In 2007, almost 220,000 men were diagnosed with prostate cancer, with the vast majority diagnosed by screening PSA. Although screening for prostate cancer has resulted in the diagnosis of more men in the early stages of their disease when prostate cancer is curable with treatment, it is not known whether men live longer as a result. In addition, the treatment of early stage prostate cancer is associated with side effects, including urinary incontinence, impotence, and bowel symptoms, which can significantly affect men's quality of life.

An ideal screening test would diagnose men with clinically significant disease, or disease that would affect quality or length of life if it were not diagnosed and treated. It is thought that PSA screening diagnoses many men with prostate cancer who would never know they had the disease were it not for a positive screening test. These men therefore often undergo the discomfort and anxiety of biopsy and treatment, with the associated side effects, without necessarily improving or prolonging their lives. In addition, PSA fails to identify many men who have potentially deadly disease and require aggressive treatment. Efforts to address the shortcomings of this test have included varying the way PSA testing is conducted, for example, by changing the definition of an "abnormal" PSA, or by decreasing the frequency of testing. Two large randomized clinical trials enrolling almost 300,000 men are currently underway, and it is hoped the results of these trials will tell us whether PSA screening saves lives. However, these trials will not tell us the best way to screen, as they can only tell us if the way they screened was effective.

This proposal describes the creation of a decision analytic model to assess not only whether PSA screening prolongs life, but also whether it improves quality of life. Decision analysis is a technique that permits the construction of a computer model that reproduces reality in order to answer a clinical question whose answer is not known. We will build a model that compares outcomes of a group of men who were screened with a group who were not screened, in order to evaluate whether screening saves or improves lives. Using a model, we will be able to test many different screening strategies simultaneously in order to establish which strategy is most effective in saving and improving life, and also which is the most cost-effective. Such models are flexible, and new information can be continually incorporated in order to ensure that it reflects the most current understanding of prostate cancer development, screening, and treatment. It can also identify areas where data is lacking and guide future clinical trials. The most effective strategies identified by the model can then be used to guide clinical practice, to set guidelines, and to shape health policy.

The training program I propose in this application is designed to provide me with the skills and knowledge necessary to create the model described above, and to further my career goals of becoming an expert clinician, decision analytic modeler, and an independently funded academic researcher. It consists of mentored research under the guidance of Dr. Philip Kantoff of Dana-Farber Cancer Institute (DFCI), an expert in prostate cancer clinical and translational research, and Dr. Michael Barry of Massachusetts General Hospital (MGH), an expert in PSA screening and decision analysis. I will collaborate with other experts in decision analysis, including Drs. James Stahl, Elkan Halpern, Shannon Swan, and Scott Gazelle at MGH's Institute for Technology Assessment (ITA), and with Anthony D'Amico at DFCI, an expert in clinical research. All are uniquely positioned to help me acquire the skills necessary for a career in health outcomes research. Specifically, I will acquire expertise in decision analysis model building and analysis through discussion with Drs. Barry, Stahl, Gazelle, and Swan. I will also develop my data analytic skills through interactions with Drs. Kantoff and D'Amico, and the health outcomes researchers and statistician at ITA listed above. I will take courses offered by the Society for Medical Decision Making and by the Harvard School of Public Health, including courses on decision analysis, cancer screening, and biomarkers. I will also continue to see patients throughout the award period and will further my clinical training by attendance at seminars, interactions with the clinical faculty at DFCI, ASCO meetings, and my own reading. I strongly believe that clinical experience enriches clinical research, and it will always be a pleasure as well as a responsibility to continue this work.