Clinical characteristics, treatment, and survival of 30 patients with gastrointestinal natural killer/T‐cell lymphoma

Abstract Background The gastrointestinal (GI) tract is the second most frequent extranasal involvement site for ENKTL. This study aimed to explore the clinicopathological features, treatment models, survival outcomes, and prognosis of gastrointestinal ENKTL (GI‐ENKTL). Methods The clinical data of GI‐ENKTL patients were extracted from the China Lymphoma Collaborative Group (CLCG) database and were analyzed retrospectively. Results A total of 30 patients were enrolled, with a male/female ratio of 4:1 and a median age of 42 years. Twenty‐nine patients received chemotherapy, of whom 15 patients received asparaginase‐based (ASP‐based) regimens. Moreover, seven received surgery and three received radiotherapy. The overall response an d complete remission rates were 50.0% and 30.0% for the whole cohort, 50.0% and 37.5% for patients treated with ASP‐based regimens, and 50.0% and 25.0% for those treated with non‐ASP‐based regimens, respectively. The median follow‐up was 12.9 months and the 1‐year overall survival rate was 40.0% for the whole cohort. For those patients in an early stage, ASP‐based regimens resulted in a superior 1‐year progression‐free survival rate compared to non‐ASP‐based regimens (100.0% vs. 36.0%, p = .07). However, ASP‐based regimens did not improve survival in patients at an advanced stage. Conclusion GI‐ENKTL still has a poor prognosis, even in the era of modern asparaginase‐based treatment strategies.

received surgery and three received radiotherapy. The overall response an d complete remission rates were 50.0% and 30.0% for the whole cohort, 50.0% and 37.5% for patients treated with ASP-based regimens, and 50.0% and 25.0% for those treated with non-ASP-based regimens, respectively. The median follow-up was 12.9 months and the 1-year overall survival rate was 40.0% for the whole cohort. For those patients in an early stage, ASP-based regimens resulted in a superior 1-year progression-free survival rate compared to non-ASP-based regimens (100.0% vs. 36.0%, p = .07). However, ASP-based regimens did not improve survival in patients at an advanced stage.
Conclusion: GI-ENKTL still has a poor prognosis, even in the era of modern asparaginase-based treatment strategies.

K E Y W O R D S
asparaginase/therapeutic use, efficacy, physiopathology, retrospective analysis, T cell lymphoma

| INTRODUCTION
Extranodal natural killer (NK)/T-cell lymphoma (ENKTL) comprises a subgroup of cytotoxic T or NK cells that originate outside the lymph nodes and is associated with Epstein-Barr virus. 1 ENKTL has a strong geographic presence in Latin American and East Asian populations (accounting for 5.0%-15.0% of all lymphomas) and is rare in European countries and America (less than 1% of all lymphoma). [2][3][4][5][6][7][8] ENKTL is almost exclusively extranodal, with 80.0%-90.0% of cases involving the upper aerodigestive tract (UADT), including the nasal cavity, paranasal sinuses, nasopharynx, Waldeyer's ring, and palate. Cases of extranasal areas commonly occur in the skin/soft tissues and the gastrointestinal (GI) tract, followed by the testis, lungs, adrenal glands, and the central nervous system. 4 The outcome of ENKTL patients is affected by multiple factors and tumor location is a determining factor in tumor malignancy.
Although the GI tract is the most common anatomic site of extranodal non-Hodgkin's lymphoma that accounts for 6.0%-23.0% of all NHLs and 0.9%-6.5% of all GI cancers, 7 information on ENKTL with GI tract (GI-ENKTL) involvement is relatively scarce. According to a previous study based on 81 patients from the Asia Lymphoma Study Group in 2013, GI-ENKTL presented with more severe clinical features and had an extremely poor prognosis. 8 However, the study did not evaluate the widely used ASP-based regimens that have been developed in recent years. In addition, previous studies had small sample sizes, were from a single institution, and had incomplete clinic data. [9][10][11][12] We, therefore, retrospectively analyzed the clinical features, treatment strategies, and outcomes of patients with GI-ENKTL using the database from the China Lymphoma Collaborative Group (CLCG).
The aim was to evaluate the impact of different treatment modalities on outcomes in GI-ENKTL.

| Patients
A total of 2640 patients with newly diagnosed ENKTL treated at 16 Chinese institutions between 2000 and 2016 in the CLCG database were retrospectively reviewed. The eligibility criteria included patients whose examination revealed GI tract involvement. Patients who were lost to follow up or had incomplete data were excluded.
Lymphomatoid gastropathy (LyGa) pathologically confirmed has been excluded by hematopathologists. Patients of all ages and both sexes were considered for inclusion. Eventually, 30 patients formed the study population. Follow-ups were conducted via an outpatient clinic and over telephone. Routine follow-up plan includes physical examinations and imaging every 3 months during the first 3 years, twice a year during the fourth and fifth year, and annually after 5 years. No patients were lost to follow-up.
The staging work-up of ENKTL included a routine physical examination, chest/abdominal/pelvic computed tomography (CT) or positron emission tomography (PET)-CT scan, nasopharyngeal biopsies, and bone marrow aspirate analyses. For cases undergoing surgery, the removed specimen tissues were used for staging. Patients who filled the following criteria were included: (1) patients who presented initially with GI symptoms with ruled out aggression in other distant parts; and (2) patients with UADT lesions who were assayed for GI tract problems at the staging work-up.

| Evaluation and definition
Clinical data, including demographics, stage, tumor location, serum lactate dehydrogenase (LDH), B symptoms, and Eastern Cooperative Oncology Group performance status (ECOG PS) were analyzed. With end-of-treatment evaluation after first-line therapy including endoscopy with visual inspection, imaging analysis (CT or PET-CT) according to the type of examination at the initial work-up, and treatment modalities, including surgery, chemotherapy regimens, and radiotherapy, were also analyzed. The Ann Arbor staging and treatment evaluation were performed by local investigators following institutional imaging protocols in accordance with local standard practice.
Baseline disease characteristics, treatment strategies and patient outcomes for patients in the entire cohort were obtained from medical records. Responses were assessed using Lugano or Cheson criteria by PET or CT scans, clinician assessment, or both. 13 Progression-free survival (PFS) was defined as the time from first treatment to disease progression, relapse, or death. The OS was measured from the date of first treatment to the date of last follow-up or death.
Prognostic stratification models were analyzed including the International Prognostic Index (IPI; including LDH level, stage, B symptoms, and regional lymph node involvement) as well as the nomogram-revised risk index (NRI), a model derived from the international multicenter analysis of data from 1582 patients receiving asparaginase treatment. One point was assigned to each age >60 year, LDH, stage II disease, ECOG score ≥2, or primary tumor invasion. Two points were assigned to stage III/IV. Patients were then stratified into four risk groups (low, 0; intermediate low, 1; intermediate high, 2; and high, ≥3). 14

| Follow-up and statistical analysis
Follow-ups were conducted via an outpatient clinic and over telephone. Routine follow-up plan includes physical examinations and imaging every 3 months during the first 3 years, twice a year during the fourth and fifth year, and annually after 5 years. No patients were lost to follow-up Over survival (OS) and PFS were estimated using the Kaplan-Meier curves and log-rank test was to assess the long-term prognosis stratified by chemotherapy regimens, prognostic factors and prediction models. Variables in the univariate analysis and other factors that had potential clinical significance were further entered into the Cox regression analysis to analyze independent risk factors. To compare the characteristics of patients, Fisher's exact test (or χ 2 test) was used.

| Treatment models
Radiotherapy could be radical treatment for limited-stage ENKTL originating from UADT. However in this GI-ENKTL cohort, radiotherapy In the cohort, high-dose therapy and autologous stem cell transplantation (HDT-ASCT) was given to one patient at stage I as consolidation therapy after CR in first-line treatment.
However, the OS was not significantly different between the two groups, with 1-year OS rate of 100.0% for the ASP regimens and 87.5% for the non-ASP-based regimens ( p = .390).

| Prognostic factors and prediction models
Univariate analysis revealed that an elevated LDH ( p = .037), stage III-IV (p = .003), and ≥2 extranodal involvement sites ( p = .013) were F I G U R E 1 Distribution of the anatomic involvement of the digestive tracts. Proportion of each part in 30 patients with gastrointestinal (GI) involvement statistically associated with poor OS, which was not associated with age >60 years ( p = .713), and B symptoms ( p = .683, Table 2). The COX analysis indicated that stage III/IV was an independent factor associated with poor survival (p = .009, HR = 2.004, 95% CI = 1.633-33.695, Table 2). The prognostic models, IPI score, and NRI score ≥2 showed a significant association with poor OS ( Figure 4).

| Comparison of GI-ENKTLs and SEER database
A total of 35 GI-ENKTL patients were identified in the SEER database. which is consistent with other studies. 12 IPI and NRI prediction models were also applicable to GI-NKTL patients.
Because it is a rare disease, there are few studies discussing the clinical features and survival outcomes according to chemotherapy regimens of GI-ENKTL. 8,9,11,12 Chemotherapy is the most basic treatment modality. Accordingly, anthracycline-based regimens were conventionally used in the past, but the efficacy was unsatisfactory, probably due to multidrug resistance caused by the expression of Pglycoprotein. 16   Our study has limitations. Given the rarity of the disease and GI involvement being less common, this study was a retrospective study.
In addition, the heterogeneity of individual characteristics and heterogeneous treatment modalities influenced the results, which highlights the need for further studies on this disease. Third, due to the small number of cases the multivariate survival analysis using COX regression model should be interpreted with caution.

| CONCLUSIONS
Even through comprehensive treatment strategies, the prognosis of the disease remained poor, especially for patients of advanced stage.
Asparaginase-based chemotherapy may be beneficial to early-stage patients with GI-ENKTL. Of note, surgical intervention was also a fea-