Quality of Life, Overall Survival, and Costs of Cancer Drugs Approved Based on Surrogate Endpoints

The lack of evidence of a clinically meaningful benefit for many cancer drugs approved by the US Food and Drug Administration (FDA) through expedited pathways raises questions about whether physicians and patients can make informed treatment decisions.¹ Kim and Prasad² reported that for 18 of the 36 cancer drugs that were approved by the FDA from 2008 to 2012 on the basis of a surrogate endpoint—typically, tumor shrinkage or progression-free survival—postmarket studies did not indicate any overall survival (OS) benefit. To determine other potential benefits of these 18 drugs, we analyzed all peer-reviewed findings and FDA review summaries for quality of life (QoL) and calculated the drugs’ annual cost to assess their value.

We obtained a list of the 18 cancer drugs approved by the FDA between 2008 and 2012 (V. Prasad, MD, MPH, and C. Kim, MD, MPH, email communication, December 2015). These drugs were not shown in postmarket studies to significantly prolong OS. Between January 11, 2016, and February 4, 2016, we searched PubMed for clinical trials using the drug’s name, cancer type, and MeSH (Medical Subject Heading) terms quality of life and treatment outcome. We also examined the FDA review summaries on fda.gov.

Because many patients with cancer are 65 years or older, we estimated annual Medicare costs for each drug using a method similar to that described by Bach.³ Calculation of the cost of drugs reimbursed under Medicare Part B (eg, intravenous drugs) was based on the October 2015 ASP Pricing Files,⁴ while calculation of the cost of drugs reimbursed under Medicare Part D (eg, oral drugs) was based on the Full Cost of Drug reported in the Medicare Plan Finder by the Humana Enhanced Prescription Drug Plan for zip code 98155.⁵ The health care costs in this zip code are representative of the average health care costs in the United States. This investigation of published reports was exempt from institutional review board approval.

Our search on PubMed identified 466 articles, 31 of which met our inclusion criteria of a clinical trial with a comparison group using a validated QoL survey instrument. We accessed FDA review summaries, which were publicly available for 15 of the 18 drugs. The QoL studies compared 7 drugs with placebo or observation groups and found that 2 drugs demonstrated worse effects on QoL, 4 drugs showed no statistical difference, and 1 drug had mixed results. Compared with another drug, 1 drug (ofatumumab) reduced OS, while 6 drugs had no OS benefit in comparison with placebo or observation groups.

The average cost of the 6 drugs with no different or worse QoL was $87 922 per year (ranging from $20 237 to $169 836). Despite having no OS or QoL benefit compared with placebo or observation, while presenting significant risks for serious adverse effects, all but 1 of the drugs retains FDA approval. (FDA rescinded bevacizumab for breast cancer in 2011.) Of the 6 drugs with QoL data comparing them to another drug, 1 drug demonstrated better QoL, 2 drugs did not differ significantly from the comparator, and 3 had mixed results. There was no apparent trend in loss to follow-up that could have confounded the results.

The estimated annual costs of the 18 drugs ranged from $20 237 for rituximab to $169 836 for cabozantinib-S-malate, and 13 drugs had annual costs that exceeded $100 000 (Table). The most expensive drug, cabozantinib, did not improve OS and worsened QoL, compared with placebo. The average annual drug costs were similar, regardless of QoL.

Requirements for the FDA’s various expedited pathways are less stringent than for other drug reviews, often including only 1 pivotal trial, fewer patients, shorter follow-up, and surrogate endpoints rather than clinically meaningful outcomes such as OS or QoL. As a condition of these cancer drug approvals, postmarket studies that evaluate OS are usually required.

Unfortunately, the randomized control design of postmarket studies can be lost due to the crossover from the control to treatment group or vice versa, raising questions about the true effect of treatment. For example, most patients in the control group of the crizotinib study subsequently received crizotinib outside the study.⁶ In several studies, when there was no statistically significant increase in OS, post hoc statistical analyses were conducted as an attempt to control for such changes in treatments. Post hoc analysis is inherently subject to confounding, however, especially when treatments are no longer random.

If a new cancer drug does not have a statistically significant OS or QoL benefit, compared with the benefit of other drugs, physicians and patients must weigh the known risks and the costs of treatment choices. However, our analysis indicates that, even when postmarket studies show the new drugs to have no clinically meaningful benefit compared with placebo or observation, most drugs retain FDA approval and remain on the market at prices comparable to those of the most expensive cancer drugs. This situation adds to the skyrocketing costs of cancer care, Medicare, and other health care programs.

Corresponding Author: Diana Zuckerman, PhD, National Center for Health Research, 1001 Connecticut Ave NW, Ste 1100, Washington, DC 20036 (dz@center4research.org).

Published Online: November 29, 2016. doi:10.1001/jamainternmed.2016.7761

Author Contributions: Drs Zuckerman and Rupp had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Zuckerman.

Acquisition, analysis, or interpretation of data: Both authors.

Drafting of the manuscript: Both authors.

Critical revision of the manuscript for important intellectual content: Both authors.

Statistical analysis: Rupp.

Administrative, technical, or material support: Both authors.

Conflict of Interest Disclosures: Dr Zuckerman reported owning stock in Johnson & Johnson. No other disclosures were reported.