Abstract
Background/Aim: Sequential therapy using chemotherapy and subsequent immune checkpoint inhibitor (ICI) treatment prolongs the survival of patients with advanced urothelial carcinoma (UC). However, no comparison data for oncological outcome between pembrolizumab and avelumab has been reported. Thus, we compared oncological outcomes between pembrolizumab as second-line therapy and maintenance avelumab therapy in patients with advanced UC. Patients and Methods: We retrospectively evaluated patients with advanced UC treated with pembrolizumab or avelumab between January 2018 and February 2023. We compared oncological outcomes after adjusting for patient characteristics. Immune-related adverse events (AEs) in each group were evaluated using the Common Terminology Criteria for Adverse Events. Results: There were 186 and 44 patients in the pembrolizumab- and avelumab-treated cohorts, respectively. After propensity score matching, 43 patients from each group were selected and analyzed. Median progression-free survival from the initiation of pembrolizumab and avelumab treatments was 126 and 139 days, respectively (log-rank test, p=0.625). Median overall survival in the pembrolizumab and avelumab cohorts were 658 days and not reached, respectively (log-rank test, p=0.249). Thirty-eight (20.4%) and 14 (31.8%) all-grade immune-related AEs were observed in 186 pembrolizumab- and 44 avelumab-treated patients, respectively (chi-squared test, p=0.112). Regarding endocrine-related AEs, 12 (6.5%) and none (0%) were observed in pembrolizumab- and avelumab-treated patients, respectively (Fisher’s exact probability test, p=0.129). Conclusion: Pembrolizumab and maintenance avelumab therapy provide equivalent oncological outcomes in patients with advanced UC. Although no significant difference was observed, there might be a potential risk of higher endocrine-related AEs due to pembrolizumab compared to avelumab maintenance therapy.
In recent decades, platinum-based first-line (1L) chemotherapy has been the only established systemic therapy for patients with metastatic or locally advanced UC (1-3). Although many second-line chemotherapy regimens have been reported to prolong survival outcomes, none of these is a definite established treatment based on Level 1 evidence (4-10). According to the open-label phase 3 prospective KEYNOTE-045 trial, pembrolizumab, an anti-programmed cell death protein-1 (PD-1) antibody, has been used as the standard second-line treatment for these patients with advanced UC (11, 12). In the KEYNOTE-045 trial, oncological outcomes were compared between patients treated with pembrolizumab and those treated with second-line chemotherapy with paclitaxel, docetaxel, or vinflunine. Accordingly, the median overall survival (OS) was 10.3 months vs. 7.4 months [hazard ratio (HR)=0.73, 95% confidence interval (CI)=0.59-0.91, p=0.002] (12). Two-year follow-up data also demonstrated the clinical benefit of using pembrolizumab in patients with advanced UC after the failure of 1L chemotherapy (11). Moreover, in real-world data, there are reports showing that the introduction of pembrolizumab prolonged the prognosis in patients with advanced UC (13, 14).
Another immune checkpoint inhibitor (ICI), the anti-programmed cell death ligand-1 (PD-L1) antibody avelumab, has also been approved as a switching maintenance therapy after 4 to 6 courses of platinum-based chemotherapy to achieve at least stable disease (SD) with 1L chemotherapy (15). The JAVELIN bladder 100 study was an open-label, phase 3 randomized trial comparing patients who received best supportive care after 1L chemotherapy. The OS in the avelumab and best supportive care cohorts were 21.4 months and 14.3 months, respectively (HR=0.69, 95%CI=0.56-0.86, p<0.001). Thus, the oncological benefit of switching avelumab to maintenance therapy is also supported by Level 1 evidence.
These options for advanced UC therapy are now expanding, leading to prolonged survival in patients with advanced UC. However, because the study design and background were different between the KEYNOTE-045 and JAVELIN bladder 100 trials, there is no data comparing the oncological outcomes of these two agents in patients with advanced UC. Clinicians must often consider which ICIs are suitable and beneficial to each patient. Therefore, we performed a retrospective multicenter study to compare the oncological outcomes of these agents using propensity score matching.
Patients and Methods
Study population and design. The clinical data of patients with metastatic UC who underwent pembrolizumab therapy and avelumab treatment after prior platinum-based chemotherapy were retrospectively obtained from 13 hospitals. The patients received pembrolizumab or avelumab therapy between January 2018 and February 2023. Patients who received pembrolizumab after neoadjuvant platinum-based treatment without systemic chemotherapy were excluded. OS in both the pembrolizumab- and avelumab-treated patients was defined as the time from ICI initiation to the last visit or death from any cause. Computed tomography (CT) was performed every 3-6 months to evaluate metastases or progression. The study was approved by the Ethics Committee of Sapporo Medical University School of Medicine and other hospitals included in the study (institutional review board number 302-211). Informed consent was obtained from the subjects through an opt-out process owing to the retrospective design of this study.
Parameters. Age at initiation of ICI treatment, Eastern Cooperative Oncology Group Performance Status (ECOG-PS), sex, history of neoadjuvant or adjuvant chemotherapy, primary site, presence of low albumin level, anemia, neutrophil-to-lymphocyte ratio (NLR), number of 1L chemotherapy cycles, best response to prior platinum-based chemotherapy, and history of primary site resection were assessed and compared between the groups. Information on AEs due to ICI was evaluated using the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.
Treatment and endpoints. Patients in the avelumab cohort were treated with avelumab (10 mg/kg) and intravenous infusion for 1 h every two weeks. For pembrolizumab treatment, 200 mg/body of pembrolizumab was intravenously infused for 30 min every three weeks or 400 mg/body every six weeks. The primary endpoint was to compare the OS between the two groups after adjusting for the patients’ clinical characteristics by propensity score matching at 1:1. The secondary endpoint was to compare the progression-free survival (PFS) and profile of immune-related AEs between the two groups. To elucidate background differences, propensity score matching analysis was performed for low albumin levels, best response to 1L chemotherapy, and the presence of liver or bone metastasis at the time of initiation of ICI therapy. Immune-related AEs have been defined in a previous report (16).
Statistical analyses. Differences between groups were statistically assessed using Fisher’s probability test and the chi-squared test for categorical variables. p<0.05 (two-sided) was considered statistically significant. The OS rate was examined using the Kaplan-Meier method, and survival in the subgroups was compared using the log-rank test. All data were analyzed using EZR ver. 1.35 (17) (Saitama Medical Center, Jichi Medical University, Saitama, Japan).
Results
Patient characteristics. The patient characteristics before propensity score matching are shown in Table I. The pembrolizumab-treated cohort included 186 patients, whereas the avelumab-treated cohort included 44. The median follow-up periods from the initiation of ICIs in the pembrolizumab and avelumab groups were 274 days [interquartile range (IQR)=115.3-676.3] and 276 days (164.5-418.0), respectively. One hundred seven of 186 (57.5%) patients treated with pembrolizumab and 7 of 44 (15.9%) treated with avelumab died of any cause during the follow-up period.
Pre-matching characteristics of patients with the advanced urothelial carcinoma according to immune checkpoint inhibitor.
Oncological outcomes. Figure 1 demonstrates the OS in these two groups before propensity score matching; avelumab-treated patients had a better OS than pembrolizumab-treated patients. However, between these two groups, the pembrolizumab cohort had higher proportions of patients having low albumin level (<3.5 g/dl), stable disease (SD) or progressive disease (PD) as best response in 1L chemotherapy, and the presence of liver and bone metastases. To elucidate possible biases according to the backgrounds of the two cohorts, we performed propensity score matching for these factors at the time of ICI initiation. Table II shows post-matching patient characteristics of the pembrolizumab and avelumab cohorts. In the post-matching cohorts, the median follow-up period of pembrolizumab and avelumab cohorts were 448.0 days (IQR=223.5-697.0) and 273.0 days (IQR=161.0-406.0), respectively. Twenty-five of the 43 (41.9%) patients treated with pembrolizumab and 7 of the 43 (16.3%) avelumab-treated patients died of any cause during the follow-up period. Figure 2 and Figure 3 show the OS and PFS of the pembrolizumab- and avelumab-treated post-matching patients at the time of ICI initiation. No significant differences were observed between the two groups. Figure 4 shows the OS from the initiation of 1L chemotherapy in the post-matching cohorts (log-rank test, p=0.221). No difference in OS was observed regardless of the evaluation from the commencement date of 1L chemotherapy or ICIs. Moreover, we evaluated prognostic factors in 44 avelumab-treated patients. Figure 5A and B show the Kaplan-Meier analyses for OS according to the 1L chemotherapy best response and neutrophil-to-lymphocyte ratio (NLR) before ICI initiation. The results were: best response better than partial response (PR) and NLR less than 3 were favorable factors in univariate analyses using the log-rank test. Of the 43 avelumab-treated post-matching patients, 18 were treated with enfortumab vedotin therapy, whereas 10 of the 43 patients were treated with enfortumab vedotin in the pembrolizumab cohort after ICI discontinuation. The median time from the end of chemotherapy to ICI initiation in the matched cohorts was 48 days (IQR=11.5-163) and 22 days (IQR=9.5-35) in pembrolizumab- and avelumab-treated patients, respectively (Mann-Whitney U-test, p=0.031).
Pre-matching overall survival rate according to treatment regimen.
Post-matching characteristics of patients with advanced urothelial carcinoma according to immune checkpoint inhibitor.
Overall survival (OS) of avelumab and pembrolizumab-treated patients after propensity score matching. ICI: Immune checkpoint inhibitor.
Progression-free survival (PFS) of avelumab and pembrolizumab-treated patients after propensity score matching. ICI: Immune checkpoint inhibitor.
Post-matching overall survival rate according to treatment regimen from the time from first-line chemotherapy initiation.
Overall survival (OS) (A) rate in 44 avelumab treated patients according to the best response of first-line chemotherapy and the level of pre-immune checkpoint inhibitor initiation neutrophil-to-lymphocyte ratio (NLR) level (B). SD: Stable disease; PD: progressive disease; CR: complete response; PR: partial response; ICI: immune checkpoint inhibitor.
Adverse events. In our study, we also compared the immune-related AEs caused by pembrolizumab and avelumab (Table III). Thirty-eight (20.4%) and 14 (31.8%) all-grade immune-related AEs were observed in 186 pembrolizumab- and 44 avelumab-treated patients, respectively (chi-squared test, p=0.112). Endocrine-related AEs, including hypo- and hyperthyroidism, hypophysitis, and adrenal insufficiency were observed in 12 (6.5%) and none (0%) of the patients treated with pembrolizumab and avelumab, respectively (Fisher’s exact probability test, p=0.129).
Detail of immune-related adverse events in the pembrolizumab and avelumab cohorts.
Discussion
In the present study, we compared OS and PFS between pembrolizumab- and avelumab-treated patients with advanced UC and concluded that both treatments resulted in equivalent oncological outcomes. Pembrolizumab has been established as a second-line treatment for advanced UC after progression to 1L platinum-based chemotherapy. In contrast, avelumab is a maintenance therapy for which the response to 4-6 courses of 1L chemotherapy was at least SD or better. Theoretically, the clinical indications of these two ICIs differ. However, in real clinical practice, we often have to consider which strategy is more suitable and beneficial, or whether both provide similar oncological benefits. Unfortunately, there are few reports comparing these treatments’ oncological outcomes to help in clinical decision-making. Miyake et al. reported the first real-world data in Japan comparing these two agents and second-line chemotherapy in terms of OS (18). The study included 27 patients treated with avelumab, 103 treated with pembrolizumab, and 74 treated with second-line chemotherapy. According to the report, avelumab was reported to have statistically better OS than pembrolizumab or second-line chemotherapy (log-rank test, p=0.014). However, no significant difference in OS was observed when only 45 pembrolizumab patients with SD or better response to 1L chemotherapy were included (log-rank test, p=0.097). Although that article provided important insights into avelumab therapy, the selection and background biases between these groups could not be eliminated. In the present study, we adjusted for patients’ backgrounds between pembrolizumab- and avelumab-treated patients. This is the first report to demonstrate equivalent oncological outcomes in pembrolizumab- and avelumab-treated patients with advanced UC using propensity score matching.
This study provides fundamental data for determining treatment strategies for patients with advanced UC. For instance, if the patient has difficulty with continuous platinum-based chemotherapy owing to AEs, switching maintenance therapy to avelumab does not deteriorate the oncological outcome compared with pembrolizumab usage after progressive disease. Since cisplatin-based chemotherapy, such as gemcitabine plus cisplatin or dose-dense MVAC (methotrexate, vinblastine, doxorubicin, and cisplatin), carries the risk of severe AEs or deterioration in the quality of life (2, 3), switching to maintenance therapy is a useful option without compromising the oncological outcome. Therefore, adherence to and preference for 1L chemotherapy may be important in determining the subsequent ICI agents. Moreover, post-hoc analyses of the JAVELIN Bladder 100 study demonstrated that maintenance avelumab therapy did not compromise the quality of life evaluated by patient-reported outcomes compared to outcomes in the best supportive care alone group (19).
According to our data, avelumab-treated patients with NLR less than 3 and best response better than PR in 1L therapy had better OS (Figure 5A and B). We also investigated the presence of liver metastasis, anemia, low albumin levels, and Eastern Cooperative Oncology Group Performance Status (ECOG-PS) as prognostic factors in univariate analyses. However, these factors did not reach statistical significance (data not shown), which may be due to the small number of deaths in the cohort of 44 avelumab patients. Furubayashi et al. also reported that no significant differences in OS were observed between maintenance avelumab treated patients with or without metastasis to any organ including primary site (20). Miyake et al. reported that patients who achieved complete or partial response to 1L chemotherapy were less likely to experience tumor relapse during maintenance avelumab therapy (18) and this finding is consistent with our data. Because of the number of patients who died during the follow-up period in the avelumab-treated cohort, multivariate analysis could not determine the prognostic factors. Sonpavde et al. reported five prognostic factor model for the survival of metastatic UC patients receiving PD-L1 inhibitors (21). In this study, ECOG-PS, liver metastasis, platelet count, NLR, and lactate levels were identified. Because atezolizumab-, avelumab-, and durvalumab-treated patients were included together, it is necessary to evaluate only the avelumab-treated cohort in the future.
Immune-related AE profiles were compared in our study. Although the two agents belong to the same ICI category, the detailed safety profiles differ between pembrolizumab and avelumab, especially with regard to endocrine-related AEs. Wang et al. have reported real-world data on the incidence of adverse endocrine events associated with anti-PD-1 and anti-PD-L1 inhibitors (22). In this study, pembrolizumab had the highest incidence of endocrine-related AEs compared to other ICIs. However, patients who experienced endocrine-related AEs demonstrated a significantly higher objective response rate than those who experienced non-endocrine AEs. A meta-analysis of the incidence of endocrine dysfunction has also been reported (23). In this report, the incidence of thyroid dysfunction induced by anti-PD-1 inhibitors was higher than that in anti-PD-L1 inhibitor-treated patients. Accordingly, differences in safety profiles should be considered when selecting suitable ICIs for patients with advanced UC. In 44 avelumab-treated patients, 14 (31.8%) showed immune-related AEs. Tomita et al. reported that 13 of 36 (36.1%) Japanese subgroup patients in the JAVELIN Bladder 100 had immune-related AEs (24) and this is consistent with our present study.
Our study has several limitations. Because pembrolizumab was approved earlier than avelumab, the follow-up period differed between the two groups. A longer duration of pembrolizumab use may lead to a larger number of immune-related AEs. However, since immune-related AEs typically occur within a short time frame after initiation (25, 26), we regard it reasonable to compare the frequency and details of immune-related AEs between the two groups. Additionally, since enfortumab vedotin became available from December 2021 in our country, avelumab-treated patients received more enfortumab vedotin therapy than pembrolizumab-treated patients, although the difference was not statistically significant.
Conclusion
The comparison of the oncological outcomes of pembrolizumab after platinum-based 1L chemotherapy and maintenance avelumab in patients with advanced UC using propensity score matching showed equivalent PFS and OS after the initiation of ICIs and 1L chemotherapy.
Acknowledgements
The present study was performed on behalf of the Sapporo Medical University Urologic Oncology Consortium (SUOC) and the following urologists conducted data collection: Masahiro Matsuki, Shunsuke Sato, Kimihito Tachikawa, Hidetoshi Tabata, Azusa Yamana, Yoshiki Hiyama, Yu Ito, Hideki Adachi, Ryo Otani, Kosuke Shibamori, Masaya Senda, Yohei Ueki, Shuichi Kato, Takashi Muranaka,, Takuto Ogasawara, Yasuyuki Sakai, Tomohiro Toyota. The Authors would also like to thank Editage (www.editage.jp) for English language editing.
Footnotes
Authors’ Contributions
Tetsuya Shindo: Conceptualization; Data Curation; Formal analyais; Investigation; Methodology; Resources; Writing-Original draft; Writing-review and editing. Kohei Hashimoto: Conceptualization; Writng-Original draft; Supervision. Atsushi Takahashi: Conceptualization; Data curation. Shintaro Miyamoto: Data curation. Yasuharu Kunishima: Data Curation. Shunsuke Sato: Data curation. Fumimasa Fukuta: Data curation. Yoshiki Hiyama: Data curation. Akio Takayanagi: Data curation. Ryuichi Kato: Data curation. Atsushi Wanifuchi: Data curation. Yohei Ueki: Data curation. Manabu Okada: Data curation. Hideki Adachi: Data curation. Ko Kobayashi: Conceptualization. Toshiaki Tanaka: Conceptualization. Naoya Masumori: Writing-original draft; Writing-review and editing; Supervision.
Conflicts of Interest
Naoya Masumori received lecture fees from Astellas, AstraZeneca, Takeda, Kissei, and Jansen Pharmaceutical, and research funding from Astellas, Takeda, Kissei, Ono Pharmaceutical, and Daiichi-Sankyo. Kohei Hashimoto received honoraria from Bayer, Jansen Pharmaceutical, Astellas, and AstraZeneca.
- Received November 15, 2023.
- Revision received December 3, 2023.
- Accepted December 4, 2023.
- Copyright © 2024 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.











