Time to revisit reduced-dose radiotherapy for PCNSL?

In this issue of Blood Advances, Tringale et al¹ report a single-institutional analysis of patients with treatment-naïve primary central nervous system lymphoma (PCNSL), summarizing changing practices in consolidation strategies and clinical outcomes between 1983 and 2020. Of the 559 patients, 385 (69%) received post-induction consolidation therapy, with either whole-brain radiotherapy (≤24Gy reduced-dose [RD], or >24Gy standard-dose [SD] WBRT) and cytarabine (Ara-C), autologous hematopoietic cell transplantation (AHCT), or non-myeloablative chemotherapy (NMC), typically Ara-C. WBRT use significantly reduced during this period, from 61% in the 1990s to 12% in the 2010s, largely replaced with AHCT or NMC (4% to 32% and 27% to 52% respectively). This temporal change in consolidation practices was seen across all prognostic groups (Memorial Sloan-Kettering Cancer Centre Recursive partitioning analysis [MSKCC RPA] classes 1-3), with a preference for NMC in higher RPA classes more recently.

These evolving patterns highlight the ongoing uncertainty regarding the role of consolidation in PCNSL. SD-WBRT has demonstrated improved progression-free survival (PFS) but not overall survival (OS) and is associated with significant neurotoxicity particularly in those aged >60 years.² Two more contemporary randomized studies, IELSG32 and PRECIS, demonstrated that replacing SD-WBRT with AHCT significantly reduced the risk of neurotoxicity without compromising efficacy.^3,4 AHCT is however associated with significant morbidity and mortality particularly in older, frailer patients.⁵ The role of NMC as consolidation is even less clear, and although data suggest it is inferior to AHCT,⁶ it is hard to draw firm conclusions given the heterogenous regimens used (both for induction and consolidation chemotherapy), as well as the difficulties of defining NMC consolidation itself; should administering Ara-C after high-dose methotrexate (HD-MTX) be considered “consolidation” or is it simply delivering induction therapy in a sequential rather than concomitant manner?

A total of 351 of 385 patients (91%) in Tringale et al’s cohort achieved a complete response (CR/CRu) after consolidation. Importantly, these patients had good long-term outcomes: 2-year PFS of 67% (95% confidence interval [CI], 62-72%), 5-year PFS of 46% (95% CI, 41-53%), 2-year OS of 85% (95% CI, 81-89%), and 5-year of OS 65% (95% CI, 59-71%), calculated from the start of consolidation, which compare favorably with studies from similar eras.^7-9 Furthermore, a 6-month landmark analysis demonstrated that receipt of any consolidation resulted in improved OS (P < .001). Overall, the findings from Tringale et al show the potential benefits of supplementing HD-MTX chemotherapy induction with additional “consolidation”, regardless of modality of the latter.

In Tringale et al’s study, patients treated with RD-WBRT (with Ara-C) achieved improved PFS compared with non-AHCT consolidation (predominantly NMC with Ara-C), particularly in the RPA 1 and RPA 2 cohorts. Morris et al first demonstrated the potential benefit of RD-WBRT after R-MPV/Ara-C (rituximab, HD-MTX, procarbazine, vincristine, followed by Ara-C) chemotherapy in 31 patients (2-year PFS, 77%; 5-year OS, 80%).¹⁰ Although those aged >60 years did worse (P = .02), importantly, significant neurocognitive sequalae were not seen, including no deaths attributable to RD-WBRT. Subsequent preliminary data from NRG-RTOG 1114, a randomized study of R-MPV/Ara-C with and without RD-WBRT (n = 91; median age, 66 vs 59 years), demonstrated significantly improved outcomes in those treated with RD-WBRT (hazard ratio, 0.51; 95% CI, 0.27-0.95; P = .015) without significant compromise in neurological function as per investigator assessment.¹¹ Although formal neuroimaging and neuropsychological outcomes are awaited, these findings, taken together with the work of Tringale et al and Morris et al, highlight not only the potential benefits of RD-WBRT compared with Ara-C alone but also raise the question of whether RD-WBRT could be a viable alternative to AHCT, one best answered via a prospective, randomized trial.

The findings from Tringale et al are contrasted by a recent retrospective study in 190 patients from 11 centers showing 2-year PFS of 55% (47%-62%) and OS of 77% (70%-83%), despite only 31% of patients receiving WBRT.¹² Best outcomes were seen in the R-MPV cohort (2-year PFS, 74%; P = .0001; 2-year OS, 82%; P = .0024), 74% of whom also received “consolidation” with Ara-C. Although this study was underpowered to assess the impact of WBRT, a task made even more difficult by the heterogenous dosing of “consolidative” WBRT (20-45Gy), the impact of HD-MTX dosing was assessed, with dose intensity and, to a lesser degree, cumulative dose, associated with improved outcomes. These findings are consistent with two other recent retrospective analyses demonstrating maintaining HD-MTX dose intensity and cumulative dose are the key to improving outcomes, particularly in older patients (≥60 years), who form the majority of patients with PCNSL not treated on a clinical trial.^9,13 Unfortunately dosing data, which would have further informed the likely impact of consolidation strategies, were not collected in Tringale et al’s study, not unreasonably given the near 40-year census period. They too, however, identified the type of induction (R-MPV) as an important predictor of outcome, even when accounting for the decade of diagnosis and consolidation strategy used. R-MPV/Ara-C remains a viable alternative to MATRix (HD-MTX, Ara-C, thiotepa, rituximab), which is challenging to deliver at full dose to patients outside of clinical trials.¹⁴ 

Tringale et al acknowledge the limitations of their findings, in particular the selection bias that is inherent to single-center studies, which may account for the younger age of their cohort than that of registry data (median age, 63 versus 67 years).⁸ Nonetheless, the authors suggest future therapies should incorporate novel agents in first-line therapy, such as Bruton tyrosine kinase (BTK) inhibitors, as well as multimodal consolidation strategies (ie, involved-site radiotherapy and chimeric antigen receptor T-cell therapy), taking into account both patient factors and response to induction to optimize outcomes. Another intriguing approach would be to design a study that incorporates novel agents into induction regimens but also builds on the concept of a “sequential response-adapted” therapeutic strategy. For example, vincristine in R-MPV/Ara-C, which has poor CNS penetration, could be replaced with a PD-1 inhibitor and/or a BTK inhibitor. Patients could then proceed to maintenance therapy with a novel agent, only proceeding to AHCT (or RD-WBRT) if there is evidence of disease relapse/progression on multimodal response assessment using MRI, PET, and circulating tumour DNA.^15,16 Such an approach could further improve outcomes: maximizing response rates and reducing toxicity by reserving AHCT/WBRT for those who truly need it.

Tringale et al have effectively outlined the evolving landscape of consolidation strategies and outcomes for patients with PCNSL and potentially reinvigorated interest in RD-WBRT.

Conflict-of-interest disclosure: M.T. declares no competing financial interests.