication groups. The characteristics of the patients with early eradication or late eradication of H. pylori are shown in Table S5 (PSM) and Table S6 (IPTW) in the Supplement. The results showed that there were no significant differences in all-cause mortality, PCa-specific mortality, and CRPC between the early eradication and late eradication subgroups (Table S7 and Figure S1).

| DISCUSSION
To the best of our knowledge, this is the first real-world evidence cohort study to investigate the association between H. pylori infection and mortality in PCa patients receiving ADT. Using PSM analysis on 62,014 PCa patients, we demonstrated that H. pylori infection was associated with lower risks of all-cause mortality and PCa-specific mortality in patients receiving ADT after adjustments for age, disease extent of PCa, comorbidities, and related medications. In addition, patients with early H. pylori T A B L E 1 Characteristics of the patients with and without Helicobacter pylori infection before and after propensity score matching There are several possible mechanisms that might explain why H. pylori infection reduces mortality in PCa patients receiving ADT. H. pylori infection may have result in a changed human microbiome environment. Gut microbiota may be able to regulate sex hormones. 11 It has been reported, for example, that changing the gut microbial environment in mice changes mice testosterone levels. 11  et al. reported that H. pylori absorbs and glucosylates androgens as membrane lipid components. 16 In addition, H. pylori was found to be associated with lower androgen activity levels and lower levels of androstanediol glucuronide (3-alphadiol-G) (AAG), a testosterone biomarker, in older men in the NHANES III. 11 H. pylori changes the gut microbiome, which may lead it to metabolize sex steroid hormones and influence their activity. 17 H. pylori also reduces androgen levels, which may have a potential benefit to ADT for PCa. Chinese populations have high incidences of gastric cancer and lower incidences of PCa than Caucasians. 18 Meanwhile, Chinese populations also have lower androgen-related parameters, such as testicular weight, than Caucasians. 19 The association of H. pylori with lower androgen levels may explain the above observations. Further studies are warranted, however, to achieve a better understanding of the effects of H. pylori infection.
Regimens for the eradication of H. pylori may affect androgen levels. PPIs, such as omeprazole, esomeprazole, or lansoprazole, may induce hepatic cytochrome P450 (CYP450), which increases testosterone metabolism and results, in turn, in reduced circulating testosterone levels. 20,21 H2 blockers, such as cimetidine, also have effects on CYP450, including further antiandrogen effects. 22 Metronidazole has been reported to reduce androgen and cause gynecomastia. 23 The antiandrogen effects of these regimens may have some benefit in ADT for patients with PCa. In a recent study, PPI inhibited apoptosis and further increased survival in hormone-sensitive LNCaP cells via the ErbB2, MAPK-ERK1/2, PI3K/Akt, and GSK-3β pathways. 24 Further studies are warranted, however, to elucidate the relationships between these regimens and ADT for PCa.
Meanwhile, H. pylori has extra-gastrointestinal phenomena, with the prostate being a possible extragastrointestinal colonization site. 7 Immune responses evoked by H. pylori infection may be associated with chronic prostatitis. 7,25 Proliferative inflammatory atrophy (PIA) could be induced by prostatitis, and PIA is a suspected risk factor lesion for PCa development. 26,27 H. pylori may stimulate chronic systemic inflammatory responses through the production of various proinflammatory cytokines, such as tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), interleukin-1β (IL-1β), interleukin-6 (IL-6), and interleukin-8 (IL-8). 28,29 These proinflammatory cytokines may in turn be associated with aggressive PCa types. [30][31][32] On the other hand, higher regulatory T cell (Tregs) expression levels in patents with H. pylori colonization have been reported. 33 Tregs may downregulate immune responses via TGF-β or IL-10 to reduce inflammatory responses. 34,35 Further studies are warranted, however, to elucidate the relationships between the cytokine effects of PCa and H. pylori.
The strength of this study is a large real-world evidence study that enrolled 62,014 PCa patients with a long follow-up period. We used the PSM analysis and IPTW analysis to ensure the accuracy of the study outcomes. The study included 3022 study subjects with H. pylori infection, with the risks of all-cause mortality and PCa-specific mortality being compared among those subjects and a matched control group. Subgroup analysis with age, hypertension, diabetes, hyperlipidemia, coronary heart disease, cerebrovascular disease, chronic kidney disease, CCI score, and disease extent of PCa was also conducted to investigate the effects of H. pylori infection on the risk of PCa mortality in order to provide substantial information for physicians.
Our study has several limitations. First, the PSA levels and Gleason scores were not available in the investigated databases. It was thus difficult to further evaluate the PCa of the included subjects. In addition, several potential confounding factors, including body mass index, educational level, intestinal microbiota, and inflammation parameters were not available in the investigated databases. Second, all the diagnoses were made using the ICD-9, ICD-10, or ICD-O-3 coding systems, such that diagnostic pitfalls may have occurred. Third, the NHI database is an administrative database, such that patients' socioeconomic status, family history, personal health behaviors, laboratory data, and image findings were not available. We did not have data to confirm the exact duration of H. pylori infection in each study subject, and the severity of H. pylori was also not available in this study. Fourth, there were no data in the investigated databases to confirm whether early or late H. pylori eradication was successful. Finally, this populationbased study was a retrospective study, the association between H. pylori, intestinal microbiota, and prostate cancer may be lack of solid collateral evidence. To achieve a better understanding of the effects of H. pylori infection, further prospective studies are warranted to investigate the relationship between PCa mortality and H. pylori in patients with ADT.

| CONCLUSIONS
This is the first real-world evidence study to demonstrate that H. pylori infection is associated with a reduced risk of mortality in PCa patients receiving ADT. This novel finding provides clinical evidence of a potential role for H. pylori infection in ADT for PCa.

AUTHORSHIP CONTRIBUTION
Liu JM and Hsu RJ thought of the concept and designed the study. Wu CT provided study materials. All authors collected and assembled data, conducted data analysis, wrote and made final approval of the manuscript.

ETHICS STATEMENT
This study was approved by the Institutional Review Board of the Tri-Service General Hospital (approval number: TSGHIRB No B-107-04). As this was a retrospective study and all data were anonymous, the Institutional Review Board department agreed with the authors that it was waived the need for informed consent.