Abstract 4530: An independent validation of a screening test using mass spectrometry for detection of hepatocellular carcinoma

Abstract

Abstract Background: Early detection is critical to improve outcome in hepatocellular carcinoma (HCC). Despite inadequate sensitivity (SS) and specificity (SP), abdominal ultrasound and alpha-fetoprotein (AFP) are considered methods of choice for HCC surveillance. As less than 30% of patients (pts) are diagnosed early enough for resection or transplantation, a test with improved SS and SP is needed. A test to detect HCC in a high-risk population from 10 uL serum, combining MALDI mass spectrometry and AFP data was developed using a dropout-regularized hierarchical machine learning approach designed to minimize overfitting in small development sets. It was previously validated in 293 high risk pts (158 HCC, 135 non-HCC) with SS/SP of 83%/84% in development and 81%/79% in validation across various etiologies and Child-Pugh classification [1]. Methods: The test was applied to an independent validation cohort of 156 pts (97 HCC, 59 non-HCC healthy controls), blinded to clinical data, the performance was assessed by SS, SP. Sub-group analyses were performed by grade and stage. Performance was compared with that of AFP by area under the curve (AUC), using the test output prior to the predefined thresholding which yields the final binary cancer/benign test result. Results: Table 1.HCC patient demographics in the independent validation setClinical characteristicsIndependent validationPatients with HCC (N=97)Median age (range), years62 (38 - 89)Median AFP (range), ng/mL4.0 (less than 1.5 - 10,000)Gender, male / female85% / 16%Hepatitis B / C / No virus / not available (NA)3% / 27% / 20% / 51%Grade I / II / III / NA17% / 22% / 12% / 50%Stage I / II / III / IV / NA12% / 14% / 33% / 26% / 14%Previous liver directed or systemic therapy yes/no44% / 54%ECOG PS 0 / 1 / 2 / 3 / NA13% / 20% / 11% / 5% / 51% In independent validation, AUC for the test output prior to thresholding was 0.979, significantly better than AFP AUC 0.915 (P<0.001). SS and SP were 88% and 100%. SS in grade I, II, and III was 75%, 76%, and 92% and in stage I, II, III, and IV was 75%, 86%, 94%, and 92%. Conclusion: Results in the independent cohort confirm performance of the test, with better SS and SP than AFP and historical ultrasound. The SS was also high in low grade and early stage disease, indicating the test’s potential to improve early detection of HCC when curative approaches are feasible. Data and samples from Data Bank and BioRepository, Roswell Park; funding by the NCI grant P30CA016056 [1] D Mahalingam et al. Hepatology 62(S1): 1135A (2015) Citation Format: Sunyoung S. Lee, Kristopher Attwood, Heinrich Roder, Senait Asmellash, Krista Meyer, Stylianos Kakolyris, Carlos Oliveira, Joanna Roder, Julia Grigorieva, Leonidas Chelis, Renuka Iyer, Devalingam Mahalingam. An independent validation of a screening test using mass spectrometry for detection of hepatocellular carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4530.