Performance of a multi-target, multi-cancer early detection (MCED) blood test in a prospectively collected cohort

BackgroundMulti-target blood tests have the potential to detect a broad range of cancer types and stages. We previously trained and independently assessed the performance of four biomarker classes for cancer detection in a retrospectively assembled case-control feasibility study. The primary aim of this study was to assess the performance of a methylation and protein (MP) classifier using samples from a large, multi-center, prospectively collected case-control study. As part of the ongoing assay development process, we also performed an exploratory analysis investigating the addition of a somatic mutation reflex approach (MP-reflex) on samples with scores below the threshold for a positive MP call but above a clear negative MP call. MethodsThis study included 6,352 samples (1,438 cancer and 4,914 non-cancer) representing all stages among 21 tumor organ types divided approximately equally between a training and a test set. The sensitivity of the MP classifier was determined at a high specificity. The MP-reflex approach was investigated on MP samples at the same specificity threshold and detectable MP signals above pre-specified lower thresholds. ResultsAt 98.5% specificity, the observed overall sensitivity for MP biomarkers was 50.9%, with sensitivities of 15.4%, 26.1%, 38.0%, 67.8%, 85.5%, and 37.5% for stages I, I and II combined, II, III, IV, and unknown, respectively. When breast and prostate cancers were excluded from the analysis, overall test sensitivity was 56.8%, with sensitivities of 17.2%, 30.7%, 48.6%, 73.5%, 86.5%, and 40.0% for stages I, I and II combined, II, III, IV, and unknown, respectively. ConclusionMethylation and protein biomarkers detected all cancer organ types incorporated in the analysis, including those without standard of care (SoC) cancer screening options and aggressive cancers with poor 5-year survival. These results demonstrate the potential clinical validity of multi-biomarker testing for the detection of several cancer types.