410 – Diagnostic Tests: Regulatory Considerations of Intermediate Outputs and Oncology Screening, and Optimal Point of Heterogeneous ROC Curves
Uncertainty of the Reported Results in Molecular Diagnostics
Jeffrey E. Vaks
Roche Molecular Systems
With qualitative PCR tests, there are two types of reported results: target not detected and target detected. With quantitative PCR tests, there are four types of reported results: target not detected; target detected, but below the lower limit of quantitation; quantitative; target detected, but above the upper limit of quantitation. The number of copies of target nucleic acid varies randomly in randomly drawn samples. Formulas based on cumulative Binomial and asymptotic cumulative Poisson probabilities, quantifying the uncertainties of the above types of reported results as functions of expected numbers of copies in test samples, are derived and graphed. It is shown that Poisson approximation errors are small, and simplified Poisson-based formulas can be used for quantifying the uncertainty of the types of reported results. The uncertainty of the quantitative results is characterized by the repeatability and reproducibility standard deviations along with the respective numbers of degrees of freedom estimated with data. This work helps in clear understanding of the uncertainty of reported results, which in-turn helps laboratorians interpreting and QC test results, clinicians diagnosing and monitoring treatment of disease, blood banks evaluating the risk of transmission of infectious agents from donors to recipients, pharma evaluating the efficacy of new medications, etc.