Development and Clinical Validation of a Large Fusion Gene Panel for Pediatric Cancers

Abstract

Gene fusions are one of the most common genomic alterations in pediatric cancer. Many fusions encode oncogenic drivers and play important roles in cancer diagnosis, risk stratification, and treatment selection. We report the development and clinical validation of a large custom-designed RNA sequencing panel, CHOP Fusion panel, using anchored multiplex PCR technology. The panel interrogates 106 cancer genes known to be involved in nearly 600 different fusions reported in hematological malignancies and solid tumors. The panel works well with different types of samples, including formalin-fixed, paraffin-embedded samples. The panel demonstrated excellent analytic accuracy, with 100% sensitivity and specificity on 60 pediatric tumor validation samples. In addition to identifying all known fusions in the validation samples, three unrecognized, yet clinically significant, fusions were also detected. A total of 276 clinical cases were analyzed after the validation, and 51 different fusions were identified in 104 cases. Of these fusions, 16 were not previously reported at the time of discovery. These fusions provided genomic information useful for clinical management. Our experience demonstrates that CHOP Fusion panel can detect the vast majority of known and certain novel clinically relevant fusion genes in pediatric cancers accurately, efficiently, and cost-effectively; and the panel provides an excellent tool for new fusion gene discovery.

Figure 1

Schematic diagram of Children's Hospital of Philadelphia (CHOP) Fusion panel design and workflow. A: Venn diagram of genes of interest in leukemia and non–central nervous system (CNS) solid tumor and CNS tumors. B: CHOP Fusion panel workflow, from sample to report. FFPE, formalin fixed, paraffin embedded.

Figure 2

Analytical performance of Children's Hospital of Philadelphia (CHOP) Fusion panel. A: Coverage of all targeted genes. B: Comparison of read-on-target rate among bone marrow (BM)/blood, fresh/frozen tissue, and formalin-fixed, paraffin-embedded (FFPE) samples. C: Comparison of deduplication ratio among bone marrow/blood, fresh/frozen tissue, and FFPE samples. D: Detection limit of CHOP Fusion panel using a serial dilution assay.

Figure 3

CRLF2/EPOR overexpression as a result of promoter fusion. Children's Hospital of Philadelphia (CHOP) Fusion panel detects the overexpression of CRLF2 (A) and EPOR (B) caused by promoter fusion consistent with RT-PCR. Top panels: Gene expression level by CHOP Fusion panel (normalized against housekeeping gene VCP) or by RT-PCR (normalized with housekeeping gene ACTB). Bottom panels: Visualization of gene expression level using Archer Analysis software. version 4

Figure 4

Clinical utility of Children's Hospital of Philadelphia (CHOP) Fusion panel in 276 clinical cases. A: Fusion-positive cases in different cancer groups: leukemia, non–central nervous system (CNS) solid tumors, and CNS tumors. B: Fusion-positive cases in different sample types: bone marrow/blood, fresh/frozen tissue, and formalin-fixed, paraffin-embedded (FFPE) samples. C: Distribution of recurrent and novel fusions detected by CHOP Fusion panel. D: Most abundant recurrent fusions detected in leukemia, non-CNS tumors, and CNS tumors. E: Clinical significance of all fusions detected. VUS, variant of uncertain significance.

Figure 5

Schematic diagram of protein domains of selected fusions, detailed fusion sequences, and Sanger confirmation. A: KIAA1549-BRAF and GIT2-BRAF in pilocytic astrocytoma patients. B: Two different BCR-ABL1 fusions in a chronic myeloid leukemia (CML) and a B-cell acute lymphoblastic leukemia (B-ALL) patient. C:CENPC-ABL1 in a B-ALL patient. Black arrows indicate the direction of reading frame; red arrows, the break points on mRNA level.