Prescribing Changes After Accelerated vs Regular Approval of Oncology Therapies

Introduction

The US Food and Drug Administration (FDA) has approved more than 200 oncology drug indications under the Accelerated Approval (AA) program.1 Indications granted AA rely on surrogate or intermediate clinical end points, with required postmarket studies to confirm benefit and support conversion to regular approval (RA) or withdrawal. Given inherent uncertainty about AA drugs and variable evidentiary quality supporting AA and subsequent RA, it is critical to understand prescribing responses after AA and subsequent RA.2

Methods

This cross-sectional study included patients who were diagnosed with advanced solid malignant neoplasms from January 1, 2011, to October 31, 2024, and received at least 1 systemic therapy. To obtain drug utilization information, we used patient-level electronic health record data from the Flatiron Health Database, curated through natural language processing–enabled abstraction from approximately 280 US cancer clinics.3 The University of Pennsylvania institutional review board determined that this study was exempt from review and the requirement of informed consent because all data were deidentified. This study followed the STROBE reporting guideline.

Information on 161 AA indications that were granted between January 1, 2011, and July 1, 2023, was obtained from publicly available FDA databses.1 ,4 We identified 29 AA indications in our dataset with at least 30 eligible patients in both the preapproval and postapproval periods (eFigure in Supplement 1). Patients were deemed eligible for an AA indication if they were aged 18 years or older and met cancer-, line of therapy–, and biomarker-specific criteria for an indication.

For each indication, the primary outcome was the mean (SD) percentage point difference in the proportion of eligible patients who started a line of therapy with the corresponding drug in the 6 months before and after AA or RA. For example, if 10% and 25% of eligible patients received the drug before and after AA, respectively, the percentage point difference was 15%. For off-label use, we examined 2 scenarios: (1) use of AA drugs in the first-line setting that were approved for a later-line indication (line-discordant); and (2) use of biomarker-specific AA drugs among biomarker-negative patients (biomarker-discordant). We used t tests to compare the average difference in prescribing changes after AA vs RA using R version 4 (R Project for Statistical Computing). The threshold for significance was a 2-sided P <.05.

Results

The cohort included 63 434 patients (median [IQR] age, 67 [59-75] years; 32 749 female [51.6%]; 47 412 receiving care at community practices [74.7%]) who received 128 917 eligible lines of therapy. Among 16 eligible AA indications (55.2%) converting to RA, prescribing increased 23 percentage points after AA (mean [SD], 6% [15%] pre-AA; 30% [18%] post-AA) and 1 percentage point after RA (mean [SD], 33% [23%] pre-RA; 34% [23%] post-RA) (difference in AA vs RA prescribing change, 22 [95% CI, 19-26] percentage points; P <p .001]) (Figure 1). AA prescribing response varied by indication; alectinib in non–small cell lung cancer had the largest increase after AA (55 percentage points). AA prescribing responses were larger for indications eventually granted vs not granted RA (23 [95% CI, 13-34] vs 7 [95% CI, 3-12] percentage points). Off-label prescribing increases after AA were small (3 percentage points for line-discordant use; 1 percentage point for biomarker-discordant use) (Figure 2).