Clinical features, central nervous system risk, and therapeutic outcomes in primary sinonasal diffuse large B-cell lymphoma: a retrospective cohort study of 94 patients

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common histological subtype of non-Hodgkin lymphoma. However, primary involvement of the sinonasal region is uncommon, accounting for roughly 1–2% of lymphomas (1,2). Primary sinonasal DLBCL (PSN-DLBCL) and extranodal NK/T-cell lymphoma are the predominant histologies in this region (ENKTL), which exhibit distinct anatomical predilections: DLBCL arises more frequently in the paranasal sinuses (3), while ENKTL typically originates in the nasal cavity (4). According to Surveillance, Epidemiology, and End Results (SEER) database estimates, the incidence of PSN-DLBCL is approximately one case per million individuals (5). The maxillary sinus is the most commonly involved subsite, followed by the nasal cavity; orbital extension is frequently observed, whereas the frontal sinus is rarely affected (6). Because the sinonasal compartment is relatively concealed, tumors may remain clinically occult until they invade adjacent critical structures, and patients frequently present with locally advanced disease.

It is well-established that the heterogeneity and clinical outcomes of DLBCL are closely associated with the primary site of involvement, as evidenced in primary breast, testicular, and mediastinal DLBCL (7,8). Nevertheless, PSN-DLBCL has been described mostly in small series and case reports, and its clinical course, patterns of dissemination, and optimal management remain incompletely defined. Clinically, these patients frequently pose a therapeutic challenge, involving critical decisions regarding the selection of standard immunochemotherapy versus more intensive treatment regimens, the potential utility of consolidative radiotherapy, and the necessity of central nervous system (CNS) prophylaxis.

We therefore conducted a retrospective cohort study of treatment-naïve patients with PSN-DLBCL to describe clinicopathologic features, treatment patterns, and prognosis, and to explore the incidence and presentation of CNS involvement and relapse; given the retrospective study design and limited events, analyses are exploratory and hypothesis-generating rather than conclusive. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0107/rc).

Methods

Study design

This retrospective cohort study enrolled consecutive, treatment-naïve patients with newly diagnosed PSN-DLBCL who received initial treatment at Beijing Tongren Hospital, Capital Medical University, between October 2009 and December 2024. PSN-DLBCL was defined as histologically confirmed DLBCL with the dominant tumor mass located in the nasal cavity and/or paranasal sinuses, and with initial clinical presentation attributable to the sinonasal lesion (for example nasal obstruction, epistaxis, facial pain or swelling). Expert hematopathologists confirmed all pathological diagnoses, and immunohistochemical verification of CD20 expression was obtained for every tumor specimen.

Diagnosis and evaluation

Baseline assessment and response evaluations were systematically performed by whole-body positron emission tomography/computed tomography (PET/CT), magnetic resonance imaging (MRI) of the head and neck, supplemented by the contrast-enhanced CT. Assessments were routinely performed at baseline, at an interim timepoint (after 2–4 cycles), and at end of treatment. Disease staging was assigned according to the Ann Arbor staging system, and treatment response was categorized based on the Lugano classification response criteria.

CNS involvement was determined using a combination of PET/CT, cranial MRI, and lumbar puncture with cerebrospinal fluid (CSF) analysis. Flow cytometric immunophenotyping of CSF was conducted. Bone marrow aspiration and biopsy were performed in patients with disseminated disease, PET/CT findings suggestive of marrow infiltration, or unexplained pre-treatment hematological abnormalities.

We defined extensive local invasion (ELI) as radiologic tumor extension beyond the bony confines of the sinonasal complex into at least two adjacent anatomical structures (e.g., orbit, skull base, pterygopalatine fossa, hard palate, or overlying skin), as assessed on imaging review.

Treatment protocols

All patients were scheduled to receive 6–8 cycles of chemotherapy unless precluded by intolerable toxicity or disease progression. The selection of the specific chemotherapeutic regimen was at the treating physician’s discretion, based on disease burden, patient performance status, and comorbidities. The chemotherapeutic regimens included R-CHOP (rituximab 375 mg/m², cyclophosphamide 750 mg/m², doxorubicin 50 mg/m², vincristine 1.4 mg/m², and prednisone 100 mg), da-R-EPOCH (rituximab 375 mg/m², dose-adjusted etoposide 50 mg/m², vincristine 0.4 mg/m², and doxorubicin 10 mg/m² via continuous infusion on days 1–4, cyclophosphamide 750 mg/m² on day 5, and prednisone 60 mg/m² on days 1–5), or Pola-R-CHP (polatuzumab vedotin 1.8 mg/kg, rituximab 375 mg/m², cyclophosphamide 750 mg/m², doxorubicin 50 mg/m², and prednisone 100 mg). The specific anthracycline agent (doxorubicin, liposomal doxorubicin, epirubicin, or pirarubicin) was selected based on individual clinical assessment. For patients with CNS involvement or prophylaxis, the treatment protocol incorporated high-dose methotrexate (HD-MTX) (3–3.5 g/m²) combined with rituximab, supplemented by intrathecal therapy with dexamethasone, cytarabine, and methotrexate. The standard cycle interval was 21 days. Doses were adjusted based on patient tolerance, with frail elderly patients typically receiving 50–75% of the standard doses.

Ethics statement

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethicsl Committee of Beijing Tongren Hospital, China (No. TREC2022-KY103). The need for informed consent was waived for this retrospective chart review, which utilized only pre-existing medical records.

Statistical analysis

All statistical analyses were performed using R software (version 4.5.0). The primary endpoints of this study were overall survival (OS) and progression-free survival (PFS). OS was defined as the time from diagnosis to death from any cause, and PFS as the time from diagnosis to disease progression, relapse, or death from any cause. Patients who were lost to follow-up or event-free were censored at the date of their last known contact.

Categorical variables were summarized as frequencies and percentages. Survival curves for OS and PFS were generated using the Kaplan-Meier method, and comparisons between groups were performed with the log-rank test, implemented with the survival and survminer packages. Median follow-up time was estimated using the reverse Kaplan-Meier method, and analysis was performed using the survival package. The cumulative incidence of relapse (CIR) and non-relapse mortality (NRM) were calculated in a competing risks framework using the cuminc function from the cmprsk package, with group differences assessed by Gray’s test.

To identify potential prognostic factors, univariate Cox proportional hazards regression models were first fitted for OS and PFS. Variables that demonstrated an association with a significance level of P<0.1 in the univariate analysis, along with factors deemed clinically relevant, were subsequently incorporated into multivariate Cox proportional hazards models. The results of these analyses were presented as hazard ratio (HR) with 95% confidence interval (CI) and were visualized using forest plots (forestplot package).

We performed a landmark analysis to evaluate the prognostic impact of interim and end-of-treatment (EOT) response on survival outcomes, with two prespecified patient cohorts defined as follows: the interim analysis cohort included patients eligible for inclusion if they received ≥2 cycles of systemic treatment and had a PFS duration of more than 60 days; the EOT analysis cohort comprised patients who completed ≥6 cycles of systemic treatment and had a PFS duration exceeding 180 days. Kaplan-Meier survival analysis was conducted to assess the association between interim/EOT response and survival, with day 60 and day 180 designated as the landmark time points for the interim and EOT survival analyses, respectively.

Associations between categorical variables, including adverse event rates and baseline characteristics, were compared using Fisher’s exact test or the Chi-squared test, as appropriate. All statistical tests were two-sided, and a P value of <0.05 was considered statistically significant.

Results

Patient characteristics and disease presentation

This retrospective analysis included 94 treatment-naïve patients with PSN-DLBCL diagnosed and treated at Beijing Tongren Hospital between October 2009 and December 2024. The median follow-up duration was 43.7 months (95% CI: 32.7–63.9). Baseline characteristics and detailed anatomical involvement patterns are summarized in Table 1. The median age at diagnosis was 68.5 years (range, 31–90 years), with 68 patients (72.3%) aged ≥60 years. A male predominance was observed, comprising 58 patients (61.7%) with a male-to-female ratio of 1.6:1.

At initial presentation, 58 patients (61.7%) had Ann Arbor stage I–II disease, while 36 (38.3%) presented with advanced stage III–IV. The distribution of International Prognostic Index (IPI) scores was as follows: low risk (0–1) in 25 patients (26.6%), intermediate risk (2–3) in 52 patients (55.3%), and high risk (4–5) in 17 patients (18.1%). The CNS-IPI risk stratification showed a similar distribution, with low, intermediate, and high risk observed in 25 (26.6%), 52 (55.3%), and 17 (18.1%) patients, respectively. Elevated serum lactate dehydrogenase (LDH; the upper limit of normal was 198 U/L before 2019 and has been 250 U/L since 2019) was present in 19 patients (20.2%), and B symptoms were reported in 25 patients (26.6%). ELI on imaging was documented in 72 patients (76.6%), reflecting the locally aggressive nature of this extranodal presentation.

Concurrent involvement of both the nasal cavity and paranasal sinuses was most frequent, occurring in 67 patients (71.3%). Isolated disease confined to the nasal cavity or paranasal sinuses alone was observed in 17 (18.1%) and 10 (10.6%) patients, respectively. Notably, local bone invasion was identified in 73 patients (77.7%). Extension beyond the PSN compartment commonly involved the orbit (56 patients, 59.6%), regional lymph nodes (38 patients, 40.4%), and Waldeyer’s ring (29 patients, 30.9%). Distant lymph node involvement was present in 15 patients (16.0%). At diagnosis, CNS involvement was detected in 9 patients (9.6%) and bone marrow infiltration in 9 patients (9.6%). All patients with CNS involvement originated from the spread of the affected local structure.

Regarding initial treatment allocation (Table 1), 57 patients (60.6%) received R-CHOP, 24 (25.5%) received da-R-EPOCH, and 6 (6.4%) received Pola-R-CHP. Seven patients (7.5%) received alternative regimens, primarily targeted therapy combined with radiotherapy, due to chemotherapy intolerance. CNS-directed prophylaxis or treatment was administered to a substantial proportion: 49 patients (52.1%) received intrathecal therapy and 20 (21.3%) received HD-MTX. Consolidative radiotherapy was delivered in 9 patients (9.6%).

Comparison of baseline characteristics between treatment cohorts revealed a significant disparity in age distribution (Table 2). The R-CHOP group comprised predominantly elderly patients, with 82.5% (47/57) aged ≥60 years compared to 41.7% (10/24) in the da-R-EPOCH cohort (P<0.001, by the Chi-squared test). Gender distribution, IPI risk categories, LDH levels, HD-MTX administration, and Eastern Cooperative Oncology Group Performance Status (ECOG PS) were otherwise comparable between groups, with no statistically significant differences identified.

Survival analysis

Survival outcomes are summarized in Figure 1. The estimated 1-, 2-, and 5-year PFS rates were 83.9%, 77.8%, and 67.0%, respectively, with corresponding OS rates of 89.4%, 83.3%, and 73.1% (Figure 1A,1B). The median PFS and OS were not reached for both endpoints [The KM-estimated median (139 months) exceeded the median follow-up time (43.7 months), reflecting the low event rate and high censoring in this cohort].

Figure 1 Survival outcomes in primary sinonasal DLBCL. (A) Kaplan-Meier estimate of PFS and (B) OS for the entire cohort (n=94). Median PFS and OS were both 139 months. (C) PFS and (D) OS stratified by treatment regimen: R-CHOP (n=57), da-R-EPOCH (n=24), and Pola-R-CHP (n=6). No significant differences were observed between groups by log-rank test (P=0.60, 0.47, respectively). (E) PFS and (F) OS stratified by primary anatomical site: isolated nasal cavity (n=17), isolated paranasal sinus (n=10), or concurrent involvement of both sites (n=67). No significant differences were identified by log-rank test (P=0.54, 0.62, respectively). (G) CIR and NRM for the entire cohort. (H) CIR and NRM by primary site location (nasal only vs. sinus only vs. both). No significant differences between groups were observed by Gray’s test (P=0.89, 0.22, respectively). (I) CIR and NRM by treatment regimen (R-CHOP vs. da-R-EPOCH). No significant differences were identified by Gray’s test (P=0.37, 0.12, respectively). CIR, cumulative incidence of relapse; da-R-EPOCH, dose-adjusted rituximab, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin; DLBCL, diffuse large B-cell lymphoma; NRM, non-relapse mortality; OS, overall survival; PFS, progression-free survival; Pola-R-CHP, polatuzumab vedotin, rituximab, cyclophosphamide, doxorubicin, and prednisone; R-CHOP, rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone.

Extent of primary site involvement (nasal cavity only, paranasal sinus only, or both) showed no significant impact on survival outcomes (Figure 1C,1D). Treatment regimen did not significantly affect PFS or OS across R-CHOP, da-R-EPOCH, and Pola-R-CHP cohorts (Figure 1E,1F).

CIR and NRM are depicted in Figure 1G. Competing risks analysis demonstrated no significant differences in CIR or NRM based on primary site location (Figure 1H) or between R-CHOP and da-R-EPOCH regimens (Figure 1I).

In our cohort, CNS relapse was documented in two patients. The first patient presented with isolated CNS relapse, with no evidence of CNS involvement at baseline, multiple extranodal lesions (adrenal gland, kidney, and testis), and a CNS-IPI score of 5. The second patient had confirmed CNS involvement at initial diagnosis and died rapidly after systemic relapse. PFS for these two patients was 373 days and 362 days, respectively. An additional two patients experienced CNS progression, both occurring in the context of concurrent systemic progression, with PFS of 40 days and 152 days, respectively.

Prognosis analysis

Univariate Cox proportional hazards analysis identified several clinical factors associated with OS and PFS (Figure 2).

Figure 2 Forest plots illustrating the results of univariate Cox proportional hazards analyses for OS (A) and PFS (B). Each plot displays the HR with corresponding 95% CI for baseline clinical features and treatment-related variables, including age (≥60 vs. <60 years), sex, Advanced stage, LDH level, IPI score, B symptoms, ELI, CNS involvement, and therapeutic interventions such as IT, HD-MTX, and radiotherapy. Variables to the right of the reference line (HR =1.0) indicate an increased risk of adverse outcomes, whereas those to the left indicate a reduced risk. CI, confidence interval; CNS, central nervous system; ELI, extensive local invasion; HD-MTX, high-dose methotrexate; HR, hazard ratio; IPI, International Prognostic Index; IT, intrathecal therapy; LDH, lactate dehydrogenase; OS, overall survival; PFS, progression-free survival.

For OS, elevated LDH (HR =2.68; 95% CI: 1.16–6.15; P=0.02) and high IPI scores (HR =3.05; 95% CI: 1.30–7.17; P=0.01) were significantly associated with inferior outcomes. Advanced stage (HR =2.12; 95% CI: 0.89–5.73; P=0.08) and CNS involvement (HR =2.42; 95% CI: 0.82–7.10; P=0.10) demonstrated a trend toward poorer OS. B-symptoms (HR =1.23; P=0.54) showed nonsignificant unfavorable trends. Conversely, radiotherapy (HR =0.28; 95% CI: 0.04–2.02; P=0.21), intrathecal therapy (HR =0.61; P=0.52), and HD-MTX (HR =0.72; P=0.60) showed protective trends without reaching statistical significance.

For PFS, high IPI scores (HR =2.31; 95% CI: 1.02–5.24; P=0.045) and increased LDH (HR =2.82; 95% CI: 1.33–5.96; P=0.007) were significantly associated with inferior PFS. Advanced stage (HR =1.84; P=0.09) and CNS involvement (HR =1.88; P=0.23) showed nonsignificant adverse trends. Intrathecal therapy (HR =0.52; 95% CI: 0.26–1.08; P=0.08) and radiotherapy (HR =0.20; P=0.12) showed protective effects, while ELI (HR =0.48; 95% CI: 0.21–0.99; P=0.046) were significantly associated with improved PFS.

Multivariate Cox proportional hazards analysis identified high IPI scores as an independent adverse prognostic factor for both OS and PFS. For OS, high IPI scores were independently associated with inferior survival (HR =3.71; 95% CI: 1.41–9.72; P=0.008). For PFS, high IPI scores similarly conferred a significantly worse prognosis (HR =3.12; 95% CI: 1.25–7.80; P=0.01), and male sex emerged as an independent predictor of shorter PFS (HR =2.56; 95% CI: 1.06–6.17; P=0.037). Although ELI and radiotherapy did not reach statistical significance, both demonstrated a protective trend for PFS (ELI: HR =0.37, P=0.052; radiotherapy: HR =0.12, P=0.053), suggesting a potential, albeit inconclusive, benefit warranting further investigation. Detailed information is presented in Table 3.

Among the 94 patients, 86 underwent interim response assessment after 2–4 treatment cycles using whole-body PET/CT and regional imaging. Of these, 51 achieved complete response (CR), 31 reached partial response (PR), and four experienced progressive disease (PD). For those with response at interim assessment, patients with CR exhibited better PFS and OS than those with PR (Figure 3A,3B), with the difference in OS being statistically significant (P=0.01); but it was not significant in PFS (P=0.054).

Figure 3 Interim treatment response evaluated after 2–4 cycles of therapy. (A,B) PFS and OS stratified by interim response: CR vs. PR. Patients achieving CR exhibited superior OS compared with those achieving PR (P=0.01), while the difference in PFS did not reach statistical significance (P=0.054). (C,D) PFS and OS for CR and non-CR patients. Much more favorable PFS and OS were observed (C: P=0.003; D: P<0.001). CR, complete response; OS, overall survival; PR, partial response; PFS, progression-free survival.

As to the end of treatment evaluation, 64 patients performed examination; 58 achieved CR, and two reached PR, with PD in four patients. Compared with non-CR patients, those who achieved CR in the end of treatment demonstrated much more favorable PFS and OS (Figure 3C,3D).

Adverse events

As presented in Table 4, the safety profile was characterized by a high incidence of hematologic toxicity, which was reported in 86 patients (91.5%). The majority of these cases (78 patients, 83.0%) were grade ≥3 neutropenia. Non-hematologic adverse events were less common. The most frequently reported was drug-induced liver injury (n=33, 35.1%), followed by pulmonary infection (n=29, 30.9%). Chemotherapy-related lung injury and cardiovascular toxicity were each documented in 10 patients (10.6%), and gastrointestinal toxicity in 8 patients (8.5%). Notably, severe (grade ≥3) non-hematologic toxicities were uncommon, with only a single occurrence (1.1%) each of pulmonary infection, cardiovascular toxicity, and drug-induced liver injury.

A comparative analysis of adverse events between the R-CHOP (n=57) and da-R-EPOCH (n=24) groups demonstrated no significant differences in the incidence of either hematologic or non-hematologic toxicities (Table 5). Rates of grade ≥3 neutropenia were comparable between the two regimens (88.9% vs. 91.7%, P=0.49). Similarly, the frequencies of gastrointestinal toxicity, pulmonary infection, treatment-induced lung injury, cardiovascular toxicity, and drug-induced liver injury did not differ significantly between the two treatment cohorts. Overall, both regimens exhibited similar safety profiles in this population.

Discussion

Consistent with prior population-based analyses, our study corroborates that PSN-DLBCL portends a relatively favorable prognosis, with 61.7% of patients presenting with limited-stage disease and a 5-year OS rate of 73.1%, a figure substantially superior to nodal DLBCL (5-year OS: 65.5%) and comparable to other extranodal sites with indolent biology. This survival advantage aligns closely with previous reports demonstrating a 5-year OS rate of 69.9% for PSN-DLBCL (9), and further extends the observation that extranodal head and neck DLBCL collectively exhibits a significantly favorable prognosis compared to nodal counterparts (10).

However, there is a tendency of CNS involvement at the time of diagnosis in PSN-DLBCL patients: our CNS involvement rate of 9.6% notably exceeds the 2–2.5% benchmark in nodal DLBCL (11,12). Shimada et al. reported that paranasal sinus involvement was associated with a significantly higher risk of secondary CNS relapse in DLBCL patients, which was not observed in our cohort (13). The discrepancy may partly reflect the limited number of patients with paranasal sinus involved in their study (n=16, with 3 CNS relapses), as well as differences in patient characteristics and the degree of CNS-IPI adjustment between the two cohorts.

Importantly, our cohort lacked the traditional high-risk extranodal pattern of multi-organ dissemination to kidneys, adrenal glands, testes, or bone marrow, sites that amplify CNS-IPI scores through hematogenous vulnerability (14). Instead, we hypothesize that the CNS predilection of PSN-DLBCL arises primarily from direct locoregional extension through cranial foramina and basilar skull erosion, which facilitates contiguous spread to the frontal and temporal lobes, rather than via vascular dissemination. This mechanism shares key features with the perineural invasion patterns well-described in primary testicular DLBCL, albeit via osteolytic pathways unique to the sinonasal anatomic structure. This proposed mechanism of contiguous spread has direct implications for clinical management and therapeutic decision-making. Intrathecal chemotherapy, which yields therapeutic concentrations predominantly within the leptomeningeal space, may be inadequate to prevent or manage parenchymal brain invasion secondary to direct locoregional tumor extension. In these clinical settings, HD-MTX, which achieves reliable penetration of the brain parenchyma, represents a theoretically more rational option for both prophylaxis and therapeutic intervention (15). In line with findings from several prior cohort studies, the overall incidence of CNS relapse in our cohort was relatively low (16). While this favorable outcome may be partially explained by the intensive CNS-directed prophylactic regimens administered to a large proportion of our patient cohort, the definitive efficacy of these strategies has yet to be validated in larger, prospective clinical trials (17).

The lack of a protective effect of HD-MTX and intrathecal chemotherapy in both univariate and multivariate Cox regression analyses may seem counterintuitive. This finding is likely explained by the fact that these CNS-directed strategies were preferentially administered to patients with baseline confirmed or high-risk CNS involvement, a patient subgroup inherently linked to worse clinical outcomes (18). As a result, the potential survival benefit of prophylaxis in standard-risk patients may have been diluted by the poor prognostic profile of this high-risk subgroup. In addition, the low overall incidence of CNS relapse across the full cohort imposes a statistical ceiling effect, which limits the statistical power to detect a significant survival benefit associated with these prophylactic interventions.

Our finding that ELI conferred PFS protection (HR =0.48, P=0.046) challenges the conventional opinion that tumor aggression portends poor outcomes. While we hypothesized that symptomatic bone destruction triggers earlier diagnosis, our data refute this: the proportion of local stage is similar between patients with ELI (40/72, 55.6%) and the whole cohort (58/94, 61.7%). Instead, we propose a tumor microenvironment immunogenicity hypothesis: invasive fronts in PSN-DLBCL may provoke robust anti-tumoral immune responses mediated by dense tumor-infiltrating lymphocytes (TILs) (19). It was also demonstrated that extranodal DLBCL with high TILs density at invasive margins exhibited a prolonged survival (20), a phenomenon mechanistically linked to IFN-γ signaling and antigen presentation upregulation (21). The sinonasal mucosa’s inherent lymphoid-rich architecture may amplify this effect, where ELI exposes tumor neoantigens to regional immune sentinels (22).

Our analysis revealed no significant difference in efficacy between the R-CHOP and da-R-EPOCH regimens. It is crucial to contextualize this finding: da-R-EPOCH cohort contained 58.3% patients <60 years versus 17.5% in the R-CHOP group (P<0.001), creating a fitness advantage that tolerates regimen intensity (23). Therefore, the similar outcomes between the two regimens may suggest that the potential therapeutic advantage of da-R-EPOCH was offset by the more favorable baseline prognosis of the younger R-CHOP patients. This equipoise underscores that R-CHOP remains a standard, effective backbone for a majority of patients with PSN DLBCL.

Radiotherapy plays a pivotal role in the management of primary head and neck malignancies, particularly for squamous cell carcinoma (24) and ENKTL (25); however, its utility in DLBCL remains controversial. Previous studies have demonstrated survival advantages with combined radiotherapy during the single- and multi-agent chemotherapy eras (5,9), yet this benefit appears attenuated in the rituximab-based immunochemotherapy era (9). Our cohort similarly failed to demonstrate a survival benefit from radiotherapy. Moreover, considering the associated locoregional mucocutaneous and osteous toxicities (26), consolidative radiotherapy is not recommended. Nonetheless, given its established efficacy, radiotherapy may serve as an alternative for chemotherapy-intolerant patients or as second-line therapy for refractory/relapse disease.

Conclusions

In our retrospective cohort, PSN DLBCL showed favourable treatment response and prognosis after immunochemotherapy. Notably, although CNS involvement was more frequent at baseline, it did not lead to higher CNS relapse rates; these findings are observational and hypothesis-generating. We postulate that such involvement mainly results from contiguous anatomical extension, especially via skull base infiltration, rather than haematogenous dissemination. High IPI scores were adverse prognostic factors. Interim and EOT assessments indicated superior survival in patients achieving CR. The lack of clear survival benefit from intensified chemotherapy, systemic HD-MTX prophylaxis, or consolidative radiotherapy may reflect limited statistical power, selection bias, and baseline imbalances, and so cannot be interpreted as definitive evidence of equivalence. Likewise, although intrathecal therapy alone appeared insufficient as a substitute for systemic HD-MTX in our series, this conclusion is tentative and context-dependent. Taken together, our study contributes real-world data by systematically characterizing clinicopathologic features, CNS assessment practices, and contemporary treatment patterns for PSN-DLBCL in a comparatively large single-centre cohort. These insights form a foundation for prospective, ideally multicentre, risk-adapted studies to rigorously assess treatment intensification, CNS-directed strategies, and radiotherapy in this rare extranodal lymphoma.

Acknowledgments

We would like to thank the Department of Pathology, the Department of Radiology and Imaging, and the Department of Ophthalmic Oncology at Beijing Tongren Hospital, Capital Medical University for their invaluable support in the preparation of this manuscript.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0107/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0107/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0107/prf

Funding: This study was supported by grants from Noncommunicable Chronic Diseases-National Science and Technology Major Project (No. 2025ZD0544300) and the National Natural Science Foundation of China (No. 82370188, to L.W.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0107/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Beijing Tongren Hospital, China (No. TREC2022-KY103). The need for informed consent was waived for this retrospective chart review, which utilized only pre-existing medical records.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Peng KA, Kita AE, Suh JD, et al. Sinonasal lymphoma: case series and review of the literature. Int Forum Allergy Rhinol 2014;4:670-4. [Crossref] [PubMed]
2. Quraishi MS, Bessell EM, Clark D, et al. Non-Hodgkin’s lymphoma of the sinonasal tract. Laryngoscope 2000;110:1489-92. [Crossref] [PubMed]
3. Dubal PM, Dutta R, Vazquez A, et al. A comparative population-based analysis of sinonasal diffuse large B-cell and extranodal NK/T-cell lymphomas. Laryngoscope 2015;125:1077-83. [Crossref] [PubMed]
4. Eriksen PRG, Clasen-Linde E, Brown PN, et al. NK- and T-cell lymphoma of the nasal cavity and paranasal sinuses in Denmark 1980-2017: a nationwide cohort study. Leuk Lymphoma 2022;63:2579-88. [Crossref] [PubMed]
5. Kanumuri VV, Khan MN, Vazquez A, et al. Diffuse large B-cell lymphoma of the sinonasal tract: analysis of survival in 852 cases. Am J Otolaryngol 2014;35:154-8. [Crossref] [PubMed]
6. Eriksen PRG, Clasen-Linde E, Nully Brown P, et al. Sinonasal B-cell lymphomas: A nationwide cohort study, with an emphasis on the prognosis and the recurrence pattern of primary diffuse large B-cell lymphoma. Hematol Oncol 2022;40:160-71. [Crossref] [PubMed]
7. López-Guillermo A, Colomo L, Jiménez M, et al. Diffuse large B-cell lymphoma: clinical and biological characterization and outcome according to the nodal or extranodal primary origin. J Clin Oncol 2005;23:2797-804. [Crossref] [PubMed]
8. Shi Y, Han Y, Yang J, et al. Clinical features and outcomes of diffuse large B-cell lymphoma based on nodal or extranodal primary sites of origin: Analysis of 1,085 WHO classified cases in a single institution in China. Chin J Cancer Res 2019;31:152-61. [Crossref] [PubMed]
9. Lehrich BM, Abiri A, Goshtasbi K, et al. Treatment Modalities and Survival Outcomes for Sinonasal Diffuse Large B-Cell Lymphoma. Laryngoscope 2021;131:E2727-35. [Crossref] [PubMed]
10. Lee DY, Kang K, Jung H, et al. Extranodal involvement of diffuse large B-cell lymphoma in the head and neck: An indicator of good prognosis. Auris Nasus Larynx 2019;46:114-21. [Crossref] [PubMed]
11. Al-Mansour M, Saif SA, Alharbi Z, et al. The Outcomes of Diffuse Large B-cell Lymphoma Patients with Synchronous and Early Central Nervous System Involvement: A Single-Center Experience. Asian Pac J Cancer Prev 2023;24:623-31. [Crossref] [PubMed]
12. Bobillo S, Khwaja J, Ferreri AJM, et al. Prevention and management of secondary central nervous system lymphoma. Haematologica 2023;108:673-89. [Crossref] [PubMed]
13. Shimada K, Ohmachi K, Machida R, et al. Secondary central nervous system involvement in patients with diffuse large B-cell lymphoma treated with rituximab combined CHOP therapy - a supplementary analysis of JCOG0601. Ann Hematol 2024;103:2021-31. [Crossref] [PubMed]
14. Jemaa S, Paulson JN, Hutchings M, et al. Full automation of total metabolic tumor volume from FDG-PET/CT in DLBCL for baseline risk assessments. Cancer Imaging 2022;22:39. [Crossref] [PubMed]
15. Chua BJG, Low CE, Yau CE, et al. Recent updates on central nervous system prophylaxis in patients with high-risk diffuse large B-cell lymphoma. Exp Hematol Oncol 2024;13:1. [Crossref] [PubMed]
16. Thieblemont C, Altmann B, Frontzek F, et al. Central nervous system relapse in younger patients with diffuse large B-cell lymphoma: a LYSA and GLA/DSHNHL analysis. Blood Adv 2023;7:3968-77. [Crossref] [PubMed]
17. Puckrin R, El Darsa H, Ghosh S, et al. Ineffectiveness of high-dose methotrexate for prevention of CNS relapse in diffuse large B-cell lymphoma. Am J Hematol 2021;96:764-71. [Crossref] [PubMed]
18. Ma J, Li Q, Shao J, et al. Central Nervous System Involvement in Patients with Diffuse Large B Cell Lymphoma: Analysis of the Risk Factors and Prognosis from a Single-Center Retrospective Cohort Study. Cancer Manag Res 2019;11:10175-85. [Crossref] [PubMed]
19. Lorencin Bulic M, Jurlina M, Müller D, et al. Tumor-Infiltrating Lymphocytes Predict Extranodal Extension and Prognosis in Regionally Advanced Oral Cavity Cancer. Diagnostics (Basel) 2025;15:2431. [Crossref] [PubMed]
20. Cho Y, Lee J, Han B, et al. Tumor-infiltrating T lymphocytes evaluated using digital image analysis predict the prognosis of patients with diffuse large B-cell lymphoma. J Pathol Transl Med 2024;58:12-21. [Crossref] [PubMed]
21. Zhu Q, Yang Y, Chen K, et al. Diffuse large B-cell lymphoma: the significance of CD8(+) tumor-infiltrating lymphocytes exhaustion mediated by TIM3/Galectin-9 pathway. J Transl Med 2024;22:174. [Crossref] [PubMed]
22. Yin G, Guo W, Liu H, et al. Characteristics of tumor infiltrating lymphocytes in sinonasal mucosal melanoma and prognosis for patients. Curr Probl Cancer 2022;46:100878. [Crossref] [PubMed]
23. Schulpen M, Beishuizen A, Chamuleau MED, et al. Survival disparities between children and adolescents and young adults for the major subtypes of non-Hodgkin lymphoma in the Netherlands: a large population-based study. Haematologica 2024;109:936-41. [Crossref] [PubMed]
24. Wang Y, Han J, Zhu Y, et al. New advances in the therapeutic strategy of head and neck squamous cell carcinoma: A review of latest therapies and cutting-edge research. Biochim Biophys Acta Rev Cancer 2025;1880:189230. [Crossref] [PubMed]
25. Tse E, Zhao WL, Xiong J, et al. How we treat NK/T-cell lymphomas. J Hematol Oncol 2022;15:74. [Crossref] [PubMed]
26. Huang XM, Zheng YQ, Zhang XM, et al. Diagnosis and management of skull base osteoradionecrosis after radiotherapy for nasopharyngeal carcinoma. Laryngoscope 2006;116:1626-31. [Crossref] [PubMed]