1 Introduction

Diffuse large B-cell lymphoma (DLBCL) has been reported to be the most prevalent subtype of non-Hodgkin's lymphoma [1]. Patients with DLBCL are mainly treated with immunochemotherapy, such as the rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) regimen. However, despite the treatment, 30%–40% of patients experience relapse or refractory disease, resulting in unfavorable prognoses [2]. Furthermore, approximately 2%–5% of patients with DLBCL have been reported to experience central nervous system (CNS) recurrence, with a median overall survival of 3.5 months following the onset of CNS recurrence [3-5].

The CNS International Prognostic Index (IPI) is the standard model for predicting the risk of CNS recurrence and incorporates both adrenal involvement points and IPI factors [4]. Patients with a CNS-IPI score of 4–6 have been reported to present a CNS recurrence risk of > 10%. DLBCL incidence at high-risk extranodal sites, such as the testes, kidneys, and breasts, can also result in a CNS recurrence risk of > 10% [6, 7]. Furthermore, patients with double-hit lymphomas have been shown to exhibit a 13% incidence of CNS recurrence within 3 years [8]. Despite notable advances in identifying and mitigating the risk of CNS recurrence utilizing molecular-typing methods, addressing financial implications and validation of reported methods in larger cohorts remain critical for their integration into standard clinical practice [9].

To reduce CNS recurrence, various treatment strategies have been developed to integrate CNS-penetrating prophylaxis into frontline therapy. Notably, intravenous high-dose methotrexate (HD-MTX) is a commonly recommended CNS prophylaxis over intrathecal therapy, as most relapses are parenchymal in nature [10-12]. Many studies have reported enhanced survival outcomes following the incorporation of this CNS prophylaxis into the standard regimen [5, 13, 14]. However, recent studies have raised concerns regarding the efficacy of HD-MTX [15-17]. Additionally, there remains a lack of consensus regarding the optimal delivery method for MTX, particularly in terms of timing.

Hence, this study aimed to assess the protective efficacy of HD-MTX against CNS recurrence and its influence on survival outcomes in newly diagnosed patients with DLBCL, which presents a high risk for CNS recurrence. Additionally, different HD-MTX treatment regimens and associated toxicities were investigated in this study.

2 Methods

3 Results

3.1 Patient Characteristics

We identified 531 patients with newly diagnosed DLBCL during the study period. Of them, 206 patients were enrolled in this study, who were diagnosed with DLBCL (October 2018–July 2023) and met the high-risk criteria for CNS recurrence (Figure 1). After excluding 49 patients owing to their excessive baseline creatinine levels and ≥ 75 years of age, 19 patients who were unable to adhere to follow-up protocols, and 2 patients with pre-existing CNS involvement, a total of 136 patients remained in the final cohort. They were then categorized into the following two groups: the non-HD-MTX group (n = 90) and the HD-MTX group (n = 46). The baseline characteristics of the different patient groups are summarized in Table 1. Notably, both groups exhibited similarities regarding autologous stem cell transplantation (ASCT) status, extranodal involvement, Ann Arbor stage, NCCN-IPI, and CNS-IPI. However, a higher proportion of younger patients were present in the HD-MTX group compared with that in the non-HD-MTX group.

TABLE 1. Baseline characteristics.

	HD-MTX group (n = 46)	Non-HD-MTX group (n = 90)	All patients (n = 136)	p
Sex
Female (%)	17 (37.0)	36 (40.0)	53 (39.0)
Male (%)	29 (63.0)	54 (60.0)	83 (61.0)	0.731
Age
Median (range), y	53.0 (20–73)	62.5 (19–74)	59.4 (19–74)	< 0.001a
Cell of origin
GCB(%)	21 (45.7)	37 (41.1)	58 (42.6)	0.612
Non-GCB(%)	25 (54.3)	53 (58.9)	78 (57.4)
Double hitb
Yes(%)	7 (7.8)	7 (7.8)	14 (10.3)	0.177
No (%)	39 (84.8)	83 (92.2)	122 (89.7)
B symptoms
Yes(%)	18 (39.1)	26 (28.9)	44 (32.4)	0.227
No(%)	28 (60.9)	64 (71.1)	92 (67.6)
Ann Arbor stage
I (%)	4 (8.7)	5 (5.6)	9 (6.6)	0.320
II (%)	5 (10.9)	5 (5.6)	10 (7.4)
III (%)	3 (6.5)	11 (12.2)	14 (10.3)
IV (%)	34 (73.9)	69 (76.7)	103 (75.7)
NCCN-IPI scores
0 ~ 1 (%)	5 (10.9)	2 (2.2)	7 (5.1)	0.161
2 ~ 3 (%)	9 (19.6)	15 (16.7)	24 (17.6)
4 ~ 5 (%)	22 (47.8)	52 (57.8)	74 (54.4)
≥ 6 (%)	10 (21.7)	21 (23.3)	31 (22.8)
CNS-IPI scores
≥ 4 (%)	28 (60.9)	69 (76.7)	97 (71.3)	0.054
< 4 (%)	18 (39.1)	21 (23.3)	39 (28.7)
Intrathecal
Yes (%)	13 (28.3)	15 (16.7)	28 (20.6)	0.114
No (%)	33 (71.7)	75 (83.3)	108 (79.4)
Involvement of the kidney and adrenal gland
Yes (%)	14 (30.4)	19 (21.1)	33 (24.3)	0.230
No (%)	32 (69.6)	71 (78.9)	103 (75.7)
Involvement of the testis
Yes (%)	10 (21.7)	9 (10.0)	19 (14.0)	0.062
No (%)	36 (78.3)	81 (90.0)	117 (86.0)
Involvement of the breast
Yes (%)	7 (15.2)	11 (12.2)	18 (13.2)	0.626
No (%)	39 (84.84)	79 (87.8)	118 (86.8)
Autologous stem cell transplantation
Yes (%)	10 (21.7)	11 (12.2)	21 (15.4)	0.146
No (%)	36 (78.3)	79 (87.8)	103 (83.7)

• Abbreviations: CNS, central nervous system; GCB, germinal center B-cell; HD, high-dose; IPI, International Prognostic Index; MTX, methotrexate; NCCN, National Comprehensive Cancer Network.
• ^a Represents statistical significance.
• ^b Denotes the presence of MYC translocation with B-cell lymphoma (BCL)2 and/or BCL6.

Notably, in the HD-MTX group, 24 patients received treatment before remission, 18 after remission, and 4 during progression. Additionally, 17 patients underwent one course, 13 underwent two courses, and 16 underwent three or more courses of HD-MTX treatment.

3.3 Survival Outcomes

In the final patient cohort, the 2- and 5-year PFS rates were 64.2% and 32.8%, respectively, whereas the 2- and 5-year OS rates were 64.9% and 54.4%, respectively. The difference in OS and PFS was statistically significant between the HD-MTX and non-HD-MTX groups (Figure 2A,B). Specifically, the 2-year PFS and OS rates in the HD-MTX group were 70.7% and 72.9%, respectively, whereas those in the non-HD-MTX group were 60.8% and 60.8%, respectively (for PFS, p = 0.013; HR: 0.506, 95% CI: 0.310–0.826; for OS, p = 0.024; HR: 0.490, 95% CI: 0.283–0.849). The results of subgroup analyses of PFS and OS indicated that patients classified as low- or intermediate-risk patients per NCCN-IPI, younger than 60 years of age, without B symptoms, who did not undergo ASCT, with germinal center B-cell (GCB) type, with CNS-IPI score ≥ 4, and of male sex showed greater benefit from the HD-MTX treatment (Figure 2C–F and Figure 3A,B). The subgroup analysis was exploratory, with no preset assumptions, and the results should be interpreted with caution.

The results of Cox regression survival analysis revealed factors influencing PFS and OS in patients at high risk for CNS recurrence. Univariate analysis identified the following factors influencing PFS and OS: stage IV disease, CNS-IPI scores ≥ 4, involvement of the testes or kidneys and adrenal glands, absence of CNS prophylaxis (intrathecal or HD-MTX therapy), and CNS recurrence. Multivariate analysis revealed CNS-IPI scores ≥ 4 and CNS recurrence as the significant factors (Table 2).

TABLE 2. Multivariate Cox regression analyses of risk factors for progression-free and overall survival.

	PFS		OS
	HR (95% CI)	p	HR (95% CI)	p
CNSIPI scores ≥ 4	3.830 (1.948 ~ 7.527)	< 0.001	3.620 (1.662 ~ 7.887)	0.001
CNS recurrence	2.724 (1.321 ~ 5.617)	0.007	3.109 (1.494 ~ 6.469)	0.002

• Abbreviations: CI, confidence interval; CNS, central nervous system; HR, hazard ratio; IPI, International Prognostic Index.

4 Discussion

CNS recurrence in patients with DLBCL considerably exacerbates the disease status, and it is correlated with shorter survival and poor prognoses [3-5]. Therefore, early identification of high-risk patients and implementation of effective preventive strategies is crucial to reduce CNS recurrence rates and improve patient survival outcomes. Although HD-MTX therapy is commonly used for CNS prophylaxis in clinical practice, its efficacy regarding disease management remains controversial. Hence, this retrospective study evaluated HD-MTX's prophylactic efficacy in patients with DLBCL presenting a high risk of CNS recurrence and compared CNS recurrence rates, PFS, and OS between patients receiving HD-MTX and those who did not.

Overall, the findings of this study indicate that compared with the patients in the non-HD-MTX group, those in the HD-MTX group exhibited a lower 2-year CNS recurrence rate and a higher 2-year PFS and OS rates. However, although differences in PFS and OS rates were statistically significant, the reduction in CNS recurrence rate was not. These results are consistent with those previously reported. For instance, Goldschmidt et al. analyzed 480 patients at high risk for CNS recurrence in Israel, of whom 130 (27%) received HD-MTX therapy. Notably, HD-MTX significantly improved both PFS and OS (p = 0.001) but could not reduce CNS relapse rates (HD-MTX vs. non-HD-MTX was 6.9% vs. 6.3%, p = 0.97). They attributed the observed survival benefits to enhanced systemic disease control rather than the prevention of CNS relapse [14]. Notably, similar mechanisms may underlie the improved survival rates observed in the HD-MTX group in the present study. Nevertheless, the potential contribution of reduced CNS recurrence to the observed survival benefit cannot be entirely dismissed, especially considering the downward trend of the CNS recurrence rate in the HD-MTX group compared with that in the non-HD-MTX group. The multivariate analysis of PFS and OS revealed the incidence of CNS recurrence as the independent risk factor influencing patient survival. However, this needs to be further validated with a larger cohort, as this single-center study may have an insufficient sample size for the low probability event of CNS recurrence.

Although the application of HD-MTX in high-risk populations for CNS recurrence may prolong survival or reduce disease progression, studies identifying specific subgroups affected by this treatment are lacking. Herein, the results of subgroup analyses on PFS and OS revealed that patients with better chemotherapy tolerance (NCCN-IPI: 0–5; age ≤ 60 years; sex: male; and absence of B symptoms) or those at higher risk of CNS recurrence (CNS-IPI ≥ 4) experienced more significant improvements in PFS and OS. Furthermore, patients who did not undergo ASCT showed greater survival benefits from HD-MTX treatment, which may be attributed to its systemic control mimicking ASCT consolidation or overlapping CNS prophylaxis effects from ASCT conditioning regimens. Patients with GCB-subtype DLBCL presented superior survival outcomes, which were associated with higher rates of double-hit lymphoma in this subgroup (17.2% in GCB vs. 5.1% in non-GCB, p = 0.022). Reportedly, the clinical characteristics of highly aggressive behavior and increased CNS recurrence rate in double-hit lymphoma may result in it benefiting from HD-MTX chemotherapy and effective CNS prophylaxis.

CNS recurrence is reported to predominantly occur in the early stages (median time for recurrence: 6–7 months) [4, 17]. Herein, the median TT-CNS was 5.5 months, which is consistent with the abovementioned range. Consequently, many medical centers have incorporated intercalated HD-MTX (i-HD-MTX) during the early cycles of the R-CHOP regimen to prevent early CNS recurrence. However, Wilson et al. reported no evidence that completing the R-CHOP regimen increased the risk of CNS recurrence compared with the i-HD-MTX regimen [18]. To the best of our knowledge, the largest study to date on HD-MTX-related adverse reactions is a retrospective study involving 258 patients performed in the United States. The patients were divided into the HD-MTX (n = 128) and non-HD-MTX (n = 130) groups. Notably, the HD-MTX group exhibited a higher incidence of grade ≥ 3 oral mucositis and increased alanine aminotransferase levels. Additionally, delays or dose reductions in the R-CHOP regimen were more frequent in the HD-MTX group compared with those in the non-HD-MTX group (31.2% vs. 16.9%) [19]. Delays in chemotherapy administration can adversely affect treatment efficacy. Recently, Song et al. conducted a multicenter retrospective clinical study on the impact of chemotherapy delays on prognosis in pediatric Burkitt lymphoma. They showed that children experiencing chemotherapy delays exceeding 7 days presented significantly lower 3-year OS rates compared with those with delays of ≤ 7 days [20]. Consequently, it was recommended that HD-MTX chemotherapy-associated delays should be restricted to ≤ 7 days whenever feasible to mitigate treatment delay-induced negative effects on therapeutic efficacy and patient prognosis.

Herein, the chemotherapy intervals of two HD-MTX regimens (HD-MTX day 6 vs. HD-MTX day 10–14) and R-CHOP were compared. Although both HD-MTX regimens had chemotherapy delays, owing to the shorter delay, the HD-MTX day 6 regimen presented superior results to the conventional HD-MTX day 10–14 regimen. Regarding hematologic and nonhematologic toxicities, both HD-MTX regimens showed no statistically significant differences compared to the R-CHOP-alone regimen. However, the incidences of high creatinine levels and oral mucositis were significantly higher with HD-MTX regimens. These results are consistent with those previously reported.

This study has some limitations. First, the small sample size, coupled with constraints inherent in nonrandom observational and regression analyses, may have resulted in imbalanced baseline characteristics across different groups. Notably, the mean baseline age was significantly lower in the HD-MTX group compared with that in the non-HD-MTX group. Typically, age is regarded as a high-risk prognostic factor in lymphoma; however, in the present cohort of patients at high risk for CNS recurrence, age showed no association with survival outcomes in either univariate or multivariate prognostic analyses. Second, baseline CNS evaluation may have influenced the study results. Herein, baseline CNS assessment was not mandatory for all patients and was conducted only for those with a CNS-IPI score of ≥ 4. Moreover, more sensitive detection methods for CNS involvement, such as measuring chemokine (C-X-C motif) ligand 13 and interleukin-10 levels in the CSF, were not implemented clinically. Hence, patients experiencing early CNS relapse may not have been identified before initiating chemotherapy, either because of missed detection or insufficient sensitivity of present detection methods. Bennett et al. [15] reported the real-world incidence of CNS relapse following baseline CSF analysis, which was conducted to rule out asymptomatic leptomeningeal involvement in patients newly diagnosed with high-risk DLBCL. They showed that among patients with CNS involvement (indicated by a positive baseline CSF analysis), the incidence of CNS disease events over 5 years was 44.4%, compared with 5.6% for patients without CNS involvement (indicated by a negative baseline CSF analysis).

In this study, definitive results regarding the CNS prophylactic efficacy of HD-MTX could not be achieved; nevertheless, the findings show the promising clinical viability of the HD-MTX administration regimen—specifically the HD-MTX day 6 regimen. This regimen effectively reduced the typical HD-MTX-associated chemotherapy delays. However, owing to the limited sample size and baseline imbalances, statistical analyses could not ascertain the optimal number of HD-MTX cycles. Future studies need to focus on validating the HD-MTX day 6 regimen and determining the optimal number of treatment cycles.

Altogether, the role of HD-MTX in preventing CNS recurrence could not be conclusively determined due to the lack of robust evidence. However, the intravenous HD-MTX day 6 regimen showed notable therapeutic potential for patients who can tolerate the treatment, such as younger individuals and those at NCCN-IPI low or intermediate risk. Despite lacking results, this study serves as a valuable reference for future studies on HD-MTX treatment and for decision-makers to offer guidance on clinical treatment decisions.

Author Contributions

Yanli Wang: conceptualization, methodology, funding acquisition, writing – original draft, project administration. Xiaolian Wen: investigation, data curation, resources, writing – review and editing. Tao Guan: data curation, methodology, resources, investigation. Hongwei Zhang: project administration, visualization, validation. Wei'e Han: resources, supervision, data curation. Min Bai: data curation, supervision, resources, investigation. Xiaolan Liu: resources, supervision, data curation, project administration. Min Zhang: data curation, supervision, resources, methodology. Liping Su: conceptualization, resources, data curation. Weihua Zhang: conceptualization, methodology, resources, writing – original draft.

Ethics Statement

The Ethics Committee of the Shanxi Medical University Affiliated Cancer Hospital (Shanxi Provincial Cancer Hospital) which is registered as Institutional Review Board at the National Medical Research Registration and Filing Information System (https://www.medicalresearch.org.cn) approved the evaluation presented here (KY2023080).

Conflicts of Interest

The authors declare no conflicts of interest.