P-200: Semi-automated Interphase FISH (iFISH) Spot Scoring in CD138-Positive Cells: Validation Study for Genetic Abnormalities Detection in Multiple Myeloma

Introduction: Genetic abnormalities detected in plasma cells are important predictors of prognosis in myeloma multiple (MM) and interphase Fluorescent in Situ Hybridization (iFISH) in selected plasma cells is the gold standard genetic technique. Manual iFISH analysis is a time-consuming, exhausting, and error-prone process that needs the participation of at least two qualified analysts. Interobserver variability can lead to misunderstandings and scoring discrepancies, while interlaboratory variability makes it difficult to compare data. Automated analysis provides an opportunity to overcome difficulties with manual analysis. Guidelines for manual fluorescent spot analysis already exist but they are not clear for automated scoring. The aim of the study was to validate the automated analysis and to compare the effectiveness of this with standard manual analysis for iFISH in selected CD138 cells

Methods: A workstation was optimized based on the manufacturer's configurations. Six commercial probes (CDKN2C/CKS1B, RB1/DLEU1/LAMP1, TP53/CEN17, FGFR3/IGH, CCND1/IGH, and IGH/MAF) were examined to detect gains, losses, and rearrangements of genes across a total of 180 slides (20 samples). Reference values proposed by the European Myeloma Network (10% for rearrangements and 20% for gains and losses) were used to compare the results of manual and semi-automated analyses. The time spent by the biologist for semi-automated and manual analyses was compared in another 10 samples. Statistical analyses were done using the Tau-b of Kendall coefficient (assess the reliability of the percentage of signal classes for each probe and classifier were employed to categorize the classes into normal or abnormal diagnoses), the Kappa coefficient (for accuracy) and paired Wilcoxon test (comparison the time spent for manual and semi-automated per probe).

Results: The first important data was that automated analysis was not effective, lacked validation and was excluded. The results for the probe CDKN2C/CKS1B showed 5 cases with abnormal signals (4 gain of CKS1B), RB1/DLEU1/LAMP1 probe showed 3 cases with deletion of all probes, TP53/17cen probe detected 1 abnormal case with trisomy 17, FGFR3/IGH probe detected 5 cases with abnormal signals (1 FGFR3::IGH) MAF/IGH probe detected 5 cases with abnormal signals ( 1 IGH::MAF) and CCND1/IGH probe detected 7 cases with abnormal signals (4 CCND1::IGH). A 100% of accuracy for normal or abnormal signals patterns was found in manual and semi-automated analysis. Less time for analysis occurred in the semi-automated analysis than in the manual one for the CDKN2C/CKS1B probes (2.3 vs 3.9 minutes, p= 0.013) and for CCND1/IGH (2.0 vs 4.7, p=0.009).

Conclusions: n/a