We reviewed the electronic medical records of patients who were diagnosed with IVLBCL at Samsung Medical Center between 2001 and 2018. The diagnosis was based on histopathological criteria including the presence of CD20-positive neoplastic B cells with large cell morphology and intravascular or sinusoidal proliferation of lymphoma cells. All patients were evaluated for staging work-ups in clinical practice including physical examinations, complete blood cell counts, serum biochemistry, computed tomography scans of chest and abdominal pelvis, bone marrow aspiration and biopsy, and ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography scans. Additional work-up procedures, such as brain or spine magnetic resonance imaging (MRI), cerebrospinal fluid (CSF) analysis, and cytological examination of pleural fluid, were performed if indicated clinically. Stage was determined using the Ann Arbor staging system, and the risk of prognosis was defined by the International Prognostic Index (IPI) consisting of age, Eastern Cooperative Oncology Group performance status, Ann Arbor stage, numbers of extranodal involvements, and serum lactate dehydrogenase (LDH) levels at diagnosis. This study was approved by the Institutional Review Board of Samsung Medical Center, Seoul, Korea, and the requirement for informed consent was waived because of the retrospective nature of the study (IRB File No: 2018-10-065).

After reviewing the medical records, clinical features at the time of diagnosis were collected. CNS involvement was defined as the presence of a lymphoma in the brain parenchyma and/or spinal cord as demonstrated by MRI, or leptomeningeal involvement diagnosed by CSF analysis at the time of diagnosis or during or after treatment. As a predictive model for prognosis as well as CNS involvement, we gathered the data required for CNS-IPI and National Comprehensive Cancer Network (NCCN)-IPI because they showed better prognostic power compared with existing standard IPI [12,13]. As most patients were treated with R-CHOP chemotherapy regimens (rituximab 375 mg/m² on day 1, cyclophosphamide 750 mg/m², doxorubicin 50 mg/m², vincristine 1.4 mg/m² [maximum 2 mg] on day 1, and prednisone 100 mg on day 1 to 5), data including the number of cycles and dose intensity (cumulative dose in mg) / (treatment duration in weeks) of R-CHOP were gathered. For patients with CNS involvement, CNS-directed therapy such as intrathecal (IT) or intravenous (IV) methotrexate (MTX) was added to R-CHOP in clinical practice. The IT chemotherapy regimens were MTX-based combination regimens (MTX 15 mg, Ara-C 30 mg/m², hydrocortisone 15 mg/m²) or MTX single therapy (MTX 15 mg IT), whereas the IV MTX regimen used 1 to 1.5 g/m² over 6 hours. These data about CNS-directed therapy were also collected. The treatment response to R-CHOP was re-determined by the Revised Response Criteria for Malignant Lymphoma [14]. High-dose chemotherapy followed by autologous stem cell transplantation (ASCT) was done in selected patients after the completion of R-CHOP chemotherapy (upfront ASCT) or salvage chemotherapy (salvage ASCT) according to the physicians’ discretions. As this was a retrospective study, adverse events and toxicities were based on a review of medical records. According to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) version 4.0, grade 3 or 4 adverse events were recorded.

Descriptive statistics are reported as proportions and medians. Intergroup comparisons were performed using Fisher’s exact test for categorical variables and the Mann-Whitney nonparametric U test for age. Progression-free survival (PFS) was calculated from the date of diagnosis to the first day of disease progression, relapse, death as a result of any cause, or last date of follow-up. Overall survival (OS) was calculated from the date of diagnosis to death or to the last date of follow-up. Survival curves were generated using the Kaplan-Meier method, and survival rates were compared using the log-rank test. Univariate and multivariate analysis was performed using Cox regression analysis. Two-sided p values < 0.05 were considered significant, and statistical analyses were performed using the statistical software package IBM PASW version 18.0 (SPSS Inc., Chicago, IL, USA).

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Forty-six patients were diagnosed with IVLBCL, and their median age at diagnosis was 62 years (range, 34 to 82). Male patients predominated (n = 26, 57%), and B symptoms were present in 31 patients (78%) (Table 1). Hepatomegaly and/or splenomegaly were observed in 26 patients (57%), whereas lymphadenopathy was less common (n = 22, 48%). Cytopenia affecting the hemoglobin level (< 11 g/dL) and/or platelet count (< 100 × 10⁹/L) were also found frequently (Table 1). The clinically presenting symptoms were diverse; however, fever was the most common symptom (27/46, 59%) (Fig. 1A). Neurological symptoms, such as diplopia and facial palsy, and altered mental functions were the second most common symptoms (15/46, 33%). Bone marrow and liver were the most commonly involved extranodal organs (54% and 48%, respectively) and were the most common sites of biopsy for diagnosis in this study (Fig. 1B). Accordingly, most patients were at stage IV disease (n = 42, 91%) (Table 1). All patients had increased levels of serum LDH, and 30 patients had extremely high levels (more than three times the normal range) (Table 1 and Fig. 1C). Sixteen patients (35%) had HLH, and four patients showed severe hypercalcemia (up to 1.35 mmol/L). Based on the IPI score, 44 patients (96%) were classed as high-to-intermediate risk or high risk, whereas 44 patients (96%) were classed as having a high-to-intermediate risk or a high risk by NCCN-IPI (Table 1). Classification according to the CNS-IPI score showed that 38 patients belonged to the high risk group, while only eight patients were in the low or intermediate risk categories (Table 1).

All patients received R-CHOP as a first-line treatment after diagnosis with a median number of six cycles (range one to eight). At the end of treatment, 31 patients achieved a complete response (CR), whereas eight patients showed disease progression (Fig. 2). Six patients died after the first or second cycle of R-CHOP, and the causes of death were treatment-related adverse events including cytopenia, infectious complications, and pulmonary hemorrhage. One patient was lost to follow-up after the completion of the first cycle. Therefore, the CR rate based on intent to treat was 67% (31/46). According to physicians’ decisions, the dosage of R-CHOP was reduced in 13 patients for elderly or frail patients. The administration of rituximab in the first cycle was delayed in 21 patients when considering the risk of infusion-related reaction and rituximab-related severe adverse events. As a result, the median dose-intensity of each drug was as follows: rituximab 200 mg/wk (range, 163 to 240), cyclophosphamide 393 mg/wk (range, 299 to 477), doxorubicin 26 mg/wk (range, 18 to 32), and vincristine 1 mg/wk. The most frequently attenuated drugs were doxorubicin and vincristine, which were reduced by approximately 20% to 25%. However, the overall response rate was not significantly different, regardless of the dose reduction or intensity (data not shown). The most common grade 3 or 4 toxicity seen was neutropenia or febrile neutropenia (n = 26, 57%). Three patients developed pulmonary hemorrhage during the first cycle of R-CHOP, especially immediately after the infusion of rituximab. Two of them recovered; however, one patient died from massive pulmonary hemorrhage and subsequent infectious complications. Infusion-related reactions associated with rituximab such as fever, chills, and dyspnea were also observed in seven patients. Upfront ASCT was done for two patients including one patient with CNS involvement at diagnosis, and these patients were still alive at the time of analysis without evidence of relapse (Fig. 3). On the other hand, the outcome of six patients undergoing salvage ASCT after relapse was poor; thus, only one patient was alive. Likewise, patients with disease progression at the end of treatment with R-CHOP showed dismal prognoses even after salvage chemotherapy except for one (Fig. 2).

At diagnosis, 11 patients had CNS involvement (24%) (Fig. 1B), and neurologic signs and symptoms were present in eight of them (8/11; 73%). Neurologic symptoms were heterogeneous, including diplopia, headache, visual defect, hearing difficulty, facial palsy, and mental changes. The diagnosis of CNS involvement was done by CSF analysis via lumbar puncture (n = 6), while the remaining five patients were diagnosed by radiologic imaging such as MRI. Four patients developed CNS involvement during or after treatment with R-CHOP. As a result, 15 of the 46 patients had CNS involvement (33%). A comparison of characteristics at diagnosis between patients with and without CNS involvement did not show significant differences (Table 2). However, all patients with CNS involvement belonged to the high-risk group according to CNS-IPI, whereas 74% of patients without CNS involvement were grouped as high risk by CNS-IPI (p = 0.04) (Table 2). CNS-directed therapy was combined with R-CHOP in 11 patients who had CNS involvement at diagnosis. Among these patients, five patients were alive without evidence of relapse. However, six patients died because of disease relapse or early mortality caused by infectious complications. In particular, one patient relapsed after CR, and all patients died from disease progression even after ASCT (Fig. 3). Likewise, the outcomes of four patients who had CNS relapse after R-CHOP chemotherapy were poor. Three of these patients died from disease progression, and only one patient was alive after being salvaged successfully by CNS-directed chemotherapy.

At a median follow-up of 47.0 months (95% confidence interval [CI], 25.0 to 69.0), the median OS was 45.0 months (95% CI, 25.8 to 64.2) (Fig. 4A). The median PFS was 23.0 months (95% CI, 4.5 to 41.5), and the OS was significantly worse in patients with relapse or disease progression (Fig. 4B). The presence of thrombocytopenia at diagnosis had a significant association with OS (p = 0.010) (Fig. 4C). As most patients (93%) had stage IV disease and belonged to the high or high-to-intermediate risk group according to the IPI and NCCN-IPI grades, these parameters could not discriminate patients with better and worse survival outcomes (data not shown). The CNS-IPI grading also failed to show a significant association with survival because 38 patients (83%) were grouped as having high risk CNS-IPI (Table 1). Thus, the univariate analysis for OS using each parameter showed that only the presence of thrombocytopenia or anemia was significantly associated with a poor OS (Table 3). The presence of HLH and CNS involvement at diagnosis were not significantly associated with poor survival outcomes (Table 3 and Fig. 4D). However, none of these parameters showed independent associations with overall survival in the multivariate analysis (Table 3).