Exploring the synergy between telomere length and genomic complexity in <scp>CLL</scp>

Telomeres are repetitive sequences that protect chromosome ends from being recognized as DNA damage during cell division. They progressively shorten as cells divide, serving as a hallmark of cellular ageing.1 This process also plays a role in cancer, where critical attrition leads to short, dysfunctional telomeres prone to chromosomal fusions, ultimately causing genomic instability. In chronic lymphocytic leukaemia (CLL), reduced telomere length (TL) associates with adverse prognostic factors, including genomic complexity, TP53 aberrations, unmutated-IGHV (U-IGHV), specific DNA methylation profiles and unfavourable outcomes even in early-stage disease.2-5 Genomic complexity in CLL, detected by chromosome banding analysis (CBA) as complex karyotypes (CK; ≥3 chromosomal abnormalities in a single clone) or by genomic microarrays (GC by GM; ≥3 copy number aberrations [CNA]), constitutes a poor prognostic/predictive factor.6, 7 Some authors reported that telomere shortening in CLL reflects a more aggressive disease phenotype associated with genomic complexity, underscoring the relevance of integrating these parameters for improved prognostication.8 Investigating the interplay between TL and CK/GC could improve our understanding of CLL pathogenesis and refine prognostic models for these challenging cases. The present study aims to characterize TL in a cohort of CLL patients enriched in cases with high genomic complexity and explore its relationship with other biomarkers to elucidate the impact of telomere dynamics in the context of genome instability. A total of 149 untreated CLL patients from 14 European institutions were included. Samples were collected at diagnosis or before treatment initiation, within a maximum of 15 months from the date of CBA (median time from CBA to sampling = 0 months; range: 0–14). Due to the purpose of the study, the cohort was enriched in CK cases (n = 60, 40.3%). Demographics are provided in Table S1. Absolute TL was assessed using monochrome multiplex quantitative-polymerase chain reaction (MMQ-PCR)9, 10 and correlated with clinico-biological parameters and outcome. Median follow-up was 66 months (range: 0–178), during which 67 patients died. The median TL of the entire cohort was 2.82 Kb (range: 2.31–5.10 Kb). When TL was assessed taking into account the quantification of abnormalities detected by either CBA or GM, a progressive TL reduction was found as complexity increased, being patients with high-CK (≥5 aberrations) or high-GC (≥5 CNA) those who displayed the shortest median TL (2.53 and 2.58 Kb, respectively) compared to non-complex cases (2.97 Kb; p<0.001 and 2.99 Kb; p<0.001, respectively) (Figure 1A,B; Figure S1). Next, when integrating the complexity detected by both methods along with TL, patients with 0–2 aberrations in both assays exhibited longer TL, whereas those with ≥5 abnormalities tended to have shorter TL (Figure S2). To explore whether TL was associated with other well-established prognostic factors in CLL, patients were classified according to the categories defined in Döhner's hierarchical model, which is based on FISH results. The shortest TL was observed in patients with del(11q)/del(17p) (median: 2.57 and 2.55 Kb, respectively) compared to the remaining groups (Figure 1C). Additionally, U-IGHV cases also showed shorter TL than those with mutated-IGHV (2.66 Kb vs. 3.31 Kb; p<0.001) (Figure 1D). Then, a more detailed assessment of how genomic complexity influences TL in CLL patients was performed by dividing the cohort according to CK status. Expectedly, patients with CK had shorter median TL than non-CK patients (2.68 Kb vs. 2.97 Kb; p<0.001) (Figure 1E). When CK was combined with other high-risk biomarkers, it was significantly associated with reduced TL in most subgroups. The only exceptions were found in the delATM subgroup, where the presence of CK did not significantly impact TL (2.63 Kb vs. 2.53 Kb; p=0.154), and in the del/mutTP53 subset. Of note, the latter included only four non-CK cases, making statistical analysis unfeasible due to the small sample size (Figure 1F–H). Similar results were observed when considering GC assessed by GM (Figure S3). Besides, we explored the association between chromothripsis and TL. Interestingly, the 15 patients with chromothripsis showed a shorter median TL than non-chromothriptic cases (2.53 Kb vs. 2.86 Kb; p=0.001), consistent with previous reports indicating that telomere attrition can lead to end-to-end chromosomal fusions, facilitating catastrophic chromosomal rearrangements.11 However, when comparing chromothriptic cases with those carrying intermediate/high-GC, TL was not significantly different (Figure S3E,F), suggesting that telomere erosion may be a shared factor underlying both chromothripsis and genomic complexity. It is also possible that chromothripsis arises from the attrition of one or a few specific chromosome ends, which may remain undetected with the global TL measurement approach employed in this study. Regarding the prognostic impact of TL, the median value of the whole cohort was used for dichotomization into short-TL (<2.82 Kb) and long-TL (≥2.82 Kb) subgroups. Supporting previous findings, short-TL was significantly associated with shorter time to first treatment (TTFT; 3 months vs. 79 months; p<0.001) (Figure 2A) and overall survival (OS; 80 months vs. median not reached; p=0.001) (data not shown) compared to patients with long-TL. Subsequently, patients were stratified according to the presence of CK, del/mutTP53, delATM and U-IGHV to assess the clinical impact of TL combined with other established factors, revealing different prognostic subgroups. Overall, in patients lacking CK, TP53 or ATM abnormalities, or U-IGHV, the presence of short-TL significantly aggravated TTFT. Notably, short-TL worsened the outcome of both del/mutTP53 and CK cases (Figure 2B,C), while no significant differences were observed in the number of cytogenetic abnormalities between short-TL and long-TL patients within the CK subgroup (median: 5 [3–19] vs. 4 [3–8]; p=0.206). These findings support the notion that TL provides complementary prognostic information beyond traditional markers. Interestingly, this impact was not observed within the delATM or U-IGHV subset (Figure 2D,E). Multivariate Cox regression analysis (MVA) including CK, del/mutTP53, U-IGHV, delATM and TL confirmed short-TL as an independent TTFT predictor, suggesting that TL measurement could improve risk stratification of these high-risk patients. Indeed, TL was the dominant covariate for TTFT, showing the highest hazard ratio (HR: 2.15; 95% CI: 1.21–3.81) (Figure 2F). No significant multicollinearity was observed (variance inflation factor [VIF] analysis; data not shown). Likewise, when patients were classified based on GC by GM, TL allowed the separation of two subsets with different prognosis within the low-GC and the intermediate/high-GC subgroups (Figure S4A). Again, short-TL retained its prognostic impact in the MVA together with del/mutTP53, while GC did not reach statistical significance (Figure S4B). In summary, our findings confirm that TL shortens as genomic complexity increases, irrespective of the detection method, supporting the hypothesis that telomere erosion contributes to the acquisition of genomic complexity.8, 12 While TL shortening correlated with delTP53, IGHV status and delATM, no significant differences in TL were observed among delATM patients based on CK status, supporting previous observations showing that CLL patients with abnormal ATM display extreme telomere erosion.13 ATM is a key regulator of the DNA damage response and telomere maintenance and its deficiency has been associated with impaired recognition and repair of dysfunctional telomeres. Our results suggest that ATM loss induces such profound telomere shortening that it masks the TL differences associated with increasing genomic complexity. Mutational analysis of ATM and BIRC3 genes might add further detail, as biallelic BIRC3 loss would provide pro-survival signals through NFKB without affecting TL or CK.14 Concerning its prognostic value, our data show that TL remains a powerful prognostic factor even when combined with established high-risk biomarkers. Nonetheless, this study has some limitations that should be considered. Although different sample types were used for TL measurement, no significant differences were observed, indicating that no technical bias was introduced (Figure S5). The study is based on a retrospective cohort highly enriched in CK cases, and certain subgroups were represented by a limited number of cases; therefore, these findings require validation in larger and more representative populations. Additionally, despite being a promising biomarker, the clinical application of TL is hindered by the lack of standardization in measurement techniques and the heterogeneity of cohorts across studies. Currently, several methodologies are available for TL assessment, including qPCR, single telomere length analysis (STELA), FISH, flow cytometry or NGS. Even though the PCR-based technique employed in this work provides relative TL measurements, these can easily be converted into absolute values. Besides, its simplicity, cost-effectiveness, ability to handle large sample numbers and no need for advanced equipment might facilitate its incorporation into clinical trials. Similar to other biomarkers, TL has been established as an independent prognostic factor in CLL patients treated with chemoimmunotherapy.3 Notwithstanding, the therapeutic management of CLL has evolved significantly with the advent of targeted therapies, and in this context, the role of TL remains uncertain. The CLL14 trial is the first study to date suggesting that the predictive impact of TL is neutralized by the venetoclax–obinutuzumab combination.15 Therefore, TL should be evaluated in prospective trials using non-genotoxic agents to clarify its value in risk stratification and guide treatment decisions. In conclusion, our study highlights the strong association between TL, genomic complexity and high-risk biomarkers in CLL, supporting its potential as a prognostic factor. Despite limitations related to cohort composition, TL emerged as an independent predictor of clinical outcome. However, future prospective studies with larger cohorts and standardized methodologies are needed to further validate the clinical utility of TL, particularly in the context of targeted therapies. SRC, AP and BE designed the research study, compiled patients data and analysed the data; SB and JS performed a great proportion of genomic microarrays; HP was involved in telomere length assessment; JCS was involved in interpretation and critical discussion of the results; GMR, CH, MJL, RC, RS, MJC, AV, PA, FB, EG, AC and FNK provided data and samples. All authors read and reviewed the final version of the manuscript. The authors want to thank María Laura Blanco, María Dolores García-Malo and Marco Antonio Moro for their contribution to the study, providing patient samples and data, and Generalitat de Catalunya (21SGR00025) and the Spanish Ministry of Health for partly supporting this work. SRC was supported by Ministerio de Ciencia, Innovación y Universidades of Spain (FPU17/00361) and Fundación Española de Hematología y Hemoterapia (FEHH-AstraZeneca). JCS is funded by Bloodwise (11052, 12036), the Kay Kendall Leukaemia Fund (873), Cancer Research UK (ECRIN-M3 accelerator award C42023/A29370, Southampton Experimental Cancer Medicine Centre grant C24563/A15581, Cancer Research UK Southampton Centre grant C34999/A18087 and programme C2750/A23669) and the Bournemouth Leukaemia Fund. This work has not received any specific funds. The authors declare no relevant conflicts of interest. This study was conducted in accordance with the Declaration of Helsinki Good Clinical Practice, and all applicable regulatory requirements, and approved by the Drug Research Ethical Committee (CEIm) of the Hospital del Mar, Barcelona (2024/11774). Patients provided informed consent for the collection of samples and data. Further data are available upon reasonable request from the corresponding author. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.