Grail, Inc. · NASDAQ:GRAL
Grail — Technical Presentation (June 2021)
13 slides from Grail's technical presentation, June 2021
Slide 1: Evaluation of Cell-Free DNA Approaches for Multi-Cancer Early Detection
Evaluation of Cell-Free DNA Approaches for Multi-Cancer Early Detection Minetta C. Liu,1 Arash Jamshidi,2 Eric A. Klein,3 Oliver Venn,2 Earl Hubbell,2 John F. Beausang,2 Nan Zhang,2 Kathryn N. Kurtzman,2 Chenlu Hou,2 Donald A. Richards,4 Timothy J. Yeatman,5,6 Allen L. Cohn,7 David D. Thiel,8 Mohan
Slide 2: Declaration of Interests
DECLARATION OF INTERESTS Minetta C. Liu MCL is an uncompensated consultant for GRAIL, Inc. The Mayo Clinic was compensated for MCL’s institutional responsibilities for the Circulating Cell-Free Genome Atlas (CCGA) study protocol oversight and advisory board activities for GRAIL, Inc. Content of this
Slide 3: Multi-Cancer Early Detection Tests: Ideal Attributes for Use at Population Scale
MULTI-CANCER EARLY DETECTION TESTS Ideal Attributes for Use at Population Scale Achieve benefits of early detection while minimizing harms: ● Low false positives: achieved through high specificity ● Localizing ability: predicts cancer signal origin to inform diagnostic work-up Circulating Cell-Free
Slide 4: Three CCGA Substudies
THREE CCGA SUBSTUDIES CCGA Substudy 1: Multiple Approaches to cfDNA Analysis for Cancer Signal Detection CCGA1 CCGA2 CCGA3 Discovery Training / Validation Validation Methylation Patterns Methylation Patterns Methylation Patterns Mutations Machine-learning classifier to Clinical Validation supporting
Slide 5: Mapping Assays to Samples, Features, and Classifiers
MAPPING ASSAYS TO SAMPLES, FEATURES, AND CLASSIFIERS Assay Sample Feature Classifier name Whole-genome bisulfite sequencing cfDNA Whole-genome methylation patterns (≈30 million CpGs) WG methylation* Targeted mutations covering 507 cfDNA Single nucleotide variants SNV genes WBCs Single nucleotide var
Slide 6: CCGA Substudy 1: Results – Participant Disposition
CCGA SUBSTUDY 1: RESULTS The Circulating Cell-free Genome Atlas (CCGA) Study Three pre-specified CCGA substudies Participant Disposition 15,254 participants (non-cancer, 44%; cancer, 56%) Substudy 1: Substudy 2: Substudy 3: Discovery analyses of 3 Assay refinement Further assay validation independen
Slide 7: CCGA Substudy 1: Results – Cancer Signal Detection Performance
CCGA SUBSTUDY 1: RESULTS Performance Cancer Signal Detection Sensitivity at 98% specificity Assay Classifier % (95% CIs) TP/Total samples, n WGBS ● WG methylation 34% (30%-39%) 158/464 ● SNV 16% *** (13%-20%) 73/464 TS ● SNV-WBC 33% (29%-38%) 155/464 WGS ● SCNA 27% *** (23%-31%) 125/464 ● SCNA-WBC 3
Slide 8: Cancer Signal Origin Prediction Accuracy
CCGA SUBSTUDY 1: RESULTS Cancer Signal Origin Prediction Accuracy Comparison to Assay Classifier Accuracy WG Methylationa WGBS WG methylation 75% (95/127) -- TS SNV-WBC 35% (44/127) p=6.5x10-12 WGS SCNA 41% (52/127) p=8x10-9 • WG methylation predicted CSO with significantly more accuracy than either
Slide 9: Circulating Tumor Allele Fraction
CIRCULATING TUMOR ALLELE FRACTION Relationship Between cTAF and Classifier Performance What? How? Why? • Estimated fraction of the • Computed from the • Allows for detection total cfDNA isolated from a detectable tumor-tissue- performance to be assessed blood sample that contained identified mutatio
Slide 10: cTAF Explains the Majority of Cancer Signal Detection Performance
CCGA SUBSTUDY 1: RESULTS cTAF Explains the Majority of Cancer Signal Detection Performance WG Methylation Cancer Detection • cTAF accounted for 72% of the Cancer Signal Strength variance in WG methylation cancer signal detection classifier scores • cTAF was an independent predictor of cancer signal
Slide 11: Clinical Limit of Detection for Each Cancer Signal Detection Classifier
CCGA SUBSTUDY 1: RESULTS Clinical Limit of Detection for Each Cancer Signal Detection Classifiera Clinical LOD: the cTAF where the probability of detecting a cancer signal was 50% • Clinical LOD mirrored relative sensitivity performances • WG methylation, SNV-WBC, and pan-feature classifiers provide
Slide 12: Choosing a Candidate for Further Development: Why WG Methylation?
CHOOSING A CANDIDATE FOR FURTHER DEVELOPMENT Why WG Methylation? Cancer Signal Accuracy of Clinical Limit of Assay Potential Detection CSO Prediction Detection • Among the top- • Predicted CSO with • Among the lowest • Significant potential performing cancer significantly higher clinical LOD for fur
Slide 13: Conclusions
CONCLUSIONS CCGA Substudy 1 Demonstrates: Clinical LOD is a useful benchmark by which to assess cfDNA-based test performance cTAF may more closely reflect cfDNA tumor biology than clinical staging or typing information WG methylation provided the most promising cfDNA approach for a blood based MCED