AUTHORS
Allison Kayne, MD, Zain Hannan, MD2, Anna Bondonese, MD3, Sukhrob Makhkamov, DO4, Anisha Jobanputra, MD5, Sean Rilley, MD6, Michael Wysota, MD7
AFFILIATIONS
1 Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, United States
2 Lankenau Medical Center, Philadelphia, Pennsylvania, United States
3 Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, United States
4 Jefferson Torresdale Hospital, Philadelphia, Pennsylvania, United States
5 Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, United States
6 Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, United States
7 Thomas Jefferson University Hospital, Philadelphia, Pennsylvania, United States
ABSTRACT
I. Background
Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma. Although frontline immunochemotherapy has improved outcomes, CNS relapse remains a rare but devastating complication associated with poor survival. The CNS International Prognostic Index (CNS-IPI) identifies patients at increased risk for CNS relapse (Schmitz et al., 2016), and patients with high CNS-IPI scores are commonly considered for CNS prophylaxis (ppx). Prior studies have identified involvement of the testes, kidney/adrenal glands, breast, bone marrow, and multiple (>3) extranodal sites as additional risk factors for CNS relapse (Boehme et al., 2009; Schmitz et al., 2016). However, the risk of CNS relapse and the role of CNS ppx in patients with extranodal sites adjacent to the CNS, including the sinonasal region, orbit, oral cavity, epidural space, and paraspinal region, remain poorly defined and these patients are not specifically represented in current risk stratification models. We evaluated patterns of CNS ppx use and clinical outcomes in this understudied population.
II. Methods
A retrospective cohort of DLBCL patients treated at Thomas Jefferson University Hospital (TJUH) from 2014–2025 was identified via EPIC SlicerDicer (~2,000 patients). Eligible patients had extranodal disease involving the oral cavity, nasal cavity, orbit, paranasal sinuses, epidural space, or paraspinal region. Patients with CNS involvement at diagnosis were excluded. Demographic, disease characteristics, CNS ppx, relapse, and survival data were collected. Statistical analysis included Fisher exact and Wilcoxon rank-sum tests, Cox proportional hazards models, and Kaplan-Meier estimates.
III. Results
91 patients met inclusion criteria, of whom 29 (32%) received CNS ppx. Median age was 62 years (IQR, 55.5–71), and 55% were male. No significant differences were seen in age, sex, race, or extranodal disease between patients who did or did not receive CNS ppx. Median follow-up was longer among patients receiving ppx (58.8 vs. 37.8 months, p=0.033). Intrathecal methotrexate (41%) and high-dose methotrexate (34%) were the most frequently used prophylactic regimens. Patients receiving ppx had significantly higher-risk disease characteristics, including IPI scores of 3–5 (59% vs. 34%, p=0.039) and Ann Arbor stage IV disease (61% vs. 31%, p=0.031), with a higher, though not statistically significant, prevalence of MYC/BCL2 double-expressor or double-hit disease (38% vs. 18%, p=0.064). Median OS was numerically longer among patients receiving ppx than those who did not (56 vs. 35 months; HR 2.34, 95% CI 0.75–7.26, p=0.141). CNS relapse occurred in 2 of 29 patients (6.9%) receiving ppx compared with 10 of 62 patients (16%) who did not receive ppx (p=0.30). In a multivariable Cox model adjusting for MYC/BCL2 status and Ann Arbor stage, ppx among patients with lower IPI scores demonstrated a trend toward improved overall survival, although this did not reach statistical significance (HR 0.073, 95% CI 0.0005–1.12, p=0.062).
IV. Discussion
In this study, patients with extranodal sites adjacent to the CNS who received CNS ppx demonstrated a trend toward longer overall survival and fewer CNS relapses compared with controls, although neither outcome reached statistical significance likely due to low sample size. Notably, the prophylaxis group had a higher prevalence of adverse disease characteristics, including higher IPI scores, stage IV disease, and MYC/BCL2 alterations. Despite this higher-risk profile, these patients experienced fewer CNS relapses, supporting further evaluation of CNS ppx in anatomically high-risk populations. Of note, recent multicenter data evaluating high-dose methotrexate ppx in high-risk DLBCL demonstrated no significant reduction in CNS relapse, questioning the benefit of routine ppx and highlighting the need for improved patient selection (Wilson et al., 2026). In our cohort, multivariable analysis demonstrated a trend toward benefit of overall survival among patients with lower IPI scores, suggesting that anatomic disease location may provide additional risk stratification beyond traditional clinical factors. Prospective studies are needed to identify specific patient subsets, including those with extranodal disease adjacent to the CNS, who may benefit most from CNS ppx.
DISCLOSURES
Allison Kayne, MD: No relevant conflicts of interest to declare.