Target wise and pound foolish: A simple technique to evaluate the trade-off between clinical benefit and economic burden of monoclonal antibodies

Purpose: The cost of cancer therapies continues to increase, raising concerns of distributive justice and sustainability of our health-care system. We propose a simple set of cost effectiveness calculations using the concept of number needed to treat (NNT) to frame this debate in objective terms for practicing physicians and policy makers. We then apply this strategy to all monoclonal antibodies approved for adult hematology/oncology indications as of July 2015. Methods: Monoclonal antibodies were identified from the "FDA Approved Drug Products". Approved hematology/oncology indications for each antibody were extracted from package inserts. All comparative studies used to support each drug's approval for each indication were analyzed. Overall survival (OS) was our preferred outcome measure. If OS data was not collected, the primary endpoint used in the trial was substituted. Average Wholesale Price was used to calculate the cost. Number Needed to Treat (NNT), "Incremental Cost" and "Cost Index" were calculated for each antibody and each disease indication. Results: Forty-eight unique monoclonal antibodies were identified. Sixteen met inclusion/exclusion criteria. Forty-seven trials provided relevant comparative data. We report Overall Survival (OS), Progression Free Survival, and alternative endpoints for each trial. In aggregate, mean NNT for an OS advantage was 12 (range 5-25), mean "Incremental Cost" per improved OS was $788,876 (range $164,399-$2,469,012), and mean "Cost Index" per improved OS was $51,382/month (range $3245 - $205,751). Drug efficacy did not correlate with drug pricing. Conclusion: The "common currency" of NNT provides a concise, clinically relevant approach for framing the debate between individual benefit and public health priorities. Some monoclonal antibodies for some hematology/oncology indications may not meet reasonable thresholds for widespread use.