Extranodal natural killer/T-cell lymphoma, nasal type

The significance of radiotherapeutic parameters




The objective of this study was to investigate the correlation between local recurrence and radiotherapeutic parameters, including dose and RT radiotherapy (RT) field.


The current study included 35 patients who were diagnosed with immunohistochemically confirmed nasal natural killer (NK)/T-cell lymphoma between 1976 and 2004. There were 21 males and 14 females, and they ranged in age from 18 years to 76 years (median, 51 yrs). The primary tumor originated in the nasal cavity in 28 patients, and 32 patients had Stage I disease. Seventeen patients received treatment solely with RT, and the remaining 18 patients received a combination of chemotherapy and RT. The median tumor dose was 50 grays (Gy) (range, 22–60 Gy). Twenty-seven patients received RT to include all macroscopic lesions, all paranasal sinuses, the palate, and the nasopharynx. Eight patients received RT to all macroscopic lesions with generous margins.


A complete remission (CR) or a CR/unconfirmed was achieved in 28 patients (80%). The 5-year overall survival (OAS) rate, disease-free survival (DFS) rate, and local control probability (LCP) were 47.3%, 42.9%, and 65.2%, respectively. Patients who received RT only to macroscopic lesions fared less well in terms of LCP (LCP 5 years, 71.9% vs. 41.7%; P = 0.007). The difference in RT field also affected both the OAS rate and the DFS rate. Patients who received RT doses ≥ 50 Gy tended to achieve favorable local control.


In the management of nasal NK/T-cell lymphoma, the RT field affected treatment outcomes. RT doses ≥ 50 Gy resulted in favorable local control. Cancer 2006. © 2005 American Cancer Society.

Extranodal natural killer (NK)/T-cell lymphoma, nasal type (formerly known as midline lethal granuloma, polymorphic reticulosis, or angiocentric immunoproliferative lesions) recently was recognized as a distinct entity of malignant lymphoma.1–4 Because this type of lymphoma often shows an angiocentric and angiodestructive growth pattern, together with a broad cytologic spectrum of atypical cells and a zonal necrosis, it was categorized as angiocentric lymphoma in the revised European–American lymphoma classification.5 These lymphomas are uncommon in the U.S. and Europe, but they are prevalent in East Asia and in certain parts of Central and South America. In a recent nationwide study of malignant lymphoma among Japanese, it was reported that NK/T-cell lymphoma accounts for approximately 2.6% of all malignant lymphomas in Japan.6 Patients with this lymphoma present either with symptoms of nasal obstruction or epistaxis due to the presence of a mass lesion or with destruction of midfacial structures. The tumor also extends to adjacent tissues, such as the nasopharynx, paranasal sinuses, orbit, oral cavity, palate, and oropharynx.

The confusing nomenclature and its rarity have prevented investigators from establishing the optimal treatment for patients with extranodal NK/T-cell lymphoma. Furthermore, the finding that much published research not only included patients who did or did not undergo immunohistochemical confirmation of NK/T-cell lymphoma7–22 but also included patients who had B-cell lymphoma served to complicate the interpretation of that research.22–30 However, the response to radiotherapy (RT) generally is so rapid and dramatic that the delivery of RT has been accepted as the preferred treatment of choice for localized disease. Several investigators have advocated the combination of chemotherapy and RT9, 18, 26, 30; however, whether the addition of chemotherapy to RT offers any survival benefits to patients is questionable in most series.11, 12, 14–17, 19–22, 24, 25, 27–29 Patients with nasal NK/T-cell lymphoma experience recurrence at various anatomic sites, including the lymph nodes, skin, liver, spleen, and bone marrow; however, local failure remains the predominant pattern of recurrence. Thus, it is indispensable to clarify the role of RT in each clinical setting. The optimal dose required to achieve appropriate local control and the volume to be treated also require clarification.

In the current study, we analyzed the effects of various factors on local control with special reference to RT parameters, such as the dose to be delivered and the fields to be covered. The objectives of this study were to investigate the correlation between RT field and local recurrence and to establish the dose-response relation that influenced the probability of local control in patients with Stage I and II nasal NK/T-cell lymphoma who were treated at our institutions.


Between July 1976 and May 2004, 38 patients with nasal NK/T-cell lymphomas received RT at Chiba University Hospital and Narita Red Cross Hospital. Of these 38 patients, 3 were excluded from the current analysis for the following reasons: palliative intent for advanced disease (n = 2 patients) and consultation at the time of recurrence (n = 1 patient). The median follow-up was 27 months (range, from 1.5 mos to 28.0 yrs), and the median follow-up of surviving patients was 11.1 years.

Histologic specimens were evaluated by an expert hematopathologist (J. T.). Immunohistochemical examination was undertaken in every patient to exclude B-cell lymphomas. Consequently, all tumors were positive for either CD56 and CD3ϵ or for T-cell markers, such as CD3, CD43, and CD45RO. For staging of their disease, patients underwent a physical examination; complete blood counts; screening blood tests of hepatic and renal function; chest radiograph; gallium scintigraphy; computed tomography (CT) scans of the head and neck, chest, abdomen, and pelvis; examination of the gastrointestinal tract; and bone marrow aspiration and/or biopsy. All patients were staged according to the Ann Arbor criteria.

Patient Characteristics

There were 21 males and 14 females, and the patients ranged in age from 18 years to 76 years (median, 51 yrs). The primary tumor originated in the nasal cavity in 28 patients and in the paranasal sinuses in 4 patients, including 2 patients with mesopharynx tumors and 1 patient with a nasopharyngeal tumor. Thirty-two patients had Stage I disease. Systemic B symptoms were present in seven patients, and lactate dehydrogenase (LDH) elevation was found in seven patients (Table 1). Informed consent was provided according to the Declaration of Helsinki.

Table 1. Patient Characteristics
CharacteristicsNo. of patients (%)
  1. SD: standard deviation.

 Range18–76 yrs
 Mean ± SD52 ± 16 yrs
 Median51 yrs
 Male21 (60)
 Female14 (40)
Primary site 
 Nasal cavity28 (80)
 Paranasal sinus4 (11)
 Pharynx3 (9)
B symptoms 
 Yes7 (20)
 No28 (80)
 Lactate dehydrogenase elevation 
 Yes7 (20)
 No28 (80)
Disease stage 
 Stage I32 (91)
 Stage II3 (9)


All patients received RT from a cobalt-60 unit or a linear accelerator with 4-megavolt (MV), 6-MV, or 10-MV photons to achieve dose homogeneity. An appropriate energy of electron field also was applied to treat the tumor behind the lens block in the photon field. Generally, the planning target volume included all macroscopic lesions, the paranasal sinuses, the nasopharynx, the upper gum, and the palate with adequate margins. Regardless of primary tumor localization, elective cervical lymph node irradiation was not delivered unless the neck was involved clinically. The most common field arrangement was two lateral opposing photon fields with supplementation between the medial canthus by appropriate energy of electron. All patients received RT with a conventional fractionation schedule at a median tumor dose of 50 grays (Gy) (range, 22–60 Gy). Seventeen patients received treatment treated solely with RT, and the remaining 18 patients received a combination of chemotherapy followed by RT. Anthracycline-containing combination chemotherapy was administered to 15 patients (Table 2).

Table 2. Treatment Characteristics
CharacteristicNo. of patients (%)
  1. RT: radiotherapy; SD: standard deviation; Gy: grays.

 RT alone17 (49)
 RT and chemotherapy18 (51)
 Yes15 (83)
 No3 (17)
 Range22.0–60.0 Gy
 Mean ± SD37.9 ± 9.3 Gy
 Median50.0 Gy
 Primary alone8 (23)
 Others27 (77)

Statistical Analysis

Tumor response was assessed by using standard criteria.31 Overall survival (OAS), disease-free survival (DFS), and the local control probability (LCP) were calculated using the method of Kaplan and Meier.32 The log-rank test was used to assess significance in univariate analysis, and the Cox proportional hazards model was used to assess significance in multivariate analysis.33


Overall Results

At the time of evaluation, 28 patients achieved complete remission (CR) or CR/unconfirmed (CRu), which resulted in an 80% CR rate (95% confidence interval [95% CI], 66–94%). Of the seven patients who failed to obtain CR or CRu, the primary tumor was controlled well in three patients. The 2-year OAS rate, DFS rate, and LCP were 57.6% (95% CI, 40.7–74.4%), 53.0% (95% CI, 36.2–69.8%), and 73.8% (95% CI, 57.9–89.6%), respectively. The corresponding values at 5 years were 47.3% (95% CI, 29.8–64.7%), 42.9% (95% CI, 25.8–60.0%), and 65.2% (95% CI, 47.3–83.2%), respectively (Fig. 1). The 5-year OAS rate, DFS rate, and LCP for the patients who received RT alone were 43.8% (95% CI, 19.4–68.1%), 43.8% (95% CI, 19.4–68.1%), and 60.2% (95% CI, 35.2–85.2%), respectively. The corresponding values for patients who received both chemotherapy and RT were 51.8% (95% CI, 27.5–76%), 43.2% (95% CI, 19.8–66.6%), and 71.9% (95% CI, 48.2–95.7%), respectively. The administration of chemotherapy did not appear to have an impact on all endpoints.

Figure 1.

This chart illustrates the curves for overall survival (solid line), disease-free survival (dotted line), and local control probability (dashed line).

At the time of the current analysis, 16 patients were alive without evidence of disease, and 14 disease recurrences were observed. Of these 14 patients who developed disease recurrence, 5 patients had locally recurrent disease, 6 patients had distant metastasis, and the remaining 3 patients experienced both local and distant or regional failure. The median time to disease recurrence was 11.6 months. The metastatic sites included the liver, spleen, lymph nodes, subcutaneous soft tissue, skin, intestine, bone marrow, and brain. Two patients who experienced local recurrence were salvaged successfully by a second course of RT. Those 2 patients developed local disease recurrence at 8.5 years and 15.5 years after their initial course of RT. There were 18 deaths during the study period, all of which were due to progressive or recurrent tumors.

RT Field, RT Dose, and Local Control

We also assessed the relation between RT field and the LCP. During the study period, 27 patients received RT that included all macroscopic lesions and sites of potential contiguous spread (i.e., all paranasal sinuses, the palate, and the nasopharynx) with adequate margins. The RT field in the remaining eight patients encompassed all macroscopic lesions with generous margins. Although there were six local recurrences in the former group, all but two patients in the latter group experienced local disease recurrence. The LCP at 5 years was 71.9% versus 41.7%, respectively (P = 0.007) (Fig. 2).

Figure 2.

This graph illustrates the local control probability as a function of radiotherapy (RT) field. Solid line: an RT field that encompassed all sinuses, the nasopharynx, and macroscopic lesions; dashed line: an RT field that included macroscopic lesions with a margin.

Next, we assessed whether there was a dose-response relation for local control. Among the 26 patients who received ≥ 50 Gy of RT, 20 patients were able to achieve local control; however, only 3 of 9 patients in the low-dose group obtained local control (Fisher exact test; P = 0.038). Figure 3 illustrates the LCP as a function of RT dose. The 5-year LCP for patients who received ≥ 50 Gy versus patients who received < 50 Gy was 69.2% (95% CI, 47.9–90.5%) and 53.3% (95% CI, 19.4–87.3%), respectively (P = 0.13). The OAS and DFS rates for patients who received ≥ 50 Gy were 47.7% (95% CI, 28.6–68.6%) and 41.6% (95% CI, 21.3–61.9%), respectively, which did not differ significantly from patients who received < 50 Gy.

Figure 3.

This graph illustrates the local control probability as a function of radiotherapy dose. Solid line: doses ≥ 50 grays (Gy); dashed line; doses < 50 Gy.

Prognostic Factors

The clinical and treatment factors that we assessed for potential prognostic impact included age, gender, primary site, B symptoms, LDH elevation, disease stage, dose of RT, and chemotherapy. However, none of those variables was identified as an independent prognostic factor for OAS and LCP. The only factor that was found to be associated with OAS, DFS, and LCP was the RT field (P = 0.027, P = 0.020, and P = 0.007, respectively). However, multivariate analysis failed to identify any prognostic factors for those three endpoints.


Extranodal NK/T-cell lymphoma, nasal type, which was recognized previously as angiocentric lymphoma, has a distinct position in the new World Health Organization classification system.1 It is characterized by CD56 and cytoplasmic CD3 expression, a germline configuration of the T-cell receptor gene, and a strong association with Epstein–Barr virus (EBV). Many groups have reported treatment outcomes for patients with nasal non-Hodgkin lymphoma.7–30 Some groups treated patients only with RT,7, 8, 10, 13, 23 but others delivered both chemotherapy and RT.9, 11, 12, 14–22, 24–30 Those studies demonstrated OAS and LCP rates at 5 years of 24–86%, and 31–84%, respectively. Although there are many suggestions in the literature regarding the management and natural history of nasal lymphoma, many of these reports included patients with B-cell lymphoma and patients who did not receive immunohistochemical examinations, which led to a great deal of confusion in relation to the roles of chemotherapy, failure patterns, and treatment outcomes.

There were eight studies, including our previous work,14 that included only patients who had immunophenotypically confirmed nasal NK/T-cell lymphoma.14, 16–22 Furthermore, of those eight studies, five included only patients who had both CD3ϵ-positive and CD56-positive expression to exclude peripheral T-cell lymphoma.17, 18, 20–22 A summary of those eight studies is provided in Table 3. According to those reports, the OAS ranged from 36.5–86%. Although the Mexican group demonstrated very surprising results,16 the remaining 7 groups reported that OAS was approximately 40%,14, 17–22 which was comparable to the results of the current study. All eight series administered chemotherapy and RT and concluded that conventional chemotherapy followed by RT appeared to be ineffective for the majority of patients and that innovative treatment modalities are needed to improve outcomes. However, Yamaguchi et al. observed that patients who received treatment with concurrent chemoradiotherapy or with RT followed by chemotherapy enjoyed favorable outcomes.18 Therefore, those authors concluded that RT followed by, or combined with, chemotherapy was best as initial treatment, and they recommended nonanthracycline-containing chemotherapy (dexamethasone, etoposide, ifosphamide, and carboplatin) based on their previous observation that nasal NK/T-cell lymphomas express P-glycoprotein.34 Since 1998, Cheung et al. also have employed concurrent chemoradiotherapy in an attempt to intensify local treatment.20 Those authors selected cisplatin as the chemotherapeutic agent in this concurrent setting. Conversely, Kim et al. also administered anthracycline-containing chemotherapy concurrently with RT in two patients.17 Furthermore, Ribrag et al. observed that two patients who were treated with alternated chemotherapy and RT achieved a CR.19 The sequence of chemotherapy and RT that will most effectively achieve a satisfactory local control rate will be resolved by future studies; however, it can be concluded that chemotherapy followed by RT is disadvantageous.

Table 3. Summary of the Literature
ReferencePhenotypeTreatmentNo.LFR5-yr OAS (%)
  1. LFR: local failure rate; OAS: overall survival; NK: natural killer; CT: chemotherapy; RT: radiotherapy; NR: not reported.

Itami et al., 199114NK or T-cellCT → RT or RT96/9NR
Aviles et al., 200016NK or T-cellRT → CT108NR86 (8 yrs)
Kim et al., 200117NK cellCT → RT17NR59 (3 yrs)
Yamaguchi et al., 200118NK cellRT → CT or CT → RT127/1239
Ribrag et al., 200119NK or T-cellRT → CT or CT → RT or RT20NRNR
Cheung et al., 200220NK cellCT → RT7931.1%37.1
Chim et al., 200421NK cellCT → RT6735/6742.5 (10 yrs)
You et al., 200422NK cellCT → RT46NR36.5
Current studyNK or T-cellCT → RT or RT3534.8%47.3

In addition to controlling systemic disease, it is indispensable to achieve high LCP in patients with localized nasal NK/T-cell lymphoma. Local recurrence rates ranged from 31–67%, and these high local failure rates led to very poor outcomes.14, 16–22 Although it is evident that RT should play an essential role in achieving local disease control, the dose to be delivered and the field to be covered have not been resolved. With regard to treatment volume, Cheung et al. administered RT to the nasal cavity and nasopharynx,20 two investigational groups encompassed all paranasal sinuses and the Waldeyer ring in addition to the nasal cavity,21, 22 and four investigational groups delivered RT to the tumor with an adequate margin.14, 17–19 The remaining Mexican investigators delivered RT with an extended field, but to our knowledge the details were not reported.16 In the current study, we observed that the patients who received RT to macroscopic lesions with a margin achieved an inferior local control rate compared with patients who received with an RT field that encompassed all paranasal sinuses, the palate, and the nasopharynx in addition to the nasal cavity. In contrast, Cheung et al. recommended meticulous CT conformal planning with the aid of magnetic resonance imaging scans to deliver RT to the macroscopic tumor with an adequate margin.20 Although the majority of reports in the literature do not mention the recommended RT field, three groups of investigators advocated that the RT field should encompass the paranasal sinuses in addition to macroscopic lesions,7, 23, 24 a recommendation that is consistent with our current observations.

Many researchers have delivered 30–60 Gy to control macroscopic lesions. In the current study, we suggested that patients who received ≥ 50 Gy had a tendency to achieve superior local control rates compared with patients who received < 50 Gy, which is well in accordance with the observation of Cheung et al.20 You et al. administered higher RT doses (54–60 Gy) and achieved an 83.3% failure-free survival rate at 5 years.22 Furthermore, although 50% of their patients did not have immunohistochemical confirmation of nasal NK/T-cell lymphoma, a Korean group demonstrated a clear dose-response relation within the range of 20–54 Gy with a plateau at doses in excess of approximately 54 Gy.13 Conversely, 2 other groups reported that 45 Gy appeared to be an effective dose for local control.16, 17 Those data indicated that it is necessary to deliver at least 45 Gy of RT to achieve a favorable local control rate; however, whether higher doses would achieve better local control rates remains unresolved.

We were able to identify that the RT field was a significant prognostic factor for LCP, DFS, and OAS. Several groups have advocated that disease stage,15, 17, 20, 22 performance status,20 B symptoms,17, 20 LDH elevation,22 and the International Prognostic Index (IPI)21, 22 are of prognostic importance. However, Aviles et al. reported that there was no evidence that the IPI was applicable in patients with nasal NK/T-cell lymphoma,16 and there have been no widely accepted prognostic factors. With regard to molecular markers, Lin et al. examined 19 true NK-lineage nasal NK/T-cell lymphomas and demonstrated that CD94 expression was a favorable prognostic factor.35 In addition, Au et al. demonstrated that the EBV DNA level at the time of presentation was correlated with disease stage and LDH, and high presentation EBV DNA levels were associated significantly with inferior DFS. Furthermore, patients with EBV DNA levels that increased further or that failed to become undetectable during treatment had significantly inferior survival. Those investigators concluded that, in patients with EBV-positive lymphomas, the plasma EBV DNA level is valuable as a tumor biomarker and for prognosis.36 The significance of these new molecular markers will be elucidated in future clinical trials.

The rarity of this type of lymphoma limits large-scale, prospective, randomized trials. However, several of our findings have important implications in the management of nasal NK/T-cell lymphoma. The high efficacy of RT in achieving a CR with an RT field that encompasses all paranasal sinuses, the nasopharynx, and the palate, in addition to macroscopic lesions, and with RT doses ≥ 50 Gy suggests that it may be advantageous to incorporate adequate RT up-front in the treatment strategy. The occurrence of failures at distant sites implies that systemic chemotherapy also should be administered. Accordingly, we have launched a prospective study to evaluate the efficacy and toxicity of concurrent chemoradiotherapy for patients with nasal NK/T-cell lymphoma.