Addressing challenges with real-world synthetic control arms to demonstrate the comparative effectiveness of Pralsetinib in non-small cell lung cancer

Abstract

As advanced non-small cell lung cancer (aNSCLC) is being increasingly divided into rare oncogene-driven subsets, conducting randomised trials becomes challenging. Using real-world data (RWD) to construct control arms for single-arm trials provides an option for comparative data. However, non-randomised treatment comparisons have the potential to be biased and cause concern for decision-makers. Using the example of pralsetinib from a RET fusion-positive aNSCLC single-arm trial (NCT03037385), we demonstrate a relative survival benefit when compared to pembrolizumab monotherapy and pembrolizumab with chemotherapy RWD cohorts. Quantitative bias analyses show that results for the RWD-trial comparisons are robust to data missingness, potential poorer outcomes in RWD and residual confounding. Overall, the study provides evidence in favour of pralsetinib as a first-line treatment for RET fusion-positive aNSCLC. The quantification of potential bias performed in this study can be used as a template for future studies of this nature.

Conflict of interest statement

S,P. receives honoraria from Boehringer Ingelheim, AstraZeneca, Roche, Takeda and Chugai Pharma; and provides consulting or advisory role for Boehringer Ingelheim, AstraZeneca, Roche, Takeda, Novartis, Pfizer, Bristol-Myers Squibb, MSD, Guardant Health, AbbVie and EMD Serono. Dr Liu has received research funding (to institution) from Alkermes, AstraZeneca, Bayer, Blueprint Medicines Corporation, Bristol-Myers Squibb, Corvus, Debiopharm, Elevation Oncology, Genentech, Lilly, Lycera, Merck, Merus, Pfizer, Rain Therapeutics, RAPT, and Turning Point Therapeutics; has served as consultant or advisory board member to Amgen, AstraZeneca, BeiGene, Blueprint Medicines Corporation, BMS, Catalyst, Daiichi Sankyo, G1 Therapeutics, Genentech/Roche, Guardant Health, Inivata, Janssen, Jazz Pharmaceuticals, Lilly, Merck/MSD, PharmaMar, Pfizer, Regeneron, and Takeda. Mr Scheuer reported receiving personal fees from Roche, receiving shares from Roche as an employee during the conduct of the study, and reported being an employee of and receiving shares from Novartis outside the submitted work. G.G.H. and A.L. reported receiving funding from Roche during the conduct of the study. S.V.R. reported receiving personal fees from Roche during the conduct of the study. F.G. has consulted or provided expert opinion for AMGEN, AstraZeneca, Bayer, BMS, Boehringer Ingelheim, Celgene, GSK, Lilly, MSD, Novartis, Pfizer, Roche, Siemens, and Takeda; has received fees from Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, BMS, Celgene, GSK, Lilly, MSD, Novartis, Pfizer, Roche, Siemens, and Takeda; and has received funding for scientific research from Amgen, AstraZeneca, Boehringer Ingelheim, BMS, Celgene, GSK, Lilly, MSD, Novartis, Pfizer, Roche, Siemens, and Takeda. V.S. reports research funding/grant support for clinical trials from AbbVie, Agensys, Alfa-sigma, Altum, Amgen, Bayer, Berg Health, Biotherapeutics, Blueprint Medicines Corporation, Boston Biomedical, Boston Pharmaceuticals, Celgene, D3, Dragonfly Therapeutics, Exelixis, Fujifilm, GSK, Idera Pharma, Incyte, Inhibrx, Loxo Oncology, Medimmune, MultiVir, Nanocarrier, National Comprehensive Cancer Network, NCI-CTEP, Novartis, Northwest Biotherapeutics, Pfizer, PharmaMar, Roche/Genentech, Takeda, Turning Point Therapeutics, UT MD Anderson Cancer Center, and Vegenics; travel support from ASCO, ESMO, Helsinn, Incyte, Novartis, and PharmaMar; consultancy/advisory board participation for Helsinn, Incyte, Loxo Oncology/Eli Lilly, Medimmune, Novartis, R-Pharma US, QED Pharma, and other relationship with Medscape. V.S. is also an Andrew Sabin Family Foundation Fellow at The University of Texas MD Anderson Cancer Center, acknowledges support of The Jacquelyn A. Brady Fund, is supported by NIH grant R01CA242845. MD Anderson Cancer Center Department of Investigational Cancer Therapeutics is supported by the Cancer Prevention and Research Institute of Texas (RP1100584), the Sheikh Khalifa Bin Zayed Al Nahyan Institute for Personalized Cancer Therapy (1U01 CA180964), NCATS Grant UL1 TR000371 (Center for Clinical and Translational Sciences), and the MD Anderson Cancer Center Support Grant (P30 CA016672).

© 2022. The Author(s).

Figures

Fig. 1. Kaplan–Meier curves for TTD, OS, and PFS for comparisons between the 1 L pralsetinib trial cohort and 1 L pembrolizumab cohort, and 1 L pralsetinib trial cohort and 1 L pembrolizumab with chemotherapy cohort.

A–C The Kaplan–Meier curves are for each endpoint TTD, OS, and PFS panels respectively for the comparison between 1 L pralsetinib versus 1 L pembrolizumab (ESS = 109 for the pralsetinib cohort, and ESS = 115 for the pembrolizumab cohort), and D–F for the comparison between 1 L pralsetinib versus 1 L pembrolizumab with chemotherapy after IPTW-adjustment (ESS = 109 for the pralsetinib cohort, and ESS = 217 for the pembrolizumab with chemotherapy cohort); the median OS for the pralsetinib cohorts could not be computed.

Fig. 2. Bias plots showing unmeasured confounding for comparisons between the 1 L pralsetinib trial cohort and 1 L pembrolizumab cohort, and 1 L pralsetinib trial cohort and 1 L pembrolizumab with chemotherapy cohort.

A Bias plots for unmeasured confounding corresponds to the comparison with 1 L pembrolizumab comparison (HR 0.38, 95% CI 0.21–0.67), B corresponds to the comparison involving 1 L pembrolizumab with chemotherapy comparison (HR 0.37, 95% CI 0.21–0.64). These graphs plot unconfounded treatment effect estimates as risk ratios (ARR adjusted risk ratio) after adjusting for a hypothetical unmeasured binary confounder over a range of confounder-exposure and confounder-outcome associations on the risk ratio scale. The colors map the strength of an unmeasured confounder (x and y axes) to the robustness of this study’s conclusions (color gradient). The worst-case strengths of measured baseline confounders are shown using HRs from the multiple imputation resulting from QBA for missing data assumptions.

Fig. 3. Kaplan–Meier curves for OS showing the robustness of the estimated hazard ratios for the comparison between 1 L pralsetinib trial cohort and 1 L pembrolizumab cohort, and 1 L pralsetinib trial cohort and 1 L pembrolizumab with chemotherapy cohort when the RWD cohorts are transformed to have increased OS.

A The Kaplan–Meier curves for OS correspond to the comparison between 1 L pralsetinib versus 1 L pembrolizumab (N = 637 for the digitised KEYNOTE cohort, ESS = 109 for the pralsetinib cohort, ESS = 115 for the pembrolizumab cohort when untransformed and at the transformation threshold), and B corresponds to the comparison between 1 L pralsetinib versus 1 L pembrolizumab with chemotherapy (N = 410 for the digitised KEYNOTE cohort, ESS = 109 for the pralsetinib cohort, ESS = 217 for the pembrolizumab with chemotherapy cohort when untransformed and at the transformation threshold) after QBA of the HRs; red indicates the digitised curve from the corresponding KEYNOTE trial (KEYNOTE-42 for pembrolizumab and KEYNOTE-189 for pembrolizumab with chemotherapy), light green indicates the untransformed weighted KM curve for the EDM cohort, dark green is the weighted KM curve for the EDM at the transformation threshold where the adjusted HR remains significant, and blue indicates the pralsetinib group’s weighted KM curve.

Fig. 4. Flowcharts describing the patient selection process for the cohorts drawn from the ARROW trial, Flatiron Health CGDB, and Flatiron Health EDM datasets.

A The flowchart corresponds to the ARROW trial, B to the Flatiron Health CGDB, and C to the Flatiron Health EDM datasets.