Survival benefit of radiotherapy on patients with early-stage extranodal nasal-type natural killer / T-cell lymphoma : an analysis of the surveillance epidemiology and end results database

Background: Extranodal natural killer/T-cell lymphoma (ENKTL) is a rare malignant lymphoid malignancy. The survival benefit of radiotherapy (RT) in earlystage ENKTL patients remains controversial. This study was conducted to investigate the prognostic factors, and evaluate survival benefit of RT in early-stage ENKTL patients. Materials and Methods: Early-stage ENKTL patients between 2004 and 2013 were searched from the Surveillance Epidemiology and End Results (SEER) database. Clinical characteristics including sex, age at diagnosis, race, marital status, era of diagnosis, histology, Ann Arbor stage and RT were summarized. Kaplan–Meier and multivariate Cox proportional hazards regression analysis were performed to investigate the independent prognostic factors for early-stage ENKTL patients. Subgroup analysis was conducted to evaluate the benefit of RT on overall survival (OS) and cancer specific survival (CSS) based on different Ann Arbor stages. Results: Patients with early-stage ENKTL were more likely to be younger, present with B symptoms. RT was more likely used for younger patients and those presented with B symptoms. Kaplan–Meier and multivariate Cox regression analysis showed that Ann Arbors stage and RT were the independent prognostic factors for survival (p < 0.05). And RT was associated with a lower risk of mortality for OS and CSS. Subgroup analysis based on different Ann Arbor stages showed that RT can decrease 66% risks of mortality in Stage I and more than 60% risks of mortality in Stage II (p < 0.005). Conclusions: RT could contribute to significant survival benefit among early-stage ENKTL patients. Significant survival benefit of RT still remained in any disease stage.


INTRODUCTION
Extranodal nasal-type natural killer/T-cell lymphoma (ENKTL) is a rare subtype of non-Hodgkin's lymphoma, which is prevalently distributed in East Asia and South America [1,2].ENKTL occurs predominantly in the nasal cavity and nasal side area and the disease is highly aggressive, is not sensitive to chemotherapy and has poor prognosis.The 5-year overall survival (OS) ranges from 40% to 90% for early-stage patients [3][4][5][6][7].
ENKTL is closely connected with the Epstein-Barr virus (EBV) infection, which assists in the diagnosis of ENKTL [8].The characteristic surface markers of NK/T cells and the specific genetic mutations can also contribute to its diagnosis.Due to unique biological behavior and various clinical manifestations [9], ENKTL has been recognized as a novel classification of lymphoid neoplasms by the World Health Organization [10].
Although several studies have explored the treatment of ENKTL in recent years, the optimal treatment is still unclear [11].Currently, radiotherapy (RT) is the most widely used treatment for patients with early-stage ENKTL.Depressingly, the clinical efficacy of RT reported was controversial [6,9,[12][13][14][15].Few prospectively designed, randomized and controlled clinical trials have been conducted to explore the survival benefit of RT for early-stage ENKTL patients and current evidence mainly relies on retrospective studies and series case reports with small samples.Additionally, most of them were conducted in Asian countries [6,9,[12][13][14][16][17][18][19][20].In view of above, we conducted a population-based analysis of patients with early-stage ENKTL from the Surveillance, Epidemiology, and End Results (SEER) database to evaluate the survival benefit associated with RT and investigate the prognostic factors for early-stage ENKTL patients.

The baseline of patient characteristics
A total of 348 patients with Stage I and II ENKTL from 2004 to 2013 were found eligible for our analysis, including 221 (63.51%) patients in RT group and 127 (36.49%) in no-RT group.Demographic and pathologic information are summarized in Table 1.A total of 123 (35.34%) patients were over 60 years old (p = 0.025) and 232 (66.67%) of them were male (p = 0.665).For patients younger than 60 years old or presented with B symptoms, they were more likely to receive RT (p < 0.05).The patients in two groups showed no significant variances with regards to other baseline characteristics including sex, race, marital status, era of diagnosis, histology and Ann Arbor stage.

Kaplan-Meier analysis for the effect of clinical characteristics on OS and CSS
Kaplan-Meier analysis was performed to calculate the OS and CSS of the early-stage ENKTL patients (Table 2).The OS (Figure 1) and CSS (Figure 2) were significantly longer in RT group than that in no-RT group (log-rank test p < 0.001), which was in favor of RT.What's more, the median OS of RT group (93.0 months, 95% CI: 56.0-NA months) was much longer than that of no-RT group (10.0 months, 95% CI: 7.0-22.0months).Ann Arbor stage was significantly associated with OS (Figure 3) and CSS (Figure 4).Sex, age at diagnosis, race, marital status, era of diagnosis, histology and B symptoms were found no significant associations with survival (p > 0.05).

Multivariate Cox proportional hazards regression analysis for identification of prognostic factors in patients with early-stage ENKTL
When we adjusted all clinical factors in the multivariate analysis, Ann Arbor stage and RT were recognized as independent prognostic factors for OS and CSS among early-stage ENKTL patients (Table 3).In the terms of OS, the mortality risks of the patients in RT group decreased more than 60% compared with that in no-RT group (HR = 0.39, 95% CI 0.28-0.54,p < 0.001).For CSS, RT was still identified as a protective factor with a much lower hazards of mortality (HR = 0.40,95% CI 0.28-0.58,p < 0.001).Ann Arbor stage was also an independent prognostic factor but negatively associated with OS and CSS.

DISCUSSION
Extranodal natural killer/T-cell lymphoma, nasal type (ENKTL) is rare, aggressive and poor prognostic.The disease is relatively common in Asia and South America but rare in Western countries .And that results in few large sample clinical trials of ENKTL.To date, there has been no standard treatment available for ENKTL.The current treatment includes chemotherapy, RT alone, chemotherapy after RT, RT after chemotherapy or concurrent chemoradiotherapy [19,23,24].Available evidence suggests that RT is preferred for early-stage NKTCL patients.Li et al. [25] reported that RT alone could significantly improve the survival outcomes of patients with Stage I ENKTL.In their study, the complete response was 95.4% (83/87) and partial response 2.3% (2/87).For all patients, the 5-year OS of 80%, progressionfree survival of 69%, and local control of 93% were observed.More recently, Yang et al. [26] conduct a largescale retrospective analysis of Stage I ENKTL patients and they are all given high-dose extended-field RT with a median radiation dose of 50 Gy (range, 20-65 Gy).The overall response to RT arrived to 97.7% (85/87) patients in a short time, recommending 50 Gy as the optimal dose for patients with early-stage disease.
However, due to the rarity of ENKTL and few large sample randomized controlled trials, the role of RT is still not well defined.Fortunately, the Surveillance, Epidemiology, and End Results (SEER) Program has been established and the relevant information of cancer patients is collected from large numbers of cancer registries which covers 9,0468 population, accounting for 28% in the United States .With large clinical information of cancer patients, it is a favorable tool to investigate rare carcinoma.In this study, we used SEER database to retrospectively evaluate the survival benefit of RT on patients with earlystage ENKTL.Since great advancements have been achieved in modern radiotherapy and the applications of rituximab in the last decade for lymphoma [27], we restricted the era of diagnosis from 2004 to 2013 to lower the risk of potential confusion.The characteristics of ENKTL patients used in our study included sex, age at diagnosis, race, marital status, era of diagnosis, histology, Ann Arbor stage, B symptoms and RT.We found RT was more likely to be used in patients younger than 60 years old and those presented with B symptoms.Kaplan-Meier analysis showed that OS was associated with Ann Arbors stage and RT (p < 0.05), but not associated with sex, age at diagnosis, race, era of diagnosis, histology and B symptoms (p > 0.05).It was consistent with the previous literature reports [12,21,22,[28][29][30][31][32].Multivariate Cox regression analysis revealed that Ann Arbor stage and RT were both significantly associated with OS and CSS (p < 0.05).Because RT contributed to significant survival benefits for both OS and CSS, we illustrated OS in the following discussion.Patients had a higher risks of mortality in Stage II (HR 1.63, 95% CI 1.17-2.27,p = 0.004) compared with those in Stage I. Interestingly, no significant mortality risks difference were observed in age at diagnosis, era of diagnosis and histology (p > 0.05).This may be associated with the rapid progression, quick    Though the present study was large populationbased, the results should be interpreted with caution since it had several limitations.First, SEER-18 cohort database does not have information about RT dosage, the timing of RT, and side effects of RT, which limits us to investigate the definite advantage of RT irrespective of diseasespecific chemotherapies or better supportive care.Second, RT group had a higher proportion of younger patients than the older (43.97% vs 19.54%, p = 0.025).Generally, the younger may maintain good physical condition with less comorbidities, which may well contribute to better OS and CSS [33].Third, the SEER-18 cohort database lacks the prognostic information (IPI score, EBV-DNA, Ki-67, lactate dehydrogenase (LDH) and so on), which are well documented independent prognostic factors for survival outcome [34,35].For example, the EBV-DNA copy numbers are associated with tumor burden and can predict independently the outcomes of ENKTL patients [8].Finally, the present research had its inherent limitations and several bias since it was a retrospective study.Further prospective studies are needed to explore in depth.Despite the potential limitations, there were better OS and CSS in ENKTL patients who underwent RT, which was enough to demonstrate the survival benefit of RT in early-stage ENKTL patients.
One should be noted that chemotherapy regimens were not brought into the survival analysis in our study.It was as a direct result of missing chemotherapy records in the SEER-18 cohort database.However, as we mentioned previously, ENKTL is an uncommon disease and its chemotherapy regimens still remain controversial.For the last decades, no effective novel drug has been available [36], and the anthracycline-based chemotherapy regimens that have high similarity are recommended for most ENKTL patients [37,38].Furthermore, the addition of chemotherapy to RT does not change the survival outcomes of patients with early-stage ENKTL [24,25,[39][40][41][42][43][44].Hence, we have reasons to assume that lack of chemotherapy information has little impact on the final conclusions of our study.
In conclusion, RT could contribute to significant survival benefit among early-stage ENKTL patients.Ann Arbor stage and RT were both independent prognostic factors for early-stage ENKTL patients .Ann Arbor stage was negatively associated with OS and CSS.RT was positively associated with OS and CSS.Subgroup analysis showed that survival benefit of RT still remained in any disease stage.

Statistical analysis
The statistical analysis was performed with R software (version 3.3.2).Baseline characteristics of patients were analyzed with Chi-squared test.Survival analysis was performed with Kaplan-Meier analysis and multivariable Cox proportional hazards regression analysis were conducted to evaluate whether clinical factors were associated with OS and CSS, respectively.The clinical factors contained sex, age at diagnosis, race, marital status, era of diagnosis, histology, Ann Arbor stage, B symptoms, and radiotherapy.Finally, subgroup analyses via Cox proportional hazards regression model were conducted based on different disease stages.All p-values were two-sided and the p < 0.05 was considered statistically significant.

Figure 1 :
Figure 1: Kaplan-Meier curves of the overall survival in patients according to RT and no-RT.

Figure 2 :
Figure 2: Kaplan-Meier curves of the cancer specific survival in patients according to RT and no-RT.

Figure 3 :
Figure 3: Kaplan-Meier curves of the overall survival in patients according to Ann Arbor stage.

Figure 4 :
Figure 4: Kaplan-Meier curves of the cancer specific survival in patients according to Ann Arbor stage.