IDENTIFYING AND OVERCOMING BARRIERS IN THE TRANSLATION OF ETIOLOGIC CANCER BIOMARKERS
Marrone, Michael T.
MetadataShow full item record
Background: With the goal to improve the translation of cancer biomarkers, this dissertation examined strategies to overcome two barriers in the practice of translational epidemiology: Aim 1) practice of multidisciplinary team science to overcome threats to validity, and Aim 2) use of quantitative metrics to determine the influence of continued investigation to improve utility of information from future biomarker investigations. Aim 1 and 2 strategies were used in Aim 3 to inform the analysis of multiple biomarkers of glycemia to better characterize the relationship between the natural history of diabetes and prostate cancer mortality. Methods: Aim 1 examined the impact of the practice of multidisciplinary team science on identifying and overcoming threats to validity. A case-study of a multidisciplinary team’s investigation of three tissue biomarkers where threats to validity were identified along with appropriate solutions through the practice of team science was carried out. In Aim 2 the impact of continued investigation of a biomarker-cancer relationship was quantified by adapting established research synthesis and clinical trial metrics – fail-safe number and conditional power analysis. To document how these metrics can be adapted to overcome the lack of utility of information from continued investigation, they were applied to a previously curated set of 98 meta-analyses of prospective studies investigating biomarkers and cancer risk. The strategies evaluated in Aims 1 and 2 were applied to Aim 3, by first identifying evidence gaps about the relationship between diabetes and prostate cancer followed by assembling a team with expertise in glycemia biomarkers and diabetes, and prostate cancer etiology. The team developed a refined strategy of incorporating multiple biomarkers of glycemia to better define normo- and hyperglycemia to investigate the relationship between the natural history of diabetes and prostate cancer in the Atherosclerosis Risk in Communities study. Results: Aim 1: Through the practice of team science, the multidisciplinary team consisting of a pathologist, cancer biologists, a biostatistician, and epidemiologist identified measurement error in the pre-analytic and analytic phase of the biomarker measurement, and was able to overcome the threats to validity by implementing appropriate corrections in the data analyses. Aim 2: Applying the fail-safe number and conditional power calculation to the 98 meta-analyses, we observed patterns in the characteristics of the existing evidence and the values of each of these metrics including the size of the observed summary estimate, the number of studies included in the observed meta-analysis, and the extent of between-study heterogeneity. Aim 3: After incorporating strategies from Aims 1 and 2, creating joint categories of the three glycemia biomarkers, and using men without diagnosed diabetes who had normal values for all three biomarkers as the reference group, men without diagnosed diabetes high on all three markers had close to a 5-fold increase in risk of prostate cancer death (HR: 4.80; 95% CI: 1.11 to 20.95). Men with diagnosed diabetes had a non-statistically significant 3-fold increase in risk of prostate cancer death. Conclusions: The inferential benefit achieved through the practice of multidisciplinary team science coupled with the adaptation of the fail-safe number and conditional power analysis to quantify the impact of continued biomarker investigation provide two strategies for the more efficient practice of translational epidemiology. Using these strategies to inform the analysis of biomarkers of glycemia and prostate cancer mortality, revealed an elevated risk of prostate cancer death in men without diagnosed diabetes with elevated glycemia and in men with diagnosed diabetes. These findings speak to the overall importance of diabetes prevention and good glycemic control in men with diagnosed diabetes.