Principal component analysis yields results comparable to those of an elaborate Boolean strategy: simplifying the assessment of measurable residual disease in chronic lymphocytic leukemia patients.

Abstract

Measurable residual disease (MRD) is becoming increasingly important in the chronic lymphocytic leukemia (CLL) context. It is of independent prognostic significance in terms of favorable progression-free and overall survival. The standardized methods used to assess CLL MRD are based on flow cytometry and real-time quantitative PCR. We here present a nine-color assay for CLL MRD with the ROR-1 marker antigen as recommended by the European Research Initiative (ERIC) on CLL; the sensitivity is at least 10-5. We used 54 samples to develop a new principal component analysis (PCA) method based on the Kaluza© "radar" presentation mode. We used a Navios flow cytometer (Beckman Coulter©). We confirmed the linearity of our method over more than five dilutions. The specificity limit was 1.3×10-6 and the lower limit of detection was 3.6×10-6. Compared to the Boolean method, the sensitivity, specificity, and positive and negative predictive values of our PCA method were 100%. When MRD was detectable, PCA and Boolean assays were in agreement (linear regression, R2=0.99). We developed a new PCA-based method for detection of CLL MRD. Our method is comparable to that of the consensus method in terms of sensitivity, and it is also much easier and faster.