Background: In recent years, CAR-T therapy has gradually emerged as a treatment option for relapsed/refractory primary central nervous system lymphoma (R/R PCNSL). However, patient responses to CAR-T treatment vary significantly, with only approximately 30% of patients achieving long-term survival. Whether molecular heterogeneity correlates with the heterogeneous treatment responses to CAR-T therapy in R/R PCNSL patients remains unreported. To explore the prognostic factors influencing CAR-T therapy outcomes in R/R PCNSL, we conducted a retrospective analysis of the genomic heterogeneity and its impact on treatment efficacy and prognosis in R/R PCNSL patients at our center.

Method: This study retrospectively analyzed 49 patients with R/R PCNSL. Targeted next-generation sequencing (NGS) was performed to profile mutations across 200 lymphoma-relevant genes. Logistic regression and Cox proportional hazards models were applied to evaluate the impact of molecular heterogeneity and baseline clinical characteristics on CAR-T treatment outcomes. A three-tiered prognostic risk stratification model was constructed by integrating genomic heterogeneity with key clinical features.

Result: This study enrolled 49 R/R PCNSL patients (median age 59 years, range 26-76; 59.2% male) including 15 relapsed (30.6%) and 34 refractory cases (69.4%), with 37 patients (75.5%) receiving CAR-T therapy. Genomic analysis identified high-frequency mutations in MYD88 (28.6%), CD79B (24.5%), PIM1 (22.4%), KMT2D (20.4%), CREBBP (16.3%), TP53 (14.3%), BTG2 (14.3%), B2M (12.2%), EZH2 (12.2%), and HIST1H1E (10.2%), involving key pathways including NF-κB signaling, cell cycle regulation, and immune evasion.

Thirty-seven patients receieved CAR-T treatment. The overall response rate to CAR-T therapy was 67.6% (25/37). With a median follow-up of 12.4 months, the median progression-free survival (PFS) was 8.2 months (95% CI: 5.6-14.3) and median overall survival (OS) was 15.1 months (95% CI: 9.8-not reached). The 1-year PFS and OS rates were 39.2% (95% CI: 24.6-53.8%) and 58.6% (95% CI: 42.1-71.8%), respectively. Among CAR-T recipients, univariate analysis demonstrated TP53 mutations (OR=0.15, p=0.008), B2M mutations (OR=0.12, p=0.018), and IELSG≥2 (OR=0.29, p=0.040) predicted poorer CR rates, while IRF4 mutations enhanced response (OR=8.21, p=0.044). TP53 mutations, B2M alterations, IELSG≥2, and MSKCC high-risk status were significantly associated with inferior OS and PFS outcomes. Notably, the more frequently mutated genes MYD88, CD79B, PIM1 showed no significant association with CR rates, PFS and OS (all p>0.20). Multivariate analysis confirmed TP53 as an independent CR predictor (adjusted OR=0.18, p=0.026). TP53 mutations and IELSG≥2 independently predicted worse PFS (HR=3.10 and 2.80) and OS (HR=3.50 and 3.00; all p<0.01).

This study developed a three-tiered prognostic risk stratification model for R/R PCNSL patients receiving CAR-T therapy based on four key prognostic factors: TP53 mutation status, IELSG score, MSKCC risk classification, and KPS performance status. The low-risk group (n=12), defined as TP53 wild-type patients with IELSG<2, KPS>60 and MSKCC low/intermediate-risk, showed excellent outcomes with 83.3% 1-year OS and median PFS not reached. The intermediate-risk group (n=15), characterized by any single risk factor (TP53 mutation or IELSG≥2 or MSKCC high-risk or KPS≤60), demonstrated intermediate survival with 46.7% 1-year OS and median PFS of 187 days. The high-risk group (n=10), consisting of TP53-mutated patients with at least one additional risk factor, had markedly poor outcomes with only 10.0% 1-year OS and significantly shorter median PFS of 54 days (stratification p<0.001).

Conclusion: Through comprehensive analysis of genomic and clinical baseline characteristics in R/R PCNSL patients receiving CAR-T therapy, this study identified TP53 mutations, IELSG≥2, and MSKCC high-risk classification as independent adverse prognostic factors. The developed three-tiered prognostic risk stratification model, integrating both genomic markers and clinical features, enables early identification of high-risk patients and provides a framework for risk-adapted precision treatment strategies.

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