Comment on the American Society of Clinical Oncology Value Statement

Abstract

Our concern is that the primary efficacy measures used in the article, such as median overall survival (OS) and progression-free survival (PFS), may not sufficiently summarize the true benefi to f some oncology drugs, especially immuno-oncology agents, and may not adequately address the issues that are foremost in the patient’s mind: What will this treatment mean for me? Will I live longer with an acceptable quality of life or will I not? The measures of median OS and PFS are established, validated benchmarks that provide important information about the impact of treating a large group of patients with a drug or therapeutic regimen. However, depending on the range of outcomes resulting from that treatment, the medians may not always convey the information that is most useful for patients who are deciding whether to invest in a costly course of therapy. Take, for example, the checkpoint inhibitor ipilimumab (BristolMyers Squibb, Princeton, NJ) in metastatic melanoma. The median OS in previously treated patients is approximately 10 months, with modest response rates in the range of 10% to 15%. 2 To be sure, a several-month improvement in OS was an important step forward in the treatment of this disease. But we believe that apatient is equally or even more likely to be interested in knowing that landmark analyses have shown that 18% to 20% of patients survive 5 years or more and that there is a plateau of survival past 3 years. Asking “Is it worth thousands of dollars for a 4-month increase in the OS of patients receiving this treatment?” is different from asking “Is it worth thousands of dollars to increase your chances of surviving 5 years from less than 10% to 18%?” Median survival differences are used to calculate a commonly used measure of value, which is the quality-adjusted life-year. Insurance providers, investigators, and public health agencies have been focused on OS and PFS medians because they have been shown to be excellent surrogates for the benefit of chemotherapeutic and targeted drugs when they are used in large patient populations. For individual patients, however, N is always one, and they may be better informed by focusing on the odds of increased survival instead of the statistical medians. There has been much discussion of achieving a“tail on the curve” of survivalwith immuno-oncology drugs, and recent published studies using the programmed death-1 antibodies pembrolizumab 3 and nivolumab 4 suggest that may indeed be the case. The assessment of variables like PFS and response rate for immune therapies is further complicated by the well-documented unconventional responses seen with these agents, with progression preceding regression and long duration of response in some cases. 5 We do not wish to argue that important measures like median OSandPFS andquality-adjusted life-years should bereplaced inall cases by assessing the benefit of a long-term survival plateau after treatment with an oncology drug or regimen. Rather, we wish to suggest that it be factored into how physicians present treatment options to their patients so that they can make the best-informed decision possible about the value of therapy for their cancer.