Solid lines connect monthly rates of utilization among patients initiating first-line therapy each calendar month. The blue dotted line marks June 1, 2018, as the time of US Food and Drug Administration (FDA) label change. Programmed death-ligand 1 (PD-L1) testing indicates the fraction of patients initiating first-line therapy who received at least 1 PD-L1 test prior to the start date of their first-line therapy. Immunotherapy includes nivolumab, pembrolizumab, atezolizumab, avelumab, and durvalumab; chemotherapy includes cisplatin- and carboplatin-based therapies.
Generalized logistic regression model adjusting for sex, age, race, history of smoking, region, Eastern Cooperative Oncology Group performance status, practice type, and time (indexed to start date of first-line therapy), showing probabilities of first-line immunotherapy, chemotherapy, and programmed death-ligand 1 (PD-L1) testing utilization. The blue dotted lines mark the time of the US Food and Drug Administration (FDA) label change. Dashed lines are extrapolations from the model as if there had been no label change. Race was included as a covariate since previous studies have documented differential treatment patterns and outcomes in bladder cancer by race. Race was self-reported by patients or classified by clinicians using the following options defined by the electronic health record: Asian, black or African American, Hispanic or Latino, non-Hispanic white, or other race. The marginal effect was the predicted change in percentage of patients using first-line immunotherapy, chemotherapy, and PD-L1 testing (prior to first-line therapy start), as the mean percentage-point difference between the observed (with FDA label change) and expected (if no FDA warning) utilization on January 31, 2019. Immunotherapy and chemotherapy models include an interaction between FDA label change and time.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Parikh RB, Adamson BJS, Khozin S, et al. Association Between FDA Label Restriction and Immunotherapy and Chemotherapy Use in Bladder Cancer. JAMA. 2019;322(12):1209–1211. doi:10.1001/jama.2019.10650
Approximately half of patients with advanced bladder cancer are ineligible for standard management with cisplatin-based chemotherapy. In 2017, based on single-group phase 2 studies, the US Food and Drug Administration (FDA) granted accelerated approvals to 2 anti–programmed cell death protein (PD-1)/programmed death-ligand 1 (PD-L1) immunotherapies, pembrolizumab and atezolizumab, for first-line treatment of any cisplatin-ineligible patients.1,2 However, data from ongoing phase 3 trials showed that patients with PD-L1–negative tumors who received immunotherapy had decreased survival relative to those receiving platinum-based chemotherapy. In June 2018, the FDA limited the indication for the 2 first-line immunotherapies to cisplatin-ineligible patients with PD-L1–positive tumors. The FDA’s decision to restrict the indication after accelerated approval based on early review of confirmatory trial data was unique.3 Given the rapid uptake of immunotherapies in oncology,4 whether or how quickly the FDA label restriction would affect clinical practice was unclear.
This interrupted time-series analysis included patients diagnosed as having advanced bladder cancer between January 1, 2016, and January 31, 2019, who were treated with at least 1 line of systemic therapy. We used data from the Flatiron Health database, derived from deidentified electronic health records of commercially and publicly insured patients with cancer receiving care in more than 280 geographically diverse US clinics.5 We excluded patients who received therapies not listed in National Comprehensive Cancer Network guidelines for bladder cancer.
Outcome utilization rates were calculated among patients initiating first-line therapy each calendar month. Multivariable logistic regression models were fit for 3 outcomes (utilization of first-line immunotherapies, cisplatin- or carboplatin-based chemotherapy, and PD-L1 testing), adjusted for race, age, smoking status, performance status, region, practice type, and year of diagnosis. Each model included an exposure variable indicating whether first-line treatment started before or after June 1, 2018, and an interaction variable with time. Because oncologists may use immunotherapies interchangeably, all immunotherapies with bladder cancer indications available at the time (pembrolizumab, atezolizumab, nivolumab, avelumab, and durvalumab) were included.
We estimated the average marginal effect of the FDA label change on the rates of each outcome evaluated on January 31, 2019 (8 months after label change). Marginal effects represent the absolute percentage-point difference between the probabilities of each outcome associated with the label change and the predicted counterfactual had there been no label change. The 95% confidence intervals for marginal effects were based on 1000 bootstrap simulations. All tests were 2-sided with α = .05 using R, version 3.3.2 (R Foundation for Statistical Computing). The Copernicus Group and University of Pennsylvania institutional review boards approved this study and granted waivers of informed consent.
Among 1965 patients with advanced bladder cancer (n = 645 initiating first-line immunotherapy; n = 1147 initiating first-line chemotherapy), the median age was 73 years, 26% were women, 74% were white, and 94% received care at community practices.
The FDA label change was associated with reversals in immunotherapy and chemotherapy trends and increased PD-L1 testing (Figure 1). Between May 2018 and January 2019, the unadjusted rate of immunotherapy use decreased from 51.9 to 30.3 per 100 patients, while rates of chemotherapy use increased from 37.0 to 60.6 and PD-L1 testing increased from 9.3 to 21.2 per 100 patients. As of January 31, 2019, after adjusting for patient and practice factors, the FDA label change was associated with a decrease in immunotherapy use (marginal effect, −37.4%; 95% CI, −32.7% to −42.1%) and an increase in chemotherapy use (marginal effect, 34.4%; 95% CI, 28.8%-40.0%) and in PD-L1 testing (marginal effect, 12.7%; 95% CI, 9.4%-15.8%) (all P < .001) (Figure 2).
The FDA label restriction on first-line immunotherapy was associated with a decrease in immunotherapy use and an increase in chemotherapy use and PD-L1 testing. Given the rapid growth of oncology therapies receiving accelerated approval,6 how regulatory responses to postmarket safety events affect use of such drugs is important. This study suggests that the FDA label changes were associated with changes in practice, even when the decision was based on emerging trial data. Limitations include that clinician adherence to the revised label change could not be determined, and other factors, including evolving clinician preferences, could contribute to trends seen. The effect of the FDA label change on clinical outcomes requires further study.
Accepted for Publication: June 28, 2019.
Corresponding Author: Ravi B. Parikh, MD, MPP, Abramson Cancer Center, 3400 Civic Center Blvd, Twelfth Floor, Philadelphia, PA 19104 (email@example.com).
Author Contributions: Drs Parikh and Adamson had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Parikh and Adamson contributed equally to this work. Drs Cohen and Mamtani served as co–senior investigators.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Parikh, Adamson, Khozin, Baxi, Cohen, Mamtani.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Parikh, Adamson, Baxi, Cohen, Mamtani.
Obtained funding: Parikh, Mamtani.
Administrative, technical, or material support: Parikh, Adamson, Khozin, Galsky, Baxi.
Supervision: Parikh, Adamson, Mamtani.
Conflict of Interest Disclosures: Drs Adamson, Baxi, and Cohen are employees of Flatiron Health, an independent Roche subsidiary. Roche manufactures atezolizumab. Dr Adamson reported receiving personal fees from the Institute for Disease Modeling, Intellectual Ventures, and Salutis Consulting. Dr Galsky reported receiving grants and personal fees from Roche, Merck, AstraZeneca, and Bristol-Myers Squibb and personal fees from Seattle Genetics, Astellas, Dracen, Dragonfly, Pfizer, EMD Serono, Incyte, Aileron, Inovio, Lilly, Janssen, and BioMotiv. Dr Baxi reported receiving personal fees from Flatiron Health. Dr Cohen reported receiving support from Flatiron Health and Roche. Dr Mamtani reported receiving personal fees from Roche. No other disclosures were reported.
Funding/Support: This work was supported by grants from the Conquer Cancer Foundation (Dr Parikh) and grants 5-T32-CA009615 (Dr Parikh) and K23-CA187185 (Dr Mamtani) from the National Cancer Institute of the National Institutes of Health.
Role of the Funder/Sponsor: The funders and Roche Group had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Melissa Lynn Estévez, MS, supported protocol development and coding of the analysis. Nicholas R. Jones, BA, and Lucy Dao-Ke He, BA, contributed to preliminary analyses. Shreyas Lakhtakia provided data visualization. Neal Meropol, MD, provided comments on the manuscript. All are employees of Flatiron Health and did not receive compensation for their roles.