br The primary limitations of this
The primary limitations of this trial are retrospective design and the low number of centers involved (2 centers). Another limitation could be the LY294002 that higher rate of IRAEs reflects higher treatment exposure and no T-cell activation. To define this point,
we analyzed the number of IRAEs according to ICI exposure. The incidence of ORAEs was higher when ICI exposure was > 3 months than when ICI exposure was < 3 months. However, after 3 months of exposure, we found approximately the same rate of pa-tients with (52%) and without (48%) IRAEs. This trend was confirmed above 6 months of exposure, with 54% patients with IRAEs and 46% without IRAE. These results suggest that higher rate of IRAEs does not reflect higher treatment exposure, suggesting higher OS and ORR are associated with IRAE occurrence more than ICI exposure. Furthermore (Table 5), we found no statistically significant difference in OS and ORR in patients who had IRAEs before 3 months and after 3 months. Patients with shorter ICIs exposure and earlier IRAEs did not have better outcomes. Prospective trials are needed to confirm these trends. One of them, published in December 2017,18 enrolled 43 patients treated with nivolumab in Japan, and found a trend towards an association between ORR and PFS and the time from the first infusion of nivolumab to IRAE occurrence. The difference between these
Table 3 Association Between IRAE Grade and Immunotherapy Efficacy
Abbreviations: CI ¼ Confidence interval; DCR ¼ disease control rate; HR ¼ hazard ratio; IRAE ¼ immune-related adverse event; NR ¼ not reached; OR ¼ odds ratio; ORR ¼ overall response rate; OS ¼ overall survival; PFS ¼ progression-free survival.
Mathieu Grangeon et al
Table 4 Association Between OS or PFS and IRAE Subtype
Abbreviations: CI ¼ Confidence interval; HR ¼ hazard ratio; IRAE ¼ immune-related adverse event; NR ¼ not reached; OS ¼ overall survival; PFS ¼ progression-free survival.
results and ours may be explained by the different population (Asian vs. Caucasian) and the type of IRAEs selected (mainly rashes, pyrexia, and diarrhea in the Japanese study).
The global efficacy of ICIs was consistent with the phase III pivotal trial results. In the overall population, the OS rate was 14 months versus 12.2 months in Checkmate 057,10 9.2 months in Checkmate 017,9 10.4 months and 12.7 months (for pem-brolizumab 2 mg/kg and 10 mg/kg, respectively) in Keynote 010,11 and 13.8 months in the OAK trial.6 The PFS was 2.6 months in our study versus 2.3 months in Checkmate 057,10 3.5 months in Checkmate 017,9 3.9 months and 4 months in Keynote 010 for pembrolizumab 2 mg/kg and 10 mg/kg, respectively,11 and 2.8 months in the OAK trial (Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial).6 The ORR was 13% in our study versus 19% in Checkmate 057,10 20% in Checkmate 017,9 18% in Keynote 010,11 and 14% in the OAK trial.6 This difference in ORR could be observed because we analyzed a real-life population, in comparison with over-selected patients enrolled in clinical trials. Thereby, these results are inter-esting because Archean/Proterozoic Era are relevant to everyday patients.
Adverse events were reported for 58% of patients in Checkmate 057,10 63% of patients treated with pembrolizumab 2 mg/kg and 66% of patients treated with pembrolizumab 10 mg/kg in Keynote 010,11 and 64% of patients in OAK.6 The toxicity rate was lower in the present study (46%) because only IRAEs were reported. The incidences of the different IRAEs were similar in the present study and the Keynote, Checkmate 057, Checkmate 017, and OAK trials. Thyroid dysfunction was observed in 13% to 19.5% of patients
There is still a lack of data or conflicting results on how to predict the response to ICIs. A first retrospective trial published in September 2017 in JAMA16 reported data from 134 patients treated with nivolumab. A total of 51% of patients had IRAEs, with a sta-tistically significant longer PFS and OS in patients with IRAEs. However, no detail was given about IRAE definition and diagnosis. In a prospective trial published in Lung Cancer in 2018,17 with 38 patients treated with nivolumab, a 30% incidence of IRAEs was found, with a better ORR for patients with IRAEs in comparison with patients without IRAEs (63% vs 7%). These 2 studies were performed in Asian patients, which could have a different toxicity profile for ICIs than for the Caucasian patients analyzed in our study. A retrospective analysis20 of 160 patients with malignant diseases (including 44% [n ¼ 71] of patients with NSCLC) reported IRAEs in 40% of patients. No difference was found for the OS rate in the overall population; however, 2 subgroups were analyzed separately: patients in clinical trials (46%; n ¼ 73) and patients off trials (54%; n ¼ 87). For the patients in clinical trials, the ORR was better for the patients with IRAEs compared with patients with no IRAE (P ¼