Tissue-specific predictive performance: A unified estimation and inference framework for multi-category screening tests

TL;DR

This paper develops statistical methods to evaluate tissue-specific performance metrics of multi-cancer detection tests, enabling more precise assessment and clinical decision-making.

Contribution

It introduces analytical techniques for estimating cancer-specific accuracy and predictive values in multi-category screening tests, addressing variability overlooked by aggregate metrics.

Findings

Method accurately estimates tissue-specific performance metrics.

Simulation validates the proposed analytical approach.

Application demonstrates practical utility on real MCED data.

Abstract

Multi-Cancer Early Detection (MCED) testing with tissue localization aims to detect and identify multiple cancer types from a single blood sample. Such tests have the potential to aid clinical decisions and significantly improve health outcomes. Despite this promise, MCED testing has not yet achieved regulatory approval, reimbursement or broad clinical adoption. One major reason for this shortcoming is uncertainty about test performance resulting from the reporting of clinically obtuse metrics. Traditionally, MCED tests report aggregate measures of test performance, disregarding cancer type, that obscure biological variability and underlying differences in the test's behavior, limiting insight into true effectiveness. Clinically informative evaluation of an MCED test's performance requires metrics that are specific to cancer types. In the context of a case-control sampling design, this paper derives analytical methods that estimate cancer-specific intrinsic accuracy, tissue localization readout-specific predictive value and the marginal test classification distribution, each with corresponding confidence interval formulae. A simulation study is presented that evaluates performance of the proposed methodology and provides guidance for implementation. An application to a published MCED test dataset is given. These statistical approaches allow for estimation and inference for the pointed metric of an MCED test that allow its evaluation to support a potential role in early cancer detection. This framework enables more precise clinical decision-making, supports optimized trial designs across classical, digital, AI-driven, and hybrid stratified diagnostic screening platforms, and facilitates informed healthcare decisions by clinicians, policymakers, regulators, scientists, and patients.