Prognostic impact of central nervous system prophylaxis and a simplified BL model in primary testicular lymphoma: a real-world multicenter study

Abstract

Purpose

Primary testicular lymphoma (PTL) is a rare but aggressive form of extranodal lymphoma with a high risk of central nervous system (CNS) relapse and poor long-term survival. However, the optimal CNS prophylaxis strategy and effective prognostic models for PTL remain unclear. This study aimed to evaluate the prognostic impact of intrathecal (IT) prophylaxis and to develop a novel, simplified prognostic model in a Chinese multicenter cohort.

Methods

We retrospectively collected data from a total of 55 patients with PTL, treated at three major tertiary hospitals in China. Multivariate Cox regression identified independent prognostic factors for overall survival (OS) and progression-free survival (PFS). A new risk stratification model (BL model, based on B symptoms and LDH levels) was constructed and validated using time-dependent C-index and calibration plots, and compared with the International Prognostic Index (IPI).

Results

IT prophylaxis reduced CNS relapse rates (9.1% vs. 36.4%, p < 0.001) and was independently associated with improved OS (HR = 0.18, p < 0.001) and PFS (HR = 0.21, p < 0.001). The BL model (B symptoms and LDH), demonstrated superior predictive accuracy compared to the IPI, with higher AUCs at 1-, 3-, and 5-year OS (0.883, 0.894, 0.854 vs. 0.656, 0.804, 0.724), and a corrected C-index of 0.798. Calibration analysis confirmed good agreement between predicted and observed survival.

Conclusion

IT prophylaxis significantly improves survival and reduces CNS relapse in PTL. The BL model provides a simple yet effective tool for individualized risk stratification, outperforming IPI and aiding clinical decision-making in PTL.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00432-025-06358-9.

Introduction

Primary testicular lymphoma (PTL) is a rare extranodal subtype of non-Hodgkin lymphoma (NHL), accounting for only 1–2% of NHLs and ~ 10% of testicular malignancies (Grainger and Cheah 2025; Gundrum et al. 2009). However it is the most common testicular cancer in men over 60 years old (Vitolo et al. 2008). It typically presents as a firm, painless testicular mass, often accompanied by hydrocele, and a high proportion of relapse (Cheah et al. 2014; Horne and Adeniran 2011). Histologically, most cases (80–98%) are diffuse large B-cell lymphoma (DLBCL), particularly the non-germinal center B-cell (non-GCB) subtype (60–96%) (Cheah et al. 2014).

PTL is characterized by aggressive behavior and a high propensity for extranodal spread, particularly to the central nervous system (CNS) (Conconi et al. 2024). While the treatment includes R-CHOP chemotherapy combined with orchiectomy, testicular irradiation, and occasionally CNS prophylaxis, optimal strategies remain controversial (Cheah et al. 2014). Notably, although intrathecal (IT) prophylaxis is often used to prevent CNS relapse, evidence supporting its benefit in PTL is limited and conflicting (Kim et al. 2014; Mannisto et al. 2019).

In addition to therapeutic uncertainties, reliable prognostic models for PTL are lacking. The widely used International Prognostic Index (IPI) shows limited discriminative power in early-stage PTL (Wittekind et al. 2019). Other models, such as the one proposed by Zhi et al. (2023), are based on large databases like SEER but lack key clinical and laboratory variables, limiting their real-world applicability. Therefore, there is a pressing need to clarify the clinical value of IT prophylaxis and to develop a simple, practical, and accurate prognostic tool tailored to PTL.

To address these gaps, we conducted a multicenter retrospective study of 55 patients with PTL in China. To our knowledge, this is one of the largest PTL cohorts reported since 2014 outside of SEER-based studies, with comprehensive treatment and laboratory data (Grainger and Cheah 2025). We focused on evaluating the prognostic value of IT prophylaxis in reducing CNS relapse and improving survival. In addition, we developed a simple two-factor prognostic model—the BL model—based on B symptoms and LDH levels, and compared its performance with the widely used IPI.

Methods

Patients

A total of 468 hospitalization records were retrieved from the electronic medical record systems using keyword searches related to “lymphoma” and “testis/epididymis/scrotum”. After de-duplication (n = 380) and exclusion of cases with secondary testicular involvement (n = 22) or uncertain diagnoses due to incomplete clinical information (n = 11), 55 patients with confirmed primary testicular lymphoma (PTL) were included from three tertiary hospitals—Peking University Third Hospital, Peking University First Hospital, and the 5th Medical Center of PLA General Hospital—between 2004 and 2022 (Fig. 1A). The World Health Organization's (WHO) 2016 classification was used for diagnosis. These medical records and study protocols conform to the ethical requirements of Peking University Third Hospital. The clinical characteristics collected include age, Eastern Cooperative Oncology Group performance status (ECOG PS) (Trotti et al. 2000), Ann Arbor stage (Wittekind et al. 2019), International Prognostic Index (IPI) (International Non-Hodgkin’s Lymphoma Prognostic Factors Project 1993), extranodal involvement, B symptoms, and bulky disease (defined as any mass greater than 7.5 cm in diameter), as well as hematological and radiological examination and bone marrow biopsy. For DLBCL, histopathological classification was performed by the Hans algorithm. Additionally, the study gathered treatment information and response information, including first-line treatment, radiotherapy of the contralateral testis, prophylactic IT chemotherapy, rituximab usage, orchiectomy, and cause of death. Eleven patients did not receive intrathecal prophylaxis for the following reasons: (1) early treatment practices when CNS prophylaxis was not routinely applied; (2) patient refusal; (3) spinal disease; and (4) failed lumbar puncture.

Fig. 1.

Treatment response and survival outcomes in PTL patients. A A schematic overview of the cohort selection, prognostic analysis of treatment modalities (including CNS prophylaxis), and construction of the BL model. B Number of patients stratified by treatment modalities and response outcomes. C Distribution of treatment responses (CR, PR, SD, PD) by treatment subgroups. D Kaplan–Meier Curve of OS in PTL Patient. E Kaplan–Meier Curve of PFS in PTL Patients. F Impact of Intrathecal (IT) Prophylaxis on OS. G Impact of Intrathecal (IT) Prophylaxis on PFS. R±CHOP/CHOP-like: rituximab ± cyclophosphamide, doxorubicin/epirubicin, vincristine and prednisone/prednisolone; DA-EPOCH ± R: Dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin ± rituximab; SMILE: dexamethasone, methotrexate, ifosfamide, L-asparaginase, and etoposide; IT: IT prophylaxis; Double IT: double intrathecal chemotherapy [methotrexate (MTX) 12.5 mg + prednisolone (PSL) 5 mg]; Triple IT: triple intrathecal chemotherapy [cytarabine (Ara C) 50 mg + methotrexate (MTX) 12.5 mg + prednisolone (PSL) 5 mg]; Combined therapy: Rituximab + contralateral testicular radiation + orchiectomy + prophylaxis with IT prophylaxis

Response evaluation and follow-up

Tumor response was classified as complete remission (CR), partial remission (PR), stable disease (SD), or progressive disease (PD) according to the International Workshop Criteria (Cheson et al. 2007). Overall survival (OS) was defined as the time from diagnosis to death. Progression-free survival (PFS) was defined as the time from treatment to either disease progression or death.

Statistical analysis

Statistical analysis was performed on follow-up data using R 4.1.0. The median follow-up was computed using the Kaplan–Meier method (Altman et al. 1995). Survival probability curves were estimated with the Kaplan–Meier method, and differences were analyzed using the log-rank test (Dinse and Lagakos 1982). Univariate and multiple Cox proportional hazards regression (backward, stepwise) were performed to identify the independent factors. Nomograms were generated and validated through ROC curve analysis. All P values were two-sided, and the results were considered significant if P < 0.05.

Results

Clinical characteristics and risk profile of PTL patients

The 55 patients diagnosed with PTL from multiple centers in China (Table 1) had a median age of 65 years (range, 23–84 years). The median follow-up was 42 months (IQR, 26–100). Thirty-two patients (58.2%) presenting with advanced-stage disease (Ann Arbor III/IV). Seventeen (30.9%) patients reported B symptoms, and 27 (49.1%) had high or high-intermediate IPI scores. Immunochemical staining was performed on all 55 patients, with 49 (89.1%) diagnosed with DLBCL, of which 39 (70.9%) were classified as the non-GCB subtype. Elevated LDH levels were found in 22 (40%) patients; 7 (12.7%) had bulky disease, and 17 (30.9%) patients had extranodal involvement.

Table 1.

Clinical characteristics of primary testicular lymphoma

Characteristics	n	%
Age	> 60	36	65.4
	≤ 60	19	34.6
Type	DLBCL	49	89.1
	NK/T	4	7.3
	Small B cell lymphoma	2	3.6
Subtype	Non-GCB	39	70.9
	GCB	10	18.2
B symptoms	Present	17	30.9
	Absent	38	69.1
IPI	Low	8	14.5
	Low-intermediate risk	20	36.4
	High-intermediate risk	14	25.5
	High risk	13	23.6
ECOG	0–1	52	94.5
	≥ 2	3	5.5
Ann-Arbor stage	I/II stage	23	41.8
	III/IV stage	32	58.2
Laterality	Bilateral	11	20.0
	Left	18	32.7
	Right	26	47.3
Bulky diseasea	≥ 7.5 cm	7	12.7
	< 7.5 cm	48	87.3
LDH	Normal (≤ 250 U/L)	33	60.0
	Elevated (>250 U/L)	22	40.0
ALB	≤ 40 g/L	21	38.2
	> 40 g/L	34	61.8
Extra-nodal involvement	Yes	17	30.9
	No	38	69.0

^aBulky disease: any mass at any site with a diameter larger than 7.5 cm

Abbreviations: IPI = international prognostic index; ECOG = Eastern Cooperative Oncology Group; LDH = lactate dehydrogenase; ALB = albumin

We conducted a prognostic analysis of each treatment plan, focusing on short-term response, long-term survival, and CNS relapse (Fig. 1A). Among the 55 patients evaluated for therapeutic efficacy (Fig. 1B), CR was achieved in 42 patients (76%), while PR was observed in 7 patients (13%). One patient (1.8%) showed SD, and 5 patients (9.1%) showed PD. The short-term treatment did not show significant differences (Fig. 1C). Forty-four (80%) patients received 4 to 8 cycles of IT prophylaxis and thirty-nine (71%) patients underwent orchiectomy (Fig. 1B). All patients received systemic chemotherapy, and 23 (43%) patients received prophylactic radiotherapy. Forty-nine (89%) patients received R ± CHOP or CHOP-like regimens, while 2 received DA-EPOCH ± R, and 2 received SMILE.

During our follow-up, 18 deaths and 19 relapses occurred. The Kaplan–Meier curves revealed that the estimated 1-year, 3-year, and 5-year OS rates were 88.9%, 73.7%, and 65.0%, respectively (Fig. 1D). The 1-year, 3-year, and 5-year PFS rates were 79.7%, 58.4%, and 54.5%, respectively (Fig. 1E).

Intrathecal prophylaxis improves survival and reduces CNS relapse

IT prophylaxis demonstrated a clear survival benefit in patients with PTL in our cohort. Among the 55 patients, 44 received IT prophylaxis, while 11 did not, and related treatment details were summarized (Supplementary Table 1). Patients in the non-IT group experienced significantly worse outcomes, with 10 deaths and disease progression observed in all 11 cases. In contrast, IT prophylaxis significantly improved OS (p < 0.01, Fig. 1F) and PFS (p < 0.001, Fig. 1G), whereas other treatment modalities showed no significant impact on survival in this cohort (Supplementary Fig. 1).

Multivariable analyses confirmed that IT prophylaxis is an independent prognostic factor for improved outcomes classified by across various treatment modalities (Supplementary Table 2), with better OS (HR = 0.18, p < 0.001) and PFS (HR = 0.21, p < 0.001). Subgroup analysis showed that IT prophylaxis significantly improved OS and PFS across multiple subgroups (p < 0.05, Supplementary Fig. 2), with notable benefits in patients with DLBCL, non-GCB subtype, KI67 ≥ 50%, unilateral involvement, stage III/IV, absence of B symptoms, ECOG 0–1, normal ALB, elevated LDH, tumor size < 7.5 cm,   ≥ 2 extranodal sites, and across various treatment modalities.

To explore the potential mechanism underlying these benefits, we analyzed CNS and systemic relapse patterns (Fig. 2A). In the non-IT group (n = 11), 10 patients relapsed (90.9%), with 4 experiencing CNS relapse (36.4%) and 6 exhibiting systemic relapse (54.5%). In contrast, the IT group (n = 44) demonstrated a lower relapse rate, with only 4 patients showing CNS relapse (9.1%) and 9 displaying systemic relapse (20.5%). Chi-square testing confirmed a significant difference between the groups (p < 0.001, Fig. 2B). Kaplan–Meier curves illustrated that IT prophylaxis markedly improves CNS relapse-free survival (p = 0.0028, Fig. 2C). Multivariable Cox analysis established IT prophylaxis as an independent factor in reducing CNS relapse risk (HR = 0.15, p = 0.009, Fig. 2D, Supplementary Table 3). No significant difference founded between double IT (methotrexate and dexamethasone) and triple IT (methotrexate, cytarabine, and dexamethasone) regimens (p = 0.914, Fig. 2E; Cox p = 0.6, Fig. 2F). These findings suggest that IT prophylaxis plays a pivotal and independent role in improving survival and reducing CNS relapse in patients with PTL.

Fig. 2.

Impact of IT prophylaxis on relapse in PTL patients. A Flowchart from IT regimen to recurrence. B Comparison of relapse rates between IT and non-IT groups. C Kaplan–Meier Curve of CNS relapse-free survival with IT prophylaxis. D Multivariable cox analysis of IT prophylaxis on CNS relapse risk. E Comparison of double IT vs. triple IT on CNS relapse-free survival. F Cox analysis of double IT vs. triple IT on CNS relapse risk

BL—a simplified two-factor prognostic model outperforms the IPI

To address the limitations of existing prognostic tools in PTL, we developed a simplified risk model based on two readily accessible clinical variables: B symptoms and serum LDH levels (Fig. 1A, Fig. 3A). Variables with p < 0.1 in univariate Cox regression were included in multivariate analysis, which identified B symptoms (HR = 5.80, p = 0.001) and elevated LDH (HR = 2.98, p = 0.044) as independent predictors of OS (Table 2).

Fig. 3.

Prognostic evaluation and validation of IPI and BL models in PTL patients. A Workflow of prognostic model development for PTL patients. B Kaplan–Meier (KM) curve for OS stratified by IPI model. C Time-dependent ROC curves for the IPI model at 12, 36, and 60 months. D KM curve for OS stratified by the BL model. E Time-dependent ROC curves for the BL model at 12, 36, and 60 months. F Time-dependent C-index curves comparing the IPI and BL models. G Calibration plot for OS using the BL model at 12, 36, and 60 months

Table 2.

Clinical factors associated with PFS and OS in MCL patients on COX analysis

Variable	Groups	Univariate COX analysis		Multivariate COX analysis
		OS HR (95% CI P value)	PFS HR (95% CI P value)	OS HR (95% CI P value)	PFS HR (95% CI P value)
Age	> 60 vs. ≤  60	0.50 (0.16–1.55, p = 0.230)	0.70 (0.28–1.71, p = 0.429)
Ann Arbor	I/II vs. III/IV	4.11 (1.27–13.30, p = 0.018)	3.10 (1.18–8.13, p = 0.022)
LDH	Normal vs. abnormal	3.60 (1.33–9.76, p = 0.012)	4.97 (2.00–12.37, p < 0.001)	2.98 (1.03–8.64, p = 0.044)	5.00 (1.79–13.96, p = 0.002)
Extranodal sites	< 2 vs. ≥ 2	2.28 (0.85–6.12, p = 0.101)	3.14 (1.34–7.35, p = 0.008)
ECOG	0–1 vs. ≥ 2	4.17 (0.90–19.26, p = 0.068)	4.65 (1.32–16.38, p = 0.017)
B symptom	Without vs. with	6.52 (2.42–17.58, p < 0.001)	4.47 (1.91–10.48, p < 0.001)	5.80 (2.08–16.21, p = 0.001)	4.38 (1.67–11.46, p = 0.003)
Laterality	Unilateral vs. bilateral	2.44 (0.78–7.56, p = 0.124)	3.26 (1.27–8.39, p = 0.014)		7.55 (2.56–22.32, p < 0.001)

Abbreviations: ECOG = Eastern Cooperative Oncology Group; LDH = lactate dehydrogenase

For benchmark, we first evaluated the performance of the IPI in this cohort. While a significant OS difference was observed between the high- and low-risk groups (p = 0.04, Fig. 3B), the intermediate groups overlapped, and other comparisons did not reach statistical significance (p > 0.05). The AUCs of the IPI for 1-, 3-, and 5-year OS were 0.656, 0.804, and 0.724, respectively (Fig. 3C), indicating modest discriminatory power.

Based on our multivariate results, we constructed a new scoring system—the BL model. Patients were stratified into four risk categories: Low-risk (L): 0 points; Low-intermediate (LI): 1 point for LDH; High-intermediate (HI): 1 point for B symptoms; High-risk (H): both factors present. The BL model demonstrated clear prognostic separation. In OS analysis (Fig. 3D), the high-risk group had significantly worse outcomes than both the low-risk (p < 0.001) and high-intermediate (p = 0.039) groups. The model yielded higher AUCs than the IPI: 0.883, 0.894, and 0.854 at 1-, 3-, and 5-year OS, respectively (Fig. 3E).

Internal validation using 1000 bootstrap resamples further supported the model's robustness. The BL model achieved a corrected C-index of 0.798 (95% CI 0.713–0.883), with time-dependent C-index values ranging from 0.846 to 0.690 across 1 to 120 months. In contrast, the IPI had a lower corrected C-index of 0.716 (95% CI 0.591–0.841), and time-dependent values between 0.793 and 0.592 (Fig. 3F). Calibration curves for 1-, 3-, and 5-year OS confirmed excellent agreement between predicted and observed survival probabilities (Fig. 3G) by 1000 bootstrap, which highlight the BL model as a simple yet powerful prognostic tool that outperforms the IPI in real-world PTL patients.

To assess the robustness of the BL score across different treatment, we performed subgroup analyses based on six key therapies: rituximab, orchiectomy, contralateral testicular irradiation, CHOP-like regimens, R-CHOP/R-CHOP-like regimens, and standard therapy. Five of six comparisons yielded statistically significant results (p ≤ 0.05, Supplementary Fig. 3). Although the standard therapy subgroup (n = 11) did not reach statistical significance, the limited sample size likely contributed to this result, and a consistent trend with the overall cohort was still observed. Time-dependent AUCs in subgroup ranged from 0.795 to 0.903, comparable to the full cohort (0.854–0.894).

Discussion

PTL remains a rare and aggressive form of extranodal lymphoma, often presenting in older males with high-stage disease and a marked propensity for CNS relapse (Dunn et al. 2002; Horne and Adeniran 2011). Existing literature primarily consists of single-arm phase II trials and retrospective studies, many of which are limited by number of cases. While large databases such as SEER include more cases, the lack of detailed variable recording limits their utility for prognostic modeling and treatment outcome analysis (Conconi et al. 2024). To our knowledge, excluding SEER-based analyses, this study represents the largest multicenter retrospective PTL cohort reported since 2014, with detailed treatment and laboratory data available for all patients (Grainger and Cheah 2025). When compared with Western cohorts, the clinical presentation of PTL in China is broadly similar in age and histology (Supplementary Table 4), but higher proportions of advanced stage, ECOG ≥ 2, and bilateral involvement have been reported (Conconi et al. 2024; Medina et al. 2023; Shah et al. 2023; Sun et al. 2022; Xu and Yao 2019). Our real-world multicenter study demonstrates that IT prophylaxis significantly improves OS and PFS and reduces CNS relapse. In addition, we developed and validated a novel prognostic model, the BL model, which outperformed the widely used IPI in risk stratification for OS.

CNS relapse represents one of the most devastating complications of PTL, with a historically reported rate of ~ 30% and poor salvage outcomes (Pollari et al. 2021). Such relapses carry a dismal prognosis, with a median survival of less than one year even after aggressive salvage therapy (Yan et al. 2021). For decades, intrathecal chemotherapy has been used empirically in PTL to reduce CNS relapse risk, but evidence for its efficacy is mixed. Some retrospective analyses, such as a 2014 single-center experience, suggested that patients treated without IT prophylaxis had relatively low CNS relapse rates, questioning its necessity (Kim et al. 2014). Another study highlighted that most CNS relapses in the rituximab era are parenchymal, potentially less preventable by IT chemotherapy alone, advocating for high-dose intravenous methotrexate instead (Mannisto et al. 2019).

Our results support the use of IT prophylaxis, which reduced CNS relapse from 36.4 to 9.1% (p < 0.001), and improved OS and PFS, with multivariate analysis confirming its independent prognostic value (OS: HR = 0.18, PFS: HR = 0.21; p < 0.001 for both). These findings align with IELSG30 (Conconi et al. 2024) and other reports advocating CNS prophylaxis (Abramson et al. 2010; Cheah et al. 2014; Holte et al. 2013; Shen et al. 2022).

Subgroup analysis further demonstrated the consistency of IT prophylaxis benefits across clinical strata (P < 0.05, Supplementary Fig. 2), particularly in patients with DLBCL histology, non-GCB subtype, unilateral involvement, stage III/IV disease, and high Ki-67. These results reinforce the relevance of IT prophylaxis in routine PTL management, especially in higher-risk groups. Given the challenges of conducting prospective trials in rare malignancies, our data provide valuable real-world evidence supporting its continued use.

While the IPI remains a widely used prognostic tool in lymphoma, it has shown limited applicability in PTL, particularly in early-stage patients, where the IPI score is often less than 2 point (Wittekind et al. 2019). Other models, such as that proposed by Zhi et al. (2023), were derived from SEER data with limited laboratory variables and included factors impractical for pre-treatment decision-making (e.g., year of diagnosis). The R-IPI, though valuable in DLBCL, also has limitations—it was developed under the assumption of uniform rituximab exposure (Sehn et al. 2006), its advantage over IPI is not consistently observed (Biccler et al. 2018; Ruppert et al. 2020), and importantly, previous PTL studies have applied only the conventional IPI, without incorporating the R-IPI for risk stratification (Sun et al. 2022; Vitolo et al. 2011; Xu et al. 2019).

Multivariate analysis identified elevated LDH and B symptoms as independent predictors of worse OS. This result is consistent with prior findings that highlighted the LDH as a marker of tumor burden and B symptoms as indicators of systemic disease activity from the MD Anderson Cancer Center (Mazloom et al. 2010) and multicenter studies (Sun et al. 2022). Based on static result and clinical meaning, we propose a new B symptom and LDH (BL) model.

In contrast, our BL model, incorporating only B symptoms and LDH, offers a streamlined and clinically accessible tool that demonstrated superior discriminative power in K-M analysis, higher AUCs (0.854–0.894), and a corrected C-index of 0.798—outperforming the IPI across all validation metrics (Fig. 3C, E).

Internal validation confirmed the robustness of the BL model, with time-dependent C-index ranging from 0.846 to 0.690 over 120 months, and calibration plots showing excellent agreement between predicted and observed survival at 1, 3, and 5 years (Fig. 3C, F). These results underscore the clinical utility of the BL model in everyday practice. To address potential bias from treatment heterogeneity, we performed subgroup analyses across six key therapeutic components (Supplementary Fig. 3), with 5 of 6 comparisons reaching statistical significance and ROC curves closely aligned with those of the overall cohort. Notably, systemic High-Dose (HD) MTX was administered only in a subset of patients, largely due to advanced age, comorbidities, concerns about tolerability, and differences in practice patterns over time. This limited our ability to validate the BL score specifically in a uniformly HD-MTX–treated cohort, and it highlights the need for further evaluation in this subgroup to better define its prognostic utility.

Taken together, our findings highlight the importance of IT prophylaxis as part of comprehensive CNS-directed management in PTL, consistent with prior evidence that testicular involvement itself constitutes a high-risk extranodal site. Contemporary guidelines and major reviews, including ESMO (Tilly et al. 2015), BSH (McKay et al. 2020) and NCCN (Andreadis 2025), consistently recommend CNS prophylaxis for all PTL patients. Against this background, the BL score is intended to capture prognostic risk rather than to determine CNS-directed interventions. In the near term, a pragmatic role of the BL score may be to refine systemic prognostic assessment and to help tailor follow-up schedules including surveillance cadence. Looking ahead, the score could also be prospectively evaluated as a stratification or enrichment tool in clinical trials that test escalation and de-escalation strategies, including those related to CNS prophylaxis.

Conclusion

This study provides important real-world evidence supporting the role of intrathecal prophylaxis in reducing CNS relapse and improving survival in patients with primary testicular lymphoma. In addition, the BL model, based on only B symptoms and LDH levels, offers a simple yet robust prognostic tool that outperforms the widely used IPI in risk stratification. Given the rarity and aggressiveness of PTL, our findings not only help refine treatment decisions but also underscore the need for prospective validation of CNS-directed strategies and pragmatic models in larger, multi-institutional cohorts. Moving forward, the integration of such tools into clinical pathways may improve risk-adapted management and long-term outcomes for this challenging lymphoma subtype.

Clinical practice points

• Primary testicular lymphoma (PTL) has a high risk of CNS relapse and poor long-term outcomes, highlighting the need for effective prophylactic strategies.

• This study supports the use of intrathecal prophylaxis as a standard component of frontline treatment in PTL to reduce CNS relapse and improve survival.

• The proposed BL model, incorporating only B symptoms and LDH levels, provides a simple and practical tool for risk stratification in PTL patients.

• These findings may support risk-adapted management approaches and improve real-world outcomes in this rare lymphoma subtype.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

NHL

Non-Hodgkin lymphomas

PTL

Primary testicular lymphoma

PT-DLBCL

Primary testicular diffuse large B-cell lymphoma

R-CHOP

Rituximab + cyclophosphamide, doxorubicin, vincristine, and prednisone

IT

Intrathecal

IPI

International prognostic index

OS

Overall survival

PFS

Progression-free survival

CR

Complete remission

PR

Partial remission

SD

Stable disease

PD

Progressive disease

LDH

Lactate dehydrogenase

BL Model

B symptoms and LDH level model

ECOG PS

Eastern Cooperative Oncology Group performance status

WHO

World Health Organization

AUC

Area under the curve

KM

Kaplan–Meier

HR

Hazard ratio

ROC

Receiver operating characteristic

GCB

Germinal center B-cell-like

DA-EPOCH

Dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin

SMILE

Steroid, methotrexate, ifosfamide, L-asparaginase, and etoposide

SEER

Surveillance, epidemiology, and end results

CSS

Cancer-specific survival

CNS

Central nervous system

Author contributions

All authors contributed to data collection, analysis, and writing of the manuscript. SL: analysis of the data and writing of the manuscript. PY: study design, data collection, analysis, and writing of the manuscript. HL, YZ, XC, FD, WW: data management, study design, data collection, data preparation, and manuscript editing. HJ: study design, data collection, analysis, and manuscript editing. SL, PY, and YZ contributed equally to this work. All authors read and approved the final manuscript.

Funding

This work was supported by the special fund of the National Clinical Key Specialty Construction Program, P. R. China (2023).

Data availability

The datasets analyzed in this study are not publicly available due to privacy restrictions but can be obtained from the corresponding author upon reasonable request.

Declarations

Conflict of interest

The authors declare no competing interests.

Research involving human participants and/or animals

The protocol for this research project was approved by the Ethics Committee of Peking University Third Hospital (Approval No. M2022254), and the study conforms to the provisions of the Declaration of Helsinki.

Informed consent

All informed consent was obtained from the subject(s) and/or guardian(s).