Program Information
Somatic Mutations Drive Distinct Imaging Phenotypes in Lung Cancer
C Parmar1*, E Rios Velazquez1 , Y Liu2 , T Coroller1 , G Cruz1 , O Stringfield2 , Z Ye7 , G Makrigiorgos1 , F Fennessy1 , R Mak1 , R Gillies2 , J Quackenbush1,4 , H Aerts1 , (1)Dana-Farber Cancer Institute, Brigham and Womens Hospital, Harvard Medical School, Boston, MA, (2) H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, (3)Tianjin Medical University Cancer Institute and Hospital, Tianjin Shi (4 )Harvard T.H. Chan School of Public Health, Boston, MA
Presentations
TH-AB-201-9 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 201
Purpose: Tumors are characterized by somatic mutations that drive biological processes, which are ultimately reflected in the tumor phenotype. Quantitative radiomics non-invasively characterizes tumor phenotypes by using a large panel of automatic image characterization algorithms. However, precise genotype-phenotype interactions through which somatic mutations influence radiographic phenotypes remain largely unknown. Here, we present an integrated analysis of four independent datasets of 763 lung adenocarcinoma patients with somatic mutation testing and quantitative computed tomography (CT) image analytics.
Methods: In univariate analysis, we selected 26 variance retaining independent and stable features by using feature reproducibility (ICC) and PCA based analysis and statistically compared their distributions between mutated and non-mutated cases. We developed multivariate radiomic signatures capable of distinguishing between tumor genotypes in a discovery cohort (n=353) by using mRMR feature selection method and random forest classifier. The performance of the signatures was then validated in the validation cohort (n=352) using AUC. All analyses were performed using Matlab-R2012b and R-3.0.2.
Results: We found sixteen and ten radiomic features to be significantly associated with EGFR and KRAS mutations respectively. We found a radiomic signature related to radiographic heterogeneity that could strongly discriminate between EGFR+ and EGFR- cases (AUC=0.69). Combining this signature with a clinical model of EGFR status (AUC=0.70) significantly improved the prediction accuracy (AUC=0.75). The highest performing signature was capable of distinguishing between EGFR+ and KRAS+ tumors (AUC=0.80) and when combined with a clinical model (AUC=0.81), substantially improved its performance (AUC=0.86). A KRAS+/KRAS- radiomic signature also showed significant, albeit lower, performance (AUC=0.63) and did not improve the accuracy of a clinical predictor of KRAS status.
Conclusion: These results suggest that certain somatic mutations drive distinct radiographic phenotypes that can be predicted using radiomics. Such radiomic-based tests can be applied non-invasively, repeatedly, and at low cost, providing an unprecedented opportunity for precision medicine applications.
Contact Email: