ICIs have revolutionized the modern treatment of advanced NSCLC. Survival advantages
have been reported in up to 20% to 40% of the study populations in various studies.
Kaplan-Meier curves and Cox proportional hazard analysis have been used to report
analysis in all the major clinical trials which have eventually led to Food and Drug
Administration approvals of ICI. This work suggests that estimations of hazard ratio
(HR) might not be representative of the effect size as disproportionately long plateau
tails of Kaplan-Meier curves may sway ICI trial results. To overcome this, Cox-TEL
adjustment method was used and data of nine major ICI trials were reevaluated to calculate
the survival benefits of ICIs over chemotherapy. The proportional hazard model divides
the study population in two groups, short-term (ST) survivors and long-term (LT) survivors,
and the differences in proportions (DPs) and HRs for ST and LT survivors were calculated.
The DP, which was representative of the incremental gain in LT survival probability,
was greatest in patients with programmed death-ligand 1 (PD-L1) expression of 1% or
more treated with nivolumab, whereas the ST-HR was nonsignificant in eight of nine
trials. Pooled Cox HR, ST-HR, and LT-DP were 0.74 (95% confidence interval [CI]: 0.68–0.82),
0.88 (95% CI: 0.82–0.94), and 0.09 (95% CI: 0.06–0.12), respectively, for NSCLC cases
with PD-L1 less than 1%, and none of the trials revealed any statistical significance
for ST-HR. In second-line setting among NSCLC cases with PD-L1 greater than or equal
to 1%, ST-HR consistently had ICI benefit with pooled ST-HR of 0.84 (95% CI: 0.78–0.91).
One of the main limitations of the analysis was the lack of LT follow-up data in most
studies. Nevertheless, this work highlights the limitation of the traditional Cox
HR model to predict survival benefit when the effect of treatment is not constant
To read this article in full you will need to make a payment