OBJECTIVES

Crossover in RCTs might cause bias when estimating the relative treatment effect. Several crossover correction methodologies may be applied, such as rank-preserving structural failure time (RPSFT), inverse probability of censoring weights (IPCW) and the two-stage method. The use of these methods adjusts survival for crossover patients to better reflect clinical practice. The aim was to validate the outcomes of crossover methodologies using RWE.

METHODS

In KEYNOTE-024, pembrolizumab was compared to chemotherapy in stage IV NSCLC with PD-L1 tumor proportion score of ≥50%. The chemotherapy arm (n=151) was corrected for crossover using the two-stage method and was compared to RWE (n=108) of stage IV NSCLC patients treated with chemotherapy. Patient characteristics of the RCT and RWE were compared on age, histology, sex and smoking status. Survival before and after crossover adjustment was compared with RWE survival using the Area Under the Curve (AUC) method up to 30 months.

RESULTS

The baseline characteristics between RWE and the RCT were similar, except for PD-L1 which was ≥50% in KEYNOTE-024, but not reported for RWE. Survival of the crossover-adjusted arm and RWE were almost identical up to 10 months. Hereafter, crossover-adjusted survival flattened, whereas RWE survival did not. The AUC resulted in 11.75 months for RWE, whereas crossover-adjusted and crossover-unadjusted survival demonstrated an AUC of 13.72 (∆AUC=1.97), and 16.01 (∆AUC=4.26) months, respectively. This shows that the crossover-adjusted chemotherapy arm reflects clinical practice more compared to the non-adjusted arm.

CONCLUSIONS

It seems that the crossover-adjusted chemotherapy arm overestimates survival compared to clinical practice. Further research is needed to determine the cause, which might be due to differences in inclusion and exclusion criteria of the RCT and RWE study, such as PD-L1 status, limited patient numbers in the tail, or methodological issues regarding the two-stage method. Validation against RWE remains important for clinically plausible survival.