Prognostic impact of proton pump inhibitors for immunotherapy in advanced urothelial carcinoma

Abstract Objective To evaluate the effects of the concomitant use of proton pump inhibitors (PPIs) and/or antibiotics (Abs) on oncological outcomes in patients with advanced urothelial carcinoma. Patients and methods We retrospectively evaluated 155 patients with advanced urothelial carcinoma who were treated with immune checkpoint inhibitors (ICIs) between August 2015 and April 2021. The concomitant use of PPI or Abs was defined as any PPI or Abs administered within 30 days before ICI initiation and during ICI therapy. The primary outcomes were the effect of PPI and/or Abs use on the objective response rate (ORR) and immune‐related adverse events (irAEs). The secondary outcomes were the effects of PPI and/or Abs use on progression‐free survival (PFS) and overall survival (OS) after ICI therapy analyzed using the inverse probability of treatment weighting‐adjusted Cox regression analysis. Results Of the 155 patients enrolled in the study, 99 (64%) were PPI users and 56 (36%) Abs users. PPI users were associated with a significantly poorer ORR than non‐PPI users (41% vs. 20%, respectively), whereas Abs use was not significantly associated with changes in ORR. The rate of irAEs was not significantly associated with the use of PPIs or Abs. Multivariate inverse probability of treatment weighting‐adjusted Cox regression analysis revealed significantly poorer PFS and OS in PPI users than in non‐PPI users, whereas Abs use was not associated with poorer outcomes. Conclusion The concomitant use of PPI may adversely affect oncological outcomes in patients with locally advanced or metastatic urothelial carcinoma treated with ICI therapy.


| INTRODUCTION
Locally advanced or metastatic urothelial carcinoma (UC) is a lifethreatening disease with a poor prognosis. [1][2][3] Although immune checkpoint inhibitors (ICIs) have significantly impacted treatment strategies in UC in both clinical trials and real-world practice, 4-8 the heterogeneity of treatment responses and outcomes is a major problem. [9][10][11] Proton pump inhibitors (PPIs) are common chronic medications used for gastroesophageal reflux and/or peptic ulcers. However, the use of PPI may cause gut microbiota dysbiosis, driven by both altered stomach acidity and direct compound effects. 12 A recent study indicated that PPI use can cause significantly poor outcomes in patients with advanced UC treated with ICI therapy, based on pooled ad-hoc analysis of data from the IMvigor210 (single-arm atezolizumab trial, n = 429) and IMvigor211 (Phase III randomized trial of atezolizumab vs. chemotherapy, n = 931) clinical trials. 13 Also, the same pooled ad-hoc analysis of data from IMvigor210 and IMvigor211 showed that concomitant use of antibiotics (Abs) can also reduce the effectiveness of cancer immunotherapies. 14 However, several studies have reported conflicting results regarding the influence of PPI on the efficacy of ICIs in advanced non-small-cell lung cancer and melanoma. [15][16][17][18] Furthermore, no previous study has investigated the association of PPI and/or Abs with prognosis in patients with advanced UC treated with ICI therapy. Therefore, we investigated the effect of PPI and/or Abs use on oncological outcomes in patients with advanced UC.

| Design and ethics statement
This retrospective, multicenter study was performed in accordance with the principles of the Declaration of Helsinki and was approved by the ethics committee of the Hirosaki University School of Medicine (2019-099) and all hospitals included in this study. Written consent was not obtained in exchange for public disclosure of study information (opt-out approach).

| Patient selection and demographics
We retrospectively evaluated consecutive 160 patients with locally advanced (cT4 or pN+) or metastatic UC treated with ICI therapy between August 2015 and April 2021 at two academic centers and three general hospitals. We verified our retrospective data before analysis. Five patients who were treated within 3 months or who had no imaging evaluation were excluded. Thus, 155 patients with advanced UC were included. The concomitant use of PPI or Abs was defined as any PPI or Abs administered within 30 days before ICI initiation and during ICI therapy. A single dose of first-generation cephem antibiotics (e.g., cephalexin via intravenous injection) were administrated to the majority of patients for peri-procedural purposes (e.g., biopsy or transurethral resection of the bladder tumor) as a standard of care. The following variables were recorded and evaluated: age, sex, Eastern Cooperative Oncology Group performance status (ECOG-PS), estimated glomerular filtration rate (eGFR), clinical stage, progression-free survival (PFS), and overall survival (OS). Tumor stage and grade were stratified based on the eight editions of the TNM classification. 19 The indication of PPI was the treatment of gastroesophageal reflux and/or peptic ulcers.

| Platinum-based first-line chemotherapy
Patients received either gemcitabine plus cisplatin; gemcitabine plus carboplatin; or methotrexate, vinblastine, doxorubicin, and cisplatin. 2,20 Cycles were repeated every 21 or 28 days until disease progression or intolerable adverse events. Because of the population difference, we used the modified cisplatin ineligibility criteria presented by Galsky et al. 21 Using the original criteria, a patient was defined as cisplatin ineligible if at least one of the following criteria was met: ECOG-PS > 1, eGFR <60 ml/min/1.73 m 2 , grade >1 hearing loss, grade >1 neuropathy, and/or New York Heart Association (NYHA) Class III heart failure. In addition, we defined the marginal criteria as being ECOG-PS 1, eGFR 50-60 ml/min/1.73 m 2 , NYHA Class II heart failure, and age >75-80 years. Patients with two or more marginal factors (such as age ECOG-PS 1 and eGFR 55 ml/ min/1.73 m 2 ) were classified as cisplatin ineligible.

| Outcomes
We divided the patients into two groups based on the concomitant use of PPI or Abs. The primary outcomes were the effects of PPI and/or Abs use on the objective response rates (ORRs) and immunerelated adverse events (irAEs). The secondary outcomes were the effects of PPI and/or Abs use on PFS and OS after ICI therapy. Fisher's exact test or the χ 2 test was used to compare categorical variables. Quantitative variables were expressed as means with standard deviations or medians with interquartile ranges. The rate of OS from ICI therapy until death was estimated using the log-rank test. The effects of PPI use on PFS and OS after ICI therapy were determined using multivariable Cox regression analyses via the inverse probability of treatment weighting (IPTW) method. The hazard ratio (HR) with a 95% confidence interval (95% CI) was calculated after controlling for potential confounders, including patient age, sex, ICI therapy treatment line, ECOG-PS at initiation of ICI therapy, tumor type, PPI use, Abs use, and exposure to radiotherapy. Table 1 presents the baseline characteristics of the patients. The median patient age was 72 (64-79) years. Of the 155 patients, 145 (94%) were treated with pembrolizumab. Overall, 56 and 99 patients were non-PPI users and PPI users, respectively. The majority of patients (n = 79, 80%) used PPIs before ICI therapy, suggesting a long-term continuous medication. There were no significant differences in baseline characteristics between the non-PPI users and PPI users. The proportions of patients with concomitant use of Abs were 38% and 51% in non-PPI users and PPI users, respectively.

| Baseline characteristics
Of the PPI users (n = 99), 79 (80%) patients had been using PPI continuously for 3 months before the initiation of ICI therapy.

| Primary outcomes
The ORR was significantly increased in non-PPI users (41%) compared that in PPI users (20%) ( Figure 1A, p = 0.005). ORR was not significantly different between the Abs users (21%) and non-Abs users (33%) ( Figure 1B, p = 0.106). The rate of irAEs was not significantly T A B L E 1 Baseline characteristics at initiation of immune checkpoint inhibitors (ICI) therapy different between the PPI users and non-PPI users (p = 0.639) ( Figure 1C).

| Effects of the concomitant use of PPI or Abs on oncological outcomes
The PFS after ICI therapy was significantly longer in non-PPI users than in PPI users (median 14 vs. 3.6 months, p = 0.002) (Figure 2A).
The OS after ICI therapy was significantly longer in non-PPI users than in PPI users (median 50 vs. 9.1 months, p < 0.001) ( Figure 2B).
The PFS after ICI therapy was not significantly different between the non-Abs and Abs users (median 6 vs. 3.6 months, p = 0.118) ( Figure 2C). The OS after ICI therapy was significantly different between the non-Abs and Abs users (median 18 vs. 8 months, p = 0.014) ( Figure 2D).

| The effect of concomitant use of PPI and
Abs on oncological outcomes ORR after ICI therapy was significantly poorer in patients using both PPI and Abs (double users, 12%) than in patients using neither (nonusers, 40%, p = 0.004) or patients using either PPI or Abs (single F I G U R E 1 Object response rate (ORR) and occurrence of immune-related adverse events (irAEs). ORR were compared between the proton pump inhibitor (PPI) users and non-PPI users (A). ORRs were compared between the antibiotic (Abs) users and non-Abs users (B). Occurrences of irAEs were compared between the PPI users and non-PPI users, and between the Abs users and non-Abs users (C)  double users (median 6.5 months) than in the non-users (median 50 months, p < 0.001) or single users (median 15 months, p = 0.015) ( Figure 3D).

| Multivariable Cox regression analyses for PFS and OS
Multivariable Cox regression analyses for PFS and OS was shown in  Figure 4D) and OS (HR: 1.05, 95% CI [0.65, 1.69), Figure 4D) after ICI therapy between the non-Abs users and Abs users. A schematic summary of present study was shown in Figure S1 as a visual abstract.

| DISCUSSION
This study investigated the prognostic impact of the concomitant use of PPI and/or Abs on oncological outcomes in patients with advanced UC treated with ICI. ORR was significantly lower in patients who used PPIs, whereas Abs use did not affect ORR. The background-adjusted PFS and OS after ICI therapy were significantly shorter in patients who used PPI, which was consistent with the results of a post-hoc analysis of clinical trials. 13 In contrast to the post-hoc clinical trial analysis, we did not observe any effects of Abs on PFS and OS after ICI therapy. The reason for this discrepancy is unclear; however, the limited effects of Abs on prognosis might be related to differences in the duration of PPI and Abs use. In our practice, we administrated a single-or 1-day dose of Abs for invasive procedures, such as biopsy or transurethral resection. Conversely, we administrated PPI continuously before and after ICI therapy if a patient had gastroesophageal reflux or peptic ulcers. However, we could not address the impact of the duration of PPI and Abs treatment because of a lack of data. Also, the cohort is underpowered to discern the differences between the peri-procedural and other prophylaxis. Although both PPI and Abs are potentially associated with gut microbiota dysbiosis, not enough information is available for the effect of both PPI and Abs on gut microbiota dysbiosis. Further study is necessary to address this issue.  22 The authors of the study developed a drug-based prognostic score that included six baseline medications (corticosteroids: 2 points, antibiotics: 1 point, and PPI: 1 point); patients with a high score had significantly poorer prognoses than those with a low score. 22 Therefore, we need to be careful for the accumulation of dysbiosis-inducing drugs in patients with ICI therapy.
The use of corticosteroids is associated with irAEs in patients treated with ICI therapy. A previous study suggested that the use of corticosteroids reduces the efficacy of ICI therapy via reduction of CD8 + T cell proliferation. 28 In addition, PPI-driven gastric hypochlorhydria was associated with the promotion of T cell tolerance and the acidic microenvironment of tumor cells, which facilitates the proliferation, progression, and metastasis of tumors. 29 Therefore, the combination of dysbiosis-inducing drugs and immune suppression agents for the drug-based prognostic score is reasonable. However, the precise mechanisms of the effect of these agents on ICI therapy remain unclear and require further study.
There is a lack of studies investigating the effects of the concomitant use of PPI or Abs and ICIs on irAEs. As many reports suggested a positive association between the efficacy and immune-related AEs, 6,30 we speculate that the concomitant use of PPI or Abs may reduce the incidence of irAEs in conjunction with attenuated therapeutic effects. However, we observed no significant association of the concomitant use of PPI or Abs with the incidence of irAEs. We found a similar proportion of irAEs in PPI users (37%), Abs users (35%), and double users (34%) ( Figure 3B). We could not explain this discrepancy, although there may be an interaction between efficacy, PPI/Abs use, and irAEs. Further research is needed to understand the interaction between efficacy and irAEs in patients undergoing ICI therapy.
Several limitations in this study should be acknowledged. First, because of the retrospective study design, we could not control for selection bias and other unmeasurable confounders. Second, the statistical analysis might be underpowered because of the small sample size. Analyses under a single population are a problem for generalization. Third, lack of data on gut microbiota and corticoste-

| CONCLUSIONS
The concomitant use of PPI may adversely affect oncological outcomes in patients with locally advanced or metastatic UC treated with ICI therapy.