CNS prophylaxis is (mostly) futile in DLBCL

The CNS International Prognostic Index (CNS-IPI) is a validated risk assessment model to help determine how high of a risk an individual patient may be for CNS relapse. It combines the traditional five IPI factors (age >60, elevated LDH, ECOG PS > 1, stage III/IV disease, and involvement of more than one extranodal site) and adds involvement of the kidneys and/or adrenal glands ([2]; Table 1). The CNS-IPI was developed by analyzing over 2100 patients in 2 large clinical trials and was validated against a data set of 1597 patients. In the CNS-IPI, patients are stratified into the low-risk (CNS-IPI 0-1), intermediate-risk (CNS-IPI 2-3), and high-risk (CNS-IPI 4-6) groups. Those in the lower risk group had a 0.6% 2-year incidence of CNS relapse, whereas the highest risk group had a greater than 10% chance. However, the CNS-IPI does not adequately capture all patients at high risk of CNS relapse, particularly those with testicular involvement, who often have limited stage disease.

Consensus guidelines from expert groups differ in their indications for treatment and the recommended route of administration. European Society for Medical Oncology (ESMO; [4]) and British Society of Hematology (BSH; [5]) recommend CNS prophylaxis in patients with high IPI or CNS-IPI or for those with high-risk extranodal site involvement, namely testicular, renal, or adrenal involvement. Both groups suggest HD-MTX as the preferred route. By contrast, the latest National Comprehensive Cancer Network (NCCN) Guidelines from the United States (version 2.2024; [6]) suggest only consideration of prophylaxis in patients with high-risk factors with either HD-MTX or IT chemotherapy.

CNS prophylaxis with HD-MTX can be given during systemic chemoimmunotherapy (intercalated HD-MTX) or at end of treatment (EOT). Previous studies have suggested that CNS recurrence happens early in the course of treatment [1]; therefore, some administer prophylaxis during the standard treatment.

On the other hand, recent studies have found no difference in CNS relapse rates between intercalated and EOT therapy but did identify a higher rate of treatment delays when therapy was given in an intercalated fashion. In a large, retrospective analysis of 1384 patients with DLBCL who received CNS prophylaxis with HD-MTX, patients on intercalated therapy did not have a difference in 3-year CNS relapse rates when compared to those in the EOT arm (5.7% vs 5.8%; p = 0.98) [7]. Those with high CNS-IPI (n = 600) had a 3-year relapse rate of 9.1%, but there was no difference between the intercalated or EOT groups. Intercalated HD-MTX, however, was associated with increased risk of R-CHOP delays, with 19.6% of patients receiving intercalated HD-MTX having delays of ≥7 days. The risk of delay was higher in older patients and those who received HD-MTX after day 10 of the cycle. Thus, if HD-MTX is planned, it should be done at the end of treatment to minimize delays in systemic treatment.

Although earlier reports suggested that HD-MTX had better penetration of the brain parenchyma and was associated with a lower incidence of relapse compared to IT [8], several larger studies have generally found no significant benefit with prophylaxis with HD-MTX, particularly in patients with complete response to R-CHOP or similar chemoimmunotherapy given with curative intent. One large retrospective evaluated 2418 DLBCL patients, including 1616 patients with a complete response (CR) to therapy. Most of the patients were high risk, with 83.4% of patients having CNS-IPI scores of 4–6. Approximately one-fourth of the patients had received HD-MTX. Patients who received CNS prophylaxis with HD-MTX did not demonstrate an improvement in CNS relapse rates, with a 5-year adjusted risk difference of 1.6% (95% CI, –1.5 to 4.4) in all patients, and 1.4% (95% CI, –1.5 to 4.1) in CR patients [9].

Another review of patients with a high risk of CNS disease also showed no difference in relapse or survival with the addition of HD-MTX. In the CNS-IPI high-risk group, 115/326 (35.3%) of patients received HD-MTX. There was no difference in relapse risk between the patients who received HD-MTX and those who did not (11.2% vs. 12.2%; p = 0.82). Multivariate and propensity score analyses did not demonstrate improvement in relapse, progression-free survival, or overall survival [10].

A single-armed study by the Nordic group incorporated both HD-MTX and cytarabine (2–3 g/m2, depending on patient age). Of the 156 patients who were treated, 7 (4.5%) experienced a CNS relapse. The baseline CNS-IPI scores were not reported. The majority of patients had an age-adjusted IPI of 2 (75%), and very few patients had involvement of >2 extranodal sites (13.5%), suggesting this group was generally at intermediate risk of CNS relapse at baseline and that the risk of relapse was not clearly mitigated by the incorporation of HD-MTX and cytarabine [11].

Intrathecal chemotherapy with methotrexate or cytarabine has also been used as CNS prophylaxis, although IT chemotherapy has poor penetration into the brain parenchyma. Previous studies were small but found that HD-MTX was associated with a lower incidence of relapse compared to IT chemotherapy [8]. More recently, several analyses found no differences between HD-MTX and IT chemotherapy.

In a large retrospective study of 1162 patients comparing CNS prophylaxis with IT and systemic methotrexate (MTX), there was no difference in the relapse rates or survival [12]. Approximately four-fifths of the patients received IT prophylaxis, while 20% received HD-MTX. Thirty percent of the patients had high-risk disease according to the CNS-IPI and approximately half were considered moderate risk. However, there was no statistically significant difference in relapse rates between the IT and HD-MTX patients (5.4% vs. 6.8%; p = 0.40), even after propensity score matching. The predicted CNS relapse rate, weighted by CNS-IPI score, was nearly identical to the observed relapse rate (5.8% vs. 5.7%).

Systemic and IT prophylaxis were also compared in a prospective, multicenter, phase III trial involving 142 patients [13]. There was no statistical difference in the 2-year cumulative incidence of CNS relapse between the IT MTX (5.5%, 95% CI 0.6–19.1) and HD-MTX arms (4.9%, 95% CI 0.2–23.9; p = 0.749). Interestingly, the median time to CNS relapse was 4.4 months in the IT MTX arm and 12.0 in the HD-MTX arm, suggesting that perhaps HD-MTX delays rather than prevents CNS recurrence. The 2-year PFS was also similar between the two groups (70.4% vs 66.4%; p = 0.571). HD-MTX was also associated with significant treatment delays of over 7 days in 59% of patients [13].

Improvements in induction chemoimunotherapy strategies and supportive care have improved outcomes for newly diagnosed patients with DLBCL through the last few decades, but treatment of secondary CNS disease remains a challenge. CNS prophylaxis has often been recommended to patients at highest risk of relapse, particularly patients with CNS 4–6 or certain high-risk sites of disease (renal, adrenal, testicular). Recent studies have generally not shown any benefit of HD-MTX or IT prophylaxis on CNS relapse or survival. In particular, intercalated HD-MTX risks substantial treatment delays compared to EOT and should generally be avoided. The studies summarized here suggest that if CNS prophylaxis is planned, IT chemotherapy has a better tolerability profile over systemic therapy without compromising overall clinical outcomes.

An algorithm proposed by Wilson and colleagues provides guidance regarding an optimal strategy for CNS prophylaxis [14]. Patients without high-risk features (i.e., CNS-IPI 4–6, ≥3 extranodal sites, or high-risk extranodal sites such as testicular, renal/adrenal, or breast) can continue with systemic therapy without any CNS prophylaxis. If any high-risk features are present, baseline evaluation of the CNS should be conducted with MRI imaging and flow cytometry of the CSF. If the workup is negative, patients should proceed to systemic therapy. Even though there is lack of evidence of efficacy of CNS prophylaxis, the patient with CR at EOT can be considered for CNS prophylaxis, especially those at highest risk with CNS 5–6, or renal, adrenal, or testicular involvement after an informed risk-benefit discussion (Fig. 1).

Proposed algorithm (Adapted from Wilson MR et al., [14]). CGA = comprehensive geriatric assessment; EPI = elderly prognosis index; EOT = end-of-treatment; CMR = complete metabolic response; sCNSL = secondary CNS lymphoma 

Full size image

One hypothesis for the lack of clear benefit of CNS prophylaxis has been that the incorporation of rituximab into initial treatment regimens has in and of itself reduced the risk of CNS relapse. However, whether or not rituximab penetrates the blood-brain barrier (BBB) for most patients remains uncertain [15]. A retrospective analysis of 435 patients at increased risk of CNS relapse (advanced stage disease and/or testicular involvement) examined differences in CNS relapse when patients were given CHOP or R-CHOP. Some patients received IT chemotherapy per institutional guidelines, and those with testicular lymphoma received scrotal radiation. In the total population, 31 patients (7.1%) experienced CNS relapse. While numbers are small, there was a trend toward fewer CNS relapses in the R-CHOP cohort (6.4% vs 9.7% at 3 years). When looking specifically at patients who were in CR following their initial therapy, 3-year CNS relapse was 2.2% in the R-CHOP treated cohort vs 5.8% in the CHOP treated cohort [16]. This supports the idea that perhaps the best “prophylaxis” for CNS relapse is giving the most effective initial therapy to achieve CR. Incorporation of circulating tumor DNA (ctDNA) to further assess depth of responses to therapy could be helpful in stratifying risk of CNS relapse [17].

In summary, CNS prophylaxis with either systemic or IT chemotherapy has not demonstrated efficacy in several large retrospective analyses. A prospective study also showed no difference in outcomes with HD-MTX and IT MTX. Ideally, the question about whether or not CNS prophylaxis should be incorporated into the initial therapy regimen for a subset of patients with DLBCL would be evaluated in a large prospective effort. This ends up being a complicated study to design for a number of reasons. Even in high-risk situations, the risk of CNS relapse is only about 10%, leading to a very large sample size. Additionally, whether or not the CNS-IPI or some other tool should be used for patient identification is not clear. Despite the data detailed above, many clinicians feels strongly about CNS prophylaxis and would feel uncomfortable randomizing patients to no prophylaxis. Alternately, many providers have moved on from CNS prophylaxis based on the data reviewed above and feel that administering any for of prophylaxis would lead to unnecessary risk without benefit.

Thus, the optimal strategy to prevent CNS progression in DLBCL remains unclear, and the optimal way to prospectively study the question is controversial as well. Identification of new biomarkers, utilizing ctDNA in the blood and/or CNS, and incorporation of novel or targeted therapies which have blood-brain barrier penetration may be important tools that can be helpful in preventing CNS relapses and will need to be studied in a step-wise fashion.