Urothelial carcinoma is an aggressive cancer with a high risk of metastatic progression. Chemotherapy plays a key role in the management of metastatic urothelial carcinoma, with, however, no possibility of cure.
1. ICIs in Urothelial Carcinoma: Rationale for Clinical Efficacy
Urothelial carcinoma (UC) appears to be a good candidate for ICIs. First, as already described, neoantigen release represents an important step in the immunogenicity of tumors. The generation of neoantigens that are recognized as ‘non-self’ by immune cells (ICs) could enhance anti-tumor immune response. Production of neoantigens is closely related to mutations occurring during tumor cell proliferation and UC carries one of the highest mutation rates among cancer types
[1][2]. Second, lymphocyte infiltration may be associated with better outcomes in non-muscle invasive UC (NMIUC) and explain why intravesical instillations of bacillus Calmette–Guerin (BCG), by enhancing lymphocyte infiltration and inducing local inflammation, were shown to prevent recurrences
[3]. It was also shown that high levels of tumor-infiltrating CD8+ lymphocytes (TILs) in muscle invasive UC were associated with better disease-free survival and overall survival (OS) than low levels of TILs
[4]. Third, some were showed that levels of PD-L1 expression in vesical UC were correlated with higher-stage, higher frequencies of postoperative recurrence and poorer survival
[5]. Based on these observations, ICIs appear as promising agents in UC, which remains a challenge for clinicians. UC is the most frequent histology and accounts for more than 90% of bladder cancer. Even if 75% of patients present with superficial and NMIUC with an excellent outcome after local treatment, invasion of muscle (MIUC) is associated with worse prognosis and a high rate of recurrence despite multimodal strategies, including cystectomy and perioperative chemotherapy
[6][7][8]. Patients with metastatic UC (mUC) have poor outcomes and, based on historical treatment, an OS that does not exceed 15 months. For many years, cisplatin-based chemotherapy remained the standard treatment in a first-line metastatic setting and second-line options were limited, including single agents, such as paclitaxel, vinflunine and docetaxel
[9][10][11].
2. ICI after Failure of Platinum-Based Therapy: Better Than Chemotherapy
ICIs have been evaluated in late-stage UC disease, after failure of platinum-based therapy; since 2016, five ICIs have been approved in this setting.
Atezolizumab (1200 mg administered intravenously (IV) every 3 weeks) received approval based on the results of the phase II IMvigor210 trial Cohort 2. This non-randomized trial enrolled 310 mUC patients who were heavily pretreated (20% received ≥3 previous chemotherapy regimens). These patients were stratified according to tumor-infiltrating immune cell (IC) PD-L1 = IC0 (<1%), IC1 (≥1% but <5%) and IC2/3 (≥5%). Atezolizumab resulted in an ORR of 16%, including 7% complete response (CR) in all patients. PD-L1 expression was associated with higher response (ORR of 28%, including 15% CR in IC2/3 patients). Interestingly, 10% of responses were also seen in PD-L1-negative patients. The median OS reached was 7.9 months for all patients and 11.9 months for IC2/3 patients
[12]. These first results were impressive in regard to the number of previous treatments these patients received and given that treatment options were limited in UC, reflecting the promising role of ICIs in advanced UC.
The randomized phase III IMvigor211 trial compared atezolizumab to standard second-line chemotherapy (vinflunine, paclitaxel, docetaxel) in 931 mUC patients after failure of platinum-based chemotherapy. The primary efficacy endpoint OS was tested in a hierarchical approach and statistical significance needed to be achieved in the IC2/3 population in order to evaluate statistically the further subgroups, such as the ITT population. Atezolizumab failed to demonstrate improved median OS compared to chemotherapy in IC2/3 PD-L1-expression patients (11.1 vs. 10.6 months; hazard ratio (HR) 0.87;
p = 0.41); there was no difference in the 1-year OS rate (46% vs. 41%, respectively). A moderate but significant difference in OS was observed in the ITT population treated with atezolizumab compared to chemotherapy (8.6 vs. 8.0 months; HR 0.85;
p = 0.038). These disappointing results were mainly explained by the fact that the OS in the chemotherapy arm, and particularly in the vinflunine arm, was higher than expected. However, despite the unmet primary endpoint, this trial confirmed the longer median duration of response (DOR) with atezolizumab compared to chemotherapy in the overall population (21.7 vs. 7.4 months, respectively), confirming the possibility of a long-lasting response in responding patients
[13]. Despite the absence of superiority of atezolizumab compared to chemotherapy and the absence of level 1 evidence, these results suggest that atezolizumab could be proposed after failure of platinum-based chemotherapy in mUC.
Nivolumab was approved based on a non-randomized phase II trial, the Checkmate 275, which enrolled 265 previously treated patients with mUC. Nivolumab (3 mg/kg IV every 2 weeks) resulted in 20% ORR for the total population. PD-L1 expression was associated with higher ORR (28.4% for patients with high (≥5%), 23.8% for patients with ≥1% and 16% for patients with low (<1%) tumor cell (TC) PD-L1 expression. The median OS was 8.74 months in all patients and was higher in patients with PD-L1 ≥1% compared to patients with <1% (11.3 vs. 5.95 months, respectively)
[14].
Pembrolizumab is the only agent that was shown to improve survival in mUC after platinum-based chemotherapy, based on the KEYNOTE-045 trial, a randomized phase III trial that compared the efficacy of pembrolizumab (200 mg IV every 3 weeks for up to 2 years) to chemotherapy (docetaxel, paclitaxel or vinflunine) in 524 mUC patients. Pembrolizumab improved OS in all patients compared to chemotherapy (10.1 vs. 7.2 months, respectively; HR 0.71, 95% confidence interval (CI) 0.59–0.86). There was no significant difference in PFS between pembrolizumab and chemotherapy in all patients (2.1 vs. 3.3 months, respectively; HR 0.98; 95% CI 0.81–1.19;
p = 0.42). Pembrolizumab also increased ORR compared to chemotherapy (21.1% versus 11.4%), with, after a follow-up of more than 5 years, a longer median DOR (29.7 vs. 4.4 months, respectively). The combined positive score (CPS) represents the percentage of PD-L1 in ICs and TCs related to the numbers of TCs; positive CPS (≥10) was not associated with a better OS, PFS or ORR in the pembrolizumab arm. OS benefit with pembrolizumab was observed in all subgroups of patients, regardless of age, ECOG performance status, prior therapy and chemotherapy choice
[15][16]. As usually observed with ICIs, pembrolizumab was better tolerated than chemotherapy, with grade 3–4 toxicities not exceeding 15% (vs. 49.4% with chemotherapy)
[15]. Pembrolizumab thus represents level 1 evidence in a second-line setting, after failure of platinum-based therapy in mUC, regardless of CPS.
Two other ICIs, durvalumab and avelumab, also received FDA approval based on Phase I/II trials. Durvalumab (10 mg/kg IV every 2 weeks) resulted in 191 mUC patients progressing to platinum-based treatment, with an ORR of 18% (3% CR); furthermore, the median OS was 18.2 months in the entire population and was higher in high PD-L1 compared to low or negative patients (20 vs. 8.1 months). Avelumab (10 mg/kg IV every 2 weeks) resulted in similar results with an ORR of 16.5%, including 4.1% of CR and 12.4% of partial response (PR). Avelumab showed clinical benefit in high-risk subgroups, including the elderly, those with renal insufficiency and upper tract patients
[17][18][19][20].
3. Optimizing Biomarker Profiles in UC
All these trials showed that only a proportion of patients could benefit from ICIs and that PD-L1 expression is not able to predict efficacy or inefficacy of ICIs in metastatic UC. Even if a high amplitude of benefit may be expected in PD-L1-positive patients, response can also be observed in PD-L1-negative patients. It was thus to concluded that PD-L1 alone is not the ideal biomarker and in current practice should not be systematically proposed to patients except in cisplatin-ineligible patients in a first-line setting. The major limitation of PD-L1 is the lack of standardization of the PD-L1 assay Different immunohistochemistry assays are used for PD-L1 scoring across the trials, using different cell types (TCs, ICs or combinations of the two), distinct cut-off values for positivity (from 1% to 25%) and distinct antibodies. Even if a retrospective analysis of 235 UC samples showed a relative concordance in PD-L1 staining when using the different antibodies (22C3, 28-8, SP142, E1L3N), standardization in the PD-L1 staining may improve the design of further clinical trials and data collection in retrospective analyses. Furthermore, PD-L1 expression is dynamic during disease evolution, changing with time and therapies, and heterogeneity exists among primary tumors and metastases
[21]. Most importantly, PD-L1 is only a single feature in the tumor microenvironment and could not reflect perfectly the antitumor capacity of the immune system.
Tumor mutational burden (TMB) is relatively high in UC compared with other cancers and reflects the production of neoantigens, which is, as described above, an important step in stimulating anti-tumoral immunity. TMB was also shown to be closely related to the immune microenvironment, suggesting that higher TMB tends to promote T-cell and NK infiltration into the tumor microenvironment
[22]. In patients with NMIUC, a high TMB was significantly associated with a higher response rate and longer recurrence-free survival after BCG
[23].
In cohort 2 of the IMvigor210 trial, TMB was higher in responding patients compared to non-responding patients (12.4 vs. 6.4 per megabase, respectively). In cohort 1 of this trial, high TMB was associated with better OS with atezolizumab compared to low TMB
[12]. In a similar way, in the Checkmate 275 trial, patients with high TMB presented higher ORR (HR 2.13; 95% CI 1.26–3.60), PFS (HR 0.75; 95% CI 0.61–0.92) and OS (HR 0.73; 95% CI 0.58–0.91) compared to patients with low TMB
[24]. The combination of TMB and PD-L1 was also found to have a better predictive value in term of PFS and OS compared to PD-L1 alone; in the IMvigor211 trial, median OS in patients with high TMB and high PD-L1 expression was higher with atezolizumab compared to chemotherapy (17.8 vs. 10.6 months)
[25][26]. More challenging is the use of TMB in clinical practice due to the lack of technical standardization in terms of the cut-off determination and the panel of genes analyzed. The phase II NCT02553642 trial is currently ongoing to evaluate the relationship between TMB and response to nivolumab/ipilimumab in advanced UC.
Tumor gene expression can more accurately describe the immune tumor microenvironment by quantifying chemokines, cytokines, or cell surface proteins than PD-L1 and TMB. The interferon-gamma (IFN-γ) signature measures the expression of up to 25 genes involved in tumor-related inflammation. In the Checkmate 275 trial, the IFN-γ gene signature score was correlated with nivolumab efficacy; among the 59 patients with a high score, 20 presented CR/PR, compared to only 19 CR/PR among the 118 patients with a low/medium score
[14]. Other gene signatures are currently evaluated in order to better identify clusters that could be associated with “hot” tumors, characterized by a high degree of immune infiltration, or “cold” tumors that could represent an immune desert. In the JAVELIN Bladder 100, expression of immune-related genes of the innate and adaptive immune system (CD8, IFN-γ, LAG3, TIGIT and CXCL9) and the number of alleles encoding high-affinity Fc gamma receptor variants predicted higher survival benefit with avelumab maintenance
[27].
In the IMvigor130 trial, the apolipoprotein B editing catalytic polypeptide (APOBEC) signature was associated with a better response with atezolizumab treatment. The APOBEC enzymes are involved in DNA repair processes and are associated with mutation signature in UC. High APOBEC mutational signature was associated with a longer survival in atezolizumab arms (monotherapy or combination plus chemotherapy) compared to chemotherapy alone
[28].
Of course, these gene signatures require validation in prospective trials. The Tumor Cancer Genome Atlas (TCGA) identified five classes of UC, depending on oncogenic mechanisms, infiltration by immune and stromal cells and histological features: luminal–papillary, luminal-infiltrated, luminal, basal–squamous and neuronal
[29][30][31]. Recently, an international consensus molecular classification reconciled the different published classification schemes, including the TCGA. Basal–squamous, stroma-rich tumors and luminal non-specified tumors present an important immune infiltration and could thus be potential candidates for ICIs
[32]. Large prospective trials will be required to evaluate the role of this classification in the prediction of ICI efficacy. The subgroup analysis of JAVELIN Bladder 100 showed that the benefit of maintenance with avelumab compared to BSC was not observed in luminal tumors but was apparent in basal–squamous (24 vs. 17.9 months), luminal infiltrated (19.9 vs. 14.3 months) and luminal–papillary tumors (22.5 vs. 13.4 months)
[33].