Integrating Optical Genome Mapping With TP53 FISH: A Synergistic Approach for Cytogenomic Analysis in Chronic Lymphocytic Leukemia

Abstract

Fluorescence in situ hybridization (FISH) is the gold standard technique for cytogenetic assessment in chronic lymphocytic leukemia (CLL). In addition, chromosome banding analysis (CBA) is recommended as part of testing to detect complex karyotypes (CK, ≥ 3 abnormalities in the same cell clone), especially as those with a high-CK (≥ 5 abnormalities) have a known worst outcome [1, 2]. Optical genome mapping (OGM) has emerged as a high-resolution technique to detect genome-wide balanced and unbalanced abnormalities, overcoming some disadvantages associated with current cytogenomic methods [3]. This study aimed to compare the effectiveness of OGM against CBA and FISH techniques in detecting poor prognostic cytogenomic biomarkers in CLL within a cohort of 102 CLL patients from two European centers, thus assessing the potential of OGM as a future routine diagnostic test.

Patients were selected to represent all the risk categories of the FISH-based Döhner hierarchical model: isolated del(13q) (n = 19), normal FISH (n = 18), trisomy 12 (n = 17), del(11q)(ATM) (n = 28), and del(17p)(TP53) (n = 20) (Table S1). OGM experiments were performed following the manufacturer's protocols, analyzed using the rare variant analysis pipeline (GRCh38/hg38 as a reference) and results visualized with the Bionano Access software (v1.7.2) (Bionano Genomics, San Diego, CA, USA). The detected structural variants (SVs) and copy number alterations (CNAs; gains and losses) were filtered out in different steps by discarding artifacts and polymorphisms, and applying filter settings based on confidence scores and size for the remaining alterations. The filtering strategy was based on criteria used in chromosomal microarrays analyses [4], and aimed to identify translocations, clinically relevant abnormalities in CLL independent of the size, and other abnormalities ≥ 5 Mb. Additionally, non-CLL abnormalities, sized between 200Kb and 5 Mb, were exclusively retained if they were part of a chromoanagenesis event or associated with other retained SV. First, we evaluated OGM's ability to detect abnormalities identified by the FISH panel and CBA. Next, we analyzed genomic complexity in patients stratified by CK status to determine a potential threshold for defining complexity by OGM. Statistical analyses were performed using SPSS v.23 software (SPSS Inc., Chicago, IL, USA). p values < 0.05 were considered statistically significant. Further details on methodology are described in Supporting Information.

Concerning cytogenomic abnormalities included in Döhner's model, OGM identified 90% (112/125) of those previously detected by FISH. Additionally, OGM detected two small deletions [one del(11q) and one del(13q)] undetectable by the standard FISH probes used in routine practice due to their small size (Figure 1A). Although OGM failed to identify 13 CNAs present in 6.5% to 17% of nuclei by FISH in some cases, it successfully detected abnormalities in similar percentages (7%–17%) in seven others (Tables S2 and S3). A visual review of whole genome CNA data was especially important for identifying these low-clonal abnormalities, as well as 17p/TP53 deletions, with 5 out of 18 (28%) detected only upon visual inspection. On the other hand, OGM provided greater resolution in characterizing the size and structural features of the deletions, offering a more detailed genomic profile compared with FISH (Supporting Information). Overall, OGM showed a good performance compared with FISH but was unreliable at detecting the presence of small clones. Thus, FISH remains essential for the identification of low-clonal alterations in populations below 15%. This limitation could be particularly relevant for the detection of low-burden del(17p), since low variant allele frequency TP53 mutations, often associated with these small clones, are recognized to contribute to resistance in chemoimmunotherapy-treated patients. Their role in targeted therapies appears less impactful but warrants further investigation.

Regarding whole genome abnormalities, OGM identified abnormalities in a greater number of patients and a higher median number of alterations compared with CBA, although differences did not reach statistical significance [89 (87%) vs. 79 (77%), p = 0.060; and 4 [range: 1–87] vs. 2 [1–15], p = 0.816, respectively]. When the ability of OGM to detect aberrations visible by CBA in individual cases was assessed, 88 patients (86%) showed concordant results with CBA, although 37 of them presented some discrepancies associated with intrinsic limitations of each technique. Likewise, 14 patients (14%) had discordant results where abnormalities, predicted to be cytogenetically visible, were only detected by one of the methods (Supporting Information).

Evaluating OGM's ability to detect global genomic complexity, patients with CK by CBA displayed a higher number of abnormalities by OGM compared with non-CK patients. In contrast, low-intermediate (3–4 abnormalities) and high-CK groups did not show significant differences (Figure 1B). Additionally, patients were classified based on the number of abnormalities found by each technique (no abnormalities, 1–2, 3–4, and ≥ 5) to compare the concordance on risk stratification based on genomic complexity [4]. OGM showed an increased proportion of patients with 3–4 abnormalities and ≥ 5 abnormalities compared to CBA (14 vs. 8, p = 0.176 and 38 vs. 17, p < 0.001, respectively). When focusing on individual cases, a good agreement was observed between methods (κ = 0.362; p < 0.001) (Figure 1C, Table S4). Therefore, parallel analyses of the abnormalities detected by CBA and OGM were also performed. Interestingly, 23/25 (92%) of CK patients by CBA showed ≥ 5 abnormalities by OGM (Figure 1C, Table S5). Similarly, OGM identified at least three abnormalities in 28 non-CK patients by CBA. Among the non-CK patients with 3–4 abnormalities by OGM (n = 13), two patients (15%) showed CNA and SV predicted to be potential CK by CBA. When focusing on the non-CK group with ≥ 5 abnormalities by OGM (n = 15), they mostly showed additional SVs and/or CNAs allowing karyotype reinterpretation and demonstrating that some abnormalities described by CBA were more complex than initially assumed (Table S6). Indeed, manual revision of OGM and CBA results suggested the presence of a potential CK in 11/15 (73%). Taking all these observations together, the cut-off for genomic complexity identification by OGM was set at 5 abnormalities. Interestingly, those patients considered complex by both techniques and those complex only by OGM showed a similar incidence of ATM deletion (43.5 vs. 53.3%, respectively, p = 0.741) and TP53 deletion (43.5 vs. 26.7%, respectively, p = 0.330) by FISH. Both groups displayed a significant enrichment compared to patients with a non-CK by CBA and OGM (17.7% ATM and 6.5% TP53 in the non-CK group, p < 0.05).

Globally, OGM showed a good performance for the detection of CK previously described by CBA with 92% sensitivity and 81% specificity, which represented an area under the ROC curve (AUC) of 0.863 (95% CI: 0.781–0.944, p < 0.001). Further, if patients predicted to be potential CK after visual revision were also considered as “true complex cases”, OGM demonstrated an enhanced performance (sensitivity: 94%; specificity: 94%) with an AUC value of 0.942 (95% CI: 0.887–0.997, p < 0.001). Further, OGM detected chromoanagenesis events not otherwise identifiable by CBA in 14 (14%) patients, who displayed a median of 21 abnormalities per case (range: 6–87). Chromothripsis and chromoplexy were the most represented phenomena with 8 (8%) and 7 (7%) cases, respectively, while chromanasynthesis was found in 3 cases (3%). Notably, one patient displayed a coexistence of chromothripsis and chromanasynthesis, and two presented all three events (Figure 1A). The regions involved in these phenomena were distributed along the genome and none of them was recurrent.

Overall, our findings align with previous studies [5, 6], confirming the good performance of OGM in the detection of genomic complexity. Moreover, using a cut-off of ≥ 5 alterations to define genomic complexity, this methodology demonstrated 92% sensitivity and 81% specificity, with 15% of non-CK cases reclassified as complex by OGM. After a manual review to identify potential complexity underestimated by CBA, sensitivity and specificity potentially improved to 94%, underscoring the value of expert interpretation alongside automated pipelines. However, the manual review has limitations, such as the potential for subjective biases and increased analysis time. The high frequency of high-risk FISH alterations identified in those patients reclassified as complex by OGM, comparable to those also CK by CBA, suggests that OGM may indeed reflect biologically significant genomic complexity, but further validation is needed. Although CBA remains the gold standard for CK detection, the capacity of OGM to uncover additional layers of complexity offers an opportunity to refine risk stratification in the era of targeted therapies for CLL.

In conclusion, OGM is a powerful tool for cytogenomic characterization in CLL, providing higher resolution for detecting CK and uncovering cryptic alterations that complement those identified by standard techniques. OGM offers significant advantages, such as independence of cell division, a high resolution (6Kb), and the ability to detect both CNA and SV in one test. However, its limitation to detect low-clonal abnormalities highlights the need of integrating it into diagnostic workflows alongside FISH for TP53 deletions testing, which remains indispensable due to its superior sensitivity. Future research, particularly prospective studies, is essential to validate these findings, establish the prognostic implications of OGM-detected complexity, and facilitate its integration into routine diagnostics to improve therapeutic decision-making in CLL.

The authors declare no conflicts of interest.