Sabit Delic, Manja Meggendorfer, Niroshan Nadarajah, Wolfgang Kern, Torsten Haferlach
MLL Münchner Leukämielabor GmbH, Germany
Objectives: The number of molecular markers used to characterize myeloid malignancies continues to constantly increase. As such, physicians and laboratories face a great unmet need to test panels of genes at a high level of sensitivity and throughput. In addition, monitoring response to therapy by analyzing mutation load of initially identified molecular markers requires extremely sensitive detection methods.
Methodology: We tested the Source instrument (RainDance, Billerica, MA) for droplet generation with subsequent next generation sequencing (NGS) or digital PCR (dPCR). For NGS we used the ThunderBolts Myeloid Panel (RainDance) composed of 53 genes with 533 amplicons. Targets of interest comprised either complete coding gene regions or hotspots. Sequencing data was generated using the MiSeq instrument (Illumina, San Diego, CA) loading 8 patients per run. 37 patients were analyzed. Results were compared to sequencing data obtained with droplet-based library preparation using the ThunderStorm (RainDance) or Access Array (Fluidigm, South San Francisco, CA) techniques.
For dPCR we used the RainDrop System (RainDance). We established 3 assays for the genes RUNX1, NPM1 and ASXL1 and analyzed 83 patients resulting in 160 samples in total. With the RUNX1 assay (mutation c.521G>A) 35 patients were analyzed. TheNPM1 assay (c.863_864insCTTG) was used to analyze 19 patients and 28 patients were analyzed with the ASXL1 assay (c.1934dupG). Droplet generation was performed with genomic DNA for RUNX1 and ASXL1 assays and cDNA for the NPM1 assay. The results were compared to known mutation loads as detected formerly by NGS (RUNX1), dPCR using the Fluidigm EP1 system (NPM1) or Sanger sequencing (ASXL1). In 76/160 samples these methods had proven the absence of the respective mutations or the presence at minimal residual disease (MRD) levels <10%. Over all samples were a mutation was detectable the mutation load ranged from 0.1% to 93%. Results: We could show that droplet generation using the Source instrument can be applied in clinical testing. NGS libraries generated by the Source instrument and analyzed on MiSeq instruments gave basically the same results as obtained by our well established routine methods. All known mutations have been identified with a comparable mutation load and good coverage. The dPCR experiments performed on the RainDrop system gave results correlating very well with the known mutation loads as analyzed by NGS, EP1 or Sanger sequencing, respectively. Conclusion: We here demonstrated that droplet-based sample preparation enabled to target 53 candidate genes for next generation sequencing in myeloid malignancies in a routine diagnostic environment. In addition, we showed that using the same instrument the droplet-based sample preparation can also be applied to dPCR for monitoring MRD. Thus the Source instrument provides a feasible 2-application droplet generation platform for both, NGS library preparation and dPCR in clinical testing.
|Back to Digital PCR|