A prognostic index for nasal‐type early‐stage extranodal natural killer/T‐cell lymphoma: A multicenter study

Extranodal natural killer/T-cell lymphoma (ENKTL) is a rare, distinctive clinicopathologic disease entity with aggressive clinical features and a strong association with the Epstein–Barr virus (EBV) infection.1, 2 Conventional prognostic indices, including the International Prognostic Index (IPI)3 and the Korean Prognostic Index (KPI),4 were devised based on the use of anthracycline-containing regimens. Thus, these indices are not relevant in the contemporary era of non-anthracycline-based chemotherapy. Recently, the prognostic index of natural killer cell lymphoma (PINK) and PINK-EBV DNA (PINK-E) was developed to guide risk-adapted treatment approaches in the modern ENKTL treatment era.5 The PINK index includes four independent risk factors: age older than 60 years, stage III/IV disease, distant lymph node involvement, and non-nasal-type disease. However, more than two-thirds of NK/T-cell lymphoma patients have stage I/II disease, more than 80% of patients have nasal-type disease, and less than 15% of patients exhibit distant lymph node involvement. Thus, the majority of ENKTL patients have nasal-type stage I/II disease without distant lymph node involvement, and these patients are in the low-risk group when stratified by the PINK index. However, treatment outcomes are unsatisfactory in a substantial proportion of patients. Thus, a new prognostic model based on non-anthracycline treatment specific for nasal-type stage I/II ENKTL might help to identify patients who will not respond well to treatment and further improve outcomes. We conducted a retrospective, multicenter study to establish a prognostic index specific for nasal-type stage I/II ENKTL, and validated our prognostic index in an independent cohort. In the training cohort, 223 patients were recruited from six hospitals between 2013 and 2017. All of the patients were positive for EBV with in situ hybridization and treated with non-anthracycline-containing regimens followed by radiotherapy. All patients were staged according to the Ann Arbor staging system and were stratified according to the IPI, KPI and PINK-E. Local invasiveness was defined in accordance with 2002 TNM classification of the American Joint Committee on Cancer.4 The validation cohort was retrospectively recruited from three hospitals in China. This study was approved by the Institutional Review Board at the Sun Yat-sen University Cancer Center. All patients provided informed consent in compliance with institutional guidelines. Patient characteristics in the training cohort vs the validation cohort are summarized in Supporting Information Table S1. In the training cohort, the majority of patients were scored as having low-risk disease according to the IPI (0-1, 96.0%), low-risk to low intermediate-risk disease using the KPI (groups 1-2, 75.8%) or low-risk disease according to the PINK-E (0-1, 90.6%). All of the patients in the training cohort received l-asparaginase-containing regimens followed by involved-field radiotherapy, 195 patients (87.4%) achieved complete remission (CR), and 22 (11.3%) patients relapsed during follow-up. After a median follow-up of 47 months, the estimated 5-year overall survival (OS) rate in the training cohort was 77.4%, and the 5-year progression-free survival (PFS) rate was 68.3%. In the univariate analysis, the prognostic factors that predicted poor OS were as follows: poor performance status (Eastern Cooperative Oncology Group [ECOG] performance status 2 to 4, P = 0.032), Ann Arbor stage II disease (P = 0.002), hemoglobin concentration less than 100 g/L (P = 0.003), serum albumin concentration less than 35 g/L (P = 0.034), local invasiveness (P = 0.013) and positive pretreatment plasma EBV-DNA (P = 0.034). In the multivariate analysis, Ann Arbor stage II disease (P = 0.017), a hemoglobin concentration less than 100 g/L (P = 0.005) and local invasiveness (P = 0.021) remained significantly associated with a poor OS. The clinical factors associated with a reduced PFS in the univariate analysis were poor performance status (ECOG performance status 2 to 4, P = 0.019), Ann Arbor stage II disease (P < 0.001), a hemoglobin concentration less than 100 g/L (P = 0.002), serum albumin concentrations less than 35 g/L (P = 0.007), local invasiveness (P = 0.019), and positive pretreatment plasma EBV-DNA (P = 0.032). In the multivariate analysis, Ann Arbor stage II disease (P = 0.001), a hemoglobin concentration less than 100 g/L (P = 0.004), and local invasiveness (P = 0.035) remained significantly associated with a poor PFS. Based on the number of risk factors (stage, hemoglobin concentration and local invasiveness), stratification of both OS and PFS was significant (Supporting Information Figure S1). We developed a new prognostic model, named the Chinese Prognostic Index for Natural Killer Cell Lymphoma (C-PINK), consisting of three risk factors (Ann Arbor stage II, hemoglobin concentration less than 100 g/L, and local invasiveness) that were significantly associated with OS and PFS. Patients were stratified according to the number of risk factors present into the three groups: group 1 (0), group 2,1 and group 3 (≥2), with 5-year OS rates of 91.9% (95% confidence interval [CI] 85.6-98.2), 71.0% (60.2-81.8), and 58.2% (41.1-75.3), respectively (P < 0.001; Figure 1A), and 5-year PFS rates of 90.6% (95% CI 82.8-98.4), 60.7% (42.1-79.3), and 42.0% (18.7-65.3), respectively (P < 0.001; Figure 1B). We applied C-PINK to an independent validation cohort, which included 109 patients. Of the 109 patients, 83 patients received asparaginase-based chemotherapy followed by radiotherapy, and 26 patients received concurrent chemoradiotherapy (CCRT) followed by non-anthracycline chemotherapy. C-PINK was also significantly associated with OS and PFS in the validation cohort (Supporting Information Figure S2A,B). We stratified our training cohort patients according to the IPI, KPI and PINK-E. The IPI could separate the patients into two risk groups with different survival outcomes (IPI, 0-1 vs ≥2, P < 0.001), although 214 (96.0%) patients were in the low-risk IPI group. The KPI was unsatisfactory for stratifying between groups 1 and 2 (P = 0.761) or between groups 3 and 4 (P = 0.837). According to the PINK-E, 202 (90.6%) patients were in the low-risk group; the PINK-E failed to stratify the survival difference between the low-risk and intermediate-risk groups (P = 0.921; Supporting Information Figure S3). The C-PINK could be used to discriminate patients classified in the low-risk group of the IPI (P < 0.001), groups 1 and 2 (P = 0.014) or groups 3 and 4 (P = 0.012) of the KPI, and the low-risk PINK-E group (P = 0.001; Supporting Information Figure S4). In summary, our new prognostic models have important implications for risk-adapted treatment approaches for patients with nasal-type stage I/II ENKTL. To improve treatment outcomes in high-risk patients, the combination of immune checkpoint inhibitors or other new agents and hematopoietic stem cell transplantation should be considered.6 However, the prognostic significance of this model should be verified by a future prospective study. The authors thank all the patients, their families, and the institutions for supporting this study. They acknowledge all medical staff, staff nurses, and research nurses at the nine medical centers, all of whom strongly contributed to the success of the study. This work is supported by the Guangdong Science and Technology Department (grant number 2017B020227002). The authors have no conflicts of interest to declare. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.