We retrospectively reviewed the medical records of patients who were diagnosed with DLBCL from 2009 to 2019, and selected patients according to the following criteria: 1) CNS relapse or progression during or after R-CHOP chemotherapy; 2) underwent ASCT after salvage treatments for secondary CNS involvement; and 3) data available for analysis of disease relapse and survival status. We excluded patients who had CNS involvement at diagnosis. As this study aimed to evaluate the role of ASCT in DLBCL patients with secondary CNS involvement, patients who did not undergo ASCT were also excluded. Patient demographics and clinical characteristics at initial presentation and at the time of secondary CNS involvement were collected, and information about the type of salvage therapy after secondary CNS involvement and ASCT was gathered. For patients who relapsed after ASCT, the patterns of relapse and post-relapse outcomes were analyzed, and survival status was last updated in April 2022. The Institutional Review Board of Samsung Medical Center approved this study (IRB No: 2022-05-035), and informed consent was waived due to the retrospective nature.

The evaluation of CNS was conducted for patients having signs and/or symptoms of suspicious CNS involvement during or after R-CHOP chemotherapy. The diagnosis of CNS involvement was based on the combination of neurologic manifestations, magnetic resonance imaging (MRI), and/or histological findings of the brain parenchyma or cerebrospinal fluid (CSF). The patterns of CNS involvement were parenchymal, leptomeningeal, and combined involvement. Parenchymal involvement was diagnosed on brain MRI or brain biopsy. If the cytologic examination of CSF showed the presence of lymphoma cells or suspicious lymphoma cells with increased protein levels, leptomeningeal involvement was diagnosed. In addition, when leptomeningeal enhancement compatible with leptomeningeal involvement was observed in brain or spine MRI of patients with neurologic symptoms or signs, leptomeningeal involvement was diagnosed.

ASCT was carried out for patients who responded to salvage chemotherapy after CNS involvement. Peripheral blood stem cells were collected during salvage chemotherapy or mobilization chemotherapy with etoposide as we previously reported for patients with primary CNS lymphoma (16). The target number of CD34+ cells was more than 3.0 × 10⁶ per kilogram of the recipient’s body weight. The conditioning regimen performed prior to ASCT was the same as that for primary CNS lymphoma, consisting of busulfan and thiotepa (busulfan 3.2 mg/kg days –9 to –5, and thiotepa 5 mg/kg days –4 to –3) as we previously reported (16, 17). Treatment response was assessed according to the involved sites. CNS response evaluation was assessed by contrast-enhanced MRI with neuroradiologic review according to the international guidelines for primary CNS lymphoma (18). If a patient had simultaneous systemic disease involvement, systemic response evaluation was also performed according to the Lugano response criteria (19).

The clinical features and treatment outcomes were analyzed, and categorical variables were evaluated using Fisher’s exact test. The time to secondary CNS involvement was defined as the time between the initial diagnosis of DLBCL to the date of secondary CNS involvement. The post-ASCT overall survival (OS) was defined as the time from the date of stem cell infusion to death from any cause, and living patients were censored at the time of analysis. The post-ASCT progression-free survival (PFS) was defined as the time from the date of stem cell infusion to the date of systemic or CNS progression or relapse or death. Survival was estimated based on Kaplan–Meier curves and compared using a log-rank test. Two-sided statistical tests yielding a P value < 0.05 were considered significant. Survival analyses were performed using IBM PASW version 24.0 software (SPSS Inc., Chicago, IL, USA).

A total of 92 patients had secondary CNS involvement at our institute between 2009 and 2019 according to the review of medical records. As the ASCT was only reimbursed for patients who were younger than 65 years old during that period, 70 patients were determined to be eligible for ASCT based on the reimbursement criteria. However, out of 70 patients who were potentially eligible for ASCT, 43 patients (43/70, 61.4%) underwent ASCT after they achieved complete or partial response to CNS-directed salvage chemotherapy and the remaining patients (n = 27) could not receive ASCT because they failed to respond to salvage therapy. Thus, we analyzed 43 patients who underwent ASCT after secondary CNS involvement. Their median age at initial diagnosis of DLBCL was 53 years (range: 23 – 66 years), and 23 patients were male (54%). The majority of patients had ECOG 0/1 (n = 38, 88%) at diagnosis, and initial stages were as follows: stage II (n = 13, 30%), III (n = 6, 14%), and IV (n = 24, 56%). However, extranodal involvement was found in most patients, and bone marrow was involved in 15 patients (35%) at diagnosis. The presence of at least one extranodal involvement was found in 43 patients (95%), and 19 patients (44%) had two or more extranodal involvements ( Figure 1 ). The involvement of a head area adjacent to the brain such as nasal cavity and paranasal sinuses was most common (n = 16) compared to other organs including liver (n = 7), gastrointestinal tract (n = 5), thorax (n = 5), soft tissue/muscle (n = 5), breast (n = 4), and kidney (n = 2). However, out of 43 patients who underwent ASCT, there was no case of DLBCL that initially had testicular involvement at diagnosis. All patients received R-CHOP as the primary treatment, and the response was as follows: CR (n = 36, 84%), PR (n = 5, 12%), and PD (n = 2, 5%). During R-CHOP chemotherapy, only eight patients received CNS prophylaxis according to physicians’ discretion in case of breast involvement or high tumor burden. Thus, six patients received IT MTX (intrathecal administration of 12 mg of MTX every three weeks) whereas two patients received high-dose MTX prophylaxis (MTX 3.5g/m² intravenous infusion every three weeks for two cycles after the completion of R-CHOP chemotherapy). However, CNS involvement occurred even in patients with stage II completely responding to R-CHOP (Patient No. 8, 25, 35, Figure 1 ). Secondary CNS involvement occurred as an isolated CNS relapse or progression in 27 patients (63%), whereas 16 patients (37%) showed concomitant systemic disease progression ( Figure 1 ). The diagnosis of secondary CNS involvement was based on various manifestations suspicious of CNS involvement, and most patients presented multiple neurologic symptoms ( Figure 2A ). Secondary CNS involvement occurred most commonly within 12 months after initial diagnosis of DLBCL ( Figure 2B ); thus, the median age at the time of secondary CNS involvement was 55.7 years (range: 23.4 – 66.8 years) as the median time to secondary CNS involvement was 10.2 months (95% CI: 5.8 – 14.7 months, Figure 2C ). At diagnosis of CNS involvement, 40% of patients had poor performance status (≥ ECOG 2), and the primarily involved sites of CNS were brain parenchyma (n = 20), leptomeninges (n = 11), both parenchyma and leptomeninges (n = 10), and spinal cord only (n =2). Patients received CNS-directed chemotherapy, mainly high-dose MTX-containing regimens as a salvage regimen. The salvage treatments that patients received after secondary CNS involvement were R-MVP (rituximab, methotrexate, vincristine, and procarbazine, n = 16), MVP (methotrexate, vincristine, and procarbazine, n = 15), ICED (ifosfamide, carboplatin, etoposide, and dexamethasone, n = 8), and other MTX-containing regimens ( Figure 2D ). The R-MPV treatment protocol consisted of rituximab (375mg/m² intravenous on day 1), MTX (3.5g/m² intravenous over 6 hours on day 2), leucovorin (15mg/m² intravenous four times per day, total 12 times during three days), procarbazine (100mg/m² orally administered for seven days at odd-numbered chemotherapy sessions), and vincristine (1.4mg/m² intravenous on day 2) for a total of five cycles at two-week intervals. Among patients who received high-dose MTX, 28 received the planned dosage (3.5 g/m²), while three received less than 3.5 g/m². The ICED treatment protocol consisted of intravenous ifosfamide 1670 mg/m² on days 1-3, carboplatin 5 x AUC on day 1, intravenous etoposide 100 mg/m² on days 1-3, and oral or intravenous dexamethasone 40mg on days 1–4 every three weeks, for four cycles.

The median number of infused CD34+ cells per body weight was 4.1 × 10⁶ cells per kg (range: 2.0 – 34.9 × 10⁶). Fifteen patients experienced grade III/IV febrile neutropenia including one patient requiring treatment in the intensive care unit. Sinus obstruction syndrome occurred in one patient, however, all those adverse events were manageable. Neutrophil engraftment (the first of three days on which the neutrophil count was maintained at 500 × 10⁶/L) and platelet engraftment (the first day free from platelet transfusion following at least seven days with a platelet count >20 × 10⁹/L) occurred around nine days after ASCT. At the median follow-up of 14.7 months (95% CI: 3.7 – 25.7) after ASCT, 17 patients did not relapse. However, the remaining 26 patients (60.5%) relapsed, and more than 60% of relapses were found (n = 16, 61.5%) within six months after ASCT. Thus, the median PFS after ASCT was 12.0 months (95% CI: 2.7 – 21.3 months, Figure 3A ). The patterns of relapse were isolated CNS, systemic, and combined relapses, and 20 patients died after relapse. Out of 26 relapsed patients, six patients were alive after subsequent salvage treatments ( Figure 3B ). Thus, the median OS after ASCT was 32.3 months (95% CI: 4.4 – 60.2, Figure 3C ). The number of survivors was significantly higher in patients receiving R-MVP after secondary CNS involvement ( Figure 3D ). Thus, the OS of patients receiving R-MVP or MVP was better than that of those receiving ICED after secondary CNS involvement, although the difference was not statistically significant ( Figure 3E ). The IT MTX was administered simultaneously to 15 patients including 11 patients with leptomeningeal involvement, however, the use of IT MTX for patients with leptomeningeal involvement did not have any impact on their survival outcome. Thus, the OS of patients who received IT MTX plus R-MVP or MVP was not significantly different from that of patients who received R-MVP or MVP alone (p = 0.255, data not shown). In case of isolated CNS relapse, R-MVP or MVP were mainly used as salvage treatments without whole-brain radiotherapy. Radiotherapy including whole-brain radiotherapy or gamma-knife was done in 13 patients, especially for concomitant systemic and CNS progression and patients who were symptomatic for their brain lesion.

The post-ASCT OS was compared by the time of onset of secondary CNS involvement after initial diagnosis. Patients with early secondary CNS involvement within 12 months showed worse OS than patients with secondary CNS involvement 12 months later, although the difference was not significant ( Figure 4A ). The comparison of survival outcomes according to the type of CNS involvement showed that the presence of leptomeningeal involvement had a negative effect on survival, although it was not statistically significant ( Figure 4B ). In addition, patients who experienced an isolated CNS relapse after ASCT showed significantly better OS than patients with systemic or combined relapse ( Figure 4C ). The onset of relapse after ASCT was also related to prognosis after ASCT, thus, post-ASCT relapse within six months after ASCT showed poor OS, and the majority of those patients with early relapse experienced systemic relapse ( Figures 1 , 4D ).