Essays on Economic Incentives and Implications of Biomarker Tests

Abstract

This thesis consists of four chapters: an introductory chapter and three research papers on specific economic aspects of personalized medicine. The approach is empirical in the first paper and theoretical in the second and third papers. In the introductory chapter, I give an account of the factors that have contributed to the slower-than-expected growth of the use of biomarker tests in clinical practice to predict drug response. There have been challenges at the scientific, regulatory, and economic levels, but there have also been some successes. The goal of this dissertation as a whole is to clarify the implications of some of these challenges and successes for the development and use of biomarker tests. In this chapter, I discuss how the research questions relate to the personalized medicine literature and summarize each of the three papers. The first paper seeks to determine the importance of the scientific complexity and predictive capability of biomarker tests. In particular, it studies how the introduction in the Norwegian health system of biomarker tests that guide cancer therapy by predicting drug response has affected the health of cancer patients. The previous literature has provided results related to the benefits to patients of cancer drugs rather than biomarker tests specifically or based on clinical trial data. This article makes two main contributions to the literature: it provides new insights concerning the effect of biomarker testing in particular on cancer treatment, and it makes use of real-world data on cancer patients. The identification strategy relies on the fact that treatments with biomarker test guidance have been introduced for different types of cancer at different points in time. We perform the analysis on Norwegian patients who were diagnosed with cancer and/or died of cancer from 2000 to 2016. We use two main objective measures of health outcomes. These are premature mortality before ages 75 and 65 and the probability of surviving three years after diagnosis. Our main results indicate that biomarker testing decreases premature mortality before ages 75 and 65 and increases the probability of surviving three years after diagnosis, but the effect of biomarker testing on survival weakens as the number of cancer drugs available increases. We also document that while an increase in the number of therapies that require biomarker testing before prescription (biomarker-guided drugs) reduces premature mortality before ages 75 and 65, an increase in the number of therapies that do not require it (nonguided drugs) increases the probability of being alive three years after diagnosis. We conclude that the potential cost per life-year gained from biomarker-guided drugs is below the threshold value used in the literature at which an intervention is considered cost-effective. The second paper investigates the impact of policies to encourage drug producers to collaborate in the development of biomarker tests to predict drug response. Economic incentives are needed because although biomarker testing has the potential to improve patients' health outcomes, it limits drug sales to patients whom the test identifies as responders. Moreover, the pricing of pharmaceuticals is inflexible in many countries, so the drug price will likely remain unchanged after biomarker testing is added to the label. Therefore, we analyze the incentives and welfare effects when the regulator can set one drug price when the test is implemented and another price when the test is not implemented and/or can subsidize drug R&D if the pharmaceutical firm agrees to collaborate in the development of the test. In the model, we consider a pharmaceutical firm that invests in developing a new drug whose price is regulated and decides on whether to allow a biomarker test to be developed for that drug. We show that the regulator faces a tradeoff between increasing the price such that the firm's incentives to invest in drug R&D increase and increasing the social cost of public funding needed to pay the higher price. We also provide the conditions under which increasing the drug price or providing a subsidy on the margin of drug R&D investment are perfect substitutes. To achieve the first-best outcome, a lump-sum tax can be used to transfer the monopoly profits of a drug developer to the government and offset the increase in the social cost of public funds. The third paper considers two drug manufacturers that face the decision of whether to use a biomarker test to select clinical trial participants and a health authority that chooses which drug to approve for the market. The firms must take into consideration the effect of the biomarker on technological R&D uncertainty and on strategic interactions due to competition for market approval. Indeed, although a biomarker test reduces potential drug sales, it increases the probability of finding statistically significant trial results and the quality of the drug, making the drug more appealing to the health authority. We show that it can be more profitable to include a biomarker in clinical trials under a duopoly than under a monopoly due to the consideration that the rival firm's product may be selected by the health authority if the firm does not use the biomarker test. This suggests that personalized medicine can be developed even without policies to encourage it that increase public expenditures. It may, however, be important to have antitrust policies in place to generate competition between drug developers.