Collections > Electronic Theses and Dissertations > Exploring Clinical and Genetic Factors to Optimize the Use of Tyrosine Kinase Inhibitors in Clinical Practice

Multi-targeted tyrosine kinase inhibitors (TKIs) are widely prescribed anticancer agents that provide significant benefit in survival across a range of cancers; however, 20-30% of individuals do not respond, demonstrating intrinsic resistance. The goal of this dissertation was to identify demographic, clinical, and genetic factors predictive of inefficacy to the multi-targeted TKIs. A retrospective analysis of 108,825 cancer patients revealed a heightened incidence of a Stevens-Johnson syndrome (SJS), a life-threatening adverse event, in cancer patients and suggested that TKIs may trigger the reaction, necessitating drug discontinuation. In a retrospective study of outcomes in routine clinical practice, chart reviews of 266 patients treated with multi-targeted TKIs identified a resistance rate of 21%. Duration of TKI treatment was significantly longer in non-resistant patients; however, there were no significant differences in demographics, tumor type treated, or TKI received. Similar rates of resistance in the subgroups suggest that there may be unidentified shared markers of resistance to these agents. A methodology study to determine the utility of archived formalin-fixed paraffin-embedded (FFPE) samples for genetic analyses was conducted. FFPE tumor samples representing a range of ages, tissue sources, and diagnoses were obtained for next-generation sequencing (NGS). We found no association between age of FFPE samples and sequencing failure; however, there was an association between DNA yield and ability to generate NGS. Both failed samples were derived from bone, suggesting that bone is not an ideal source for FFPE-derived DNA. To identify genetic predictors of intrinsic resistance, cancer patients treated with multi-targeted TKIs were classified as resistant or non-resistant, and tumor FFPE samples were used to generate NGS and copy number variation (CNV). NTRK1, KDR, TGFBR2, and PTPN11 were more commonly mutated in resistant patients and CDK4, CDKN2B, and ERBB2 demonstrated differential patterns of CNVs between groups. In combined analysis using random forest and a decision tree, CNV in CDK4 and CDKN2B were the most important features, and alone could differentiate 55% of individuals as resistant or non-resistant, thus implicating the cyclin-dependent pathway as an important factor in resistance to multi-targeted TKIs. This dissertation presents foundational evidence for the personalization of multi-targeted TKI prescribing and management.