Sara E. Nunnery, Andrew E. Fintel, W. Clay Jackson, Jason C. Chandler, Michael O. Ugwueke, and Mike G. Martin
Andrew W. Hahn, Smith Giri, Dilan Patel, Heather Sluder, Ari Vanderwalde, and Mike G. Martin
With the advent of widespread tumor genetic profiling, an increased number of mutations with unknown significance are being identified. Often, a glut of uninterpretable findings may confuse the clinician and provide little or inappropriate guidance in therapeutic decision-making. This report describes a method of protein modeling by in silico analysis (ie, using computer simulation) that is easily accessible to the practicing clinician without need for further laboratory analysis, which can potentially serve as a guide in therapeutic decisions based on poorly characterized tumor mutations. An example of this model is given wherein poorly characterized KIT, PDGFRB, and ERBB2 mutations were discovered in a patient with treatment-refractory metastatic transitional cell carcinoma of the renal pelvis. The KIT and PDGFRB mutations were predicted to be pathogenic using in silico analysis, whereas the ERBB2 mutation was predicted to be benign. Based on these findings, the patient was treated with pazopanib and achieved a partial response that lasted for 7.5 months. We propose that in silico analysis be explored as a potential means to further characterize genetic abnormalities found by tumor profiling assays, such as next-generation sequencing.