Functional Comparison of Different Exome Capture-based Methods for Transcriptomic Profiling of Formalin-Fixed Paraffin-Embedded Tumor Samples

Abstract

BackgroundThe need for fresh frozen (FF) tissue limits implementing RNA sequencing (RNA-seq) in the clinic. The majority of clinical samples are processed in clinical laboratories and stored as formalin-fixed, paraffin-embedded (FFPE) tissues. Exome capture has recently emerged as a promising approach for RNA-seq from FFPE samples. Multiple exome capture platforms are now available. However, their performances have not been systematically compared.MethodsTranscriptomic analysis of 32 FFPE tumor samples from 11 patients was performed using three exome capture-based methods: Agilent SureSelect V6, TWIST NGS Exome, and IDT XGen Exome Research Panel. We compared these methods to TruSeq RNA-seq of fresh frozen (FF-TruSeq) tumor samples from the same patients. We assessed the recovery of clinically relevant biological features, including the expression of key immune genes, expression outliers often associated with actionable genes, gene expression-based subtypes, and fusions using each of these capture methods.ResultsThe Spearman’s correlation coefficients between global expression profiles of the three capture-based methods and matched FF tumor samples, analyzed using TruSeq RNA-seq, were high (rho = 0.72-0.9, p < 0.05). There was a significant correlation between the expression of key immune genes between individual capture-based methods and FF-TruSeq (rho = 0.76-0.88, p < 0.05). All three exome capture-based methods reliably detected the outlier expression of actionable genes, including ERBB2, MET, NTRK1, and PPARG, initially detected in FF-TruSeq. In urothelial cancer samples, the Agilent assay was associated with the highest molecular subtyping agreement with FF-TruSeq (Cohen’s k = 0.7, p < 0.01). Both Agilent and IDT detected all the clinically relevant fusions which were initially identified in FF-TruSeq.ConclusionAll exome capture-based methods had comparable performance and concordance with FF-TruSeq. These findings provide a path for the transcriptomic profiling of vast numbers of FFPE currently stored in biobanks. For specific applications such as fusion detection and gene expression-based subtyping, some methods performed better. By enabling the interrogation of FFPE tumor samples, our findings open the door for implementing RNA-seq in the clinic to guide precision oncology approaches.