Background: Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma with significant pathological heterogeneity. Age at diagnosis is a well-established prognostic factor of DLBCL. It is incorporated into the International Prognostic Index (IPI) as a dichotomized variable, with age over 60 classified as an unfavorable risk feature, yet its relationship to the molecular features of DLBCL remains incompletely defined. Prior studies have reported age-associated variation in the frequency of specific genetic alterations, subtype prevalence, and cytogenetic complexity, though the sample sizes were limited with few younger patients included.

Methods: To explore age differences in tumor molecular landscape, we aggregated molecular data from 1,239 DLBCL patients from eight published studies with available clinical and somatic mutation or copy number aberration (CNA) data, with patient ages ranging between 2 and 93 years old, including 25 patients ≤ 18 years, 115 patients between 19 and 39 years, and 1,099 patients ≥ 40 years. Mutation frequencies in established DLBCL driver genes were evaluated for associations with age at diagnosis using a mixed-method statistical analysis combining three tests: logistic regression, Fisher's exact test, and two-sided t-test. Overall survival in DLBCL patients were assessed across age groups by Kaplan–Meier curves and log-rank test. Cox proportional hazard models were used to estimate hazard ratios (HRs) to identify the prognostic significance of select clinical variables and DLBCL driver genes.

Results: Of the 180 DLBCL driver genes analyzed, 16 were found to mutate at a different frequency by age of diagnosis. These included both well-known oncogenes, namely BCL10, BCL2, CD79B, MYC, MYD88, PIM1, and tumor-suppressor genes, namely FAS and SIN3A. Mutations in DDX3X, MYC, and SIN3A were much more prevalent in younger patients ≤ 18 years of age. In contrast, BCL2, PIM1, and MYD88 mutations were more frequent in patients aged ≥ 40 years. Of the 51 CNA regions analyzed, none were statistically significant in association with age at diagnosis. In survival analysis, older patients had poorer survival, with patients aged ≥ 70 having the highest mortality. In multivariable Cox regression models, age at diagnosis was one of the most significant predictors of survival (HR= 1.04, p-value = <0.001) in DLBCL patients, only second to initial treatment response. The other significant clinical variables were high IPI scores and cell of origin subtypes. In analysis of the age-associated driver genes, KLHL14, which was more frequently mutated in older patients, was prognostic for overall survival. Patients with KLHL14 mutations had the worst survival outcomes, which remained significant after adjusting for age (HR= 1.68, P = 0.012).

Conclusion: In a large DLBCL patient population pooled from multiple studies, we identified somatic mutational differences in multiple known driver genes by patient age at diagnosis, which signifies the impact of age in defining DLBCL molecular pathology. Our analyses identified age at diagnosis as among the most significant independent predictors of survival in DLBCL patients, consistent with previous findings.

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