Background:
Although improvements in overall survival (OS) remain the Oncologist’s treatment goal, because effective post-trial therapy exists, true OS gains are increasingly difficult to quantify. This raises concerns as payers, unlike regulatory agencies, may be less accepting of progression-free survival (PFS) in the absence of validated relationship. Payers quantify life years gained (LYG) to determine cost per LYG; a metric used to establish cost-effectiveness. The AXIS study provides an example of this challenge.

Method:
AXIS study randomised 723 patients (361, axitinib; 362 sorafenib) with metastatic renal cell carcinoma following failure of 1 prior systemic treatment. The primary endpoint was PFS with OS as a secondary endpoint. Study treatment was continued until evidence of disease progression, intolerability or withdrawal of consent. Patients who discontinued randomised treatment could receive subsequent treatment based on investigator-opinion. Patients were followed-up until death.

Results:
From AXIS, median PFS for axitinib was 6.7 months versus 4.7 months with sorafenib (HR = 0.665, 95% CI 0.544 - 0.812; P<0.0001). Median PFS reported for axitinib-treated cytokine refractory patients was 12.1 months and median PFS 4.8 months for axitinib-treated sunitinib refractory patients. Median OS was not reached at the time of analysis as 30% of patients had died, however estimated survival probability at 12 months was similar (66% for axitinib, 67.8% for sorafenib).
Axitinib phase II trials reported 16 months median PFS for axitinib-treated cytokine refractory patients and 7 months median PFS for axitinib-treated sunitinib refractory patients. Median OS reported for axitinib-treated cytokine refractory patients was 30 months and 14 months for axitinib-treated sunitinib refractory patients.

Discussion:
With demonstrated PFS but diluted OS gains, post-progression survival (PPS) is scrutinised. PPS is influenced by factors like crossover and non-randomised post-progression treatments that introduce unmeasurable bias. This requires complex statistical analysis using patient level data. Further research is required in this area.

Conclusion:
An OS gain is diluted in clinical trials due to heterogeneity in SPP. In studies with PFS as the primary endpoint a conservative estimate of OS gain can be estimated, however this is subject to bias if SPP estimations are likely based on nonrandomised, single arm studies.