Virtual Library

Start Your Search

M.K. Nesline



Author of

  • +

    MINI 22 - New Technology (ID 134)

    • Event: WCLC 2015
    • Type: Mini Oral
    • Track: Biology, Pathology, and Molecular Testing
    • Presentations: 2
    • +

      MINI22.05 - Quality Control Process for NGS to Minimize False Positives (ID 2989)

      17:10 - 17:15  |  Author(s): M.K. Nesline

      • Abstract
      • Presentation
      • Slides

      Background:
      Next generation sequencing (NGS) has exceptional sensitivity, but at the expense of false positives. This can result in a less than optimal positive predictive value and eventually the futile treatment of patients. We have developed a unique set of quality control filters for both Ion Torrent and Illumina that minimize false positives, but have little negative impact on sensitivity. To address this paradoxical association of sensitivity and false positives, we developed a dual platform methodology of NGS using both the Ion Torrent and Illumina to solve this classical dilemma.

      Methods:
      A series of filters were developed to determine quality cutoffs for variant calls to minimize false positives that included the minimum quality score threshold (QUALT), minimum percent variant reads (MPVR), minimum variant reads (MVR), minimum variant reads threshold (MVRT), minimum variant allelic frequency threshold (MVAF), minimum variant reads positive predictive value (MVR-PPV), and systematic errors (SE). A parallel system of using the MiSeq and PGM to sequence all specimens within an IT systems control and a Classify Callsmatrix solution for mutational analysis was designed. Unique cohorts of patients with prior exome sequencing as part of TCGA were used as gold standard controls with matching fresh frozen and FFPE samples.

      Results:
      Table 1 provides the results of filters developed to maximize sensitivity versus PPV. Using our targeted sequencing panel the PGM consistently outperformed the MiSeq for the standard performance characteristics of sensitivity and PPV for both frozen and FFPE samples. Both platforms have systematic false positives that are unique and gene specific.

      Table 1 Platform Tissue VAF setting QUAL Cutoff MVRT Cutoff MVAF Cutoff Mean Sensitivity Range Sensitivity Mean PPV Range PPV
      PGM FF 0.2% None None None 100% 93-100% 88% 70-96%
      PGM FF 0.2% >99 >=20 >.035 99% 93-100% 95% 78-100%
      PGM FFPE 0.2% None None None 99% 93-100% 58% 2-94%
      PGM FFPE 0.2% >99 >=21 >.018 97% 63-100% 92% 40-100%
      MiSeq FF 1% None None None 97% 79-100% 49% 31-66%
      MiSeq FF 1% >99 >=5 >.017 95% 66-100% 82% 66-95%
      MiSeq FFPE 1% None None None 94% 43-100% 10% 2-37%
      MiSeq FFPE 1% >99 >=10 >.028 92% 39-100% 62% 6-100%
      Table 2 provides the results for dual platform sequencing which show a marked reduction in false positives while maintaining sensitivity.
      Table 2 FF FF FF FF FFPE FFPE FFPE FFPE
      SNV(s) SNV(s) Indels Indels SNV(s) SNV(s) Indels Indels
      Percent VAF Percent VAF Percent VAF Percent VAF
      Assay Sensitivity 99.8% 2.87% 100.0% 2.90% 98.3% 3.56% 100.0% 3.60%
      Assay PPV 97.5% 2.87% 91.0% 2.90% 96.7% 3.56% 91.0% 3.60%


      Conclusion:
      Single platform NGS is plagued by false positives. Dual platform sequencing is a reliable method of diminishing false positives with minimal to no impact on sensitivity.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

    • +

      MINI22.08 - Development of a Protein Viewer for Displaying Variants of Unknown Significance in Relation to Actionable Mutations and Protein Domains (ID 2917)

      17:25 - 17:30  |  Author(s): M.K. Nesline

      • Abstract
      • Presentation
      • Slides

      Background:
      Next-generation sequencing (NGS) can be used to interrogate multiple areas of the tumor genome. Several hot-spot panels have been developed to identify variants amenable to targeted therapies and enrollment into clinical trials. Variants of unknown significance (VUS) in the vicinity of hot-spots are routinely discovered. To better understand these obscure VUS, we built a Protein Viewer that displays the relationship of known actionable variant(s) to the VUS.

      Methods:
      We developed a web-based protein viewer that can be deployed across multiple browsers. The tool supports the visual representation of 23 genes which are interrogated by our NGS platform. We used the longest mRNA transcript (hg19) to define the protein domains. All actionable variants as reported by an knowledge database were included, with the selected VUS differentially highlighted. VUS is defined as a non-actionable variant that is not reported in dbSNP.

      Results:
      Approximately 50% of all stage III and IV lung cancer patients tested by our NGS platform have one or more VUS. After the variant information is loaded in the Protein Viewer, a two-dimensional image of the full length protein with actionable variants and VUS is displayed (Figure 1). The Viewer is utilized at RPCI to present cases at our molecular tumor board for quick visualization and discussion. Figure 1 Figure 1: Protein Viewer with a PIK3CA VUS harboring a Q546H (pink) in a lung adenocarcinoma. Top panel with PIK3CA exons 2-21 boundaries (vertical lines) with protein domains (blue rectangles along axis). Bottom panel with the zoom feature which allows more discreet visualization of the VUS, a neighboring Q546K actionable variant (green), and additional actionable variants for ovarian cancer (green rectangles).



      Conclusion:
      Understanding the relationship of VUS to protein domains and proximity to previously known actionable sites is a potentially powerful way to evaluate and determine whether a patient might be a candidate for targeted therapy. Because the exact effect of the VUS on the function of the protein is still impossible to discern (tyrosine kinase inhibitor sensitivity/ resistance/no effect), the next generation of protein viewers should incorporate 3D and protein folding/domain interaction prediction capabilities.

      Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.