Should Drug Approvals Be Based on Progression-Free Survival?

What are the costs and benefits of approving drugs based on PFS?

Cancer drugs are often approved on the basis of progression-free survival (PFS) improvements, even though this is considered a surrogate endpoint for more robust clinical outcomes. A new study using non–small-cell lung cancer (NSCLC) as an example found varying degrees of costs and benefits associated with granting early access to drugs based on PFS-in many cases, the costs outstripped the benefits.

“In drug approval and coverage decisions, waiting for the most complete data on long-term effectiveness can delay access by years,” wrote study authors led by Darius N. Lakdawalla, PhD, of the University of Southern California in Los Angeles. While overall survival (OS) is generally considered the “gold standard” outcome, using PFS can be measured far more quickly.

“PFS is not perfectly correlated with OS,” the authors wrote. “In some cases, drugs with PFS benefits fail to show OS benefits.” To weigh the benefits of using PFS, the authors created a probabilistic decision model based on a hypothetical population of NSCLC patients. Results of this modeling analysis were published online ahead of print in JAMA Oncology.

They used 27 published clinical trials in order to conduct the analysis, and calculated an expected OS benefit conditional on PFS greater than 0, 1, 2, and 3 months. None of the studies used had a PFS benefit above 3.52 months, so 3.5 months was considered an upper bound for this analysis.

In the 27 trials, a positive PFS outcome correctly predicted a positive OS outcome 71% of the time. Among those trials, the mean OS benefit was 1.25 months. In the other “failed” trials, there was a mean OS reduction of 0.14 months.

Depending on the number used for “value of a statistical life year” (VSLY), the benefit of using PFS changed. For example, the medium-value parameter of $200,000 per VSLY, and $22,000 for incremental treatment cost, granting early access to any drug with positive PFS benefit is worse than waiting for the OS results. However, if you shift the cutoff to 1 month of PFS benefit, it does become beneficial to grant early access to the drug.

If 3 to 3.5 months of PFS benefit are used, the results are essentially always positive and worth granting access to the drug; with this parameter, access would generate incremental social value of between $38,000 and more than $1 million per newly treated patient per month.

“Our calculations reflect the societal perspective,” the authors wrote. “The payer perspective will tilt away from early access based on PFS, whereas the patient perspective will tilt toward early access.”

In the case of NSCLC, the authors noted that granting early access for all drugs with positive PFS only makes sense in cases where incremental treatment costs is toward the lowest end of the range. “The use of PFS in early-access decisions holds both risks and rewards that must be carefully and systematically balanced by society,” they concluded.

In an accompanying editorial, authors led by Yu Shyr, PhD, of Vanderbilt University, pointed out that different types of agents might require slightly different forms of this cost-benefit analysis. Chemotherapy agents, for example, may need that 1-month PFS benefit cutoff to justify early access, whereas development of targeted agents “may benefit from a decision model that incorporates any superiority in PFS, with non-inferiority in OS.”