Comparing rates of immune checkpoint inhibitor toxicity in different treatment settings (neoadjuvant, adjuvant, and palliative) in non-small cell lung cancer (NSCLC): A meta-analysis of phase III clinical trials.

Abstract

8645 Background: Immunotherapy has created a paradigm shift in the treatment of NSCLC. There is growing interest in the utility of immunotherapy alone or in combination with chemotherapy for NSCLC in the peri-operative setting with practice changing results demonstrated in several pivotal phase III clinical trials. Nevertheless, immune related adverse events (irAEs) are the main barrier in the care of patients treated with immune checkpoint inhibitors (ICIs) with grade ≥ 3 irAEs seen in 10-15% of patients. The present meta-analysis aims to compare rates of irAEs across different treatment settings in NSCLC to aid in individual patient treatment decisions. Methods: We extracted data from selected published phase III clinical trials which have led to the approval of PD-1/PDL-1 ICIs in the neoadjuvant, adjuvant, peri-operative and in the palliative setting. We excluded trials which did not have supplemental data on irAEs included with published results and trials of CTLA-4 inhibitors. A sensitivity analysis was performed to assess whether number of cycles correlated with toxicity rates. Results: Pooled data from 15 published phase III NSCLC clinical trials including 5918 patients were included. This included 1 neoadjuvant, 2 peri-operative, 3 adjuvant, and 9 palliative studies. One of the peri-operative trials included data on irAEs which occurred during the neoadjuvant/surgery phase only and this data was pooled within the neoadjuvant group. The ICI drugs included nivolumab, pembrolizumab, atezolizumab, durvalumab, cemiplimab +/- chemotherapy. This included a total of 2523 stage 1B-3C and 3395 stage IV patients. We found a 9.5% rate of grade 3-5 and 35.8% rate of all grade irAEs in palliative trials (n=3395). In contrast, neoadjuvant ICI (n=572) had a 3.5% rate of grade 3-5 and 16.8% of all grades irAEs. Peri-operative ICI (n=797) had a 5.4% rate of grade 3-5 and 24.47% all grades irAEs. Adjuvant ICI (n=1550) had a 7% rate of grade 3-5 and 38.52% rate of all grade irAEs. The relative risk (RR) between the palliative and the neoadjuvant setting of a grade 3-5 irAE is 2.35 (95% CI:1.54 to 3.59; p = 0.0001) and the RR of any irAE is 2.03 (95% CI: 1.69 to 2.43; p=0.0001). The following median number of ICI cycles was seen: neoadjuvant = 3.5, adjuvant = 18, peri-operative = 18, palliative = 10. Lower number of cycles did correlate with lower toxicity in the neoadjuvant setting but this was not seen in other treatment settings. Conclusions: Neoadjuvant, peri-operative and adjuvant ICI’s are associated with a lower rate of irAEs than that seen in palliative trials. A neoadjuvant-only approach is associated with the lowest rates of irAEs and this may be due to the lower number of cycles received. These findings have implications for individual patient treatment decisions and support the use of neoadjuvant immunotherapy in NSCLC.