Predictors of survival of natural killer/T-cell lymphoma, nasal type, in a non-Asian population: a single cancer centre experience

Background Natural killer/T-cell lymphoma (NKTCL), part of T-cell and NK-cell neoplasms in the World Health Organisation (WHO) classification, is an aggressive lymphoma with poor prognosis more predominantly seen in Asian and South American countries. This study evaluates the factors associated with survival among patients with newly diagnosed NKTCL in Peru. Methods Information was abstracted from medical records (MR) for all NKTCL patients >13 years of age at the Instituto Nacional de Enfermedades Neoplasicas (INEN) between 2002 and 2011. The estimate of the survival curves was performed by the Kaplan-Meier method, and the difference was computed by the log-rank test. Results Around 226 MR were reviewed, 153 met the selection criteria, the median age was 40 years (14–84). The median progression-free survival (PFS) was 20 months, five year PFS was 42.6%, univariable analysis (UA) showed statistical significance (p < 0.05) for male sex, non-nasal primary site, advanced clinical stages, B symptoms, poor performance status, regional nodal involvement (RNI). In the multivariate analysis the only poor prognostic factors was primary non-nasal (Hazard ratio (HR) = 2.40, 95% confidence interval (CI) = 1.43– 4.02, P = 0.01). The median overall survival (OS) was 49 months, five year OS was 48.9%, UA showed statistical significance for non-nasal primary site, advanced clinical stages, B symptoms, lactate dehydrogenase (LDH) > normal, RNI and local tumour invasion. In the multivariate analysis, primary non-nasal was the only poor prognostic factor with HR = 2.57, 95% CI = 1.37–4.83, P = 0.03. Conclusions In Peru, OS of NKTCL is similar to other countries. This result suggests that non-nasal NKTCL is the only poor prognostic factor of OS and PFS.

The aim of this retrospective study is to describe predictors of OS and PFS as well as clinical and pathological features of patients with NKTCL treated at our centre.

Patients
The study population included all newly diagnosed patients >13 years of age with a pathological confirmation of NKTCL seen at the National Institute of Neoplastic Diseases (INEN) in Lima,Peru between January 2002 and December 2011. Patients with other previous cancers, positive serology for HIV, and diagnosis of aggressive NK cell leukaemia, or incomplete MRs were excluded. The diagnosis of NKTCL was based on the WHO 2008 classification of haematopoietic and lymphoid tissues [1]. All cases were reviewed by an expert panel in lymphomas to confirm diagnosis.

Laboratory findings and staging
Haematological tests, including complete blood count, liver, and renal function tests, and LDH were performed. Local tumour invasion was defined differently according to the two subtypes. The nasal NKTL was defined in accordance with 7th ed., 2010 TNM classification of the American Joint Committee of Cancer. Any nasal NKTCL with T3 or greater were considered as local invasive in the analysis. For non-nasal NKTCL, the definition of local invasiveness differed according to primary sites. For gastrointestinal NKTCL, local invasiveness referred to T4 lesion in TNM system. In NKTCL primarily involving soft tissue such as muscle or skin, invasion of neurovascular structure or bone invasion was considered as local invasion. Regional nodal involvement was defined as the invasion of lymph nodes corresponding to N1, N2, or N3 of the primary lesion based upon TNM staging system. Accordingly, M1 nodes at TNM system were not categorised as regional lymph nodes in the analysis [16].
The staging was based on modified System Cotswolds Ann Arbor [26]. Performance status was evaluated according to the Eastern Cooperative Oncology Group (ECOG) scale [27]. The response was evaluated based on the revised response criteria for lymphomas [28].

Statistical analysis
A descriptive analysis of the information through frequencies, percentages, and measures of central tendency were performed. OS was defined as the time from the date of diagnosis to date of last visit or death from any cause, and PFS from the initiation of treatment to disease relapse/progression, last follow-up or death from any cause, whichever occurred first. The overall and PFS were estimated with www.ecancer.org ecancer 2016, 10:688 the Kaplan-Meier method and differences were tested using the log-rank test. The Cox proportional hazard models were used to identify predictors of survival of NK/T-cell lymphomas. A level of p <0.05 was considered for a statistical significance. The multivariate analysis was performed with all factors with statistical significance in the univariate analysis. SPSS version 22.0 was used for statistical analysis.
Chemotherapy regimen used was cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) in seven patients (early stages n = 2 , advanced stages n = 5) and CHOEP-14 (CHOP + etoposide) in 18 (early stages n = 9) patients. No one patient received SMILE regimen (dexamethasone, methotrexate, ifosfamide, l-asparaginase, and etoposide) as it was included as a salvage treatment just at the end of 2013.
The chemotherapy regimens used either as monotherapy, prior, concurrent, or followed by radiotherapy were CHOP in 33 (44.5%), CHOEP-14 in 40 (54.1%), and other regimen in one case (1.4%), totalling 74 patients with this type of treatment.
Twenty-five patients received chemotherapy only (early stages n = 11), ranging the number of cycles of chemotherapy between 1 and 8 with a median of 2.5 cycles. Among the patients in early stages receiving chemotherapy followed by radiotherapy with number of cycles ranging between 1 and 8, we find 69.6% (23 out of 33) received six cycles. It was also noted that only one patient with advanced disease received this modality of treatment.

Treatment response
The early stages had response rates as follows:

Progression-free survival
With a median follow-up time of 18 months, the median PFS time was 20 months (95% CI 0-41), and the five year PFS rate was 42.6%. The variables that showed statistical significance towards a worse outcome in the univariate analysis were male sex, non-nasal primary site, advanced clinical stages, presence of B symptoms, poor performance status (PS), and nodal involvement. The median PFS for the nasal primary was 79 months (95% CI 14-143) and the five year PFS was 50.9% as shown in Figure 1 and Table 2. The six variables with statistical significance were studied by multivariate analysis, showing that the only independent poor prognostic factor was primary nonnasal site (HR = 2.40, 95% CI = 1.43-4.02, P = 0.01) ( Table 2).

Overall survival
Seventy-five patients (49%) had died by the time of the analysis. With a median follow-up of 48 months (1-143 months). The median OS time was 49 months (95% CI 0-98) and the five year OS rate was 48.9% ( Figure 2). The factors that showed statistical significance towards a worse outcome in the univariate analysis were non-nasal primary site, advanced clinical stage, presence of B symptoms, increased LDH, nodal involvement and local tumour invasion. For nasal primary the median OS time was not reached and the five year OS rate was 55.8% as shown in Table 3 and Figure 3. The multivariate analysis showed that the only poor prognostic factor was primary non-nasal site (HR = 2.57, 95% CI = 1.37-4.83, P = 0.03) ( Table 3).
We performed a sub-analysis for early disease and we found the median OS for nasal primary (117, 87.3%) was not reached and that for non-nasal primary (

Discussion
We present here 153 NKTCL patients who fulfilled our selection criteria diagnosed over a period of ten years. To the best of our knowledge it is the largest Latin American study in NKTCL patients [4][5][6][7][8] and the second largest in the whole American continent [9][10][11][12]. The number of patients were similar to some seen in multicentre studies [2,13,15,[20][21][22][32][33][34][35][36]. The present study shows that the clinical characteristics of nasal NKTCL in Peru are similar to those described in other Latin American as well as Asian and Western countries [2,3,6,8,9,29,30]. The treatment used in our population was according to the clinical staging, with radiotherapy only or concurrent with chemotherapy for early stages, and chemotherapy only for advanced disease, as is described in the international literature [31]. L-asparaginase-based regimen (SMILE treatment) was not part of the treatment, as this was incorporated in our institution just in 2013 [37].
In this series, the PFS was 20 months which is similar to reported by Huang et al at 18 months [21]. However, the five year PFS was 29.8% which is lower than our study where the five year PFS was 42.6%. This discrepancy might be because of a lower proportion of patients with presence of B symptoms in our study (36% versus 54%, respectively) and also could have influenced the higher percentage of patients with early stages in our study (83% versus 79.5%, respectively) . In contrast, Lee et al [16] showed a five year relapse-free survival (RFS) of 60%. This result is different because this was calculated on patients who had achieved complete remission.
In our study, we show that patients with NKTCL have poor survival, with a median OS of 49 months. In other studies, the median OS has ranged between 30 and 50 months [2,16,21,34]. However, when we consider the survival for non-nasal NKTCL this was nine months, different from data reported by Au et al [3] where extranasal cases had a median OS of four months. This was probably because 19% of those patients did not receive treatment because of advanced disease unlike our study where all patients received treatment.
Regarding the cumulative probability of survival at five years, our study shows a 49% OS rate, whereas in other studies it ranged between 39% and 50%. [13,16,21,34,37,38]. When evaluating the non-nasal primary this was 16% which is comparable to the study of Au et al [3]. Only a non-nasal primary site was an independent adverse predictor in the multivariate analysis which we find is also the same variable identified in other studies [3,12,16,32]. The primary site of this type of lymphoma had been evaluated previously but with the term upper aerodigestive tract NKTCL. This included lymphomas confined to nasal cavity, nasopharynx, larynx, pharynx, and oral cavity [16]. In our study nasal type refers only to nasal cavity.
In our study, we found that the primary site is more important than clinical stage as independent prognostic risk factor, and this finding was reported by Au et al's study. Even non-nasal primary with apparently localised disease had poor prognosis [3]. The biological distinction between these two subgroups remains unknown, hence necessitating future studies with genetic and epigenetic profiling. Other prognostic factors have been evaluated in NKTCL patients such as Ki-67 expression, EBV viral load, lymphocyte and monocyte counts [19,21,25,33]. In the last years c-Myc expression, beta-2 microglobulin, CD30-CD38 expression, the albumin to globulin ratio [38][39][40][41][42][43][44], CD56 expression, higher levels of HLA-DR negative, CD33, CD11b myeloid-derived suppressor cells (MDSCs), CD14 monocytic MDSCs, independent adverse prognostic scores have also been used for evaluation [45]. However, in our study we could not evaluate these factors either because of incomplete baseline data or because of lack of tests for these factors in our centre.

Conclusion
In conclusion, in Peru the OS for NK/T-cell lymphoma is similar to other Latin American as well as Asian countries. The results suggest that non-nasal NKTCL is the only poor prognostic factor of OS and PFS which we find is even more important than the clinical staging itself. This poor prognostic factor is seen in early stages as well. It is important to conduct multicentre prospective studies including the most important clinical, laboratorial, pathological, and viral prognostic factors in order to make an accurate prognostic index. www.ecancer.org ecancer 2016, 10:688 www.ecancer.org ecancer 2016, 10:688