Central nervous system involvement at initial diagnosis of extranodal NK/T-cell lymphoma: a retrospective study of a consecutive 12-year case series

Patients with central nervous system (CNS) involvement at initial diagnosis of extranodal NK/T-cell lymphoma (ENKTL) are exceedingly rare, and the clinicopathologic features of CNS involvement have not been well characterized. In this study, we reviewed 662 patients with ENKTL from August 2008 to September 2019. Their clinical and pathological features, treatments, and survival outcomes were analyzed. The median follow-up time was 72 months. Nine of 662 (1.4%) patients were diagnosed with CNS involvement. Among them, the median age was 37 years, and seven patients were male. All patients had positive EBV-DNA, and three patients were asymptomatic at the time of diagnosis with CNS involvement. Common extranodal involved sites included bone, paranasal sinuses, breast, kidney, adrenal gland, and bone marrow. All patients were positive for cytoplasmic CD3ε, cytotoxic granule proteins, and EBER and negative for CD20. All patients received intrathecal chemotherapy and at least one cycle of systemic chemotherapy. Seven patients had died and two were still alive by the last follow-up. The median overall survival (OS) in patients with CNS involvement at initial diagnosis of ENKTL was 9 months, and the 1-year OS was 44.4%. Five patients achieved a complete response after asparaginase-based chemotherapy; two were still alive, one died of systemic progression, one died of ENKTL-associated hemophagocytic syndrome, and one died of treatment-related infections. In conclusion, CNS involvement at initial diagnosis of ENKTL is extremely rare with poor prognosis. There is no standard treatment, and asparaginase-based chemotherapy combined with intrathecal chemotherapy might yield good efficacy.


Introduction
Extranodal NK/T-cell lymphoma (ENKTL) is an aggressive non-Hodgkin lymphoma (NHL) associated with Epstein-Barr virus (EBV) [1]. The incidence of ENKTL is higher in Asia and South America than in Europe [2,3]. In China, ENKTL accounts for approximately 6-11% of all lymphomas and accounts for approximately 28-56% of mature NK/T-cell lymphomas [4]. Unlike other types of lymphomas, NK/T-cell lymphoma is almost entirely extranodal. Approximately 80% of ENKTL cases occur in the upper aerodigestive tract (nasal cavity, nasopharynx, paranasal sinuses, and palate), with the nasal cavity being the typical site of involvement, and approximately 20% of them occur in extranasal sites, including the skin, gastrointestinal tract, testis, and salivary glands [1,5]. Patients of ENKTL with central nervous system (CNS) involvement at diagnosis are exceedingly rare; therefore, the clinicopathologic features of CNS involvement in ENTKL have not been well characterized.
In terms of treatment for patients with ENKTL, historically, conventional anthracycline-based regimens, such as CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) and CHOP-like regimens, yielded poor outcomes, with a 5-year progression-free survival (PFS) rate of 41.6-47.6% and a 5-year overall survival (OS) rate of 54.5-60.6% [6]. Therefore, L-asparaginase-based regimens were proposed and achieved promising survival outcomes. Those regimens, including SMILE (dexamethasone, methotrexate, ifosfamide, L-asparaginase, and etoposide), P-Gemox (pegaspargase, gemcitabine, and oxaliplatin), and LVDP (L-asparaginase, etoposide, dexamethasone, and cisplatin), yielded objective response rates (ORRs) of 77-83% and a 5-year OS of 55-67% for patients with ENKTL [7][8][9]. However, for ENKTL patients with CNS involvement, due to the limitation of the blood-brain barrier (BBB), treatment methods varied and depended largely on each individual case. To find an optimal treatment method for future cases, additional case data for ENKTL patients with CNS involvement is needed.
To date, there have been few studies about CNS involvement in ENTKL, and the incidence of CNS involvement ranges from 0.5 to 6.6% [10][11][12]. However, most cases in their study are CNS relapse and progression, and there are few patients with CNS involvement at initial diagnosis. In addition, most clinical trials excluded those patients with CNS involvement, and the optimal therapy for them is still unknown. Therefore, to better understand the characteristics of patients with CNS involvement at initial diagnosis of ENKTL, we reviewed 662 cases with ENKTL in this study and explored the frequency, clinical and pathological features, treatments, and survival outcomes in those patients with CNS involvement at initial diagnosis of ENKTL.

Patients
We performed a retrospective study of patients with newly diagnosed ENKTL from August 2008 to September 2019 by reviewing electronic medical records at the West China Hospital of Sichuan University. Patients eligible for inclusion in this study needed to meet the following criteria: (1) pathologically and immunohistochemically confirmed diagnosis of ENKTL according to the 2008 WHO classification of lymphomas [13]; (2) absence of other malignancies; (3) no previous antitumor treatment; (4)

Diagnosis of CNS involvement
The CNS involvement of ENKTL patients was defined as lymphoma involving the brain parenchyma, spinal cord, eyes, cranial nerves, and leptomeninges, and all cases included in this study needed to meet at least one of the following criteria: (1) malignant cells found in the cerebrospinal fluid (CSF) by cytology and/or flow cytometry (FCM) analysis; (2) CNS involvement found by magnetic resonance imaging (MRI), computed tomography (CT), or positron emission tomography/computed tomography (PET/CT); or (3) brain biopsy histopathologically confirmed as ENKTL. At the time of initial diagnosis, CSF assessment, brain biopsy, and brain imaging examination were not performed in all patients, and, therefore, the real incidence of CNS involvement might be underestimated.

Patient characteristics, treatment, and efficacy evaluation
Clinical symptoms and laboratory results were collected at the time of diagnosis, such as age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS), B symptoms, complete blood count (CBC), lactate dehydrogenase (LDH), plasma EBV-DNA, and bone marrow involvement. Imaging examination, including CT of the head, neck, chest, abdomen, and pelvis, and PET/CT, was applied for staging and response assessment. PINK-E scores were recorded in this study. Treatment regimens were recorded in detail. The treatment efficacy was assessed according to the response criteria for malignant lymphoma, including complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD) [14]. OS was calculated from the time of initial diagnosis to death or the last follow-up.

Statistical analysis
Statistical analysis was performed by SPSS version 22.0 (SPSS Inc, Chicago, IL). All categorical variables were presented as frequencies and percentages. Survival results were analyzed by the Kaplan-Meier method, and survival curves were compared using the log-rank test. Statistical significance was considered if the two-sided p value was less than 0.05. The swimmer plot was constructed using R (version 4.2.1).
Among them, nine cases (1.4%) had CNS involvement at initial diagnosis of ENKTL (Tables 1 and 2). Of them, the median age was 37 years (range, 20-62 years), and only one patient was older than 60 years. Six patients were male and three were female. Six patients had B symptoms, and only one patient was positive for bone marrow involvement. All nine patients had positive EBV-DNA. Except for three patients who were asymptomatic at the time of diagnosis with CNS involvement, the other six patients presented with various symptoms, including cranial neuropathy, blurred vision, motor aphasia, headache, fatigue, and vomiting. Except for two patients who had primary CNS ENTKL ( Fig. 1), the other seven patients were systemic ENKTL involving CNS. Special extranodal involved sites in this study included bone (n = 4), paranasal sinuses (n = 3), breast (n = 1), kidney (n = 1), adrenal gland (n = 1), and bone marrow (n = 1). CNS involvement was diagnosed by positive histopathology and cytology in four patients and by positive FCM in three patients. Two patients were diagnosed by positive MRI.

Immunohistochemical phenotype
ENKTL was diagnosed by adequate immunohistochemistry that included CD20, cytoplasmic CD3ε, CD5, CD56, cytotoxic granule proteins (granzyme B, GrB; T-cell intracellular antigen 1, TIA-1), and EBV-encoded small RNAs (EBER) in situ hybridization. The immunohistochemical results of nine patients are listed in Table 3. All nine cases were positive for cytoplasmic CD3ε, positive for cytotoxic granule proteins, and negative for CD20. Seven of these nine patients (77.8%) were negative for CD5 and positive for CD56. Three cases were positive for CD30. The median proliferation index Ki-67 was 60% (range, 15-85%). All patients had positive EBER.
After the first-line chemotherapy, four cases received second-line or further treatment due to disease progression. Among them, case 1 received L-GDP (L-asparaginase, gemcitabine, cisplatin, and dexamethasone) as the second-line chemotherapy, case 2 received SMILE as the second-line chemotherapy, case 3 received LVDP as the second-line chemotherapy, and case 7 received HLH-94 protocol as the second-line treatment. After the second-line treatment, only case 1 achieved remission, but this case did not undergo HSCT due to financial limitations. After 14 months, case 1 relapsed again and received further L-GDP treatment. However, after three cycles of treatment, the patient's condition progressed further, and L-ICE (L-asparaginase, ifosfamide, carboplatin, and etoposide) was administered. After two cycles of treatment, this patient was diagnosed with hemophagocytic syndrome, and the family made the decision to abandon further treatment. The specific treatment and remission profiles of those nine patients are shown in the swimmer plot (Fig. 2).

Survival outcomes of patients with CNS involvement at initial diagnosis of ENKTL
The median follow-up for the entire cohort was 72 months (95% CI, 67.8-76.2 months). Among the 662 patients, 268 died. For nine patients with CNS involvement at initial diagnosis of ENKTL, the survival outcome was bad. Among them, seven patients died. Three of them died of systemic progression, two died of ENKTL-associated hemophagocytic syndrome, one died of treatment-related infections, and one died of acute kidney failure. Five patients achieved CR after first-line treatment. Of them, two patients had disease relapses after 12 months, and two sustained CR by the last follow-up. Three patients  had disease progression during the initial treatment. One patient with kidney involvement achieved PR after the initial treatment but died of acute kidney failure. The median OS of patients with CNS involvement at initial diagnosis of ENKTL was 9 (95% CI, 0-20.7) months, and the 1-year OS was 44.4% (Fig. 3a). Notably, the median OS of patients who achieved CR after first-line chemotherapy was 56 (95% CI, 13.7-98.3) months, while the median OS of patients who did not achieve CR was 3 (95% CI, 0-20.7) months (Fig. 3b). In addition, the median OS of seven patients with secondary CNS involvement was longer than that of two patients with primary CNS involvement (25 months vs. 3 months), but there was no significant difference between the two groups (p = 0.074) (Fig. 3c).
In the 653 patients without CNS involvement at initial diagnosis of ENKTL, the median OS was longer than those with CNS involvement at initial diagnosis of ENKTL (not reached vs. 9 months, p < 0.001) (Fig. 3d). Moreover, for the 653 patients without CNS involvement at initial diagnosis of ENKTL, there was a significant difference in the median OS between early-stage and advanced-stage patients (not reached vs. 24 months, p < 0.001). This significant difference was also found in patients with low-risk PINK-E versus patients with high-intermediate-risk PINK-E (not reached vs. 31 months, p < 0.001), and patients with upper aerodigestive tract involvement versus patients with non-upper aerodigestive tract involvement (not reached vs. 40 months, p < 0.001).

Review of reported cases with CNS involvement at initial diagnosis of ENKTL
We identified 18 patients with CNS involvement at initial diagnosis of ENKTL from literature reviews [10][11][12]15]. A total of 27 cases were collected, and the baseline features of those cases are listed in Table 5. For all patients, the median age was 48 (range, 13-83) years, and most patients were male. In most cases (18,66.7%), tumor lesions were located in the non-upper aerodigestive tract and initially presented with an advanced stage. Of the 22 patients with survival data, 63.6% died of progressive disease, and the median survival time was 7.5 (range, 0.5-139) months.

Discussion
ENKTL is an uncommon subtype of NHL, with a survival rate of 5 years ranging from 45.3 to 68.9% [6,16,17]. The frequency of NHL involving the CNS was 4% among all CNS malignant diseases, and the most common subtype of NHL involving the CNS was diffuse large B-cell lymphoma (DLBCL) [18,19]. ENKTL involving the CNS is extremely rare and has been reported in only a few studies. Therefore, the clinical features, treatment, and survival outcome of this disease are unknown. To the best of our knowledge, this is the largest cohort of CNS involvement at initial diagnosis of ENKTL to date.
In our report, we reviewed 662 patients with ENKTL and found that their the 5-year OS was 58.9%. The longterm survival outcome was similar to that in previous studies. Kim et al. reported that a non-anthracyclinebased treatment yielded a 3-year OS of 59% in newly diagnosed ENKTL patients [20]. In another study, Yamaguch et al. showed the survival data of 358 ENKTL patients receiving non-anthracycline chemotherapy, and the 5-year OS was 56% [21]. In addition, the median OS of early-stage ENKTL patients, PINK-E low-risk patients, and upper aerodigestive tract involved patients was much longer than that of the comparison group. This result was consistent with prior reports [3,20].
Moreover, we found nine cases (9/662, 1.4%) with CNS involvement at the initial diagnosis of ENKTL, and its incidence was actually higher than that in previous studies. In a retrospective study, 414 patients with ENKTL were analyzed, and the incidence of CNS involvement at initial diagnosis of ENKTL was 1.2% (5/414) [12]. In a recent report, 1019 patients with ENKTL were included, and only 5 patients (0.5%) had CNS involvement at the initial diagnosis of ENKTL [10]. The reason for the difference in the incidence of CNS involvement may be because we included patients with ocular involvement as part of CNS involvement. Our definition of CNS involvement was consistent with the Eighth International Conference on Malignant Lymphoma [22]. In our series, 77.8% (7/9) of patients were diagnosed by positive histopathology, CSF cytology, or CSF flow cytometry. Clinically, ENKTL with CNS involvement predominantly occurs in males, and it can occur in all ages (range, 13-83 years) with a median age of 48 years, which is similar to the entire cohort of ENKTL (median age of 45 years).
In addition, we found that although three advanced-stage patients were asymptomatic in our series, they were diagnosed with CNS involvement by positive imaging or CSF analysis. Patients who had extracranial site involvement, such as involvement of paranasal sinuses, breast, kidney, adrenal gland, bone, and bone marrow, often had CNS involvement. This finding was consistent with previous reports. In 2016, Schmitz et al. suggested that patients with ≥ 2 extranodal lymph node involvement, advanced stage, and kidney and adrenal gland involvement had a higher risk for CNS involvement in DLBCL [23]. Another study from Korea for ENKTL demonstrated that the primary site of involvement was the non-upper aerodigestive tract, Ann Arbor stage III/IV, and advanced NK/T-cell lymphoma prognostic index (NKPI) increased the risk of CNS involvement [24]. In a recent study, Kim et al. reviewed 652 patients with ENKTL and found that 45 patients with CNS relapse. They revealed that extranodal  [25]. Combined with our series, for patients with advanced-stage disease or involvement of the paranasal sinuses, breast, kidney, adrenal gland, bone, and bone marrow, we suggest that CNS involvement should be excluded by imaging and/or CSF examination and that CNS prophylaxis should be considered. Currently, the standard therapy has not yet been established for ENKTL patients with CNS involvement due to the rarity of cases. Systemic chemotherapy plays a critical role in the management of malignant lymphoma with CNS involvement. However, its efficacy is limited by the blood-brain barrier (BBB). MTX was found to cross the BBB at high doses, and MTX doses over 3 g/m 2 yielded antitumor levels in the CSF. This result was proven in real-world studies. Cai reported that 11 out of 14 patients with ENKTL involving the CNS achieved CR. Of them, two patients received HD-MTXcontaining regimens [10]. Nevel et al. demonstrated that methotrexate-based regimens were effective for patients with CNS ENKTL, and eight out of ten (80%) patients achieved ORR (CR rate, 50%) after treatment [11]. In addition, because asparaginase could induce the specific apoptosis of NK lymphoma cells, asparaginase has been the backbone of chemotherapy regimens for ENKTL over the past decade [26]. In our series, eight patients received asparaginase-based chemotherapy; of them, six patients achieved ORR (CR rate, 62.5%). This result was consistent with Cai's report. In their study, four out of eleven patients (33.6%) had CR after asparaginase-based chemotherapy [10]. Since PD-ligand 1 (PD-L1) expression is commonly observed in NK/T lymphoma cells, PD-1 inhibitors have shown great efficacy in the treatment of newly diagnosed advanced, relapsed, or refractory ENKTL, and are also expected to be effective in CNS ENKTL [27,28]. In Cai's abstract, two patients who achieved CR received PD-1 inhibitors as maintenance treatment with the durations of response of 26 and 72 months, respectively [10]. In the current study, two patients received PD-1 inhibitors, and one patient had CR, but another patient had disease progression. Moreover, since CSF is a reservoir for malignant lymphoma cells, all patients in our study received intrathecal chemotherapy. Li et al. reported that ENKTL patients who received IT had better survival outcomes [12]. As an exclusive treatment, radiotherapy plays an important role in early-stage ENKTL, but its role in CNS ENKTL is unclear. In terms of consolidative therapy, it has been reported that long-term survival has been improved by hematopoietic stem cell transplantation (HSCT) in patients with DLBCL involving the CNS [29]. A retrospective study of 47 patients with ENKTL that evaluated the survival outcomes of HSCT showed that among patients with CR before transplant, the 5-year PFS was significantly higher in the HSCT group than in the non-HSCT group (87% vs. 68%, p = 0.027) [30]. Nevel et al. reported that one patient with CNS ENKTL achieved long-term survival after HSCT, and HSCT is also expected to be effective in CNS ENKTL [11]. Therefore, except for methotrexate-based chemotherapy, asparaginase-based chemotherapy combined with intrathecal chemotherapy could be considered a candidate treatment option for patients with CNS involvement at initial diagnosis of ENKTL. In addition, immunotherapy is needed for further study of CNS ENKTL.
We found that patients with CNS involvement at initial diagnosis of ENKTL had a poor prognosis. In the current study, the median OS of those patients was 9 months, and the 1-year OS was 44.4%. The long-term survival outcomes were comparable to those of previous studies. In a retrospective study of four patients with CNS involvement at initial diagnosis of ENKTL, Nevel et al. reported that the median OS was 4.3 months [11]. A similar result was found in another study with a median OS of 6 months [12]. In addition, compared to CNS relapse, relatively longer OS in patients with CNS involvement at initial diagnosis of ENKTL was observed. We reviewed 20 cases with CNS relapse of ENKTL from literature reviews and found that their median OS was 4.3 (range, 1-20) months. However, the 22 patients with CNS involvement at initial diagnosis of ENKTL had a median survival of 7.5 (range, 0.5-139) months [10][11][12].

Conclusion
To conclude, patients with CNS involvement at initial diagnosis of ENKTL are extremely rare, and patients with ENKTL advanced-stage disease or involvement of the paranasal sinuses, breast, kidney, adrenal gland, bone, and bone marrow may have CNS involvement. Currently, the survival outcomes of ENKTL patients with CNS involvement are poor, but asparaginase-based chemotherapy may yield potential efficacy in some cases. However, this study was a single-center retrospective study, which may lead to selection bias. Therefore, further multicenter studies or the accumulation of more cases are required to validate our results.