Analytical and Clinical Validation of Duoseq, a Novel Assay for Rapid, on-Site Clinical DNA and RNA Sequencing of Hematologic Malignancies

Introduction:

Next generation sequencing (NGS) has become a critical component of the workup of malignancies. NGS can provide important diagnostic information including mutations, chromosomal copy number alterations and translocations (from DNAseq) as well as gene expression and fusions (from RNAseq).

Incorporation of NGS in the clinic remains difficult. DNAseq and RNAseq remain distinct assays with challenging bioinformatics and turnaround times (TAT) that can last several weeks. Duoseq is a novel assay for combined DNA- and RNAseq in a single workflow that detects single nucleotide variants (SNVs), insertions/deletions (indels), splice site variants (SSVs), structural variants (SVs) and Epstein-Barr virus (EBV) status. Initially implemented for hematologic malignancies, Duoseq assesses nearly all known gene drivers (N=475) and is compatible with both fresh-frozen and formalin-fixed, paraffin-embedded (FFPE) tissues. It is available as a reagent kit and includes cloud-based bioinformatics, enabling any clinical lab with an Illumina sequencer to perform NGS in-house with a TAT of 2-3 days.

Methods/Results:

Duoseq was validated for concordance with commercial DNA- and RNAseq kits (New England BioLabs (NEB)). Sequencing libraries were prepared on DNA alone, RNA alone and total nucleic acid (TNA) from cell lines in accordance with kit protocols and captured using the Duoseq panel. Results were equivalent between Duoseq and individual (NEB) library preparations with nearly complete concordance in DNA variants (99.4%, P<10 ^-6) and RNA expression (R ²=0.97, P<10 ^-6). Separately, copy number alteration calls across 42 cases had high concordance between Duoseq and Affymetrix array approaches (P<10 ^-6).

Analytical validation was performed at two CLIA laboratories to establish assay precision and reproducibility, sensitivity (LoD), and specificity. Analytical validation was performed on well characterized and control samples that were selected to harbor an array of clinically important variants at known allele frequencies.

Precision and reproducibility testing confirmed Duoseq's ability to produce accurate results independent of external variables. We tested inter-lab reproducibility, intra-day and inter-day repeatability and technician-to-technician reproducibility. In each case, Duoseq achieved >98% precision and reproducibility for each variant class (SNVs, indels/SSVs and SVs).

We confirmed the LoD using TNA from the same samples diluted in normal TNA to create 3-point curves that spanned our expected LoD. We confirmed the assay to be 100% accurate down to ≥5% mutant allele fraction (MAF) for SNVs, ≥10% MAF for indels/SSVs, and ≥20% tumor purity for SVs. Specificity was additionally assessed by including normal cell lines in each run to determine the incidence of cross-contamination indicated by false positives. Analytical specificity was 100%, indicating no instances of sample cross-contamination.

We assessed the clinical accuracy of Duoseq using FFPE myeloid and lymphoid specimens (N=153) collected at both laboratories and characterized via orthogonal methods including Foundation Medicine (SNVs and indels/SSVs), fluorescent in situ hybridization (translocations) and in situ hybridization (EBV status). Duoseq achieved a positive predictive agreement of 98% for SNVs (N=139), 96% for indels/SSVs (N=55), 95% for translocations (N=282), and 100% for EBV status (N=58). Cell of origin (COO) for diffuse large B cell lymphoma (n=42) using Hans criteria and had high concordance (90%) with the COO calls rendered from Duoseq expression data.

Given Duoseq's array of measurements and short TAT, we plan a potential application in “match” clinical trials where patients are enrolled in treatment arms based on molecular profiling. In ongoing work to test a potential trial workflow, we enrolled four institutions that will provide cases to a central site for pathology review and molecular profiling enabling trial arm assignment. Specimen enrollment is anticipated to be completed to allow for results to be presented at the meeting.

Conclusions:

Duoseq is comparable to current methods and provides a cost and time effective option for library preparation.The large gene panel enables testing a broad range of clinical features in a single assay, including genes required for diagnosis and prognostication in heme malignancies. These results support clinical use of Duoseq.