Specificity and Universal Cancer Detection using Plasma James G Herman MD Jeff Wang PhD The UPMC Hillman Cancer Center The University of Pittsburgh The Johns Hopkins University Highly Prevalent Tumor Specific Methylation Defined for Lung CA ID: 1040040
Download Presentation The PPT/PDF document "DNA Methylation Detection for Early Lung..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. DNA Methylation Detection for Early Lung CancerSpecificity and Universal Cancer Detection using PlasmaJames G. Herman, M.D.Jeff Wang, Ph.D.The UPMC Hillman Cancer CenterThe University of PittsburghThe Johns Hopkins University
2. Highly Prevalent Tumor Specific Methylation Defined for Lung CA6 gene panel with individual 77-93% prevalence in stage I lung cancerDerived from demethylation unmasking studies of lung cancerWrangle et al, Clinical Cancer Research, 2014Improved Sensitivity using Integrated Methylation AnalysisNet Result is ~25-100 fold improvement in amplifiable DNA VJ Bailey et al, Clinical Chem, 2010, VJ Bailey et al, Clinical Chem, 2010Improved Methods for DetectionNext generation methylation detectionBailey VJ et al, Genome Res, 2009, Pisanic II TR et al, NAR, 2015, Performance of Methylation Markers Studied in Case-Control CohortHulbert A et al, Clinical Cancer Research 2017. Innovations for Improved DNA Methylation Detection
3. BloodSensitivitySpecificityPPVNPVAUC95% CICDO165%74%86%46%0.68(0.58 - 0.77)TAC176%78%90%57%0.78(0.70 - 0.86)HOXA733%94%93%36%0.60(0.51 - 0.69)HOXA981%52%81%52%0.62(0.52 - 0.73)SOX1771%86%93%54%0.78(0.70 - 0.86)ZFP4281%58%83%55%0.66(0.56 - 0.75)CD01, TAC1, SOX17 91%64%86%74%0.77(0.68 - 0.86) SputumSensitivitySpecificityPPVNPVAUC95% CICDO178%67%90%45%0.70(0.57 - 0.84)TAC184%79%94%57%0.84(0.74 - 0.94)HOXA763%92%97%40%0.77(0.67 - 0.86)HOXA977%42%83%32%0.56(0.41 - 0.69)SOX1784%88%96%59%0.84(0.75 -0.94)ZFP4288%62%90%58%0.73(0.60 - 0.87)TAC1, HOXA7, SOX17 93%79%94%75%0.89(0.80 - 0.98)Methylation Detection in Plasma and SputumStage I Lung Cancers and Surgical ControlsHulbert A et al, Clinical Cancer Research 2017
4. Validation of Plasma DNA Methylation Detection246 patients with screen detected pulmonary nodules referred to surgeonStage I: Nodules 3.0 cm or less, 163 cancer, 83 benignChen et al, Clinical Epigenetics, 2020
5. Comparison: stage 1 lung tumors (up to 5 cm included) CancerSEEK (Sensitivity of 43%)Performance of Plasma DNA Methylation DetectionStage I NSCLCa according to T size (all ≤ 3.0 cm)CDO1, SOX17, TAC1T 0 - 1.0 cm 71% 82% 83% 69% 0.81 (0.69 - 0.93)
6. Screen Detected Pulmonary Nodules: Refining Risk based on DNA Methylation DetectionProbability Screen Detected Nodule is Cancer 65 y/o Man, 1.0 cm solid nodule, Lower lobe, no emphysema Brock Model: 3.5% cancer riskRefinement of Risk using Methylation Detection (90% Sensitivity, 71% Specificity) from Chen et al 2020Positive test with + Likelihood Ratio of 3.10 = 10% Cancer ProbabilityNegative test with – Likelihood Ratio of 0.14 = 0.5% Cancer Probability (Baseline risk of screened cohort)
7. What if Methylation Test is Applied to At Risk Cohort?Probability of Lung Cancer in screened population NLST (55-74): 649 Lung Ca/26309 = 2.5% or .83%/scan or yearly Nelson (age 50-74): Lung Cancer in men (0.9%/scan ~ yearly) So about .8-.9% Risk of Lung Cancer per year Colon Cancer: 65 yo Caucasian ex smoker Male, 0.9%/5 year or .2% yearly Prostate Cancer: 65 yo Male PSA 2.0 5% 4 year risk, Refinement of Risk using Methylation Detection for 0.9% risk lung cancer (90% Sensitivity, 71% Specificity) from Chen et al 2020Positive test with + Likelihood Ratio of 3.10 = 3 % Cancer ProbabilityNegative test with – Likelihood Ratio of 0.14 = 0.15% Cancer Probability (Baseline risk of screened cohort)But most positives would not be lung cancer. Other sources??
8. SOX17 Methylation TCGA 19 MalignanciesSOX17 Methylation First Reported in Colorectal CancerData Source: TCGA
9. CDO1 Methylation TCGA 19 MalignanciesCDO1 Methylation Reported in Breast CancerData Source: TCGA
10. Universal DNA Methylation Detection LociDiscovery using 5 core cancers (Lung Adeno and Squamous, Colorectal, Breast, and Prostate) Heat Map of DNA Methylation of 32 cancer-specific probes in normal and 19 tumor histologiesData Source: TCGADanilova L, Wrangle J, Herman JG, Cope LEpigenetics. 2021
11. Example 1Near Universal Cancer Methylation LociPoor AML, Pancreatic Data Source: TCGA
12. Example 2:Near Universal Cancer Methylation LociPoor Kidney (Clear and Papillary)
13. Example 3Near Universal Cancer Methylation LociBetter Kidney Cancer Detection
14. Universal DNA Methylation Detection LociWhat is Minimal Number of Loci Needed for Efficient Detection?When used for Blood Based Detection, what is the Source?Data Source: TCGADanilova L, Wrangle J, Herman JG, Cope LEpigenetics. 2021
15. Cancer-Specific DNA Methylation Markers Distinguish Five Core Human MalignanciesStill Require Lack of Methylation in Normal TissueDanilova L, Wrangle J, Herman JG, Cope LEpigenetics. 2021
16. Cancer-Specific DNA Methylation Loci Derived from Five Core Human Malignancies Can Classify 19 Human MalignanciesDanilova L, Wrangle J, Herman JG, Cope LEpigenetics. 2021
17. DNA Methylation distinguishes cancer phenotypesConfusion matrix of the validation set (n = 5527) of cancer type prediction for 19 cancer typesSimilar Predictive Accuracy if 5 other cancers used for loci identification
18. Low Grade Glioma Identifier: Lowest Handing Fruit Strong Hypermethylator Phenotype (IDH mutations) Numerous Loci
19. Acute Myeloid Leukemia: Another Easy IdentificationUnique Biology from Epithelial Tumors
20. Melanoma: Another Easy IdentificationUnique Biology from Epithelial TumorsOverlap with another Mesenchymal Phenotype?
21. Cervical Biomarker: Head and Neck Cancer OverlapShared HPV status
22. Lung Squamous CancerHead and Neck Cancer, Bladder, Esophageal OverlapSquamous Histology
23. Lung AdenocarcinomaLittle overlap with Squamous Histology
24. Colorectal Cancer: 4 Frequent Methylated Genes with some Gastrointestinal Cancer Overlap
25. Prostate Detection: Breast Overlap Shared Hormone Driven Adenocarcinomas
26. DNA Methylation distinguishes cancer phenotypesConfusion matrix of the validation set (n = 5527) of cancer type prediction for 19 cancer types.What number of Loci Provide AcceptableTumor Identification?
27. AcknowledgementsJHU EDRN LabJeff WangTom PisanicAlex StarkChristine O’KeefePittsburgh EDRN LabBrenda DiergaardeJian-Min YuanDavid WilsonSona JoyceUniversal Methylation DetectionLuda DanilovaJohn Wrangle Leslie Cope