RNA-seq outperforms DNA methods in detecting actionable cancer mutations

Hospital for Sick Children in Toronto researchers are reporting that targeted RNA sequencing can detect clinically actionable alterations in 87% of tumors and provide decisive findings where DNA-seq either fails, returns no variant, or is not informative.

Cancer treatments have seen tremendous improvements in recent years, in part due to highly specific targeting and diagnostic techniques.

DNA-based methods dominate molecular cancer diagnostics but struggle to detect gene fusions and assess splice site consequences. RNA sequencing enables sensitive fusion detection and direct assessment of transcript-level disruption caused by splicing mutations.

In the study, “Clinical utility of targeted RNA sequencing in cancer molecular diagnostics,” published in Nature Medicine, researchers implemented a targeted RNA-seq assay to evaluate 2,310 tumors across central nervous system, solid, and hematopoietic cancers.

Of the 2,310 tumors submitted for analysis, 110 (4.8%) failed quality control, primarily due to insufficient RNA. Among samples meeting quality thresholds, sequencing was successful in 99.6%.

Oncogenic variants were identified in 74% of tumors, with an additional 13% classified as pertinent negatives—bringing total clinical utility to 87%.

Among positive cases, 40% carried single nucleotide variants (SNVs) or indels, 32% fusions, 2.5% other structural variants, and 5% showed multiple classes. Fusions were 1.8 times more common in children, while SNVs were 1.6 times more common in adults.

Among 103 samples with matched DNA sequencing, RNA-seq detected 93.3% of oncogenic variants. Variant allele frequencies between RNA-seq and DNA-seq showed strong correlations.

Among tumors sequenced with diagnostic intent, 37 received new diagnoses, representing 1.9% of evaluable cases. An additional 11 tumors were reclassified based on the presence or absence of diagnostic alterations, bringing the total to 48 revised diagnoses.

Central nervous system tumors accounted for 30 of the 48 revised diagnoses. In 13 cases, RNA-seq reclassified ependymoma as glioma, astroblastoma, or paraganglioma. Seven low-grade gliomas were re-diagnosed as diffuse midline glioma, H3 K27-altered, based on identification of H3K27M mutations.

Five solid tumors were reclassified, including one initially diagnosed as metastatic Wilms tumor, re-identified as clear cell sarcoma of the kidney following detection of a BCOR internal tandem duplication. Another case, previously tested with negative fusion results, was revised to dermatofibrosarcoma protuberans after RNA-seq identified a canonical COL1A1::PDGFB fusion.

Eight tumors were reclassified not by detection of a new alteration but by ruling out expected diagnostic features. In each case, RNA-seq excluded defining molecular markers of the histologic diagnosis, leading to re-evaluation of tumor classification.

In total, 94 of 104 patients considered for targeted therapy received treatment based on RNA-seq findings. MAPK pathway inhibitors, tyrosine kinase inhibitors, and immune checkpoint therapies were most commonly used.

Fusions alone explained oncogenesis in 86% of fusion-driven tumors, many of which lacked additional alterations. RNA-seq succeeded across all tumor types and tissue formats, including degraded FFPE samples.

Researchers conclude that targeted RNA-seq can serve as a stand-alone molecular diagnostic with high yield, minimized cost, and broad clinical applicability.