Dataset Information

Machine learning-based prediction of microsatellite instability and high tumor mutation burden from contrast-enhanced computed tomography in endometrial cancers.

ABSTRACT: To evaluate whether radiomic features from contrast-enhanced computed tomography (CE-CT) can identify DNA mismatch repair deficient (MMR-D) and/or tumor mutational burden-high (TMB-H) endometrial cancers (ECs). Patients who underwent targeted massively parallel sequencing of primary ECs between 2014 and 2018 and preoperative CE-CT were included (n?=?150). Molecular subtypes of EC were assigned using DNA polymerase epsilon (POLE) hotspot mutations and immunohistochemistry-based p53 and MMR protein expression. TMB was derived from sequencing, with >?15.5 mutations-per-megabase as a cut-point to define TMB-H tumors. After radiomic feature extraction and selection, radiomic features and clinical variables were processed with the recursive feature elimination random forest classifier. Classification models constructed using the training dataset (n?=?105) were then validated on the holdout test dataset (n?=?45). Integrated radiomic-clinical classification distinguished MMR-D from copy number (CN)-low-like and CN-high-like ECs with an area under the receiver operating characteristic curve (AUROC) of 0.78 (95% CI 0.58-0.91). The model further differentiated TMB-H from TMB-low (TMB-L) tumors with an AUROC of 0.87 (95% CI 0.73-0.95). Peritumoral-rim radiomic features were most relevant to both classifications (p???0.044). Radiomic analysis achieved moderate accuracy in identifying MMR-D and TMB-H ECs directly from CE-CT. Radiomics may provide an adjunct tool to molecular profiling, especially given its potential advantage in the setting of intratumor heterogeneity.

SUBMITTER: Veeraraghavan H

PROVIDER: S-EPMC7575573 | biostudies-literature | 2020 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning-based prediction of microsatellite instability and high tumor mutation burden from contrast-enhanced computed tomography in endometrial cancers.

Veeraraghavan Harini H Friedman Claire F CF DeLair Deborah F DF Ninčević Josip J Himoto Yuki Y Bruni Silvio G SG Cappello Giovanni G Petkovska Iva I Nougaret Stephanie S Nikolovski Ines I Zehir Ahmet A Abu-Rustum Nadeem R NR Aghajanian Carol C Zamarin Dmitriy D Cadoo Karen A KA Diaz Luis A LA Leitao Mario M MM Makker Vicky V Soslow Robert A RA Mueller Jennifer J JJ Weigelt Britta B Lakhman Yulia Y

Scientific reports 20201020 1

To evaluate whether radiomic features from contrast-enhanced computed tomography (CE-CT) can identify DNA mismatch repair deficient (MMR-D) and/or tumor mutational burden-high (TMB-H) endometrial cancers (ECs). Patients who underwent targeted massively parallel sequencing of primary ECs between 2014 and 2018 and preoperative CE-CT were included (n = 150). Molecular subtypes of EC were assigned using DNA polymerase epsilon (POLE) hotspot mutations and immunohistochemistry-based p53 and MMR protei ...[more]

PMID: 33082371

Similar Datasets

Project description:PurposeMicrosatellite instability-high (MSI-H) endometrial carcinomas are underpinned by distinct mechanisms of DNA mismatch repair deficiency (MMR-D). We sought to characterize the clinical and genetic features of MSI-H endometrial cancers harboring germline or somatic mutations in MMR genes or MLH1 promoter hypermethylation (MLH1ph).Experimental designOf > 1,100 patients with endometrial cancer that underwent clinical tumor-normal sequencing, 184 had MSI-H endometrial cancers due to somatic MMR mutations or MLH1ph, or harbored pathogenic germline MMR mutations. Clinicopathologic features, mutational landscape, and tumor-infiltrating lymphocyte (TIL) scores were compared among MMR-D groups using nonparametric tests. Log-rank tests were used for categorical associations; Kaplan-Meier method and Wald test based on Cox proportional hazards models were employed for continuous variables and survival analyses.ResultsCompared with patients with germline (n = 25) and somatic (n = 39) mutations, patients with MLH1ph endometrial cancers (n = 120) were older (P < 0.001), more obese (P = 0.001) and had more advanced disease at diagnosis (P = 0.025). MLH1ph endometrial cancers were enriched for JAK1 somatic mutations as opposed to germline MMR-D endometrial cancers which showed enrichment for pathogenic ERBB2 mutations. MLH1ph endometrial cancers exhibited lower tumor mutational burden and TIL scores compared with endometrial cancers harboring germline or somatic MMR mutations (P < 0.01). MLH1ph endometrial cancer patients had shorter progression-free survival (PFS) on univariate analysis, but in multivariable models, stage at diagnosis remained the only predictor of survival. For stage I/II endometrial cancer, two-year PFS was inferior for patients with MLH1ph endometrial cancers compared with germline and somatic MMR groups (70% vs. 100%, respectively).ConclusionsMLH1ph endometrial cancers likely constitute a distinct clinicopathologic entity compared with germline and somatic MMR-D ECs with potential treatment implications.

Dataset Information

Machine learning-based prediction of microsatellite instability and high tumor mutation burden from contrast-enhanced computed tomography in endometrial cancers.

Publications

Machine learning-based prediction of microsatellite instability and high tumor mutation burden from contrast-enhanced computed tomography in endometrial cancers.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets