Systematic review and meta-analysis of the accuracy and applicability of blood-based multi-cancer early detection (MCED) in the general population.

Abstract

3069 Background: Globally, cancer results in 10.08 million deaths per year. A single blood-based screening tool that detects multiple cancer types could significantly reduce cancer burden. We systematically reviewed and statistically examined both the accuracy and applicability of blood-based MCED tests to strategize their utilization in improving cancer detection. Methods: Original articles were searched from PubMed, Cochrane, and Embase for blood-based screening tests analyzing multiple cancer types and asymptomatic human subjects. We excluded studies with small sample sizes (n<30), hypothesis-generating/diagnostic tests, and non-blood-based tests. For cfDNA-based assays, measurements of diagnostic accuracy were pooled for meta-analysis. Results: Of 1,074 records identified and screened, 15 case-control and 4 cohort studies were analyzed, most of which utilized cfDNA-based diagnostic tests. Twelve cfDNA studies selected for meta-analysis had pooled sensitivity of 0.623 (95%CI 0.517 - 0.719) and specificity of 0.975 (0.942 - 0.990). Summary ROC curve shows variability in sensitivity and specificity between studies. However, sensitivity and specificity were not affected by study type, gender, or assay type. Sensitivity was higher for advanced staged cancers (III/IV 0.774 (0.697 - 0.837)) than early-stage cancers (I/II 0.503 (0.366 - 0.639)). Among cancer types, no significant differences were detected. Lastly, false positive and false negative rates were 0.025 (0.010 - 0.058) and 0.447 (0.438 - 0.456), respectively. Conclusions: Given high sensitivities and specificities, MCED tests show promise as additional screening tools. However, in the general population, misdiagnosis burden from false positive and negative rates is anticipated. Although multiple barriers exist to their application in the clinic, MCED tests may improve patient outcomes for cancers with no conventional screening tools and provide additional credibility when combined with existing tools. Future prospective studies with large and diverse populations are warranted. [Table: see text]