Efficacy and safety of programmed cell death-1/programmed cell death ligand-1 inhibitors in advanced urothelial malignancy: A systematic review and meta-analysis.

INTRODUCTION: Programmed cell death-1/programmed cell death ligand-1 (PD-1/PDL-1) inhibitors are the newest class of approved drugs for advanced urothelial cancer (AdUC). This review aims to collate the evidence for their efficacy and safety in various treatment settings.
METHODS: Extensive search of databases was performed (updated May 2018) and the protocol was registered on PROSPERO (CRD42017081568). The review was conducted according to the Preferred Reporting Items for Systematic reviews and Meta-Analysis statement. STATA (v 12) and Revman 5.3.5 were used for data analysis.
RESULTS: Ten nonrandomized, open-label clinical trials were included in this review. PD-1/PD-L1 inhibitors were used as second-line, stand-alone in eight trials and as first-line in cisplatin-ineligible in two trials. Heterogeneity was observed for study design, PDL-1 testing methods, cutoff criterias used and translational markers evaluated. The pooled objective response rate (ORR) was 18.2% (95% confidence interval [CI] 15.1-21.2, n = 1785) with PD-1/PDL-1 inhibitors in second-line settings as compared to 12.6% (95% CI 10.3-14.9, n = 736) with second-line chemotherapy and 23.7% (95% CI 19.9-27.4, n = 489) with PD-1/PDL-1 inhibitors as first-line therapy in cisplatin-ineligible patients. The median progression-free survival and overall survival was similar with PD-1/PD-L1 inhibitors in both second- and first-line treatment settings (1.5-2.9 vs. 2.0-2.7 months and 7.9-18.2 vs. 15.9 months) and second-line chemotherapy (3.3-4.0 months and 7.4-8 months). Odds of achieving ORR was 0.10 (95% CI 0.03-0.31, n = 229) in the second-line, stand-alone setting with a combined positive score (CPS) cutoff of 25% and was 0.34 (95% CI 0.19-0.62, n = 265) with a CPS cut-off of 10% in first-line setting in the cisplatin-ineligible.
CONCLUSIONS: PD-1/PDL-1 inhibitors appear to be promising in the treatment of AdUC and CPS may be a potentially reliable biomarker for predicting response but needs validation. Caution needs to be exercised until more data are available on imAEs and further studies are required to prove their worth as the standard of care.

INTRODUCTION

Urothelial cancer (UC) is the ninth-most common malignancy reported worldwide.[12] Advanced UC is highly lethal. The current standard of care, platinum-based combination chemotherapy, was approved three decades ago.[34] The median overall survival (OS) with cisplatin-based therapy is 12-15 months and is about 9 months for carboplatin-based treatment.[5] Patients who do not respond to the first-line treatment have a median OS of only 5–7 months with the available second-line treatment options.[6] The commonly used second-line chemotherapy, paclitaxel or docetaxel are effective in about 10% of the patients, and the response is often partial and short-lived.[78910] Therefore, there has been an unmet need to develop new therapies for these patients.

With years of research, it was found that the malignant cells might escape immune detection by exploiting the immune checkpoint, programmed cell death-1/programmed cell death ligand-1 (PD-1/PDL-1) pathway that suppresses T-cell responses. Therefore, it was conceptualized that blocking PD-1/PDL-1 pathway may cause prolonged T-cell activation and tumor rejection.[111213] Monoclonal antibodies against PD-1 and PDL-1 are the newest class of drugs available for the treatment of advanced urothelial malignancy (AdUC). The first drug in this class, Atezolizumab, was approved by the US-Food and Drug Administration (FDA) in May 2016. Subsequently, four others, namely nivolumab, durvalumab, avelumab, and pembrolizumab have been approved. All of them have been labeled as second-line agents to be used if disease progresses with platinum-based regimens. Only atezolizumab and pembrolizumab are indicated for use as first-line agents when cisplatin use is contraindicated.[14] These drugs are expected to change the treatment landscape of AdUC. Therefore, it is essential to critically examine the extent of therapeutic benefit and the predictive markers of response as well as the adverse treatment outcomes with this class of agents. The objective of this systematic review is to synthesize the existing evidence assessing the overall efficacy and safety of the currently approved PD-1/PDL-1 inhibitors in AdUC.

METHODS

The protocol was registered on PROSPERO (International Prospective Register of Systematic Reviews, CRD42017081568) and is available on the University of York website. Identified reports were reviewed according to the Consolidated Standards of Reporting Trials. The Preferred Reporting Items for Systematic Reviews and Meta-analysis were adopted for conducting the review.

Literature search

Three authors SD, RR, and NJ independently performed the electronic database searches starting from January 1, 1980 to December 31, 2017 and were updated through May 2018. The databases included the Cochrane Central Register of Controlled Trials (CENTRAL; Wiley Cochrane Library), MEDLINE (Ovid®), EMBASE®(Elsevier), and Science Citation Index Expanded through Web of Science™. The clinical trial registries, namely ClinicalTrials.gov and World Health Organization International Clinical Trials Registry Platform (http://apps.who.int/trialsearch) were also searched. The search terms included both text and medical subject heading (MeSH terms) [Supplementary Table]. Hand searching of the cross-references of the important studies was conducted to ensure that all relevant studies were identified. Google scholar and the conference proceedings were searched for additional information. The studies obtained by searching the literature were collated using the reference manager (EndNote X8), and duplicates were removed.

Search strategy

Click here for additional data file.

Selection criteria

We included studies meeting the following criteria: (1) adult patients with metastatic or locally advanced UC with Eastern Cooperative Oncology Group performance status ≤2; (2) intervention trials evaluating the efficacy and safety of PD-1 or PDL-1 inhibitors used alone but not in combination with other treatments; (3) noncomparative or comparative clinical studies; (4) i – At least one of the outcomes was assessed: objective response rate (ORR), median progression-free survival (PFS) and OS,[15] ii – Additional efficacy outcomes such as median time to response, median doses of the drug required for response, median duration of response, disease control rate, PFS rate, and OS rate have been assessed, iii – Adverse events (AE) especially immune-mediated AEs (imAE) have been assessed for evaluating the safety of PD-1/PDL-1 inhibitors; and (5) the median follow-up duration was 5 months or more.

Subgroups

The subgroup analyses were conducted for outcomes as per the PDL-1 expression status on the tumor cells or immune cells or both.

Data extraction

Two authors SD and RR independently reviewed the manuscripts and abstracted the data against the predefined inclusion criteria. Data abstraction from the eligible studies was performed using a pilot-tested standardized data abstraction form. The extracted information included study identification, authors, phase of clinical trial, location, duration, design, participant characteristics, clinical setting, details of the intervention, sample size, duration of follow-up, outcome measures for both safety and efficacy, diagnostic tests employed to assess PDL-1 positivity status, translational biomarkers investigated and the sub-group analyses. All the information was compiled in standardized tables and disagreements were resolved by consensus. The analysis was finalized after thorough discussion among the review authors (SP, RM, AKM, and SKS).

Assessment of risk of bias

We searched for an appropriate tool to assess the risk of bias for the nonrandomized, noncomparative, single-arm studies.[16] We also reviewed the studies based on the published data, nature of the outcomes assessed and the method of assessment. The heterogeneity in participant characteristics, interventions, and the PDL-1 diagnostic tests used were also noted. The Cochrane risk of bias tool (RoB 2) was used to evaluate the quality of randomized controlled trials (RCT).[17]

Data synthesis and analysis

The Cochrane Handbook was consulted for data extraction.[18] STATA (v12; StataCorp, USA) and Revman 5.3.5 (The Nordic Cochrane Centre, Copenhagen, Denmark)[19] were used to analyze the data. The observed treatment effect was reported as pooled mean with 95% confidence interval (CI). I2 statistics was used to assess for heterogeneity amongst the included studies and a value ≥ 50% was considered significant for using random effect model. However, we used random effect model for all pooled analysis considering the clinical heterogeneity between the studies. The reported means for the treatment arms were pooled in STATA to calculate the median effect size with 95% CI. Categorical outcomes were analyzed using Cochrane Mantel–Haenszel method in Revman. The observed probability of achieving treatment response was reported as odds ratios (OR) with 95% CI. The data were presented in a descriptive manner when the computation of a pooled effect size was not feasible. The subgroup analysis was performed for only ORR but not others due to nonavailability of adequate data. Egger's plot was used to assess publication bias and Galbraith plot to investigate for heterogeneity.

RESULTS

Quantity of evidence identified

One thousand two hundred and eighty-four relevant studies were identified by systematic search of the databases, and the full texts of 115 studies were examined for eligibility. A total of 10 clinical trials including 3010 participants, conducted in 11 countries in North America and Europe were included in this systematic review. The systematic process of study selection is described in Figure 1.

Figure 1

Study flow diagram programmed cell death-1/programmed cell death ligand-1 in advanced urothelial cancer

Characteristics of the included studies

Two treatment categories were recognized (1) PD-1/PDL-1 inhibitors as second-line standalone treatment after disease progression with platinum-based chemotherapy: eight studies[202122232425262728](2) PD-1/PDL-1 inhibitors as first-line treatment in cisplatin-ineligible: two studies,[2930]. Studies were excluded if combination therapies were evaluated[31] or the outcomes of interest were not reported, or the data for urothelial malignancy could not be retrieved from a basket type of trial design.[32] Adaptive and pragmatic trials with shorter follow-up were also excluded.[33] The summary information for the included studies is presented in Table 1.

Table 1

Study characteristics

PD-1 inhibitors as second-line standalone agent
Drug	Study ID location and duration	Study design	Sample size	Participant characteristics	Study interventions	Median Follow-up duration	Main outcome measure
Nivolumab	Sharma et al., 2016[20] Trial ID: NCT01928394 (CheckMate032)Study Location: 16 sites in 5 countriesStudy Duration: June 5, 2014–April 24, 2015	Multicenter, randomized (unclear), allocation by interactive voice response system, open-label, phase 1/2 study, 2-stage, multi-arm (for the purpose of this review the data has been used from a single arm)	78	Age: 18 years or olderDisease: Histologically or cytologically confirmed carcinoma of the renal pelvis, ureter, bladder, or urethra	Nivolumab monotherapy3 mg/kg IV infusion every 2 weeksAdditional arm : Nivolumab + ipilimumab	9 months	Primary Outcome: Confirmed investigator-assessed ORSecondary outcome: Safety
Nivolumab	Sharma et al., 2017[21] Trial ID: NCT02387996. (CheckMate 275)Study Location: 63 sites in 11 countriesStudy Duration: March 9, 2015-October 16, 2015	Multicenter, nonrandomized, open-label, Phase-2, single arm	270	Age: 18 years or olderDisease: Histological evidence of metastatic or surgically unresectable locally advanced urothelial carcinoma	Nivolumab3 mg/kg by 1-h IV infusion every 2 weeks	6 months	Primary Outcome: OR in all treated patientsSecondaryOutcomes : PFS and OS and investigator-assessed OR
Pembrolizumab	Bellmunt et al., 2017[27] Trial ID: NCT02256436Study Location: 120 sites in 29 countriesStudy Duration: November 5, 2014-November 13, 2015	Multicenter, randomized, open-label, phase 3, two arm study (KEYNOTE 045)	Total - 542Pembrolizumab Group - 270, Chemotherapy Group - 272	Age: 18 years or olderDisease: Histologically or cytologically confirmed urothelial carcinoma of the renal pelvis, ureter, bladder, or urethra	Intervention arm: Pembrolizumab 200 mg IV every 3 weeksControl arm: Paclitaxel: 175 mg/m2 of body-surface area, IV every 3 weeksDocetaxel: 75 mg/m2 Vinflunine: 320 mg/m2 Randomized treatment allocation	14.1 months	Primary Outcome: OS and PFSSecondary Outcome: ORR, duration of confirmed response and Safety in the total population
PD-1 inhibitors as second-line standalone agent
Atezolizumab	Rosenberg et al., 2016[22] Trial ID: NCT02108652Study Location: 77 sites from North America and EuropeStudy Duration: May 13, 2014-November 19, 2014	Multicenter, nonrandomized, phase 2, global, single-arm, two cohort trial	Cohort 1: Cisplatin-ineligible patientsCohort 2: Recurrent disease after platinum-based chemotherapy (n=310, included in the review)	Age: 18 years or olderDisease: Histologically or cytologically documented locally advanced or urothelial carcinoma (including renal pelvis, ureter, urinary bladder, or urethra)	AtezolizumabFixed dose of 1200 mg IV administered on day 1 of each 21-day cycle	11.7 months	Primary outcome: ORRSecondary outcome: Duration of response, PFS, and safety
Atezolizumab	Powles et al., 2018[28] Trial ID: NCT02302807Study location: 217 academic medical centers and community oncology practices mainly in Europe, North America, and the Asia-Pacific regionStudy duration: January 13, 2015-February 15, 2016	Multicenter, randomized controlled, open-label, phase 3, two-arm trial (IMvigor211)	TOTAL 931Atezolizumab Group - 467Chemotherapy Group - 464	Age: 18 years or olderDisease: Metastatic urothelial carcinoma who had progressed after platinum-based chemotherapy	Atezolizumab1200 mg IV administered every 3 weeksVinflunine 320 mg/m2 IV or Paclitaxel 175: mg/m2 IV or docetaxel 75 mg/m2 IV (Every 3 weeks)	17.3 months	Primary outcome: OSSecondary outcome: Investigator-assessed ORR PFS, duration of response, safety and patient reported outcomes
Avelumab	Apolo et al., 2017[25] Trial ID: NCT01772004Study location- 24 sites in the USA and EuropeStudy duration: Ongoing. Start Date - January 31, 2013	Multicenter, nonrandomized, open-label, phase 1b, multiple ascending-dose trial	44	Age: 18 years or olderDisease: Eligible participants had metastatic urothelial carcinoma of the renal pelvis, ureter, urinary bladder, or urethra, as confirmed by histology or cytology	Avelumab10 mg/kg dose by 1-h IV infusion once every 2 weeks	16.5 months (As of March 19, 2016)	Primary outcome: Safety and tolerabilitySecondary outcome: Best OR, duration of response, PFS, OS, and evaluation of PD-L1 expression
Durvalumab	Massard et al., 2016[24] Trial ID: NCT01693562Study location: 70 centers worldwideStudy duration: August 28, 2014-November 10, 2015	Multicenter, nonrandomized, open-label, single-arm, phase-1/2, dose-escalation and dose-expansion study	61 (40 PD-L1+ve and 21 PD-L1 -ve	Age: 18 years or olderDisease: Histologically or cytologically confirmed inoperable or metastatic transitional-cell urothelial carcinoma	Durvalumab10 mg/kg every 2 weeks through IV infusion	4.3 months	Primary outcome: Safety on the basis of assessment of adverse effects and serious adverse effectsSecondary outcome: ORR
Durvalumab	Powles et al., 2017[23] Trial ID: NCT01693562 Study location: 60 sites in 9 countriesStudy duration: Ongoing. Start date- August 29, 2012	Multicenter, nonrandomized, open-label, single-arm, Phase 1/2 study	191	Age: 18 years or olderDisease: Histologically and/or cytologically confirmed urothelial carcinoma were eligible for inclusion	Durvalumab10 mg/kg (Q2W), IV infusion, every 2 weeks for up to 12 months	5.78 Months (as of October 24, 2016	Primary outcome: ORRSecondary outcome: Duration of response, time to response, change in target lesion size, disease control rate (confirmed complete response or partial response or stable disease for ≥6 weeks), PFS and OS
PD-1 inhibitors as first-line agent in cisplatin-ineligible
Pembrolizumb	Balar et al., 2017 (a)[29] Trial ID: NCT02335424Study location: 91 academic medical centers in 20 countriesStudy duration: February 24, 2015-August 8, 2016	Multicenter, nonrandomized, single-arm, phase 2 study (KEYNOTE-052	370	Disease: Age group - 18 years or older, cisplatin-ineligible patients with locally advanced and unresectable or metastatic urothelial cancer of the renal pelvis, ureter, bladder, or urethra	Pembrolizumab200 mg IV every 3 weeks	5 months	Primary Outcome: OR in patients with PD-L1-expressing tumorsSecondary Outcome: OS, PFS
Atezolizumab	Balar et al., 2017 (b)[30] Trial ID: NCT02108652Study location: 47 centers across 7 countries in North America and EuropeStudy duration: June 9, 2014-March 30, 2015	Multicenter, nonrandomized, single-arm, phase 2 trial(IMvigor210)	119	Disease: Participants aged 18 years or older with previously untreated locally advanced or metastatic urothelial cancer who were cisplatin ineligible	Atezolizumab1200 mg IV dose every 21 days	17.2 months	Primary outcome: Independently confirmed ORRSecondary outcomes: Investigator-assessed ORR; duration of response, PFS, and OS

OR=Objective response, PFS=Progression-free survival, OS=Overall survival, ORR=Objective response rate, IV=Intravenous. PD-1=Programmed cell death-1, PDL-1=Programmed cell death ligand-1

Majority (8/10) of the included studies were non-randomized, open-label, single-arm, early phase clinical trials (I/II).[202122232425262930] Two of the studies were phase-III open label randomized controlled trials (RCT) for pembrolizumab[27] and atezolizumab,[28] comparing them with second-line chemotherapy. The trials included participants with lymph node, viscera, and liver metastases but excluded the participants with brain metastasis. The major criterion for cisplatin ineligibility was poor renal function (>70% of the participants); others were hearing loss and peripheral neuropathy. The median duration of treatment reported in five out of ten included studies was 2.8–3.6 months (minimum of 0 to a maximum of 35.1 months). The median follow-up period for the trials evaluating these therapies ranged from 5.8 to 17.3 months. Response Evaluation Criteria in Solid Tumors (RECIST) criteria version 1.1 was used to assess efficacy in all the studies. In majority of the trials, the treatment assessment was performed by a set of evaluators masked to the treatment allocation. Whenever these assessments were made by the principal investigator themselves, they were assisted by an Independent Central Review Committee. All trials coded the AE as per the Medical Dictionary for Regulatory Activities terminology. The severity of AE was classified according to the National Cancer Institute Common criteria for AEs version 4.0. PD-1/PDL-1 expression status in the tumor tissue was assessed by immunohistochemistry in all the included studies. However, it was not a mandatory requirement for inclusion. The staining antibodies, diagnostic testing methods, criterion for evaluation and the cutoff criteria for PDL-1 positivity varied across the studies. Exploratory translational biomarker and mutation analysis was reported only in four out of the ten clinical trials.[20222830]

We were unable to assess the risk of bias for the eight nonrandomized, noncomparative, single-arm intervention trials due to lack of a predefined tool for bias assessment.[202122232425262930] However, the risk of bias in these studies was judged as moderate, based on the objective nature of the outcomes and the blinded review committee assisted assessment of the ORR in most of the studies. The RCTs[2728] were judged to have mild-to-moderate risk of bias due to the open-label administration of the intervention and unmasked status of the outcome assessors [Supplementary Figure 1].

Risk of bias

Click here for additional data file.

Efficacy outcomes

Objective response rate

PD-1/PDL-1 inhibitors as second-line standalone: All the eight studies[202122232425262728] reported the ORR. The pooled ORR was 18.2% (95% CI 15.1–21.2, n = 1785) [Figure 2]. Two of the included studies were RCTs and included a second-line chemotherapy (vinflunine/paclitaxel/docetaxel) group as the control arm. The pooled ORR achieved with second-line chemotherapy was 12.6% (95% CI 10.3–14.9, n = 736).

Figure 2

Summary of forest plot for the objective response rate achieved with programmed cell death-1/programmed cell death ligand-1 treatment in advanced urothelial cancer

ORR with atezolizumab was found to be no better than second-line chemotherapy (13.4% [95% CI 10.5–16.9] vs. 13.4% [95% CI 10.5–16.9]),[28] whereas pembrolizumab was found to be significantly better than chemotherapy (21.1 [95% CI 16.4–26.5] vs. 11.4% [95% CI 7.9–15.8])[27] in the settings of a RCT.

PD-1/PDL-1 inhibitors as first-line treatment in cisplatin ineligible: There were two studies in this group.[2930] The estimated pooled ORR was 23.7% (95% CI 19.9–27.4, n = 489) [Figure 2].

Progression-free survival

PD-1/PDL-1 inhibitors as second-line standalone treatment: Seven out of eight studies in this group[2021222325262728] (n = 1612) reported the median PFS, which ranged from 1.5 to 2.9 months [Table 2]. The reported median PFS with second-line chemotherapy in two RCTs ranged from 3.3 to 4 months (n = 736).[2728]

Table 2

Median progression-free survival and overall survival with programmed cell death-1/programmed cell death ligand-1 inhibitor use in Advanced Urothelial Cancers

Study ID	Study drug	Number of participants	Median PFS (months)	95% CI	Median OS (months)	95% CI
PD-1 inhibitors as second-line standalone agent
Sharma et al., 2016[20]	Nivolumab	78	2.8	1.9-2.6	11.3	8.7-NE
Sharma et al., 2017[21]	Nivolumab	270	2.0	1.5-5.9	9.7	7.3-16.2
Bellmunt et al., 2017[27]	Pembrolizumab	270	2.1	2-2.2	10.3	8-11.8
PDL-1 inhibitors as second-line standalone agent
Rosenberg et al., 2016[22]	Atezolizumab	310	2.1	2.1-4.1	7.9	6.6-9.3
Powles et al., 2018[28]	Atezolizumab	467	2.1	2.1-2.2	11.1	8.6-15.5
Apolo et al., 2017[25]	Avelumab	44	2.9	1.5-4.4	13.7	8.5-NE
Massard et al., 2016[24]	Durvalumab	61	NR	-	NR	-
Powles et al., 2017[23]	Durvalumab	191	1.5	1.4-1.9	18.2	8.1-NE
PD-1 inhibitors as first-line agent in cisplatin-ineligible
Balar et al., 2017 (a)[29]	Pembrolizumab	370	2.0	2.0-3.0	NR	-
PDL-1 inhibitors as first-line agent in cisplatin-ineligible
Balar et al., 2017 (b)[30]	Atezolizumab	119	2.7	2.1-4.2	15.9	10.4-NE

NE=Not estimable, thereby meaning that the endpoint was not reached for some of the patients who were surviving at the closure of the study, NR=Not reported, PFS=Progression-free survival, CI=Confidence interval, NE=Not estimable, PD-1=Programmed cell death-1, PDL-1=Programmed cell death ligand-1, OS=Overall survival

Neither atezolizumab (2.1 months [95% CI 2.1–2.2] vs. 4.0 months [95% CI 3.4–4.2])[28] nor pembrolizumab (2.1 months [95% CI 2.0–2.,2 vs. 3.3 months [95% CI 2.3–3.5])[27] demonstrated advantage over the second-line chemotherapy for PFS in the RCT settings.

PD-1/PDL-1 inhibitors as first-line treatment in cisplatin ineligible: The reported median PFS in the two clinical trials[2930] ranged from 2.0 to 2.7 months (n = 310) [Table 2].

Overall survival

PD-1/PDL-1 inhibitors as second-line standalone treatment: Six out of the eight studies (n = 1612) reported median OS.[202122232527] The upper limit of survival was not reached till the completion of the study in three trials,[202325] hence was not estimable (NE). The median OS reported in the studies ranged from 7.9 to 18.2 months, and the quality of evidence was low [Table 2]. The median OS reported with second-line chemotherapy in both the RCTs was 7.4 to 8 months (n = 736).[2728]

OS did not significantly differ between atezolizumab and second-line chemotherapy (11.1 months [95% CI 8.6–15.5] vs. 10.6 months [95% CI 8.4–12.2]) in RCT settings[28] whereas pembrolizumab was reported to have a better median OS than second-line chemotherapy (10.3 months [95% CI 8.0–11.8] vs. 7.4 months [95% CI 6.1–8.3]).[27]

PD-1/PDL-1 inhibitors as first-line in cisplatin ineligible: OS was reported only for atezolizumab (n = 119) but not for pembrolizumab.[30] The reported median OS was 15.9 months (95% CI 10.4 to NE) [Table 2].

Other measures related to efficacy

The median time to response was reported in six out of eight trials in the second-line setting, and it ranged from 1.4 to 3.2 months.[2123242530] Similarly, it was reported in both the trials in the first-line setting and ranged from 2 to 2.1 months.[2930] None of the trials reported the median doses required to achieve the response. The median duration of response was not reached in eight out of ten trials, thereby meaning, an ongoing response in a subset of participants at trial closure.

Disease control rate data could be retrieved from seven out of eight trials in the second-line treatment setting.[20212223242528] The average disease control rate at 12 weeks was 42.1% ±8.8%. Out of the two RCTs, one[28] reported the disease control rate with second-line chemotherapy as 48% at 12 weeks. In the first-line treatment setting, atezolizumab[30] was reported to have the disease control rate of 46.7% at 12 weeks, but for pembrolizumab, this data was unavailable.[29]

The PFS and OS rates were not reported uniformly across the clinical trials. In the second-line setting, PFS rate was 21% (nivolumab),[21]16% (durvalumab),[23] and 19.1% (avelumab)[25] at 12 months. The OS rate at 12 months was 46%, 55%, 54.3%, 46.2%, and 41.2% with nivolumab, durvalumab, avelumab, atezolizumab, and second-line chemotherapy, respectively. None of the trials in the first line treatment setting reported PFS outcomes.

Subgroup analysis as per the programmed cell death-1/programmed cell death ligand-1 positivity status in the tumor tissue

Three out of ten studies assessed the expression of PDL-1 on tumor cells only,[202125] three on tumor-infiltrating immune cells only,[222830] and the rest on both types of cells and macrophages and expressed it as the combined positive score (CPS).[23242729] Studies used different cutoff criteria for PDL-1 positivity.

Programmed cell death ligand-1 expression on tumor cells

Cutoff as 1%

Nivolumab, evaluated in second-line setting, with PDL-1 expression cut-off of 1%.[2021] The ORR was achieved in 25.7% (95% CI 20.2–31.7) of the participants with PDL-1 expression of >1% as compared to 16.9% (95% CI 11.3–22.6) with <1% expression [Figure 3]. However, the odds of achieving objective response were similar (OR = 0.65, 95% CI 0.35–1.53, n = 413) in both the groups [Figure 4]. The median PFS was 5.5 months in >1% group as compared to 2.8 months in <1% group.[21] The median OS was 11.3 to 16.2 months in >1% group but 5.9 to 9.9 months in <1% group.[2122] No differences were observed in the outcomes when assessing atezolizumab in first-line setting with PDL-1 expression cut-off of 1% (23% vs. 21% in >1% PDL-1 expression group and <1% expression group, respectively).[30]

Figure 3

Summary of forest plot for the objective response rate achieved with programmed cell death-1/programmed cell death ligand-1 treatment in advanced urothelial cancer by programmed cell death ligand-1expression status

Figure 4

Forest plot for the odds of achieving objective response with programmed cell death-1/programmed cell death ligand-1 inhibitors in advanced urothelial cancer by programmed cell death ligand-1 expression status

Cutoff as 5%

Avelumab evaluated in second-line standalone setting with PDL-1 expression of 5% as cutoff.[25] ORR was achieved in 53.8% (95% CI 26.4–81.1) participants with >5% expression as compared to 4.2% (95% CI 0.8–7.6) in <5% expression [Figure 3]. The odds of achieving ORR were found to be marginally higher with higher PDL-1 expression (OR = 0.04, 95% CI 0.0–0.36, n = 37) [Figure 4]. The median PFS (12.1 months, 95% CI 2.8-NE vs. 1.7 months, 95% CI 1.5–3.0) and OS (NE, 95% CI 8.5-NE vs. 12.1 months, 95% CI 2.7-NE) were higher in participants with PDL-1 expression >5% as compared to <5%.[25] However, no differences in ORR (28% vs. 23%), median OS (12.3 vs. 19.1 months) and OS rate at 12 months (52% vs. 59%) were observed for atezolizumab in the first line setting in >5% compared to <5% expression group.[30]

Programmed cell death ligand-1 expression on tumor-infiltrating immune cells

Cutoff 1%

Atezolizumab evaluated in the second-line[22] and first-line treatment setting[30] using 1% expression as cutoff. The ORR was achieved in 17.9% (95% CI 13–24) participants with >1% expression as compared to 8% (95% CI 3–15) with <1% expression in the second-line setting whereas the corresponding numbers were 23.7% (95% CI 15–35) and 20.5% (95% CI 9–36) in the first-line setting. The odds of achieving ORR with >1% PDL-1 expression was found to be better (OR = 0.39, 95% CI 0.17–0.86, n = 310) in second-line but not in the first-line setting (OR = 0.44, 95% CI 0.16–1.22) [Figure 4]. The PFS and OS data were not available for this sub-group.

Cutoff 5%

Data were available for atezolizumab evaluated in the second-line standalone setting[2228] as well as first-line setting using 5% PDL-1 expression as the cutoff.[30] In the second-line setting, ORR was achieved in 24.3% (95% CI 18.5–30.1) participants with >5% expression as compared to 9.8% (95% CI 7.3–12.2) in <5% expression group [Figure 3] and the odds of achieving ORR was 0.34 (95% CI 0.22–0.52, n = 772) [Figure 4]. However, in the first-line treatment setting, the ORR (28% vs. 21%) and the odds of response were similar (OR = 0.67, 95% CI 0.26–1.69, n = 71).[30] The median PFS and OS were reported only by Rosenberg et al.[22] There was no difference in PFS between the groups. However, the median OS appeared to be longer with PDL-1 expression >5% (11.4 months, 95% CI 9-NE) as compared to <5% (6.6 months, 95% CI 4.4–8.8).

Combined positive score

Cutoff 10%

This data wasobtained from the study by Balar et al.[29] The ORR achieved with pembrolizumab, in the first-line setting, was 39% versus 20% and the odds of attaining ORR was also marginally higher in the group with CPS >10% compared to <10% (OR = 0.34, 95% CI 0.19–0.62, n = 265) [Figure 4]. The PFS and OS data were not available for this subgroup.

Cutoff 25%

Two trials evaluating durvalumab[2324] used this cutoff criterion. The pooled ORR in the second-line treatment setting was 28.8% (95% CI 21.1–36.4) with CPS >25% compared to 4.4% (95% CI-0.4–8.6) in <25% group [Figure 3]. The odds of achieving ORR was somewhat higher in participants with high CPS (OR = 0.10, 95% CI 0.03–0.31, n = 229)[2324] [Figure 4]. Only Powles et al.[23] reported median PFS and OS. The median PFS was 2.1 months (95% CI 1.4–2.8) versus 1.4 months (95% CI 1.3–1.5) and the median OS was 20 months (95% CI 11.6 to NE) versus 8.1 months (CI 3.1 to NE) in CPS expression >25% as compared to <25%.

Safety outcomes

The safety analysis included all the 2268 participants from 10 studies. None of the studies reported the safety outcomes by PDL-1 expression status but only for the total cohort of participants treated. A mean of 69.1% ±7.9% participants experienced treatment-related AE, out of which 14.0% ± 5.8% were serious (Grade 3/4/5) in nature. Three studies reported no treatment-related deaths[222425] and the average number of death was 0.8% ± 0.8%. The mean rate of treatment discontinuation was 4.6% ±2.5%, and the most common reason was AEs due to the study drug. Dose disruption or delay in treatment administration was reported to be as high as 34%.[29] The most common treatment-related AE was fatigue (21.5% ± 7.7%). The other common AEs were pruritus (12% ±7.5%), skin rash (7.1% ± 5.1%), diarrhea (8.6% ±3.3%), nausea (9.3% ± 3.1%), decreased appetite (8.2% ± 3.7%), arthralgia (3.9% ± 4.2%), and asthenia (5.1% ± 4.6%). No infusion-related AEs were reported for nivolumab, atezolizumab. These were 0.8% for pembrolizumab; 1%–3.2% for durvalumab and highest with avelumab (20.4%) [Table 3].

Table 3

Common drug-related adverse events reported in the included clinical trials of programmed cell death-1/programmed cell death ligand-1 inhibitors in advanced urothelial malignancy

Study ID	Drug	Any AE*	Grade-1/2 AE	Grade-3/4/5 AE	Death due to AE	AE leading to Rx discontinuation	Fatigue	Infusion-related reactions	Asthenia	Arthralgia	Nausea	Decreased appetite	Diarrhea	Vomiting	Pyrexia	Constipation, gastrointestinal	Skin rash	Pruritus	Anaemia
PD-1 inhibitors as second-line standalone agent
Sharma-2016 (n=78)	Nivolumab	81	59	22	2.56	3	36	NR	NR	12	13	NR	NR	NR	NR	1	18	29	10
Sharma-2017 (n=270)	Nivolumab	64	46	17	1	5	17	NR	5	NR	7	8	9	NR	6	9	6	9	NR
Bellmunt-2017 (n=266)	Pembrolizumab	61	46	15	1.5	5.6	15	0.8	6	NR	11.3	8.6	10.1	NR	NR	5.7	NR	19.5	4.2
PDL-1 inhibitors as second-line standalone agent
Rosenberg-2016 (n=310)	Atezolizumab	85	69	16	0	4	30	NR	NR	8	14	13	8	6	9	2	7	10	4
Powles-2018 (n=459)	Atezolizumab	69	NR	20b	1	3	17	NR	13	NR	10	12	11	3	9	6	9	12	7
Apolo-2017 (n=44)	Avelumab	66	59.1	6.8	0	9.1	20.5	20.5	11.4	NR	11.4	4.5	9.1	NR	NR	NR	9.1	6.8	NR
Masard-2016 (n=61)	durvalumab	69	64	4.9	0	1.6	13.1	3.2	6.6	6.6	6.6	8.2	9.8	NR	6.6	NR	NR	3.3	NR
Powles-2017 (n=191)	Durvalumab	68	61a	6.8	1	1.6	19.4	1	NR	5.8	6.8	9.4	8.4	NR	5.8	NR	7.3	5.2	1.5
PD-1 inhibitors as first-line agent in cisplatin-ineligible
Balar-2017(a) (n=370)	Pembrolizumab	62	46	14c1d0.2e	0.2	5	17	NR	5.2	2.2	7.2	9.2	8	NR	4.2	2.2	9.2	14.2	2.2S
PDL-1 inhibitors as first-line agent in cisplatin-ineligible
Balar-2017(b) (n=119)	Atezolizumab	66	50	16b	0.8	8	30	NR	3	4	5	9	12	5	5	NR	5	11	5

Data are represented as percentage of the events of AEs in the study population. *Any AE includes AEs of all grades (1-5). aAny grade, bGrade-3/4, cGrade-3, dGrade-4, eGrade-3/4/5. AE=Adverse events, NR=Not reported

All the trials specifically looked for immune-related AE (imAE) induced by the PD-1/PDL-1 inhibitors. The onset of such imAE was reported to range from within a few weeks from the start of the treatment to several months after the end of the treatment. Skin and skin related AEs were the most common imAE [Table 4] and first ones to appear as reported in these trials. The other common imAE were hepatitis (4%–21.6%; 7/10 trials), pneumonitis (2.3%–12%; 9/10 trials), arthralgia (2.2%–12%; 6/10 trials), and thyroiditis or parathyroiditis (0.54%–8%; 5/10 trials). An immune-mediated inflammatory change in the central nervous system was reported with nivolumab (2%), atezolizumab (4%), and pembrolizumab (1.2%–1.8%). Uveitis was reported with avelumab (2.3%) and pembrolizumab (0.3%). The other imAE were infections, myalgia and hypersensitivity, and multi-organ dysfunctions [Table 4].

Table 4

Immune-related adverse events reported in the included clinical trials of programmed cell death-1/programmed cell death ligand-1 inhibitors in advanced urothelial malignancy

Study ID	Drug	CNS (encephalitis/aseptic meningitis/hypophysitis/)	Eye Uvitis	Thyroiditis/parathyroiditis	Breathing difficulty/pneumonitis	Liver (hepatitis)/autoimmune hepatitis)	Kidney (nephritis)	Adrenal insufficiency	Pancreatitis/autoimmune diabetes/rheumatoid arthritis	Colitis	Mucositis	vasculitis	Peripheral neuritis/	Hematological (anemia, thrombocy topenia, neutropenia)
PD-1 inhibitors as second-line standalone agent
Sharma et al., 2016[20] (n=270)	Nivolumab	NR	NR	8	4	4	3	NR	NR	9	NR	NR	NR	NR
Sharma et al., 2017[21] (n=78)	Nivolumab	2	NR	NR	5	24	1	NR	6	1	NR	NR	NR	15
Bellmunt et al., 2017[27] (n=266)	Pembrolizumab	1.2	NR	11	6.4	NR	1.6	0.8	NR	3.4	NR	NR	1.2	0.4
PDL-1 inhibitors as second-line standalone agent
Rosenberg et al., 2016[22] (n=310)	Atezolizumab	NR	NR	NR	12	4	NR	NR	NR	2	NR	NR	NR	NR
Powles et al., 2018[28] (n=459)	Atezolizumab	4	NR	NR	4	NR	NR	NR	NR	NR	3	NR	NR	6.2
Apolo et al., 2017[25] (n=44)	Avelumab	NR	2.3	6.8	2.3	6.8	NR	NR	2.3	NR	NR	NR	NR	NR
Massard et al., 2016[24] (n=61)	Durvalumab	NR	NR	NR	NR	NR	1.6	NR	NR	NR	NR	NR	NR	NR
Powles et al., 2017[23] (n=191)	Durvalumab	NR	NR	NR	3.6	21.6	1	NR	NR	NR	NR	NR	NR	1.5
PD-1 inhibitors as first-line agent in cisplatin-ineligible
Balar et al., 2017 (a)[29] (n=370)	Pembrolizumab	1.8	0.27	1.81	2.54	14.16	3.89	1.54	6.08	2.27	1.35	0.27	NR	2.27
PDL-1 inhibitors as first-line agent in cisplatin-ineligible
Balar et al., 2017 (b)[30] (n=119	Atezolizumab	NR	NR	10	6	14	2	NR	NR	1	NR	1	NR	11

Figures are expressed are in percentages of the total participants; Miscellaneous AEs include: hypersensitivity, multiple organ dysfunction syndrome, infection (Influenza-like illness, upper respiratory tract infection, pneumonia, urinary tract infection), muscle spasm, muscle weakness, back pain, facial paralysis, lichen planus, hyperhidrosis, tumor flare, stomatitis, myalgia, abdominal pain, decreased weight. PD-1=Programmed cell death-1, PDL-1=Programmed cell death ligand-1, NR=Not reported, CNS=Central nervous system

We tried to analyze the dose-response relationship between the drugs and the AE. The individual PD-1/PDL-1 inhibitor drugs were used as per the doses and intervals recommended by the manufacturers irrespective of second-line or first-line setting. The dosing schedule of nivolumab (3 mg/kg 1-h intravenous [IV] infusion every 2 weeks), pembrolizumab (200 mg IV every 3 weeks), atezolizumab (1200 mg IV administered every 3 weeks), avelumab (10 mg/kg 1-h IV infusion every 2 weeks), and durvalumab (10 mg/kg 1-h IV infusion every 2 weeks) varied significantly. As reported in majority of the clinical trials, the dose discontinuation, reductions and the change in the intervals were at the discretion of the treating physician. These are not described in sufficient detail so as to understand the type of AE leading to alteration in dosing. Therefore, the available evidence is insufficient to provide any meaningful insight in this regard.

Programmed cell death ligand-1 testing methods

The diagnostic tests used for evaluating PDL-1 positivity status in the tumor tissue were (i) PDL-1 IHC 28-8 PharmDx detecting PDL-1 on tumor cells (Dako), (ii) SP-142 detecting PDL-1 on immune cells (Ventana), (iii) PDL-1 IHC 22C3 PharmDx (Dako) (iv) SP-263 (Ventana) detecting on tumor cells, macrophages, and immune cells. They were developed and validated alongside the development of nivolumab, atezolizumab, pembrolizumab, and durvalumab-avelumab, respectively. Two of the tests use rabbit monoclonal antibodies (SP-142 and SP-263) and the other two use mouse antibodies (22C3 and 28-8 pharmDx) All the tests require 4–5 mm thick fresh frozen paraffin-embedded tissue sections. The individual tests used specific cutoff criteria, namely IHC 28-8 PharmDx used 1% (tumor cells); SP-263 used 5% (tumor cells); SP-142 used 1, 1–5 and 5% (immune cells); and PDL-1 IHC 22C3 PharmDx used CPS score 10% or 25%.

Translational biomarker analysis

Exploratory translational biomarker and mutation analyses were reported in four of the ten clinical trials evaluating nivolumab[20] and atezolizumab.[222830] These included histological subtypes, mutation load and high-quality gene expression profiling for relevant helper as well as effector T-cells, their chemokines and cell surface markers. The TCGA histologic types that were likely to have a response to PD-1/PDL-1 inhibitors were (i) basal subtype-1 for nivolumab[20] and (ii) luminal cluster-II subtype for atezolizumab.[2230] The basal subtype-1 patients, who responded to nivolumab had higher interferon-ϒ, CXCL9/CXCL10, and CD8 expression.[22] The response to atezolizumab also correlated with increased trafficking of CD8+ cells and higher CXCL9/CXCL10 expression.[22] Higher median tumor mutation load was associated with a higher rate of response and increased median OS with atezolizumab.[222830]

Publication bias and assessment of heterogeneity

Egger's plot did not reveal significant publication bias [Supplementary Figure 2]. However, Galbraith plot demonstrated significant heterogeneity [Supplementary Figure 3].

Egger's plot for publication bias

Click here for additional data file.

Galbraith plot for heterogeneity

Click here for additional data file.

DISCUSSION

PD-1/PDL-1 inhibitors are undoubtedly the most exciting addition to the therapeutic armamentarium of AdUCs, a field which has not witnessed major breakthroughs for almost three decades. The experience with these new class of agents is short, though they have demonstrated some promise by achieving better ORR in the participants with high PD-1/PDL-1 expression in the tumor tissue, in both, cisplatin-based regimen failure and cisplatin ineligibility. However, when the patients were treated irrespective of their tumor PD-1/PDL-1 expression status, the response rate was only marginally higher in the PD-1/PDL-1 treatment group in the second-line settings (18.2% vs. 12.6%, respectively compared to that of chemotherapy) and in the first-line setting in cisplatin ineligible (ORR in PD-1/PDL-1 treatment group was 23.78%; no comparative data for alternative chemotherapy).

We also evaluated the OS and PFS, since ORR is not universally regarded as the best indicator of anti-cancer drug efficacy, even though the FDA recognizes it as a valid surrogate marker for drug approvals in short-term single-arm clinical trial when there is an unmet need.[15] The median PFS does not appear to be better with PD-1/PDL-1 inhibitors compared to the second-line chemotherapy (1.5–2.9–3.3–4.0 months). The median PFS was found to be similar with PD-1/PDL-1 inhibitors, in first-line and second-line setting (1.5–2.9 vs. 2.0–2.7 months) in the single arm studies. However, the available data indicates that the median OS may be longer with use of these agents as compared to second-line chemotherapy (7.9–18.2 vs. 7.4–8.0 months). Nevertheless, we consider the evidence as weak as it was inconsistent across the studies.[2728] Moreover, the possibility of a selection bias in an open label single arm early phase clinical trials cannot be ruled out. Of the two RCTs, Powles et al. did not find any difference between PFS and OS,[28] but Bellmunt et al. did report a marginal increase in OS with PD-1/PDL-1 inhibitors although the PFS was similar to chemotherapy[27]. This could be due to the non-stringent entry criteria (the patients in the PD-1/PDL-1 treatment groups did not have a cutoff for tumor PDL-1 expression to enter into the clinical trial) or a purely drug specific effect. Overall, there was a notable heterogeneity in the clinical response to the drugs.

It is also unclear at this point, whether the short-term benefit with PD-1/PDL-1 inhibitors would translate into long-term survival benefit if the patients are treated irrespective of the PDL-1 expression status in the tumor tissue. Further, the remaining period of survival beyond the PFS (till the total period of survival, which is reported be 6–16 months or more), may be burdened with the long-term imAE induced by these drugs, further compromising the quality of life. These drugs are expensive; hence a modest gain in PFS with an extended OS with poor quality of life, is likely to reduce quality-adjusted life years (QALY) gained along with an added cost of treatment. None of the clinical trials included in this systematic review and meta-analysis addressed this issue. The QOL as well as QALY are important outcomes which need to be considered before these drugs are designated as the standard of care.

In terms of efficacy outcomes when stratified by PDL-1 expression status, PDL-1 expression on the tumor cells alone or immune cells alone with a cutoff of 1% failed to predict ORR consistently. Even when the cutoff for expression was increased to 5%, there was an inconsistent indication for predictability of response.[222528] CPS seems to have emerged as a better predictor of response to treatment with PD-1/PDL-1 inhibitors in all settings.[232429] However, this biomarker was employed only in studies evaluating pembrolizumab but not the other agents. None of the studies mentioned how many of the patients had null expression status, and their outcomes. Currently, it is believed that lack of expression of PDL-1 should not preclude treatment with these agents as some of the patients with negative expression have also exhibited responses.[34] It could have been interesting to compare the outcomes in PD-1/PDL-1-negative versus positive patients, with respect to safety and survival. Another limitation is that four different testing methods have been developed parallel to the development of PD-1/PDL-1 inhibitors either as companion or complementary diagnostic test. However, skepticism has been expressed over the interchangeability of the tests, as they may differ in their sensitivities.[123435] In addition, there are concerns over the dynamic and heterogeneous nature of the PDL-1 expression, which may vary in different regions of the tumor and between the primary tumor and the metastatic lesions.[363738] The data on PDL-1 testing as a prognostic and predictive biomarker are still evolving. Although there is no consensus at present regarding the best PDL-1 testing method and the cutoff to define positivity, CPS as a predictive biomarker needs further validation in a larger cohort of patients and across all the approved agents.

PD-1/PDL-1 inhibitors have been proposed to be better tolerated due to their targeted action unlike the other immune checkpoint inhibitors such as CTLA-4. CTLA-4 targets the cytotoxic T-lymphocyte antigen reducing the T-cell function at a proximal step unlike the PD-1/PDL-1, which inhibit the T-lymphocyte function at a later stage, which may explain their better tolerability.[39] In this systematic review, approximately 70% of the patients reported treatment-related AEs and about one-fourth of them were serious in nature, though the rate of treatment discontinuation was only 5%–10%. It is worth appreciating that the median duration of treatment was as short as 2.8–3.6 months (reported only in 5 studies); but the total duration of treatment ranged from 0 to 35.5 months. Thereby meaning that the majority of the patients could not tolerate the drug beyond 3–4 months, which itself indicates poor tolerability. Some of the patients could tolerate therapy for as long as 35.5 months, however, it is unclear whether it is the general health of the patients (ECOG performance status) or PDL-1 positivity status or the response to the disease that determines tolerability to the drug. These details have not been reported in the published literature. Similarly, it is unclear at this time that if a dose-response relationship exists for the AEs.

The imAEs were as high as 21% and included several organ systems. Skin rash was the commonest and first to appear, but other more severe ones affecting critical neuroendocrine organs appeared over a period of weeks to months. The course of these imAEs and their association with dose of PD-1/PDL-1 inhibitors is unknown. There is no consensus at this time, whether their onset is an indication to stop further treatment. The available evidence is also inadequate to suggest whether the imAEs have a bearing with the extent of PDL-1 expression similar to efficacy or not. Moreover, there is no data on the adverse effects in tumors without PD-1/PDL-1 expression, which could have helped us to understand the true expression of imAEs. The iatrogenic AEs such as arthralgia and myalgia may also potentially compromise the quality of life during the extended period of survival. Further research is also required to characterize the imAEs and define strategies to manage them effectively. Moreover, all the included studies involved management of UCs as a whole, while tumors at specific sites such as upper tract urothelial tumors or bladder urothelial tumors might have differential effects and efficacy and this could also be one of the future areas of further investigation.

Several biomarkers like the TCGA subtypes (basal subtype-1 and luminal-II); higher tumor mutation load and higher expression of CD8, CXCL9, CXCL10, interferon-ϒ seem to predict better response to the PD-1/PDL-1 inhibitors. However, the data are available from limited number of studies and only for two of the agents. Therefore, further exploration of molecular markers is required to generalize the use of biomarkers for this class of agents. There are several ongoing studies with combination of targeted therapies; concurrent radiotherapy; with extensive biomarker evaluation that may reveal some interesting findings in the future.[40414243444546474849]

In this systematic review, we were unable to apply Grading of Recommendation Assessment, Development, and Evaluation[50] to rate the quality of evidence since the risk of bias for eight of the ten included studies could not be assessed due to unavailability of a validated tool. Drugs and biologic agents being developed for advanced stages of malignancy are invariably tested in nonrandomized single arm interventional studies, where there is no control arm. This is a more intuitive study design in this scenario than a RCT. Since the new drug approvals and subsequent treatment of patients are based on the data obtained from such studies, there is an unmet need to devise an instrument (tool) and validate it to assess the of risk of bias in randomized single arm interventional studies to improve the robustness of systematic review process.

Future research is required to address several aspects of PD-1/PDL-1 inhibitor use in AdUC. Assays for the assessment of PD1/PD-L1 expression on both tumor cells and immune cells need to be standardized. It is imperative to characterize the responder profile with biomarkers upfront before therapy initiation to maximize the benefits obtained. The optimum duration of therapy, discontinuation criteria and imAEs also need more investigation. The impact on the quality of life of the patients who respond to therapy and have long OS even with disease progression need more clarity to make an informed choice about the treatment. The health economics aspects of treatments also need to be better understood.

CONCLUSIONS

PD-1/PDL-1 inhibitors appear to be promising in the treatment of AdUCs with higher PDL-1 expression and the concept of a cutoff value for the same is still evolving. CPS may be considered as a potentially reliable biomarker for predicting response to these agents but need further validation as a universal biomarker for making treatment decisions. Caution needs to be exercised until more data are available on the long-term imAEs and further studies are required to prove their worthiness as a standard of care.