# Genomic risk factors for central nervous system relapse in patients with diffuse large B-cell lymphoma

## TL;DR

This study identifies genomic risk factors that can predict central nervous system relapse in patients with diffuse large B-cell lymphoma, potentially improving treatment strategies.

## Contribution

The study introduces a genomic classifier that outperforms existing clinical tools in predicting CNS relapse risk in DLBCL patients.

## Key findings

- MYD88 L265P and CDKN2A loss, along with mutations in TET2, ARID1A, and INO80, are independently linked to CNS relapse risk.
- The genomic classifier achieved an AUROC of 0.91, significantly better than the IPI (0.77) or IPI+COO (0.81).

## Abstract

Central nervous system (CNS) relapse is associated with poor survival, and remains an unmet challenge in patients with diffuse large B-cell lymphoma (DLBCL). Identifying patients at high risk of CNS relapse and offering prophylactic treatment could improve patient prognosis.

Here, we studied 234 patients with DLBCL using open patient-level clinical and sequencing data to explore risk factors for CNS relapse. Patients were divided into Cohort A (CNS involvement at baseline), Cohort B (CNS recurrence), and Cohort C (patients without secondary CNS involvement and with a follow-up interval > 3 years). We investigated the risk factors for CNS relapse in Cohorts B + C.

Genetic alterations with statistical significance, determined by univariate analysis, and an incidence rate ≥ 5%, together with clinical factors, correlated with CNS relapse risk in a multivariate analysis. Multivariate logistic regression analysis revealed that concomitant MYD88 L265P and CDKN2A loss (p = 0.012), TET2 mutation (p = 0.037), ARID1A mutation (p = 0.010), and INO80 (p = 0.002) were independently correlated with a high risk of CNS relapse after adjusting for the IPI risk groups, B symptom and cell of origin (COO). The classifier that integrated genomic risk factors was superior in predicting CNS relapse (area under the receiver operating characteristic curve [AUROC]: 0.91) compared with the IPI (AUROC: 0.77, p < 0.001) or IPI in combination with COO classifiers (AUROC: 0.81, p = 0.013).

This study identified several genomic alterations as risk factors for CNS relapse.

The online version contains supplementary material available at 10.1007/s44313-025-00087-1.