Detection of rare oncogenic fusions through concurrent DNA and RNA next-generation sequencing in a pan-cancer clinical setting

Background: Concurrent DNA and RNA testing is the gold standard for detection of oncogenic gene fusions with an associated targeted therapy. However, rare fusions without such therapies can also impact patient care by informing diagnosis, prognosis, or potential resistance to therapy. Targeted panel DNA assays are typically optimized for a small number of fusion genes, making use of RNA testing particularly salient for rare fusions. Here we quantify the benefit of concurrent DNA and RNA testing over DNA alone for such fusions, in a large real-world dataset of 74,182 patients with advanced cancer. Methods: We retrospectively analyzed deidentified records from the Tempus multimodal database of metastatic solid tumor samples from non-pediatric patients who underwent testing with both targeted panel DNA-next generation sequencing (NGS) and whole-exome RNA-NGS assays (Tempus xT and xR, respectively). Fusions both on and off the xT panel without an associated targeted therapy were considered for the study. Fusions were identified through the Tempus bioinformatic workflow, which runs RNA-NGS and DNA-NGS pipelines concurrently, followed by pathologist review. Results: Of the 74,182 patients in the cohort, 4,776 (6.4%) had at least one oncogenic fusion without an FDA-approved targeted therapy, with a total of 4,845 such fusions in the dataset. Of these fusions, 85.0% (4,119/4,845) involved genes present on the xT panel, with 56.6% (2,744) involving genes optimized for detecting rearrangements on the xT panel. Across all fusions in the study, 2,561 (52.9%) were detected by RNA-NGS only. For rare fusions involving genes on the xT panel and those with optimized detection in xT, 44.7% (1,841/4,119) and 23.9% (656/3,744) were detected in RNA only, respectively. The most common fusions in this dataset were TMPRSS2-ERG (n=2,247) with 23.4% detected only in RNA. RNA-only detection rates were particularly high for ESR1-CCDC170 (211/213, 99.1%) and PTPRK-RSPO3 (280/280, 100%). TMPRSS2 is optimized for fusion detection by xT, ESR1 is on xT but not targeted for rearrangements, and PTPRK- RSPO3 does not appear on xT. Conclusions: We show that RNA testing greatly enhances detection of rare fusions without an associated FDA-approved targeted therapy in a real-world setting. This benefit of RNA testing is most apparent for fusions not on a DNA-NGS targeted panel, but fusions targeted on the xT panel also showed increased detection with concurrent RNA testing. Among the fusions with high rates of RNA-only detection, ESR1-CCDC170 is associated with more aggressive ER+ breast cancers, and PTPRK-RSPO3 may modulate WNT signaling in colorectal cancer. Other clinically relevant fusions also showed enhanced detection from RNA-NGS. Overall, these findings suggest RNA testing could enhance detection of rare oncogenic fusions to better inform patient care.