Date of Graduation
School of Public Health
Occupational & Environmental Health Sciences
Two unresolved issues in the treatment of non-small cell lung cancer are the assessment of risk of recurrence beyond the use of tumor stage alone, and the selection of an effective chemotherapeutic agent for patients with similar tumor morphology. A prognostic model able to identify high or low-risk patients with a high degree of accuracy can be used to inform clinicians on potential improvements to the current clinical practice. Clinical presentation, pathology, demographics, and genomics have been independently verified as influencing survival. A comprehensive model capable of incorporating multiple predictors into a unified measure has the potential to simplify risk assessment and more accurately model the determinants of patient outcome.;In order to accomplish this, patient characteristics including tumor stage, grade, patient race, age, COPD status ,and sex were assessed using Cox proportional hazards modeling across combinations of surgical, radiological, and chemotherapeutic treatments. A comprehensive model combining these factors was created and showed superior prognostic ability when compared to stage alone. In order to identify miRNA markers for chemoresponse, this patient data was then compared with information on miRNA expression from both a clinical cohort and the NCl-60 anti-cancer screen. A set of predictive and prognostic miRNA were selected by measuring the association between miRNA expression and disease-specific patient survival. The sets of significant miRNA were seen to have strong associations with mechanisms of apoptosis and cell-cycle control in an analysis of networked molecules.;The results show that a comprehensive model lends itself to a more accurate assessment of patient risk, and that these improvements persist across a variety of patient profiles and treatment modalities. Additionally, miRNA expression appears to play a role in patient response to chemotherapy when assessed across categories of disease progression. Multiple miRNA showed significant associations with disease-specific survival in the population analysis. These associations were able to be corroborated in the clinical and cellular data, demonstrating that this approach may be useful for identifying broad patterns of genomic expression which influence sensitivity and resistance to chemotherapy, and hold promise in further developing clinical tools for prediction. It was shown that the large, well-annotated, and diverse patient sample derived from registry and administrative data can be leveraged to approach two of the major unresolved issues in the treatment of non-small cell lung cancer.
Putila, Joseph J., "Comprehensive Model of Lung Cancer Prediction and Prognosis" (2012). Graduate Theses, Dissertations, and Problem Reports. 4910.