Consistent performance of highly multiplexed RNA fusion detection with low RNA input and limit of detection.

Abstract

e15029 Background: Gene fusions are usually caused by chromosomal rearrangements and frequently associated with various cancers. Detection of fusion mutations is an important part of cancer management, and are usually detected in RNA purified from biopsies of fresh tissue or FFPE sections. However, the amount of RNA from these samples is usually limited. We developed a new, PCR-based fusion detection method, which allows detection of unknown and novel fusion mutations. We demonstrate its consistent performance across various RNA inputs, and sequencing depths, in terms of technical performance and limit of detection. Methods: A multiplex PCR RNA fusion panel was designed to cover known RNA fusion targets and unknown fusion targets for target enrichment via single primer multiplex PCR. SeraSeq fusion Reference RNA was used to evaluate the limit of detection of the method. The amplicons were about 110 bp on average to allow for target amplification from challenging fragmented samples such as FFPE RNA. After reverse transcription, the panel was used in a multiplex PCR to amplify the targets, and a final PCR was used to add sequencing adapters and sample indexes. The resulting libraries were sequenced on Illumina sequencers. The data was analyzed for the detection of known fusion variants across varying initial inputs of targets, as well as the performance in terms of mapping and on-target rates, as well as the presence of human ribosomal RNA. Results: We found that the fragmented RNA length had little effect on the performance of the libraries. The required numbers of PCR cycles were optimized based on testing with various amounts of starting RNA input material. We observe a linear relationship between the library yields and RNA input amounts from 10 to 50ng, which had R1 mapping rates above 95% and 98%, respectively. We report the detection of all twelve fusion variants, which are targeted by our panel and present in the Seraseq Fusion RNA Mix v4 reference standard, and which are covered with an average sequencing depth of 0.2M reads, or 3600 reads/amplicon, across all assays. Even at inputs as low as 3 ng, all 12 fusions were typically detected. In general, the results for individual fusions among the different replicates were concordant, with limited observed variance in reads across some fusion junctions between assays and replicates. Over 90% of the reads supported the fusion calls from all inputs (ranging from 1.6 to 50 ng). Over 97% of the reads supported the fusion calls between 12 and 50 ng. Conclusions: A multi-lab validation confirmed the above results, and were also comparable when using an additional distinct CleanPlex fusion panel. To conclude, our CleanPlex fusion technology allows for reliable and simultaneous detection of 12 designed clinically relevant RNA fusions even at low input amounts.