ICE polychemotherapy is an efficient salvage treatment in primary central nervous system lymphoma: an LOC network study

TO THE EDITOR:

Primary central nervous system lymphoma (PCNSL) is a rare subtype of diffuse large B-cell lymphoma (DLBCL) with a poor prognosis.¹ First-line (1L) treatment relies on a multiagent chemoimmunotherapy induction regimen containing high-dose (HD) methotrexate, followed by a consolidation phase. Recent studies demonstrate that consolidation with HD chemotherapy (HDC) followed by autologous stem cell transplantation (HDC-ASCT) improves prognosis in fit patients who achieve at least a partial response (PR) after induction.^2-4 This is the case even in patients of advanced age, provided that their general condition allows it.^5,6 Nevertheless, 25% to 30% of patients are refractory to 1L therapy, whereas 10% to 20% and >50% eventually relapse,⁷ depending on whether 1L includes HDC-ASCT. Patients with relapsed/refractory PCNSL (R/RPCNSL) have a poor median overall survival (OS) of 6.8 months,¹ representing an unmet medical need.⁸ 

Here, we report a multicentric retrospective study of immunocompetent patients with rrPCNSL treated with the ICE (ifosfamide, carboplatin, and etoposide) regimen.

Between 2010 and 2022, a total of 96 adult patients with rrPCNSL who were treated with the ICE regimen at 5 French centers of the Lymphomes Oculo-Cérébraux (LOC) network were included in the analysis. The study was conducted in accordance with the Declaration of Helsinki and received approval from the French national expert network for LOC Institutional Review Board (CNIL 913170). The inclusion criteria were as follows: age ≥18 years; PCNSL initial diagnosis confirmed by histology of brain biopsy or cytology of cerebrospinal fluid or vitrectomy; PCNSL relapse required the same criteria as diagnosis or only imaging in case of early relapse (<2 years); and negative whole-body computed tomography scan or fluoro-deoxy glucose (FDG)–positron emission tomography scan. Patients with histologies other than DLBCL, HIV infection (supplemental Methods), posttransplant lymphoproliferative disorder, and secondary central nervous system (CNS) localization of systemic DLBCL were excluded. Additional data regarding ICE modalities, response criteria, and statistics are available in supplemental Data.

The principal characteristics of the entire cohort are presented in Table 1. At the time of ICE initiation, the median age was 65 years (range, 34-84; interquartile range [IQR], 56-69), with 17% of patients aged >70 years. The median time between PCNSL diagnosis and ICE initiation was 7.6 months (IQR, 4.8-14.3). Seventy percent of patients were primary refractory (21%) or progressed during 1L therapy after initial response (49%). Seventy-three percent of the cohort received ICE in 2L. The median time from the last therapy initiation was 5.3 months (IQR, 3.1-9.8). All patients were previously exposed to HD methotrexate, 95% to HD cytarabine, 48% to rituximab, 7% to whole-brain radiotherapy (WBRT), 6% to HDC-ASCT, 4% to lenalidomide, and 3% to ibrutinib.

The median number of ICE cycles was 4 (range, 1-7). Most adverse events were hematologic, with 80% of patients experiencing grade ≥3 neutropenia, including 32% with febrile neutropenia and 90% with grade 3 thrombocytopenia. Three patients experienced encephalopathy, which was fully reversible in each case (supplemental Table 1). Twenty-seven patients (28%) received reduced doses of ICE (ranging from 50%-90% of the full dose) from the first cycle due to Eastern Cooperative Oncology Group performance status (PS), comorbidities, and/or renal failure. These patients exhibited a higher median age than the rest of the cohort (69 vs 61 years; P < .0001).

The best responses obtained after ICE were complete response (CR) and PR in 37% and 33% of cases, respectively (Table 1). The median time to the best response was 1.4 months (IQR, 1.0-2.8). Twenty-three percent of patients exhibited primary resistance to ICE therapy. It is noteworthy that among the 66 responders, 15 (23%) exhibited further progression during ICE therapy at a median of 4 ICE cycles. The response rates were higher when ICE was used in 2L vs 3L or more, with CR/PR/stable disease + progressive disease observed in 38%/36%/25% and 32%/24%/44%, respectively (P = .09). The addition of rituximab did not affect the rate of response. Forty-four patients (46%), who achieved PR/CR after salvage therapy, received HDC-ASCT (thiotepa/busulfan/cyclophosphamide, n = 22; thiotepa/BCNU, n = 14; thiotepa/busulfan, n = 3; busulfan/melphalan, n = 1; not available, n = 4), with a median time from ICE initiation to HDC-ASCT of 4.0 months (range, 2.2-40.3), either directly after ICE (n = 29) or after a further line of treatment (n = 15). These patients were significantly younger than the rest of the cohort (median age, 59 vs 67 years; P < .0001), had lower Eastern Cooperative Oncology Group PS (33 vs 63% with PS >1; P < .0004), received ICE more frequently in 2L (93 vs 56%; P = .0004), and were significantly more frequently in CR/PR after ICE (91 vs 50%; P < .001; supplemental Table 2).

With a median follow-up of 10.1 months (IQR, 4.9-35.1) from ICE initiation, the median progression-free survival (PFS) and OS were 3.4 (95% confidence interval [CI], 2.8-4.9) and 8.4 months (95% CI, 6.6-14.4), respectively (Figure 1A-B). Seventy-one patients (75%) relapsed. Sixty-eight patients (71%) died, mainly due to lymphoma progression (55/68 [81%]; supplemental Table 1).

Univariate analysis identified age (>60 vs ≤60 years; hazard ratio [HR], 1.8; 95% CI, 1.03-3.13; P = .04), PS (>1 vs 0-1; HR, 3.69; 95% CI, 2.15-6.33; P < .0001), best response to ICE (CR/PR vs progressive disease/stable disease; HR, 0.4; 95% CI, 0.24-0.67; P < .0005), and the use of HDC-ASCT (yes vs no; HR, 0.17; 95% CI, 0.09-0.32; P < .0001) as the strongest predictive factors for OS (supplemental Table 3). The median OS was 66.6 months (95% CI, 33.9 to not applicable, [NA]) for patients who received HDC-ASCT after ICE, compared with 5.3 months (95% CI, 4.7-7.3) for those who did not receive one (HR, 0.22; 95% CI, 0.13-0.36; P < .0001; Figure 1C). The type of HDC-ASCT (thiotepa/busulfan/cyclophosphamide vs others) did not affect prognosis, neither did reduced doses of ICE. Median OS for patients in CR or PR after ICE was significantly longer for those who further received HDC-ASCT than those who did not (41.8 [95% CI, 10.8 to NA] vs 8.8 months [95% CI, 6.7 to NA]; P = .004; not reached [95% CI, 43.5 to NA] vs 6.8 [95% CI, 4.8-12.7] months; P < .0001, respectively; Figure 1D-F; supplemental Figure 2). All patients who did not receive HDC-ASCT relapsed (n = 51), with a median PFS of 2.3 months (95% CI, 1.7-3). When considering specifically patients receiving ICE and HDC-ASCT in 2L, the median OS was 66.6 months (95% CI, 26.6 to NA), corresponding to a 2-year OS of 67%. Only PS (>1 vs 0-1; HR, 3.97; 95% CI, 1.90-8.28; P = .0002) and HDC-ASCT (yes vs no; HR, 0.20; 95% CI, 0.08-0.54; P = .001) significantly influenced OS in multivariate analysis (supplemental Table 3).

Our study is subject to several limitations inherent to its retrospective design, introducing biases related to the heterogeneity of previous treatments received and patient characteristics. Nevertheless, this report presents, to our knowledge, the largest real-life retrospective cohort of patients with rrPCNSL who received HD ifosfamide–based regimens. Previous studies exploring the efficacy of HD ifosfamide–based regimens in rrPCNSL included a maximum of 27 patients.^9-11 The 70% overall response rate (ORR) observed in our cohort compares favorably with the ORRs observed in previous studies that used HD ifosfamide–based regimens^10-12 or targeted therapies, such as ibrutinib or lenalidomide, which yielded ORRs of 35% to 67%.⁷ However, in the absence of consolidative HDC-ASCT, ICE did not result in long-term disease control, even in patients who achieved CR. For patients who were eligible for HDC-ASCT and responded to ICE, the long-term PFS and OS rates were notably high. This reinforced the notion that consolidation (HDC-ASCT if not performed in 1L, WBRT, or chimeric antigen receptor [CAR] T-cell therapy) or maintenance (for frail patients) are of importance in patients with rrPCNSL. Apart from patients refractory to 1L therapy, the characteristics of our cohort do not precisely correspond with those of patients who might currently be eligible for ICE treatment, because very few patients (n = 6) have undergone 1L consolidative HDC-ASCT in our cohort. Nevertheless, the CR/PR rate was 50% in this subgroup, which suggests exploring the benefit of ICE in this specific context. The recent development of CAR T cells offers another consolidative strategy for patients with PCNSL who relapsed after HDC-ASCT, and ICE may serve as a valid bridging therapy. Recent data highlighted that the response status before CAR T-cell therapy was the most significant factor for post–CAR T-cell therapy outcomes.¹³ 

In conclusion, this work suggested that ICE may be a viable therapeutic option for relatively fit patients (age <75-80 years, with no significant comorbidities), when used in combination with HDC-ASCT for patients who are refractory to 1L therapy or as a bridge to another consolidative strategy, such as WBRT or CAR T-cell therapy, for patients who relapsed after HDC-ASCT.

Acknowledgments: This study was supported by grants from Institut National du Cancer- Direction Générale de l'Offre de Soins-INSERM_12560 (Sites de Recherche Intégrée sur le Cancer CURAMUS is financially supported by the French National Cancer Institute, the French Ministry of Solidarity and Health, and INSERM, with financial support from Institut Thématique Multi-Organismes Cancer-Aviesan).

Contribution: L.P., A.G., S.C., and D.R.-W. designed the research, analyzed data, and wrote the manuscript; L.P., A.G., C.S., K.H.-X., I.B., M.B., V.M., N.A., M.O., M.U., M.L.G.-T., L.S., N.G., M.-P.M., T.G., C.H., S.C., and D.R.-W. recruited patients; and all authors reviewed and approved the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Sylvain Choquet, Service d’Hématologie Clinique, AP-HP, Hôpital Pitié-Salpêtrière, 47-83 Bd de l’Hôpital, 75651 Paris Cedex, France; email: sylvain.choquet@aphp.fr; and Damien Roos-Weil, Service d’Hématologie Clinique, Sorbonne Université, AP-HP, Hôpital Pitié-Salpêtrière, 47-83 Bd de l’Hôpital, 75651 Paris Cedex, France; email: damien.roosweil@aphp.fr.