Qian and colleagues extensively analyzed the available treatment options, such as high-dose methotrexate (HD-MTX) alone or as a component of various MTX-based chemotherapy regimens, whole-brain radiotherapy (WBRT), and surgery. We agree with the authors that HD-MTX should be included in the first-line therapy; according to our knowledge, however, the best available evidence suggests that HD-MTX should be administered in association with high-dose cytarabine to improve both progression-free survival (PFS) and overall survival (OS), as previously suggested [3].

An important goal of PCNSL treatment is survival prolongation with minimal toxicity, especially neurotoxicity. The first randomization of the phase II IELSG32 (the International Extranodal Lymphoma Study Group-32) trial was designed to determine whether rituximab and thiotepa could improve the efficacy of first-line treatment comprising HD-MTX plus HD-cytarabine (MATRix regimen). The complete response (CR) rate among patients receiving HD-MTX plus HD-cytarabine (control arm, arm A) was 23%, compared to 30% in the arm receiving rituximab (arm B) and 49% in the arm receiving both rituximab and thiotepa (group C); here, a multivariate analysis confirmed an independent association between the induction arm and CR rate [4]. The recently published second randomization was designed to investigate the efficacy of WBRT or autologous stem-cell transplantation (ASCT) as a consolidation therapy after induction for patients with chemosensitive PCNSL. Out of 122 eligible patients, 118 were randomly assigned to receive WBRT (group D) or ASCT (group E); both strategies were effective and yielded significantly improved CR rates after induction, with 2-year PFS rates of 80% and 69%, respectively [5]. As expected, hematological toxicity was more common in ASCT arm, while neuropsychological tests demonstrated cognitive impairments in attention and executive functions among patients receiving WBRT [5].

ASCT was previously shown to be high effective as a consolidation therapy with manageable toxicity in phase II trials of patients with chemosensitive PCNSL patients; consequently, an international phase III study is ongoing and will randomize patients to receive ASCT or conventional chemoimmunotherapy [6].

Because the outcomes of PCNSL are relatively worse than those of systemic DLBCL, novel agents are under investigation. The Bruton tyrosine kinase inhibitor ibrutinib was recently shown to be highly effective against relapsed/refractory PCNSL in a phase I trial [7]. Additionally, chimeric antigen receptor-modified T-cell therapy exhibited anti-tumor activity against a heavily pre-treated DLBCL with CNS localization [8].

We would also like to address the important topic of the treatment of elderly patients with PCNSL, who account for an important proportion of the total population of patients with PCNSL. The MATRix regimen is not applicable for patients >70 years old; additionally, WBRT is expected to cause significant neurotoxicity, and a recently reported study demonstrated no improvements in survival in the last decades [9]. A recent meta-analysis confirmed that treatment with a combination of HD-MTX and alkylating agents such as procarbazine or temozolomide was associated with improved survival. Although WBRT may also improve survival, it is associated with a significant risk of early or late neurotoxicity; therefore, previous authors suggested avoiding WBRT or using it only as a strategy for relapsed disease [10]. Among elderly patients with PCNSL, promising results and manageable toxicity were reported with HD-MTX when administered in association with rituximab, procarbazine, and lomustine (lomustine was omitted via protocol amendment because of infectious complications); specifically, 38 (35.5%) and 15 of 107 patients (14%) achieved a CR and PR, respectively, while the 2-year PFS and OS were 37.3% and 47%, respectively [11].

At our institution, we have worked for many years in the field of optimal PCNSL management and have retrospectively analyzed a consecutive series of 20 patients diagnosed and treated during the period from 2005 to 2016. Younger and more fit patients received HD-MTX with cytarabine either alone or in combination with rituximab with or without thiotepa (rituximab 375 mg/m² days −5 and 0, MTX 3.5 g/m² day 1, cytarabine 2 g/m²×2, days 2–3, thiotepa 30 mg/m² day 4; MATRix regimen) every 3 weeks for up to 4 cycles. Elderly and unfit patients received HD-MTX and temozolomide. Induction was followed by WBRT with a minimum dose of 36 Gy or ASCT as consolidation. Responses were assessed after the 2nd and 4th cycles according to the 2005 IPCG Response Criteria [12]; toxicity was defined according to the NCI-CTCAE criteria after each course of treatment. PFS was measured from date of treatment initiation to the date of relapse or progression requiring subsequent treatment, while OS was measured from the date of treatment initiation to the date of death. Statistical analyses were performed using MedCalc software, v2.0 (MedCalc, Ostend, Belgium). Survival was analyzed using the Kaplan-Meier method and the global log-rank test.

The median age was 61.5 years; additionally, 13 of 20 patients (65%) had an Eastern Cooperative Oncology Group Performance score of ≥2 and 4, 14, and 2 patients had low, intermediate, and high International Extranodal Lymphoma Study Group scores, respectively, as shown in Table 1. Five of 20 patients (25%) discontinued treatment early (after 1 cycle) because of disease progression or toxicity and were analyzed separately in survival curves. Fifteen patients received at least 2 courses of treatment; among them, 10 were responders (66.6%) and 5 (33.3%) achieved a CR, as reported in Table 2. Two patients older than 70 years who received HD-MTX and temozolomide experienced progressive disease (PD). Grade 3–4 hematological toxicity was reported in all cytarabine-treated cases, and fatal infections were observed in 3 of 20 patients (15%) as presented in Table 3. All 10 responders received WBRT (7 patients) or ASCT (3 patients) as consolidation; 3 patients receiving WBRT and 1 receiving ASCT exhibited improved responses from PR to CR. Neurotoxicity was reported in 5 patients (4 received WBRT). The median PFS and OS improved among patients receiving at least 2 courses of treatment, compared to the early discontinuation group (12 vs. 2 mo, P=0.03; 10 vs. 4 mo, P=0.01). In the first group, the estimated 5-year PFS (Fig. 1) and OS were 35% and 38%, respectively.

Our study had some limitations, most notably the retrospective study design and small sample size, although these were comparable to previously published experiences [1314]. Although our cohort of 20 patients could not adequately represent the general population, our 10-year experience may help to confirm the difficulties experienced in clinical daily practice regarding the adequate treatment for PCNSL because of issues related to age, diagnostic delays, and a poor performance status. In conclusion, the presented data have led us to confirm that the treatment outcomes in real-life practice are somewhat different from those in clinical trials. We suggest that the optimal induction therapy should include HD-MTX, HD-cytarabine, rituximab, and thiotepa in younger and fit patients, while WBRT and ASCT are both effective as consolidation therapies.