*2.4. Durvalumab (Imfinzi; AstraZeneca;Cambridge, United Kingdom)*

#### 2.4.1. Trial

Durvalumab is a human mAb that binds PD-L1 and also provided encouraging results on clinical response with respect to the tumor expression status of PD-L1 and was approved in 2017. The phase I/II 1108 trial evaluated patients who had received prior platinum-based chemotherapy [48]. At the median follow-up time point of 4.3 months, the ORR was 31% (95% CI, 17.6–47.1%) in all patients. PD-L1 expression was performed using the Ventana PD-L1 (SP263) assay. The proportion of tumor cells with PD-L1 membrane staining was partitioned based on defined intervals of <1%, 1–4%, 5–9%, 10–14%, ... , 90–99%, 100%. [49] Strikingly, patients in the PD-L1-positive cohort, defined as ≥25%, had an ORR of 46.4% (95% CI, 27.5–66.1%). In contrast, patients with a score of <25% were considered as PD-L1- negative cohort and had an ORR of 0% (95% CI, 0.0–23.2%).

However, updated results with a median follow-up of 5.8 (0.4–25.9) months showed that the ORR was now 17.8% (95% CI, 12.7–24.0%) in all patients [31], much lower than previously reported in the trial that led to its approval (Table 1) [48]. In addition, the ORR in the PD-L1-positive cohort dropped from 46.4% to 27.6% (95% CI, 19.0–37.5%). Nonetheless, PD-L1-positive patient ORR was still notably higher than in patients that were PD-L1-negative 5.1% (95% CI, 1.4–12.5%). The median OS was 18.2 (8.1—not estimable) months in the total population. The OS was 20.0 (11.6—not estimable) months and 8.1 (3.1—not estimable) months for the PD-L1-high and PD-L1-negative expressing cohorts, respectively. The 1-year survival rate was 55% (44–65%) in all patients. Survival rates for PD-L1-expression subgroups were 63% and 41% for PD-L1-high and PD-L1 low/negative groups respectively. Based on these first results, the FDA granted accelerated approval to durvalumab for the same indication as described for nivolumab [50]. This study was significant as, for the first time, the results indicated that patient tumors should most likely contain substantially high levels of PD-L1. PD-L1 expression cut-offs at 1–10% may be insufficient for identifying patients who will respond to ICI therapy.

Durvalumab is currently being investigated in combination with the ICI mAb tremelimumab that targets CTLA-4 (another immune checkpoint receptor) in a few different clinical trials and is reviewed in [51]. Unfortunately, in one significant trial, the phase III DANUBE trial evaluating durvalumab plus tremelimumab in unresectable, metastatic bladder cancer patients did not meet the primary endpoints for improving OS versus standard-of-care chemotherapy [52]. The trial is evaluating the efficacy of durvalumab in the first-line treatment of both cisplatin-eligible and -ineligible patients with metastatic bladder cancer. The trial arms are durvalumab monotherapy, durvalumab plus tremelimumab, and cisplatin and gemcitabine or carboplatin and gemcitabine chemotherapy. In addition, patients whose tumors were PD-L1-positive did not benefit from durvalumab plus tremelimumab. This was surprising since high PD-L1 was set at the high cut-off of ≥25% of only tumor cells expressing PD-L1. Taken together, these results are puzzling. On one hand, durvalumab effectiveness is associated with PD-L1 expression as a monotherapy in the second-line but not in combination with an additional non-overlapping ICI in patients with metastatic bladder cancer in a first-line setting. The DANUBE trial is a post-approval commitment from AstraZeneca in agreement with the FDA from the accelerated 2017 approval, and it is unclear what actions regarding its approval and/or use will follow. Result details have yet to be published or presented at the time of this review.
