The role of CNS prophylaxis in the frontline setting for patients with DLBCL

Introduction

Although the incidence of central nervous system (CNS) relapse is rare (up to 5%) in diffuse large B-cell lymphoma (DLBCL),¹ it remains a clinical challenge with poor prognostic outcomes.² CNS prophylaxis is used as a precautionary measure for patients identified as high-risk for CNS relapse; however, there is no clear standard of care or strong evidence for its use in frontline settings. CNS International Prognostic Index (CNS-IPI) is the current standardized model for defining high-risk CNS relapse factors, although it does have major limitations.²

Below, we discuss the role of CNS-IPI in estimating CNS relapse risk; the efficacy of two types of CNS prophylaxis, intrathecal (IT) prophylaxis and high-dose methotrexate (HD-MTX), in mitigating the risk of CNS relapse; as well as the future directions in this area, as presented by Orellana-Noia at the Society of Hematologic Oncology (SOHO) 2022 Annual Meeting.²  

CNS International Prognostic Index (CNS-IPI)

CNS-IPI is the most used prognostic model for estimating CNS relapse risk, validated in the British Columbia Cancer Agency Lymphoid Cancer Database—a large population-based data set of patients with DLBCL treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone or prednisolone (R-CHOP) based therapy.³

This model consists of the following six risk factors: age >60 years, elevated lactate dehydrogenase, Eastern Cooperative Oncology Group (ECOG) Performance Status >1, Stage III/IV disease, more than one extranodal site, and involvement of renal and adrenal glands, which characterizes patients into three risk categories based on a scoring system: low (0–1 points), intermediate (2–3 points), and high risk (4–6 points). Amongst the validation set, the 2-year CNS rates of relapse were 0.8%, 3.9%, and 12% for low-, intermediate-, and high-risk groups respectively.³

Whilst CNS-IPI has been shown to predict CNS relapse, it has significant limitations. Firstly, it does not fully establish all patients in the high-risk population, and secondly it is not indicative of the allocation of prophylactic measures.² Another major limitation is the inability to capture other clinical risk factors such as testicular involvement, and molecular features such as double-hit status and genetic subtypes, which could predict a CNS-relapse risk in patients with DLBCL.^2,3

A retrospective analysis evaluating the involvement of several extranodal sites in CNS relapse found that compared to CNS-IPI, two or more extranodal sites yielded higher specificity and overall accuracy. The 3-year cumulative incidence of CNS relapse was 15.2% in a group of 144 patients (9%) with three or more extranodal sites.⁴

A substudy analysis of the phase III GOYA study investigating the impact of CNS-IPI in combination with cell-of-origin (COO) subtypes revealed an independent association of the activated-B-cell-like (ABC) and unclassified COO subtypes with CNS relapse, for both univariate and multivariate analyses.⁵ The multivariate analysis showed a significantly higher CNS relapse rate amongst patients with a high CNS-IPI score vs low CNS-IPI score (hazard ratio [HR], 4.0; 95% confidence interval [CI], 1.3–12.3; p = 0.02), ABC subtype vs germinal center B-cell–like (HR, 5.2; 95% CI, 2.1–12.9; p = 0.0004), and unclassified COO subtype vs GCB (HR, 4.2; 95% CI, 1.5–11.7; p = 0.006). The combination of CNS-IPI with ABC and unclassified COO improved prediction of CNS relapse risk, with a reported 15.2% in the high-risk subgroup.⁵

CNS prophylaxis

Intrathecal prophylaxis

A multicenter study (N = 690) in patients aged ≥70 years with DLBCL who were treated with R-CHOP was carried out across eight centers in the UK between 2009 and 2018, aiming to elucidate the value of IT prophylaxis in this patient subset.² Whilst the use of CNS-IPI was validated in this cohort, there was no significant difference in the CNS relapse incidence between those who received IT prophylaxis compared with those who did not when adjusting for CNS-IPI (HR, 1.34; 95% CI 0.46–3.86; p = 0·59). Additionally, infection-related toxicity was evident in this analysis, with a higher (44.4%) readmission rate for all-cause infections amongst patients in receipt of IT prophylaxis compared to those not receiving it (25.2%) p < 0.001.⁶

A subsequent UK-based meta-analysis of stand-alone IT prophylaxis, including three trials and ten retrospective series (N = 7,357), reported that median IT prophylaxis was 11.9% and most CNS relapses were parenchymal. Overall, there was no benefit of IT prophylaxis in reducing the risk CNS relapse for any of the studies reported.²

High-dose methotrexate as CNS prophylaxis

The effectiveness of HD-MTX as CNS prophylaxis was evaluated in a multicenter retrospective study in adult patients aged 18–70 years (N = 906) with DLBCL who were part of the Alberta Cancer Registry from 2012 to 2019.⁷ Between the years 2012 of 2015, high-risk criteria for treatment with HD-MTX was an elevated lactate dehydrogenase, ECOG >1, and > 1 extranodal site involvement. After 2015, criteria included a CNS-IPI of 4–6, double-hit lymphoma, and testicular involvement. Among patients identified as high-risk for CNS relapse (n = 326), around a third (n = 115) received HD-MTX, demonstrating that identification of high-risk patients does not always indicate the uptake of CNS prophylaxis.⁷

CNS relapse risk occurred in 44/906 patients, with similar CNS relapse risk rates between high-risk patients who received HD-MTX and those who did not (11.2% vs 12.2%; p = 0.82). A propensity score analysis, which accounted for differences in baseline risk factors, and multivariate analyses accounting for CNS-IPI score, double-hit lymphoma, and testicular involvement found no significant effect of HD-MTX on CNS relapse, progression-free survival, and overall survival.⁷

The results from an international, multicenter, retrospective study in a total of 2,300 patients across 21 centers and 3 continents, previously reported on the Lymphoma Hub, showed that among the high-risk DLBCL subset (CNS-IPI 4–6, double-hit lymphoma, or testicular or breast involvement) there was no significant reduction in CNS relapse for patients receiving prophylactic HD-MTX vs no HD-MTX treatment (7.8% vs 8.9%).²

An international multicenter retrospective study on the impact of timing of HD-MTX on CNS relapse in patients with DLBCL (N = 1,384) who received at least one dose of HD-MTX, previously reported on the Lymphoma Hub, investigated the effect of intercalated vs end-of treatment administration on CNS relapse.⁸ In this study, no difference was found between the intercalated HD-MTX (n = 749) and end-of-treatment (n = 635) approaches, with 3-year CNS relapse rates of 5.7% and 5.8%, respectively. Moreover, the CNS relapse rates were similar to historical controls for high-risk groups. ^2,8

IT-MTX vs IV HD-MTX approaches

A multicenter analysis of patients (N = 1,162) with DLBCL, conducted between 2013 and 2019, investigated the effect of route of CNS prophylaxis administration on CNS relapse risk. Patients were given at least one cycle of anthracycline based therapy plus either IT (n = 894) or intravenous (IV) prophylaxis (n = 236), with the remaining 32 patients switching route due to toxicity. Results showed no significant difference in CNS relapse rates between the IT and IV prophylaxis routes (5.4% vs 6.8%; OR, 1.28; 95% Cl, 0.71–2.30; p = 0.4). After propensity score matching for several variables, differences between the IT and IV routes converged, and CNS relapse rates were more similar (5.0% vs 5.6%); this demonstrates that route of administration is not a determining factor for effective treatment.²

Future directions

The introduction of novel markers such as ctDNA could shift the prognostic and treatment paradigm of CNS relapse in the future.² The Lymphoma Hub has previously reported on the emerging role of ctDNA, highlighting two studies which demonstrate the potential of CSF ctDNA analysis as a diagnostic and prognostic indicator of CNS lymphomas. Another study (N = 217) involving six treatment centers assessed the value of pretreatment ctDNA and ctDNA dynamics during therapy on disease outcomes.⁹

Pretreatment levels of ctDNA were identified in 98% of patients and coincided with IPI and total metabolic tumor volume scores, demonstrating its potential as a prognostic biomarker of disease presence and disease burden. Moreover, pretreatment ctDNA levels were shown to be prognostic of EFS and OS outcomes, with high ctDNA levels significantly associated with inferior EFS rates in the frontline (p = 0.007) and relapsed/refractory setting (p = 0.01) when compared with low ctDNA levels at a threshold of 2.5 log hGE/ml. During early therapy, early molecular response (EMR; 2-log drop in ctDNA after one cycle) and major molecular response (MMR; 2.5-log drop after two cycles) were also predictive of EFS and OS outcomes, with EMR and MMR significantly associated with superior EFS outcomes in the frontline setting (EMR: EFS, 83% v 50%; p = 0.0015; MMR: EFS, 82% v 46%; p < .001) for the first validation set.⁹

Conclusion

Based on the studies mentioned, it is still uncertain whether CNS prophylaxis can reduce CNS relapse rates in high-risk patients with DLBCL regardless of the route of administration. Randomized trials are urgently required to produce robust and conclusive evidence for the use of prophylaxis in frontline therapy. Combining the CNS-IPI with other clinical and molecular risk factors, alongside the potential prognostic value of pretreatment ctDNA, could increase the sensitivity, specificity, and reliability of prognostic models and improve CNS treatment strategies for future patients.