Abstract
Introduction: Identifying cytogenetics abnormalities (CAs) is essential to define prognosis in patients with newly diagnosed multiple myeloma (NDMM). Emerging technologies such as Optical Genome Mapping (OGM) can detect a wide range of structural variants (SVs) and copy number variations (CNV) and may be complementary to conventional methods. However, clinical reports often focus on well-characterized and high confidence alterations, while a large volume of additional genomic data generated by OGM remains unexplored. Bioinformatics approaches are needed to fully exploit its potential, automate analysis, and uncover abnormalities with potential prognostic significance.
Methods: Sixty-two patients with NDMM from four Spanish centers with NDMM diagnosed between June 2023 and May 2025 were included in the study. OGM and FISH were performed on CD138-selected cells from bone marrow samples in every patient. In a subsequently collaborative phase, 42 samples were selected based on post-selection plasma cell purity >75% and adequate OGM quality metrics, including N50 molecule length (≥ 230 kbp), effective coverage (≥ 300 kbp), and MAP rate (>76%). An automated pipeline (AP) was developed to detect SVs (gains, deletions, and translocations) and CNVs (aneuploidies) which are previously reported. A total of 38 different CAs were evaluated: 7 aneuploidies, 7 gains, 15 deletions and 9 translocations. Concordance between manually reported CAs and those detected by the AP was assessed across varying confidence thresholds. This AP was implemented in R and included not only variant detection functions but also integrated tools to visualize the results. Paired data were compared using the Wilcoxon signed-rank test. This study was approved by the Ethics Committee of the coordinating institution.
Results: Among 42 selected individuals for automated data analysis, 23 were female (55%) and the median age was 64 years (range, 39-83). Aneuploidies were detected in 40 cases (95%), including monomies in 27, trisomies in 27, tetrasomies in 8 and pentasomies in 3. The median number of aneuploidies per case was 3. Trisomies were most frequently observed in chromosomes 9, 15, and 19, whereas monosomy was most detected in chromosome 13. The most frequent SVs included 1q21 gains (n=22), del16q (n=17), del4q (n=16), Xq gains (n=15), 1p32 deletions (n=6), and IGH (n=13) and MYC translocations (n=10). IGH partners included CCND1 (n=6), FGFR3 (n=3), MAF (n=2), MAFB (n=2) and CCND3 (n=1). Deletion 17p13 was identified in 4 cases. Chromoanagenesis events have not yet been evaluated by the AP. Focusing on MYC rearrangements, translocations with a confidence score >0.9 were identified in 10 cases, 5 of which had not been previously reported. Subsequent manual review confirmed these findings, and FISH using MYC break-apart probe was positive in 3 of the 4 samples available for analysis. Among high-risk cytogenetics features, only MYC translocations were significantly associated with a higher number of SVs (mean 9.4 vs. 6.8; p= 0.04). Other alterations, including 1q21 gain associated with del(1p32), del(17p13), and IGH::MAFB, showed higher mean SV counts but did not reach statistical significance. Finally, overall concordance between manual and automated detection remained stable (~89%) across confidence scores ranging from 0.01 to 1. As the threshold increased, variants uniquely detected by the AP decreased (from 10.1% to 4.8%), while those detected without bioinformatics increased (from 1.3% to 6%).
Conclusions: New genomic technologies such as OGM generate large volumes of information, from which potentially relevant alterations may be missed through manual interpretation. We demonstrated that a bioinformatically supported pipeline can reveal variant detection by systematically analyzing the full, unfiltered dataset. Bidirectional learning between clinicians and bioinformaticians may fully unlock the diagnostic and prognostic potential of OGM in multiple myeloma.
