Concurrent RNA- and DNA-NGS Testing for the Detection of Clinically-Relevant Fusions In Pediatric Solid-Tumor Patients

Background
Identifying structural rearrangements and gene fusions is critical to providing high quality, precision-driven clinical care for many types of pediatric cancers. RNA next-generation sequencing (NGS) provides a functional readout of the genome that enables superior detection of chimeric transcripts and novel driver fusions, particularly when genomic breakpoints reside in complex intronic regions. This study demonstrates the benefit of concurrent DNA- and RNA- NGS for fusion detection in a real world pediatric cohort of 1,050 patients, one of the largest such studies to date.

Methods
We used the Tempus de-identified multimodal database to select a cohort of pediatric solid tumor cancer patients (all stages) who received successful DNA (Tempus xT, 648 gene panel with enhanced detection of structural variants [SVs] in 22 genes) and RNA (Tempus xR, whole-transcriptome) NGS sequencing. All patients had a minimum tumor purity of 20% and were aged 0-21 at the time of sample collection (n = 1,050). All assessed fusions appeared on clinical reports.

Results
Overall, we detected a fusion in 35.1% of patients (n = 369). The top 3 cancer types in our cohort were soft tissue sarcoma (n = 321), brain/CNS cancer (n = 280), and bone cancer (n = 120), and fusion prevalence in these types was 53% (n = 170), 30.4% (n = 82) and 30% (n = 36), respectively. The fusion types with the highest overall prevalence were: EWSR1 (9.0%, n = 95), BRAF (6.1%, n = 64), PAX3-FOXO1 (2.7%, n = 28), ALK (2.2%, n = 23), and RET (2.0%, n = 21) fusions. In assessing the subset of genes that appear on both the DNA- NGS and RNA-NGS panel, 303 patients harbored one of these fusions and 38.6% (117/303) of those fusions were detected only via RNA-NGS. Among fusions with highest prevalence, the percentage detected only via RNA-NGS ranged from 3.1% (EWSR1 fusions) to 100% (PAX3- FOXO1); BRAF (53/64, 82.8%) and ALK (4/23, 17.4%) fusions both had comparatively high proportions detected only via RNA-NGS. RNA-NGS alone detected gene fusions in 65 additional patients where neither partner appears on the DNA-NGS panel. Considering only fusions associated with an indication-matched FDA-approved targeted therapy, we observed a prevalence of 9.0% (94/1050), and of these, 46.8% (44/94) were detected only by RNA-NGS. Overall, 49.3% (182/369) of fusion-positive patients would have been missed if RNA-NGS were not performed, representing 17.3% of the total cohort.

Conclusions
Pediatric solid tumors are frequently driven by structural rearrangements and gene fusions that are difficult to characterize using DNA sequencing alone. This study demonstrates that performing combined DNA-NGS and RNA-NGS substantially improves the identification of patients with a clinically relevant fusion in a large real-world data set.