The efficacy and safety of PD-1/PD-L1 immune checkpoint inhibitors in treating advanced urothelial cancer: a meta-analysis of clinical trials

Survival outcomes in advanced urothelial cancer (UC) are dismal. Over the past years, immunotherapy remains an evolving treatment modality for these patients. This meta-analysis was performed to comprehensively evaluate the efficacy and safety of immune checkpoint inhibitors. For this purpose, 18 clinical trials comprising a total of 3,144 patients were identified from the PubMed database up to September 2020. Overall, the objective response rate (ORR) to PD-1/PD-L1 inhibitors was 0.20 [95% confidence intervals (CI) 0.17–0.23]. Furthermore, the pooled 1-year overall survival (OS) and 1-year progression-free survival (PFS) rates were 0.43 (95% CI 0.33–0.53) and 0.19 (95% CI 0.17–0.21), respectively. The summary rates of any-grade and grade ≥3 adverse events (AEs) were 0.66 (95% CI 0.58–0.74) and 0.13 (95% CI 0.09–0.18), respectively. Among the different subgroups, PD-1/PD-L1 inhibitors elicited a promising ORR in patients with lymph node-only metastasis compared to those with visceral metastasis (0.41 VS. 0.17). Additionally, patients with primary tumor in the lower tract had higher ORR compared to those with primary tumor in the upper tract (0.24 VS. 0.15). Briefly speaking, this immunotherapy protocol showed an encouraging efficacy and acceptable safety profile in the treatment of advanced UC. Moreover, our findings provided potential clinical significance for patients with lymph node-only metastasis or primary tumor in the lower tract. However, these exciting findings need further confirmation.

The efficacy and safety profile of PD-1/PD-L1 inhibitors are the major concern related to immunotherapy. Recently, a meta-analysis conducted by Zhang et al., including studies performed before July 2019, reported that the pooled ORR of immune checkpoint inhibitors was 0.20, and the 1-year OS and 1-year PFS rates were 0.50 and 0.17, respectively. The summary frequencies of any-grade and grade ≥3 AEs were 0.65 and 0.11, respectively [8]. However, 6 other studies on the association between immune checkpoint inhibitors and advanced UC were carried out last year. Thus, we systematically collected available published data and performed an updated meta-analysis to investigate the efficacy and safety of PD-1/PD-L1 inhibitors in the treatment of advanced UC patients. The outcomes were then compared across subgroups stratified by different PD-L1 expression levels, studied drugs, and metastasis or primary tumor locations.

Literature search
We conducted a thorough search of the PubMed database to identify the relevant literature until October 2020, using the following research terms: "metastatic bladder cancer" OR "metastatic urothelial carcinoma" OR "bladder cancer" OR "transitional cell carcinoma" AND "PD-L1" OR "PD-1" OR "immunotherapy" OR "immune checkpoint inhibitor" OR "Pembrolizumab" OR "Atezolizumab" OR "Avelumab" OR "Durvalumab" OR "Tislelizumab" OR "Nivolumab" [8]. The search was focused on human studies, without restriction on language. We also checked for relevant articles and their references to search all eligible literature. Two authors (Y.W. and K.F.X.) independently screened the literature for eligibility and any disagreements were resolved by reaching a consensus.
The exclusion criteria were as follows: (1) Duplicates; (2) Lack of required data (3) case reports, reviews, ecological analyses and off-topic studies, etc. Besides, if multiple studies were conducted from the same or overlapping cohort, only the most informative one was included.

Data extraction
Two authors (Y.W. and Y.Z.F.) independently extracted data from the selected studies using a standardized data collection form. Any discrepancy was resolved by discussing and reaching a consensus. The extracted information was: the name of first authors, the publication year, phase of research, use of drugs in the trial, median follow-up time, PD-1/PD-L1 inhibitors used as the first line or the second line, the control group of each clinic trail, dosage of drugs, number of recipients, age of participants, ORR, 1-year PFS rate, 1-year OS rate, rates of any-grade and grade ≥3 AEs.

Outcomes and quality assessment
The outcome measures included the ORR, 1-year OS rate, 1-year PFS rate, rates of any-grade and grade ≥3 AEs. Quality assessment of the studies was conducted independently by two authors (F.L and Y.J.D) based on the Jadad score by RevMan 5.3 [9], and diverging opinions were resolved by discussion.

Statistical analysis
In this meta-analysis, we presented evaluation indicators with percentages and its 95% confidence intervals (95% CI). Both the fixed-and random-effects methods were used to estimate the overall association. Statistical heterogeneity among the included studies was measured by the Q-statistic (Statistical significance was set at P < 0.05) and I 2 statistic [10,11]. We calculated the pooled ORR, 1-year PFS rate and 1-year OS rate with 95% CI to evaluate the efficacy profile of PD-1/ PD-L1 inhibitors [8]. Similarly, we computed the overall rates for any-grade and grade ≥3 AEs to evaluate the safety of immune check point inhibitors.
Subgroup analyses were conducted to measure possible sources of heterogeneity on the basis of different PD-L1 expression levels, PD-L1/PD-1 inhibitors, studied drugs, visceral or lymph node-only metastasis, and primary tumor in the upper or lower tract. Sensitivity analyses were designed to evaluate the robustness of the results. In addition, Egger's test and Begg's were utilized to assess for potential bias [12]. All statistical AGING analyses were performed using RevMan 5.3 (Cochrane Collaboration, Oxford, UK) and the "meta" package in the R software 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria). A two-tailed P value <0.05 was considered statistically significant.

Literature search results
A flow chart of our selection process was illustrated in Figure 1. A total of 1,409 articles were identified after our search. Of those, 33 were considered to be preliminary selected articles for further review after excluding duplicate articles and screening the titles and abstracts to determine their relevance. After a full-text review of the remaining 33 articles, 7 articles were excluded due to the fact that they did not report relevant outcomes. Meanwhile the remaining 8 articles were excluded since their participant cohorts overlapped with other studies. Finally, we included a total of 18 articles in our meta-analysis [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29] (Figure 1).

Study
Year
We used Begg's and Egger's tests to conduct asymmetry tests and measure the publication bias. The Begg's test did not establish evidence of publication bias after analysis of the ORR (P = 0.195), 1-year OS rate (P = 0.593) and 1-year PFS rate (P = 0.298). Likewise, the Egger's test did not point out evidence of publication bias with respect to AGING the ORR (P = 0.653), 1-year OS rate (P = 0.493), and 1-year PFS (P = 0.266).

Safety assessment
The rates of any-grade and grade ≥3 AEs were used to gauge the safety of PD-1/PD-L1 inhibitors in the treatment of metastatic UC. The pooled rates of anygrade and grade ≥3 AEs rates are presented in Figure 4. The summary outcomes for any-grade and grade ≥3 AEs were 0.66 (95% CI 0.58-0.74) and 0.13 (95% CI 0.09-0.18), respectively. Obvious heterogeneity was found in the pooled estimation of the rate of any-grade AEs (I 2 = 95.7%, P < 0.0001) and grade ≥3 AEs (I 2 = 93.5%, P < 0.0001). Thence, subgroup analysis based on PD-L1/PD-1 inhibitors was performed to explore the sources of heterogeneity. The significant evidence of AGING publication bias was not indicated by the Egger's and Begg's tests.

DISCUSSION
Advanced UC patients have a poor prognosis. Currently, platinum-based drugs are the therapeutic mainstay for these patients and there has been a lack of effective second-line drugs [31]. Patients with advanced UC still have a lack of effective treatment regimens to slow the disease's progression long enough for the development of immunotherapy strategy [32,33]. In recent years, PD-1/PD-L1 immune checkpoint inhibitors remains an evolving treatment modality for advanced UC [34]. To date, FDA has approved 6 immune checkpoint inhibitors for the treatment of advanced UC patients who were previously treated with standard chemotherapy and for those ineligible to the standard chemotherapy.  Table 3. The outcomes of the any-grade and ≥3 grade AEs rates of PD-1/PD-L1 inhibitors. To make a further analysis of the safety and efficacy of PD-1/PD-L1 inhibitors in treating advanced UC, we performed an up-to-date meta-analysis. In this updated meta-analysis, 18 studies comprising a total of 3, 144 patients diagnosed with advanced UC were included to explore the efficacy and safety of PD-1/ PD-L1 inhibitors in the treatment of these patients. Overall, the average ORR for PD-1/PD-L1 inhibitors was 0.20 (95% CI 0.17-0.23, Figure 2A). Furthermore, the pooled 1-year OS and 1-year PFS rates were 0.43 and 0.19, respectively. Thus, PD-1/ PD-L1 immune checkpoint inhibitors elicited promising efficacy ( Table 2). The underlying mechanism of action of PD-1/PD-L1 immune checkpoint inhibitors in the treatment of advanced UC could be the fact that PD-1/PD-L1 antibodies prevents the immune escape of tumor cells by blocking the binding of PD-1 on T cells to its ligand on tumor cells. The rates of any-grade and grade ≥3 AEs were used to evaluate the drugs' safety profiles. The overall rate of any-grade AEs did not demonstrate a statistically significant difference in the PD-1 group compared with PD-L1 group. Noticeably, the pooled rates of grade ≥3 AEs in the PD-1 and PD-L1 groups were 0.18 and 0.11, respectively. However, the related mechanism for this finding is unclear.
Substantial heterogeneity was detected in our metaanalysis due to different PD-L1 expression levels, PD-L1/PD-1 inhibitors, studied drugs, visceral or lymph node-only metastasis, and either in the upper or lower tract. We conducted subgroup analyses to investigate the sources of the observed heterogeneity across studies. A large part of the detected heterogeneity may be explained by stratified analysis, which is based on differences in interventions across various studies, locations of metastases and primary tumors. Notwithstanding, we have confirmed an absence of significant publication bias in this meta-analysis either with the Begg's tests for each study. In addition, our sensitivity analyses revealed similar and robust results. This study contains several important strengths that have been briefly mentioned below. This is an updated systematic epidemiologic assessment of the safety and efficacy of PD-1/PD-L1 inhibitors in treating advanced UC patients. Our summary analysis of 18 studies involving 3, 144 patients with advanced UC provides a more stable association and reliable estimation. Furthermore, the findings observed in subgroup analyses grouped by location of metastasis or primary tumor have a promising benefit for the clinical management of those patients.
Withal, there are several potential limitations in our current study that need to be taken into account when interpreting the results. First and foremost are the limitations inherent to the majority of included studies, which were prone to have potential performance bias AGING because most of them were different phase of clinical trials, and a larger number of RCTs on PD-1/PD-L1 inhibitors in treating advanced UC patients have not been conducted. Secondly, substantial heterogeneity was observed in this present analysis; although numerous subgroup analyses were conducted, the possible sources of heterogeneity were not identified. Thirdly, the included studies had various classifications of PD-L1 expression levels by different staining cut-off values, which might have an impact on the patient populations and mislead the true summary estimation. In summary, this updated meta-analysis not only confirmed the efficacy and safety of PD-1/PD-L1 inhibitors in treating advanced UC patients but also provided potential clinical significance for patients with lymph node-only metastases or primary tumors located in the lower tract. Nevertheless, further investigation mainly via RCTs is needed to confirm these findings.

Availability of data and materials
All data generated or analyzed during this study are included in the published articles.

AUTHOR CONTRIBUTIONS
Yu Wang, Kunfeng Xie and Yunze Fang contributed to the data acquisition and data interpretation process. Yu Wang, Fei Li and Yuejun Du performed the statistical analyses and drafted the manuscript. Wanlong Tan and Lina Hou reviewed the results and provided help for paper writing. All the authors gave final approval of the submitted manuscript and reached an agreement to be equally responsible for every aspect of this work.