OS is considered the gold standard trial endpoint, particularly for health technology assessment. However OS faces challenges – from subsequent therapy bias to needing long trials that delay patients’ access to promising medicines. PFS is often used either instead of, or alongside, OS – by regulators, clinicians, payers, and more recently, value frameworks in oncology. PFS is without a standardized measure. We examine the extent of differences between independent central review (ICR) and investigator assessed (INV) PFS. We aim to increase understanding of potential variability in PFS measurement, relevant associations and possible causal factors to inform appropriate use of PFS in payer and clinician decision-making.
We searched Clinicaltrials.gov for ‘progression free survival’ and ‘cancer’, filtering for interventional phase 2 or 3 studies with results. Studies were extracted and the primary and secondary outcomes filtered for ICR and INV based PFS. We searched PubMed with the same criteria; full articles were reviewed and studies reporting for ICR and INV based PFS included. For comparative trials, we calculated difference in median PFS between intervention and control arms for ICR and INV based PFS. For single arm trials, the difference between ICR and INV based PFS was calculated where both were reported.
Of 365 studies from clinical trials.gov; 48 reported ICR based PFS and 45 reported INV. 6 studies reporting both were included. Of 49 studies from PubMed; 21 were included. There was 1 duplicate. The majority of studies were comparative (23/26), in solid tumors (21/26), and published in the last 5 years (21/26).
Calculating the PFS gain at the median, the difference between the ICR based gain and the INV based gain ranged from 0.1 to 4.3 months. In 9 comparisons the gain with ICR was greater than the gain with INV. In 6 comparisons the difference in PFS gain was ≥2 months, a difference of up to 54% of the gains alone.
ICR and INV based PFS produce different estimates of PFS gain in clinical trials, but it remains uncommon for studies to report both ICR and INV based PFS. Both measures should be required, to improve consistency of comparison across trials and transfer of trial results to real world practices and decision-making.
Clinical trial identification
Legal entity responsible for the study
PRMA Consulting Ltd
C.F. Jones, J.F. Soto Barrientos, G. Monnickendam: Employee of PRMA Consulting who sponsored this research.