2022 Volume 45 Issue 5 Pages 590-595
The gut microbiome influences tumor response to immune checkpoint inhibitors (ICIs). The proton pump inhibitors (PPI) significantly impair diversity of the gut microbiota and can affect the efficacy of ICIs. Therefore, the present study aimed to evaluate the influence of PPI on survival in patients with metastatic or unresectable urothelial carcinoma receiving pembrolizumab. We conducted a retrospective cohort study of patients with metastatic or unresectable urothelial carcinoma receiving pembrolizumab. The use of PPI was defined as any administration for ≥30 d within 60 d prior and/or 30 d after treatment initiation. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan–Meier method, and Cox proportional hazards regression analysis was performed to investigate prognostic factors based on patient characteristics. Seventy-nine patients were included in the analysis, and 34 patients (43.0%) received PPI. There were no significant differences in OS and PFS between PPI users and nonusers (median OS: 8.2 months vs. 11.2 months, hazard ratio (HR): 1.36, 95% confidence interval (CI): 0.75–2.42, p = 0.296; median PFS: 3.5 months vs. 5.1 months, HR: 1.63, 95% CI: 0.95–2.80, p = 0.069). In the multivariable analysis, PPI use was not associated with OS (HR 0.80, 95% CI 0.40–1.56, p = 0.526) or PFS (HR 1.44, 95% CI 0.79–2.60, p = 0.233). In conclusion, the estimated effect size of PPI use on survival in Japanese patients with metastatic or unresectable urothelial carcinoma treated with pembrolizumab was not reproducible.
Pembrolizumab is an immune checkpoint inhibitor (ICI) administered to patients with metastatic or unresectable urothelial carcinoma. A randomized controlled study1) showed that the objective response rate of pembrolizumab was 21.1% in patients with advanced urothelial carcinoma. Pembrolizumab was significantly associated with prolonged overall survival (OS) in patients with no liver metastases and those with tumor programmed death ligand-1 (PD-L1) positive score >1%.1)
Several recent studies have demonstrated that the gut microbiome modulates antitumor immune response.2–4) In case of melanoma, the responders treated with anti-programmed cell death 1 (PD-1) therapy showed significantly higher diversity of the gut microbiome than non-responders.2,3) A preclinical study indicated that administration of Bifidobacterium to mice inoculated with bladder cancer cells delayed tumor growth.5) Moreover, a combined treatment of Bifidobacterium and anti-PD-1 monoclonal antibody increased tumor-specific T cell response. Therefore, the gut microbiome plays an important role in the efficacy of ICI therapy.
Proton pump inhibitors (PPIs) change the composition of the gut microbiota.6–8) Additionally, the diversity of the gut microbiome in PPI users is significantly lower than that in PPI nonusers. A pooled analysis reported that PPI use was associated with shorter OS and progression-free survival (PFS) than PPI nonuse in patients with non-small cell lung cancer treated with atezolizumab.9) The analyses of individual patient data showed that PPI use was associated with shorter OS and PFS in patients with advanced urothelial carcinoma treated with atezolizumab.10) However, effects of PPI on the therapeutic efficacy of pembrolizumab remains unknown in patients with metastatic or unresectable urothelial carcinoma. In this study, we evaluated the possibility that PPI affects survival outcomes in patients with metastatic or unresectable urothelial carcinoma who received pembrolizumab therapy.
We conducted a retrospective cohort study in patients with metastatic or unresectable urothelial carcinoma who received pembrolizumab at the Kobe University Hospital between May 2017 and December 2020. Eligible patients were aged ≥20 years and received at least one infusion of pembrolizumab with a standard dose of 200 mg/body every three weeks or 400 mg/body every six weeks. The exclusion criteria were: concomitant autoimmune diseases, PPI use as needed, and inability to track PPI prescription history. This study was conducted according to the principles of the Declaration of Helsinki and approved by the Ethics Committee of the Kobe University Hospital (B210135). We obtained following information from the electronic medical charts of each patient: age, sex, performance status (PS), tumor histology, PPI type, smoking status, primary urothelial cancer site, adverse drug reactions (ADRs) with pembrolizumab, presence of liver metastases, history of operation, number of prior treatment regimens, clinical laboratory test values (hemoglobin (Hb), albumin (Alb), and neutrophil-to-lymphocyte ratio (NLR)), and concomitant use of steroids. PPI use was defined as any administration for ≥30 d within 60 d prior and/or 30 d after treatment initiation.4,7) PS was estimated according to the Eastern Cooperative Oncology Group system.
The primary endpoint was OS, which was defined as the time from the initiation of treatment to death from any cause. The secondary endpoint was PFS. PFS was defined as the time from the initiation of treatment to disease progression or death from any cause. The follow-up period was 36 months for both OS and PFS. Response to treatment was assessed as per the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines (version 1.1). ADRs were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.
Sample Size EstimationWe performed log-rank test to select a sample size to determine the difference in OS according to PPI use. Based on the results from a previous study by Hopkins et al.10) we set the following parameters: a study accrual period of 3 years, event rates of 0.50 in the placebo group during the 1-year follow-up period, minimum 0.4 years of follow-up, and pooled HR of 2.10. When the ratio of PPI users and nonusers was 1 : 1, 80 patients were required to detect a significant difference assuming an α of 0.05 and power of 0.8. Thus, considering that 5% of the patients were excluded, 84 patients were included in the present study.
Statistical AnalysisContinuous variables are presented as medians (range), and categorical variables as frequencies and proportions. The Kaplan–Meier method was used to calculate OS and PFS, and differences in survival were evaluated using log-rank test. Cox proportional hazard regression was performed to examine the prognostic factors associated with OS or PFS, and estimating the hazard ratio (HR) and 95% confidence interval (CI). The explanatory variables examined were selected referring to predictive factors and prognostic factors associated with survival outcomes in previous studies: age, sex, PPI use, PS, smoking status, history of operation, Hb, Alb, NLR, and liver metastasis. Explanatory variables with an effect p-value <0.10 in the univariate analysis were included in the multivariate analysis. PPI use was included in the multivariate analysis regardless of the p-value in the univariate analysis. p-Value <0.05 (two-sided) was considered to be statistically significant. All statistical analyses were performed using R 4.0.3 software and packages “Hmisc” (version 4.5-0) for sample size calculation.
Between May 2017 and December 2020, 83 patients were included in the study. Of these, four patients with autoimmune diseases were excluded. There were no patients who received PPI as needed or were unable to track PPI prescription history. Therefore, 79 patients were included in the analysis. The median duration of follow-up was 7.2 months. Thirty-four patients (43.0%) received PPIs. PPIs included lansoprazole, esomeprazole, vonoprazan, and rabeprazole (Table 1).
| Patient characteristics | PPI (−) (n = 45) | PPI (+) (n = 34) |
|---|---|---|
| Male, n (%) | 35 (77.8) | 24 (70.6) |
| Age (years), median (range) | 71 (57–86) | 72 (56–87) |
| Primary urothelial cancer site, n (%) | ||
| Upper urinary tract | 27 (60.0) | 19 (55.9) |
| Lower urinary tract | 17 (37.8) | 15 (44.1) |
| Upper and lower urinary tract | 1 (2.2) | 0 (0) |
| Histology, n (%) | ||
| Pure cell | 40 (88.9) | 28 (82.4) |
| Others | 5 (11.1) | 6 (17.6) |
| Prior lines of therapy | ||
| 1 | 29 (64.5) | 27 (79.4) |
| 2 | 15 (33.3) | 7 (20.6) |
| 3 | 1 (2.2) | 0 (0) |
| PS, n (%) | ||
| 0–1 | 40 (88.9) | 25 (73.5) |
| 2–4 | 5 (11.1) | 9 (26.5) |
| History of operation, n (%) | 25 (55.6) | 14 (41.2) |
| Smoking status, n (%) | 26 (62.2) | 22 (64.7) |
| Concomitant use of steroids, n (%) | 1 (2.2) | 4 (11.8) |
| Liver metastases, n (%) | 8 (17.8) | 7 (20.6) |
| Laboratory test values | ||
| NLR (%), median (range) | 2.7 (0.6–21.7) | 4.2 (1.2–46.3) |
| Alb (g/dL), median (range) | 3.9 (1.2–4.8) | 3.6 (2.0–4.3) |
| Hb (g/dL), median (range) | 10.7 (7.8–15.3) | 10.6 (6.1–15.0) |
Abbreviations: Alb, albumin; Hb, hemoglobin; NLR, neutrophil-to-lymphocyte ratio; PPI, proton pump inhibitor; PS, performance status.
There were no significant differences in OS and PFS between PPI users and nonusers (median OS: 8.2 months vs. 11.2 months, HR: 1.36, 95% CI: 0.75–2.42, p = 0.296; median PFS: 3.5 months vs. 5.1 months, HR: 1.63, 95% CI: 0.95–2.80, p = 0.069) (Figs. 1, 2). Univariate Cox proportional hazards regression analyses for OS and PFS were performed (Tables 2, 3). Poorer PS, lower Hb, lower Alb, higher NLR, and presence of liver metastases were associated with shorter OS with an effect p-value <0.10. Poorer PS, lower Hb, lower Alb, higher NLR, and presence of liver metastases were associated with poorer PFS, with an effect p-value <0.10. These factors and PPI use were included in the multivariate analysis as covariates. In the multivariable analysis, poor PS (HR 4.83, 95% CI 1.97–11.3, p < 0.001), lower Alb (HR 0.49, 95% CI 0.29–0.86, p = 0.010), higher NLR (HR 1.10, 95% CI 1.03–1.17, p = 0.022), and presence of liver metastases (HR 6.08, 95% CI 2.57–13.7, p < 0.001) were associated with shorter OS. Poorer PS (HR 3.16, 95% CI 1.42–6.59, p = 0.006) and the presence of liver metastases (HR 2.66, 95% CI 1.18–5.44, p = 0.010) were associated with worse PFS. PPI use failed to have an effect on OS (HR 0.80, 95% CI 0.40–1.56, p = 0.526) and PFS (HR 1.44, 95% CI 0.79–2.60, p = 0.233).
Abbreviations: CI, confidence interval; HR, hazard ratio; OS, overall survival; PPI, proton pump inhibitor. p-Value was evaluated using log-rank test.
Abbreviations: CI, confidence interval; HR, hazard risk; PFS, progression-free survival; PPI, proton pump inhibitor. p-Value was evaluated using log-rank test.
| Characteristics | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | p-Value | HR | 95% CI | p-Value | |
| Sex (female) | 1.57 | 0.79–2.93 | 0.185 | — | — | — |
| Age (years) | 1.01 | 0.97–1.05 | 0.503 | — | — | — |
| PPI use | 1.36 | 0.75–2.42 | 0.303 | 0.80 | 0.40–1.56 | 0.526 |
| PS (2–3) | 3.72 | 1.83–7.07 | <0.001 | 4.83 | 1.97–11.3 | <0.001 |
| Smoking status (current smoker) | 0.70 | 0.39–1.31 | 0.259 | — | — | — |
| History of operation (yes) | 0.97 | 0.54–1.74 | 0.919 | — | — | — |
| Hb (g/dL) | 0.80 | 0.67–0.95 | 0.013 | 1.09 | 0.85–1.40 | 0.501 |
| Alb (g/dL) | 0.53 | 0.38–0.76 | 0.001 | 0.49 | 0.29–0.86 | 0.010 |
| NLR | 1.12 | 1.07–1.19 | <0.001 | 1.10 | 1.03–1.17 | 0.022 |
| Liver metastases (yes) | 5.86 | 2.83–11.6 | <0.001 | 6.08 | 2.57–13.7 | <0.001 |
Abbreviations: Alb, albumin; CI, confidence interval; Hb, hemoglobin; HR, hazard ratio; NLR, neutrophil-to-lymphocyte ratio; PPI, proton pump inhibitors; PS, performance status.
| Characteristics | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | p-Value | HR | 95% CI | p-Value | |
| Sex (female) | 1.56 | 0.84–2.75 | 0.154 | — | — | — |
| Age (years) | 1.01 | 0.97–1.05 | 0.707 | — | — | — |
| PPI use | 1.63 | 0.95–2.79 | 0.076 | 1.44 | 0.79–2.60 | 0.233 |
| PS (2–3) | 3.67 | 1.95–6.80 | <0.001 | 3.16 | 1.42–6.59 | 0.006 |
| Smoking status (current smoker) | 0.82 | 0.47–1.46 | 0.492 | — | — | — |
| History of operation (yes) | 1.24 | 0.72–2.15 | 0.429 | — | — | — |
| Hb (g/dL) | 0.84 | 0.71–0.99 | 0.035 | 0.98 | 0.78–1.22 | 0.855 |
| Alb (g/dL) | 0.63 | 0.46–0.91 | 0.015 | 0.70 | 0.44–1.19 | 0.181 |
| NLR | 1.08 | 1.03–1.12 | 0.002 | 1.03 | 0.98–1.08 | 0.179 |
| Liver metastases (yes) | 3.16 | 1.56–5.94 | 0.002 | 2.66 | 1.18–5.44 | 0.010 |
Abbreviations: Alb, albumin; CI, confidence interval; Hb, hemoglobin; HR, hazard ratio; NLR, neutrophil-to-lymphocyte ratio; PPI, proton pump inhibitors; PS, performance status.
Treatment-related ADRs of any grade in patients with PPI use and nonuse are shown in Table 4. Grade 3 events occurred in four patients (11.8%) in the PPI use group and three patients (6.7%) in PPI nonuse group. None of the patients experienced grade 4 events. Six (17.6%) and four (8.9%) patients in the PPI users and nonuser groups, respectively, showed treatment-related discontinuation of pembrolizumab, including ADRs and progression disease. The most common ADRs of any grade between PPI users and nonusers were interstitial pneumonia (5.9 and 13.3%) and rash (5.9 and 13.3%), respectively.
| Adverse drug reactions | PPI (+) n = 34 | PPI (−) n = 45 | ||||||
|---|---|---|---|---|---|---|---|---|
| Any grade | Grade1 | Grade 2 | Grade 3 | Any grade | Grade 1 | Grade 2 | Grade 3 | |
| Interstitial pneumonitis | 2 (5.9) | 1 (2.9) | 1 (2.9) | 0 (0) | 6 (13.3) | 3 (6.7) | 2 (4.4) | 1 (2.2) |
| Rash | 2 (5.9) | 0 (0) | 2 (5.9) | 0 (0) | 6 (13.3) | 1 (2.2) | 4 (8.9) | 1 (2.2) |
| Liver failure | 2 (5.9) | 0 (0) | 0 (0) | 2 (5.9) | 1 (2.2) | 0 (0) | 1 (2.2) | 0 (0) |
| Colitis | 2 (5.9) | 0 (0) | 1 (2.9) | 1 (2.9) | 1 (2.2) | 0 (0) | 0 (0) | 1 (2.2) |
| Hyperthyroidism | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 4 (8.9) | 2 (4.4) | 2 (4.4) | 0 (0) |
| Pancreatitis | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 1 (2.2) | 1 (2.2) | 0 (0) | 0 (0) |
| Hypothyroidism | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 1 (2.2) | 0 (0) | 1 (2.2) | 0 (0) |
| Adrenal insufficiency | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 1 (2.2) | 0 (0) | 1 (2.2) | 0 (0) |
| Hypopituitarism | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 1 (2.2) | 0 (0) | 1 (2.2) | 0 (0) |
| Myasthenia gravis | 1 (2.9) | 0 (0) | 0 (0) | 1 (2.9) | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
All data indicated as n (%).
The present study showed that PPI use had no significant influence on survival outcomes in patients with metastatic or unresectable urothelial carcinoma receiving pembrolizumab. Moreover, PPI use was not a prognostic factor in Cox proportional hazards regression analysis. Several studies9–15) have reported the impact of PPI use on the clinical outcomes of ICI therapy. A previous retrospective study15) showed PPI use to be associated with worse OS and PFS in patients with locally advanced or metastatic urothelial carcinoma treated with ICIs. However, 70% of the patients in the previous study were received anti-PD-L1 atezolizumab and durvalumab. Therefore, it might be inappropriate to compare the results of the present study with those of a previous study.11) A pooled analysis10) reported that PPI use was associated with significantly worse OS and PFS in patients with advanced urothelial carcinoma treated with atezolizumab; whereas pembrolizumab was used in the present study. Furthermore, the influence of the gut microbiota on treatment efficacy differs between anti-PD-1 and anti-PD-L1 monotherapy.16) A preclinical study2) demonstrated that responders to anti-PD-L1 therapy had an abundance of Faecalibacterium spp. and Ruminococcaceae family than non-responders. In contrast, an association between responders to anti-PD-1 therapy and the abundance of gut microbiota, such as Akkermansia muciniphila, Alistipes indistinctus, and Enterococcus hirae was observed.4) Furthermore, PPI use was associated with an increase in abundance of Enterobacteriaceae, Enterococcaceae, and Lactobacillaceae, and a decrease in abundance of Bifidobacteriaceae and Ruminococcaceae, a bacterial group related to response to PD-L1 therapy.6,17) We hypothesized that the microbial composition related to response to pembrolizumab therapy remains unchanged by concomitant use of PPI. In addition, a meta-analysis showed that the survival outcome of concomitant PPI and ICI treatment differed by cancer types.18) The results of our study may support their perception. However, the mechanisms by which microbial diversity and composition are altered by adding PPI to anti-PD-1 or anti-PD-L1 therapy is unclear. A preclinical study showed that the composition of the gut microbiome differed between Japanese and individuals of other races, indicating that differences in the hydrogen metabolism pathway could affect the microbial composition.19) In the present study, no significant difference was observed in survival outcomes possibly because the composition of the gut microbiomes in the patients was peculiar to that of Japanese population. Additionally, we could not investigate the microbial composition of the patients; therefore, further studies are needed.
Univariate and multivariate Cox regression analyses suggested that PPI use was not a prognostic factor in patients with metastatic or unresectable urothelial carcinoma treated with pembrolizumab. Further, poor PS, lower Alb, higher NLR, and presence of liver metastases were independent factors influencing OS; also, poor PS and the presence of liver metastases were independent factors influencing PFS. Previous studies20–24) have demonstrated that poor PS, the presence of liver metastasis, lower Alb, and higher NLR were significant prognostic factors in patients with advanced urothelial carcinoma, consistent with results of the present study. These factors, rather than PPI use, may strongly contribute to survival.
A few grade 3 ADRs occurred in the PPI use and nonuse groups in patients receiving pembrolizumab, but none of the patients experienced grade 4 events. Thus, the addition of PPI to pembrolizumab is generally manageable.
In our study, we hypothesized a HR in sample size estimation based on the report which indicated the effect of PPI on survival in patients with urothelial cancer treated with atezolizumab.10) Hopkins et al.10) showed 1-year HR of 2.1 from the Kaplan–Meier curve for the participants whose PD-L1 expression was ≥5% of tumor-infiltrating immune cells despite the overall HR of 1.95. The effect size by PPI use was expected to be larger than the overall HR from the reported effect size, which evaluated the association between OS of Japanese patients treated with ICI and gut microbiota affected by PPI.6,25) Therefore, we estimated the HR of 2.1 according to PPI use in patients with urothelial cancer treated with pembrolizumab.
This study has some limitations. First, we could not examine PD-L1 expression due to the retrospective study design. A meta-analysis showed that higher expression of PD-L1 associates with objective response rate, while PD-L1 expression cannot predict OS in patients with urothelial carcinoma.26) The predictive value of PD-L1 expression remains unclear. Therefore, further studies should evaluate whether expression of PD-L1 influences the efficacy of concomitant use of pembrolizumab with PPI. Second, the concomitant medications that are well known to affect microbiota composition, such as antibiotics, steroids, and metformin may be associated with survival8); however, we could not perform statistical analysis because only a few patients received these drugs. Third, this was a small-scale, retrospective, observational study, and the number of patients recruited was insufficient to achieve the sample size estimated in the study protocol. However, we considered that this slight insufficiency of the sample size may not impact the primary endpoint.
In conclusion, our study showed that the PPI-caused effect size estimated from the relevant study on survival may not be reproduced for Japanese patients with metastatic or unresectable urothelial carcinoma received pembrolizumab. Furthermore, PPI use failed to qualify as a prognostic factor for OS or PFS, in addition to the previous known prognostic factors such as PS, Alb, NLR, and liver metastases. Thus, further research is needed to identify the composition of the gut microbiota related to clinical responses to concomitant administration of pembrolizumab and PPI. Additionally, it remains unclear whether concomitant medications that modulate the gut microbiome have an impact on the response to pembrolizumab. We believe that substantiating the influence of concomitant medications on clinical responses will help in optimizing the immune checkpoint therapy.
The authors declare no conflict of interest.
