Genomic and epigenomic profiles to distinguish pulmonary enteric adenocarcinoma from lung metastatic colorectal cancer.

Abstract

e13528 Background: As an extremely rare variant of lung adenocarcinoma, the diagnosis of pulmonary enteric adenocarcinoma (PEAC) remains challenging in the clinic due to shared morphological and immunohistochemical features with lung metastatic colorectal cancer (mCRC). Current differentiation of PEAC and mCRC mainly relies on clinical history and pathological examination while which still remain risks of misdiagnosis. Due to their distinct treatment regimens, effective molecular markers are essential for accurate diagnosis. However, comprehensive molecular features of PEAC is still poorly understood. Methods: We performed whole-exome sequencing and targeted bisulfite sequencing of 23 PEAC and 20 mCRC and matched normal tissue to improve molecular characterization. For DNA methylation profiling, differentially methylated regions (DMR) were analyzed by comparing PEAC with normal lung tissue and with mCRC. We also trained machine learning methods to distinguish PEAC from mCRC and validated the classifier in an independent cohort with 10 PEAC and 10 mCRC. Results: Mutations of KRAS, APC, and EGFR, alterations of chromosome arms 13q, 14q and 18p were found to be the major differential genetic alterations between PEAC and mCRC (P < 0.05), yet not enough to aid clinical diagnosis. For epigenomic profile, we identified 524 DMRs (false discovery rate ≤0.05) which were further reduced to 30 DMRs according to importance rank by the random forest algorithm. Based on these DMR features, we developed a diagnostic classifier that correctly classified 95.1% of patients in this discovery cohort. We further validated this predictive model in the validation cohort, with a prediction accuracy of 90.0%. We demonstrated its clinical application in two cases with difficulties to diagnosis by traditional methods. Conclusions: We have illustrated the unique genetic and methylation profiles of PEAC and mCRC. Our approach for disease classification may have a substantial impact on diagnostic precision and therapeutic decision for PEAC.