Strategies for validation of a low-volume BCR::ABL p190 assay in the clinical laboratory

Abstract

Abstract Background The low prevalence of samples positive for the BCR::ABL p190 (e1a2) fusion transcript (<1/week) can be a barrier for molecular test validation. The Cepheid Xpert BCR-ABL Ultra p190 Assay is a single-use cartridge-based RUO assay designed for the quantitative detection of BCR::ABL p190 from fresh specimens. The following study describes our validation strategy to overcome this barrier for implementing a quantitative molecular LDT. Methods This study was conducted in a hospital clinical laboratory setting between 2019 to 2023. RNA was extracted and frozen from peripheral blood (PB) and bone marrow (BM) samples that were sent in parallel to a reference lab for BCR::ABL p190 testing. High positive diagnostic specimens (>10% p190/ABL1) were presumptively identified as BCR::ABL FISH positive and p210 negative. Diagnostic specimens were diluted in known negative PB or BM to achieve %p190/ABL1 in various quantitative ranges. Reverse transcription, amplification, and quantitative real-time PCR were performed in a single reaction format (Cepheid Xpert cartridge®). For accuracy determinations, reported results of clinical samples from the reference lab were compared to the obtained values from our validation using extracted RNA only. Banked proficiency testing (PT) material was used to supplement qualitative and quantitative accuracy. For precision, Invivoscribe (IVS) control material (San Diego, CA) diluted in seven levels (12.5% to 0.0031%; 4-fold increments) and Maine Molecular Quality Controls (MMQCI, Saco ME) at 10%, 1%, 0.1%, 0.02%, 0% were run by at least three technologists and rotated amongst the 16 instrument modules. To evaluate linearity, IVS controls at 15 levels (100% to 0.003%; 2-fold increments) were assayed in triplicate. Limit of detection (LoD) was the minimum concentration at which the detection rate is 100% in 10 independent runs. For analytical specificity, high-positive p210 specimens (>10%) were evaluated for detection of p190. Results a total of 169 samples were analyzed (33 BM, 32 PB, 5 PT samples, 73 IVS controls, and 26 MMQCI controls). We found a correlation (linearity) between reference lab results and clinical/ control samples (R-squared ~0.88, P < 0.01). Overall Cohen’s kappa concordance for qualitative accuracy analysis was 0.80 (CI 95% 0.65-0.95). Sensitivity, specificity, and accuracy were 91%, 92%, and 91%, respectively. This assay showed precision in all levels of dilution (IVS and MMQCI samples) as previously established. LoD was 0.02%. Samples with high level of p210 /ABL remained negative with the p190 assay. Conclusion in the setting of low-volume assays optimized for fresh samples, such as for the detection of BCR::ABL p190, dilutions of clinical samples, banked frozen RNA extracts, PT samples, and commercial controls are important tools for validation of a quantitative LDT.