Introduction:

Immunochemotherapy remains the cornerstone of treatment for Mantle Cell Lymphoma (MCL). However, 25% of patients experience early progression, with survival rates of less than two years. Current prognostic tools, such as the MCL International Prognostic Index (MIPI), and poor prognostic histological and genetic features are insufficient for stratifying patients into individualized therapeutic strategies. This study aimed to identify biomarkers for high-risk MCL patients using an integrated analysis of clinical and biological factors.

Methods:

We analyzed data from 299 patients enrolled in the LyMa phase 3 trial, with a focus on high-risk patients, defined by refractoriness to immunochemotherapy or relapse within 12 months post-autologous stem cell transplantation. We used optical genome mapping (OGM) on frozen samples, alongside whole-exome sequencing (WES), RNA sequencing, and DNA methylation arrays analyses on FFPE tumor biopsies to identify genetic, transcriptomic and epigenetic alterations. Machine learning models, including random forest analysis and Partial Least-Squares Discriminant Analysis (PLS-DA), were employed to predict high-risk MCL status.

Results:

Among the 299 patients, 31 (10.4%) were identified as high-risk (HR) with a median overall survival of 8.5 months after relapse. HR patients exhibited significantly higher levels of LDH, higher‐risk MIPI scores (45% vs. 16%, p<0.001), Ki‐67 >30% (71% vs. 31%, p<0.001) and blastoid/pleomorphic histology (32% vs. 9%, p<0.001). In multivariate analysis, only high‐risk MIPI score, and Ki‐67 >30% were associated with HR MCL. These factors were insufficient to specifically capture HR patients, as one-third of long-term responders would have been misidentified as high-risk.

The high-risk (HR) subgroup displayed a greater burden of complex genetic alterations, with significantly increased frequencies of TP53 alterations (OR 25.4, p < 0.001), CDKN2A deletions (OR 4.5, p = 0.015), RB1 deletions (OR 4.9, p = 0.024), MYC gains (OR 5.8, p = 0.047), and MIR17HG gains (OR 11.8, p = 0.013). To improve predictive accuracy, an integrative analysis combining well-established prognostic markers with gene alterations assessed by WES, was performed. Random forest analysis achieved a test accuracy of 91% when predicting HR MCL status, with a ROC AUC of 96%. The sensitivity was 84% and the specificity was 96%, with a misclassification rate of 14%. The most influential features included the Ki-67 index, histological subtype, TP53 alterations, MIPI score, and gains of MYC and MIR17HG.

Unsupervised Uniform Manifold Approximation and Projection (UMAP) analysis of gene expression profiling on 49 FFPE samples, including 15 HR MCLs, showed that HR MCLs tended to cluster together, but the distinction was not perfect. Supervised analyses, using PLS-DA, indicated potential overfitting, suggesting that transcriptomic signals alone are insufficient for perfect discrimination. In contrast, DNA methylation analysis of 29 FFPE samples, including 12 HR MCLs, revealed a distinct epigenetic signature that robustly discriminated HR MCLs from control cases. Supervised approaches (PLS-DA) identified differentially methylated probes (DMPs, n=225) that perfectly discriminated HR MCL from controls. Importantly, this epigenetic signature was validated in an independent cohort (Barcelona cohort, n=64).

To explore the genome-wide impact of DNA methylation on gene expression, we performed correlation analyses between promoter methylation and transcriptomic data across all protein-coding genes. A subset of genes showed significant correlations, with a predominant inverse relationship in HR cases, absent in controls, indicating that promoter hypermethylation may drive transcriptional deregulation in this subgroup. Notably, CHL1, a tumor suppressor, and KLHL6, associated with chemoresistance, demonstrated strong inverse correlations between methylation and expression, supporting their involvement in HR MCL pathogenesis.

Conclusion: This study provides an integrated characterization of high-risk MCL, identifying a novel epigenetic signature that outperform traditional prognostic markers. Our baseline epigenetic approach may enhance patient stratification and support the development of personalized therapies. These results support the combined analysis of genetic and epigenetic features to capture MCL's full biological complexity.

This content is only available as a PDF.
2025
Sign in via your Institution