Performance of a Multi-Cancer Detection Test as a Tool for Diagnostic Resolution of Symptomatic Gynecological Cancers

Abstract

<h3>Study Objective</h3> Assess performance of a multi-cancer detection (MCD) test. <h3>Design</h3> Circulating Cell-free Genome Atlas (CCGA; NCT02889978) is a prospective, longitudinal, case-control study. Samples from the second CCGA substudy were used to validate performance of an MCD test, which uses targeted methylation-based cell-free DNA sequencing and machine learning classifiers to detect cancer signal, including from clinically presenting gynecologic cancers, and predict cancer signal origin (CSO). <h3>Setting</h3> N/A. <h3>Patients or Participants</h3> MCD test sensitivity and CSO prediction accuracy were evaluated in a subgroup of participants with clinically presenting cancers (CPCs). Specificity was assessed in noncancer participants and a subgroup with significant nonmalignant conditions, including endometriosis. <h3>Interventions</h3> N/A. <h3>Measurements and Main Results</h3> Specificity was 99.5% (95% confidence interval [CI]: 98.2-99.9%; 396/398) for the noncancer group and 93.8% (71.7-99.7%; 15/16) for the significant nonmalignant conditions noncancer subgroup. Overall sensitivity was 66.4% (62.2-70.3%; 344/518) for CPC participants, 50.0% (15.0-85.0%; 2/4) for cervical, 70.6% (46.9-86.7%; 12/17) for ovarian, and 25.0% (13.8-41.1%; 9/36) for uterine cancers and varied by stage (Table). CSO prediction accuracy was broadly consistent with CSOs for CPC participants with cancers detected (excluding those with multiple, unknown primaries, or "other" cancer; 91.7% [88.3-94.3%; 300/327]). <h3>Conclusion</h3> Consistent with the overall CCGA study, this MCD test performed with high specificity and accuracy of CSO prediction in this subgroup analysis of CPCs, including gynecological cancers. Findings support potential use of this test for diagnostic workup of symptomatic disease.