Can second-generation multitarget stool DNA panels reliably detect colorectal cancer and advanced precancerous lesions?
Zubin D Gagrat,M. Domanico,M. Krockenberger,Bridget L Bourne,David K Edwards,Christine M Leduc,Abhik Bhattacharya,Keith D Fourrier,Michael B Matter,P. Limburg,J. Kisiel
DOI: https://doi.org/10.1200/jco.2022.40.4_suppl.063
IF: 45.3
2022-02-01
Journal of Clinical Oncology
Abstract:63 Background: Population-based colorectal cancer (CRC) screening can reduce mortality by detecting and removing advanced precancerous lesions (APL) and early-stage invasive disease. One guideline-included strategy is the multi-target stool DNA test (mt-sDNA), which combines detection of methylation DNA markers (MDMs), KRAS mutations, and fecal hemoglobin. Since the mt-sDNA pivotal study was conducted, novel biomarkers have been discovered. A panel of highly discriminant MDMs ( LASS4, LRRC4, PPP2R5C, and reference marker ZDHHC1) was identified through a blinded, case-control study of archival specimens. Here, we evaluated the performance of this novel mt-sDNA panel, combined with fecal hemoglobin, in archival stool samples weighted to early-stage CRC and prospectively collected APL, simulating a screening population. Methods: The verification study featured 777 samples—210 cases (112 CRC [49 stage I, 38 stage II, 17 stage III, and 8 stage IV] and 98 APL) and 567 controls (176 non-APL and 391 colonoscopy-negative)—from three trials (NCT01397747, NCT01260168, and NCT02503631). Median APL size was 12 mm (interquartile range: 10 mm to 15 mm), with 86.7% adenomas (n = 85) and 13.3% sessile serrated polyps (SSPs; n = 13). The average age was 65.5 years for cases (57% men) and 63.2 for controls (47% men). Samples were processed through homogenization, targeted MDM capture, bisulfite conversion, and MDM quantitation using Long‐probe Quantitative Amplified Signal (LQAS). Fecal hemoglobin was quantified using enzyme-linked immunosorbent assay (ELISA). Samples were split into stratified 75%/25% training-testing sets and model outcomes were cross-validated 1,000 times. Results: Mean performance from the cross-validation analysis is summarized in the table below. Overall sensitivity was 95.2% for CRC and 57.2% for APL, with specificities of 89.8% (no CRC/APL) and 92.4% (no neoplasia). Subgroup analyses showed high sensitivity for early-stage CRC, with 93.4% at stage I and 94.2% at stage II. By APL subtype, sensitivity was 82.9% for high-grade dysplasia, 73.4% for villous lesions, 49.8% for tubular lesions, and 30.2% for SSPs. Conclusions: These data support high sensitivity and specificity for a second-generation mt-sDNA panel. A multicenter pivotal trial evaluating the panel is underway (NCT04144738). [Table: see text]
Medicine