DecisionDx^®-Melanoma Analytic Validation

Strong Reproducibility and Technical Reliability Across Multiple Measures

The DecisionDx-Melanoma test has strong reproducibility and technical reliability across all measures evaluated in a published study (Cook 2018). The study evaluated performance metrics of the test on 7,023 cutaneous melanoma tumor samples in Castle Biosciences’ CLIA-certified, CAP-accredited laboratory setting, including concordance of test results from samples subjected to repeat testing, inter-assay, inter-instrument, sample stability and reagent stability.

As molecular biomarkers for melanoma are integrated into patient management, preservation of tumor tissue is critical. The DecisionDx-Melanoma test can be run using as little as 5ng of RNA from a formalin-fixed, paraffin-embedded (FFPE) primary tumor and has achieved a 99% technical success rate across more than 53,600 clinical samples in a recent 4-year period (2017-2020). The high technical success rate reflects consistent performance that compares favorably to the analytic validity of genomic classifiers for breast, prostate and thyroid cancers.

Reproducibility and reliability

99% Inter-Assay Concordance

Reproducibility and reliability of the DecisionDx-Melanoma test performed on primary tumor FFPE specimens met or exceeded the requirements for a clinically applied prognostic test. Inter-assay concordance on 168 specimens run on two consecutive days was 99% and matched probability scores were significantly correlated (R² = 0.96). Inter-instrument concordance was 95% and probability scores had a correlation R² of 0.99 (p<0.001)