Treatment of extranodal NK/T cell lymphoma, nasal type

INTRODUCTION — Extranodal natural killer (NK)/T cell lymphoma (ENKL), nasal type is a rare, predominantly extranodal subtype of non-Hodgkin lymphoma. Most cases have an NK cell phenotype, but a small minority are derived from cytotoxic T cells. The tumor cells are latently infected with Epstein-Barr virus (EBV). ENKL was formerly called lethal midline granuloma or angiocentric lymphoma.

ENKL causes vascular damage and tissue destruction. It usually presents in the upper aerodigestive tract (nasopharynx), but it may also involve other extranodal sites (extranasal ENKL).

The topic discusses the treatment of ENKL.

The clinical presentation, pathologic features, and diagnosis of ENKL are presented separately. (See "Clinical manifestations, pathologic features, and diagnosis of extranodal NK/T cell lymphoma, nasal type".)

MEDICAL EMERGENCIES — Some patients with ENKL present with medical emergencies such as hemophagocytic syndrome, infections, or bleeding that require immediate intervention.

EBV-related hemophagocytic syndrome may be manifest as high fevers, maculopapular rash, central nervous system symptoms, hepatosplenomegaly, lymphadenopathy, cytopenias, coagulopathy, abnormal liver function tests, and/or extremely high serum ferritin levels [1-5]. Patients require urgent treatment with etoposide plus glucocorticoids and other treatments, as described separately. (See "Treatment and prognosis of hemophagocytic lymphohistiocytosis".)

PRETREATMENT EVALUATION — Pretreatment evaluation should include clinical evaluation, laboratory studies, imaging, fiberoptic endoscopy, and bone marrow examination. Medical fitness and comorbid conditions should be evaluated to assess the tolerance for treatment.

Men and women of child-bearing potential should receive counseling about the potential effect of treatment on their fertility and options for fertility-preserving measures. (See "Fertility and reproductive hormone preservation: Overview of care prior to gonadotoxic therapy or surgery".)

Clinical evaluation — History should assess "B" symptoms (ie, fevers, drenching sweats, unexplained weight loss) and complaints related to involvement of the upper aerodigestive tract or other sites of extranodal involvement. Physical examination should include visual inspection of the nasal and oral cavities and palpation of buccal and gingival mucous membranes, the floor of the mouth, tongue, tonsillar fossae, palate, and posterior pharyngeal wall.

Laboratory studies — Laboratory studies include:

●Complete blood count (CBC) with differential count.

●Serum chemistries, including electrolytes, renal function tests, uric acid, liver function tests, and lactate dehydrogenase (LDH).

●Infectious serology, including hepatitis B virus and HIV testing. (See "Hepatitis B virus reactivation associated with immunosuppressive therapy".)

●Plasma Epstein-Barr virus (EBV) deoxyribonucleic acid (DNA) by quantitative polymerase chain reaction (PCR), as a baseline value for assessment of response to treatment. (See 'Response assessment' below and 'Prognosis' below.)

Imaging — Pretreatment imaging should include:

●Positron emission tomography/computed tomography (PET/CT).

●CT and magnetic resonance imaging (MRI) of the nasal cavity, hard palate, anterior fossa, and nasopharynx; we favor obtaining both studies, because CT offers superior bony definition and MRI provides superior soft tissue delineation.

●CT of neck, chest, abdomen, and pelvis with contrast.

●Echocardiogram or radionuclide ventriculography (RVG) to assess cardiac function. (See "Clinical manifestations, diagnosis, and treatment of anthracycline-induced cardiotoxicity" and "Risk and prevention of anthracycline cardiotoxicity".)

Endoscopy and bone marrow examination — The following studies are required to complete the staging evaluation:

●Flexible fiberoptic endoscopy enables examination and biopsy of the nasopharynx, oropharynx, hypopharynx, and larynx. Endoscopy can detect asymmetries, bleeding, mucosal irregularities, vocal cord mobility, and pooling of secretions and direct biopsies to confirm involvement at suspect sites. Random nasopharyngeal biopsies should be performed if no lesions are seen.

Mirror examination is an acceptable substitute only if fiberoptic endoscopy is not available.

●Bone marrow examination – Bone marrow involvement by ENKL is uncommon, but it is important to perform a bone marrow examination because treatment varies according to disease stage. Although malignant lymphoid aggregates are rare, the bone marrow should be considered to be involved with ENKL if lymphoid aggregates are positive for EBV-encoded ribonucleic acid (RNA; EBER)-1 [6]. (See "Clinical manifestations, pathologic features, and diagnosis of extranodal NK/T cell lymphoma, nasal type", section on 'Pathologic features'.)

MEDICAL FITNESS — The Eastern Cooperative Oncology Group (ECOG) performance status (PS) scale (table 1) should be used to assess medical fitness. We judge medical fitness based on functional status and comorbid conditions, rather than a specific age threshold. For older patients in whom medical fitness is uncertain, we obtain geriatric consultation and/or evaluate the patient as discussed separately. (See "Pretreatment evaluation and prognosis of acute myeloid leukemia in older adults", section on 'Pretreatment evaluation'.)

●Medically fit – Medically fit patients are those with ECOG PS 0 to 1 and no comorbid illnesses, such as heart disease, pulmonary processes, kidney dysfunction, or liver disease that would compromise their ability to tolerate intensive combination chemotherapy.

●Less fit – Less-fit patients generally have ECOG PS 2 and/or comorbid illnesses that would enable radiation therapy or less intensive combination chemotherapy but would not permit intensive combination chemotherapy.

●Frail – Frail patients have impaired functional status (eg, ECOG PS ≥3) or comorbid illnesses that would not permit treatment beyond supportive care to palliate symptoms.

STAGING — For ENKL, staging distinguishes between localized disease versus advanced disease. Sites of involvement are identified by physical examination, fiberoptic endoscopy, bone marrow examination, and imaging and then are compiled using the Lugano modification of the Ann Arbor staging system (table 2). (See 'Pretreatment evaluation' above.)

ENKL is classified as follows:

●Localized disease – Localized disease includes nasal ENKL with stage I and contiguous stage II disease. All cases of extranasal ENKL, except stage I cutaneous extranasal ENKL, are treated as advanced disease:

•Stage I(E) – Single extranodal involvement (usually of the upper aerodigestive tract) with or without contiguous involvement of adjacent structures (eg, tumor with an extension to the nasal cavity, maxillary antra, anterior or posterior ethmoid sinus, sphenoidal sinus, orbit, superior alveolar bone, cheeks, superior buccinator space, or hard palate).

•Stage II(E) – Stage I(E) disease plus contiguous cervical lymph node involvement.

●Advanced disease – Advanced disease includes nasal ENKL with noncontiguous stage II(E) disease, stage III, and stage IV. All cases of extranasal ENKL, except stage I cutaneous extranasal ENKL, are treated as advanced disease:

•Stage II(E) with noncontiguous disease (eg, disease of the upper aerodigestive tract plus involvement of supradiaphragmatic lymph node regions, other than cervical nodes, Waldeyer's ring, oropharynx, or hypopharynx).

•Stage III(E) (rare) – Stage I(E) disease (usually of the upper aerodigestive tract) with cervical lymph node involvement plus infradiaphragmatic lymph node involvement and/or involvement of the spleen.

•Stage IV

-Stage I(E) disease (usually of the upper aerodigestive tract) plus isolated distant involvement

-Any involvement of the liver or bone marrow

•Extranasal ENKL – All cases of extranasal ENKL (except stage I cutaneous extranasal ENKL) are classified as advanced disease. For patients presenting with non-nasal lesions, it is important to exclude occult nasal involvement (by examination, endoscopy, random biopsies, and imaging, as described above) to substantiate the diagnosis of extranasal ENKL [7]. (See 'Pretreatment evaluation' above.)

PROGNOSIS — The prognostic index of natural killer lymphoma (PINK) and PINK-E systems are useful for assessment of prognosis in patients with ENKL. The CNS-PINK model has been created to predict risk for central nervous system (CNS) recurrence of ENKL. (See 'Prognostic scoring systems' below.)

Assessment of prognosis is an aid in pretreatment discussion of goals of care. (See 'Goals of care' below.)

Prognostic factors — Prognosis is largely related to the location and stage of disease at diagnosis. Numerous studies, using various treatment approaches, have reported that localized disease is associated with better outcomes than advanced disease [8-14]. In general, the prognosis of ENKL, nasal type is superior to that of ENKL, non-nasal type [1]. Some studies have identified performance status, presence of constitutional symptoms, and other clinical or biochemical features as prognostic factors, but these models were generally based on therapies that would not currently be considered optimal [1,14-16].

Other risk factors for inferior outcomes in patients with ENKL include:

●Epstein-Barr virus (EBV) – Levels of EBV in the plasma or bone marrow correlate with patient outcomes and may serve as an indirect measure of tumor burden. As an example, in a prospective study of 69 patients with localized ENKL treated with primary radiation therapy (RT), with or without chemotherapy, patients with pretreatment plasma EBV DNA levels >500 copies/mL had lower estimated three-year progression-free survival (PFS; 52 versus 79 percent) and overall survival (OS; 66 versus 97 percent) compared with patients with lower levels of EBV DNA [17]. After completing therapy, patients with undetectable EBV DNA had higher estimated rates of three-year PFS (78 versus 51 percent) and OS (92 versus 70 percent) than patients with detectable EBV DNA. In a study of patients treated with SMILE chemotherapy, those with high pretreatment plasma EBV DNA levels (>100,000 copies/mL) showed lower response to SMILE, poorer prognosis, and a higher rate of adverse events [18]. Other studies also report an association between EBV biomarkers and adverse outcomes [19-21].

●Other biomarkers – A retrospective study identified elevated serum soluble interleukin-2 receptor as a strong predictor of short OS and PFS [22]. In other studies, expression of cutaneous lymphocyte antigen (CLA) or cyclooxygenase 2 (COX-2), and loss of expression of granzyme B protease inhibitor 9 (PI9) have each been associated with worse outcomes [23-25].

Prognostic scoring systems — Useful prognostic scoring systems for ENKL are:

●Prognostic index of natural killer lymphoma (PINK) – PINK is a prognostic index for ENKL useful for patients treated with non-anthracycline-based chemotherapy [26]. In a retrospective analysis of 527 patients, the following were identified as adverse prognostic factors: age >60 years, stage III or stage IV disease, distant lymph node involvement, and non-nasal type disease. PINK assigns one point for each risk factor and stratified patients into three risk groups, with different rates of three-year OS:

•Low risk (0 point) – 81 percent

•Intermediate risk (1 point) – 62 percent

•High risk (≥2 points) – 25 percent

●PINK-E – EBV DNA (detected by reverse transcriptase polymerase chain reaction [RT-PCR]) was identified as an additional adverse prognostic factor, in a subset of 328 patients in the above study who had data for EBV DNA [26]. PINK-E (for patients with EBV DNA data) stratifies patients into three categories, with different rates of three-year OS:

•Low risk (0 to 1 point) – 81 percent

•Intermediate risk (2 points) – 55 percent

•High risk (≥3 points) – 28 percent

●CNS-PINK – The CNS-specific prognostic index (CNS-PINK) model was developed to assess risk of CNS involvement, based on the extent of extranodal involvement and the PINK score, using a training cohort of 399 patients and a validation cohort of 253 patients [27]. It has not yet been independently validated outside of this study. CNS-PINK is calculated as follows:

•Extranodal involvement:

≤1 site (0 points)

≥2 sites (1 point)

•PINK score:

-Low (0 points)

-Intermediate/high (1 point)

The CNS-PINK score is calculated as the total of the points above:

•CNS-PINK Low risk (0 to 1 point)

•CNS-PINK High risk (2 points)

The international prognostic index (IPI) (table 3), which is commonly used to stratify risk for other types of aggressive non-Hodgkin lymphoma, has only limited application with ENKL.

GOALS OF CARE — Goals of care should be established in discussions between the patient and clinicians at the time of diagnosis and periodically through the course of the disease. Application of one of the ENKL prognostic scoring systems is useful for considering likely outcomes. (See 'Prognostic scoring systems' above.)

Goals of care are influenced by medical fitness, disease stage, prognosis, and individual values and preferences. Assessment of medical fitness and staging of ENKL are discussed above. (See 'Medical fitness' above and 'Staging' above.)

Examples of goals of care for patients with ENKL, according to medical fitness follow:

●Medically fit – For medically fit patients, the goal of care is to achieve long-term disease control with the possibility of cure. This generally requires achievement of a complete response (CR) using intensive chemotherapy and/or radiation therapy and, for some, post-induction hematopoietic cell transplantation (HCT).

Initial management of medically fit patients is discussed below. (See 'Medically fit patients' below and 'Medically fit with advanced disease' below.)

●Less fit – For some less-fit patients, achievement of long-term disease control is a reasonable goal, but for most, treatment to achieve a CR should be a goal in order to relieve symptoms, improve the quality of life, and potentially extend survival.

Initial management of less-fit patients is discussed below. (See 'Less-fit patients' below and 'Less fit with advanced disease' below.)

●Frail – For frail patients, the goal of care is to relieve symptoms and improve the quality of life with supportive care (eg, transfusions, antibiotics, analgesics); other aspects of care of the frail patient are discussed below. (See 'Less-fit patients' below and 'Less fit with advanced disease' below.)

LOCALIZED DISEASE — Localized disease comprises stage I, contiguous stage II nasal type ENKL, and stage I cutaneous extranasal ENKL. (See 'Staging' above.)

Management should take place in a center that has expertise with ENKL, and we encourage participation in a clinical trial, when possible. Treatment should begin promptly, but it is important to first stabilize a patient's condition and control complications. Examples of such complications include hemophagocytic syndrome, uncontrolled infection, or significant bleeding. (See 'Medical emergencies' above.)

Management of localized disease depends on the patient's medical fitness. (See 'Medical fitness' above.)

Medically fit patients — For medically fit patients with localized disease, we suggest combined modality therapy (CMT; ie, chemotherapy plus radiation therapy [RT]), rather than chemotherapy alone or RT alone. Compared with either modality alone, CMT provides superior long-term survival and local disease control with acceptable levels of adverse effects.

Selection of a particular CMT regimen is discussed below. (See 'Choice of CMT protocol' below.)

Numerous retrospective studies indicate that outcomes after CMT are superior to those of patients treated with RT alone or chemotherapy alone. However, no prospective, controlled studies have directly compared these approaches. Furthermore, most studies that compared treatment approaches used suboptimal chemotherapy (eg, anthracycline-containing regimens) and some studies used RT protocols that are no longer considered optimal. Descriptions of preferred chemotherapy and RT approaches are presented below. (See 'Preferred CMT protocols' below and 'Radiation therapy alone' below.)

Adverse effects of CMT primarily reflect toxicity of the treatment components. Grade ≥3 hematologic toxicity is common with chemotherapy (administered alone or as a component of CMT), but fatal febrile neutropenia is uncommon with CMT. The predominant grade ≥3 nonhematologic toxicity associated with CMT is RT-associated mucositis; other severe nonhematologic adverse effects are uncommon.

Examples of informative studies that compared CMT with chemotherapy alone or RT alone include:

●CMT versus chemotherapy alone

•A retrospective study reported that among 143 patients with localized ENKL who received a median of three cycles of various asparaginase-based chemotherapy regimens, addition of post-chemotherapy RT (ie, treatment with CMT) was associated with superior outcomes [28]. Compared with chemotherapy alone, CMT achieved superior rates of complete response (CR; 73 versus 59 percent), two-year overall survival (OS; 90 versus 49 percent), and two-year progression-free survival (PFS; 87 versus 37 percent). Among 59 patients who initially achieved partial response (PR) with chemotherapy, RT converted 29 (49 percent) to CR. Among all patients in the study who achieved CR after chemotherapy, addition of RT was associated with superior two-year OS (91 versus 60 percent). RT was identified as an independent factor associated with better OS and PFS in a multivariate analysis. This study did not report toxicity with the various approaches.

•An uncontrolled, retrospective study reported outcomes in patients who were treated with non-anthracycline-based chemotherapy (ie, asparaginase-, platinum-, or gemcitabine-based regimens) alone or as a component of CMT [29]. Compared with chemotherapy alone, CMT was associated with superior OS (hazard ratio [HR] 0.36; 95% CI 0.24 to 0.54) and PFS (HR 0.38; 95% CI 0.27 to 0.54). Using anthracycline-based regimens, CMT was also superior to anthracycline-based chemotherapy alone; however, anthracycline-based chemotherapy is considered inadequate for treatment of ENKL, as discussed below. (See 'Choice of CMT protocol' below.)

●CMT versus RT alone – In a retrospective study of 105 patients with localized ENKL, outcomes were comparable after CMT (primarily using anthracycline-based chemotherapy) versus RT alone [9]. Comparing CMT versus RT, there were no significant differences in rates of CR (85 versus 83 percent, respectively), five-year OS (76 versus 66 percent), and five-year PFS (61 percent for both).

Choice of CMT protocol — Selection of a CMT protocol for medically fit patients with localized ENKL is influenced by prompt access to initiation of RT:

●When RT can begin within two weeks, we suggest CMT using an RT-first approach (ie, either concurrent chemoradiation [CCRT] or RT-first sequential CMT), rather than chemotherapy-first CMT protocols (ie, either chemotherapy-first sequential CMT or "sandwich therapy," in which chemotherapy precedes and follows RT). No well-controlled prospective studies or randomized trials have directly compared various CMT protocols, but retrospective and small prospective studies generally report similar toxicity profiles but superior outcomes using CCRT or RT-first sequential CMT compared with chemotherapy-first approaches.

Preferred RT-first CMT protocols are described below. (See 'Preferred CMT protocols' below.)

For localized ENKL, treatment with CCRT or RT-first sequential CMT is associated with ≥80 percent CR and long-term survival in at least two-thirds of patients; these outcomes are generally superior to outcomes associated with chemotherapy-first CMT protocols. As an example, a retrospective study of 65 patients with localized ENKL compared RT-first CMT versus chemotherapy-first CMT; superior outcomes were associated with RT-first CMT, including five-year OS (90 versus 49 percent, respectively) and five-year disease-free survival (DFS; 79 versus 40 percent) [30]. Grade ≥3 hematologic toxicity is common with all CMT protocols and radiation-associated mucositis is the predominant serious nonhematologic adverse effect.

●For settings where RT cannot begin within two weeks, we consider chemotherapy-first sequential CMT or sandwich CMT regimens to be acceptable alternatives. (See 'Alternative CMT approaches' below.)

Preferred CMT protocols — When RT can begin within two weeks, we consider either CCRT or RT-first sequential CMT to be acceptable, as described above. (See 'Choice of CMT protocol' above.)

Localized disease frequently presents as a small mass confined to the nasal cavity, and early RT generally produces a rapid response. RT should be delivered as involved-site RT (ISRT), rather than using larger fields [31]. For patients who are treated with curative intent, we favor highly conformal dose distribution (eg, intensity-modulated RT or image-guided RT) to reduce exposure of adjacent tissues and limit long-term complications.

No specific RT-first regimen is clearly superior, and the preferred approach varies by institution. We consider the following RT-first CMT approaches to be acceptable:

●CCRT with reduced dose DeVIC – For CCRT, we favor ISRT 50 to 50.4 gray (Gy) with three concurrent cycles of reduced dose DeVIC. Treatment with reduced (ie, two-thirds) dose DeVIC includes dexamethasone 40 mg/day intravenously (IV) on days 1 to 3, etoposide 67 mg/m² IV over 2 hours on days 1 to 3, ifosfamide 1 g/m² IV over 3 hours on days 1 to 3, and carboplatin 200 mg/m² IV over 30 minutes on day 1 [32].

A prospective study of 33 patients treated with RT and concurrent DeVIC reported 78 percent CR; five-year PFS and OS were 67 and 73 percent, respectively [32,33]. Grade ≥3 neutropenia was nearly universal, but there were no episodes of febrile neutropenia or infections; anemia and thrombocytopenia were reported in 15 and 11 percent, respectively. The most common grade ≥3 nonhematologic toxicity was radiation-associated mucositis in 30 percent. RT with concurrent DeVIC was associated with five-year PFS and OS rates of 61 and 72 percent, respectively, in another retrospective study of 169 patients with localized ENKL [22].

●CCRT with cisplatin, followed by VIDL – An acceptable alternative CCRT regimen is ISRT 40 to 44 Gy with concurrent cisplatin (30 mg/m² IV weekly for four weeks), followed by two cycles of VIDL (etoposide 100 mg/m² IV on days 1 to 3, ifosfamide 1200 mg/m² IV on days 1 to 3, dexamethasone 40 mg days 1 to 3, and L-asparaginase 4000 international units [IU]/m² every other day from days 8 to 20) [34]. This regimen was associated with 87 percent CR and estimated five-year PFS and OS were 73 and 60 percent, respectively [34]. Grade ≥3 hematologic toxicity was frequent during VIDL chemotherapy, L-asparaginase toxicity was manageable, and there were no treatment-related deaths.

Studies using other CCRT-based regimens have reported similar outcomes [35-38].

●RT-first sequential CMT with m-SMILE – An acceptable RT-first CMT approach is RT 45 Gy followed by two to three cycles of modified SMILE (m-SMILE; methotrexate 2 g/m² IV over 6 hours on day 1, leucovorin 15 mg x 4 IV or orally on days 2 to 4, ifosfamide 1500 mg/m² IV on days 2 to 4, mesna 300 mg/m² x 3 IV on days 2 to 4, dexamethasone 40 mg/day IV or orally on days 2 to 4, etoposide 100 mg/m² IV on days 2 to 4, and pegaspargase 1500 to 2500 IU/m² intramuscularly [IM] or IV on day 8) [39]. A retrospective study that included 11 patients with localized ENKL who received m-SMILE (nine received two cycles and two received three cycles) reported that, with a median follow-up of two years, all were alive with no evidence of disease; one patient progressed 20 months after the start of therapy at an adjacent site [39].

Anthracycline-based chemotherapy, such as CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) or related regimens, is inadequate for treatment of ENKL. A retrospective multicenter study that included 2226 patients with localized ENKL reported that, compared with patients who received anthracycline-based CMT, patients treated with a variety of non-anthracycline-based CMT regimens (mostly asparaginase-based, but some platinum-based or gemcitabine-based regimens) had superior five-year OS (77 versus 65 percent; HR 0.64; 95% CI 0.54 to 0.77) and five-year PFS (68 versus 54 percent; HR 0.65; 95% CI 0.56 to 0.76) [29]. Other uncontrolled studies also reported that outcomes were superior when ENKL is treated with non-anthracycline-based CMT regimens [32,34,35,39,40].

Alternative CMT approaches — When RT cannot begin within two weeks, we treat with one of the following chemotherapy-first CMT protocols. No specific regimen is clearly superior, and the preferred approach varies by institution. For these protocols, when the initial chemotherapy yields a CR, the RT clinical target volume should include the pretreatment gross tumor.

●Chemotherapy-first sequential CMT with SMILE – The SMILE regimen is similar to m-SMILE (described above), but it includes seven doses of L-asparaginase per cycle instead of pegaspargase [41]. (See 'Preferred CMT protocols' above.)

In a multicenter retrospective study of 29 patients with localized disease, chemotherapy-first CMT with SMILE chemotherapy (median, four cycles, followed by RT) was associated with 75 percent overall response (OR) and 65 percent CR [42]. With median follow-up of 31 months, newly diagnosed patients had estimated 47 percent five-year OS and 60 percent four-year DFS. Adverse effects of SMILE included grade ≥3 neutropenia and thrombocytopenia in 67 and 42 percent, respectively.

●Sandwich therapy with P-GEMOX – Sandwich therapy with P-GEMOX includes two cycles of P-GEMOX (pegaspargase 2500 IU/m² IM on day 1, gemcitabine 1250 mg/m² IV on day 1, oxaliplatin 85 mg/m² IV on day 1, every two weeks), followed by ISRT 50 to 50.4 Gy, followed by two to four additional cycles of P-GEMOX [43]. A single-center study of 38 patients (that included all stages of disease) treated with P-GEMOX sandwich therapy reported 92 percent OR and 87 percent CR; with median follow-up of 16 months, rates of one-year PFS and OS were both 87 percent. Grade ≥3 toxicity included anemia, thrombocytopenia, and neutropenia in 26 to 41 percent of patients, hypertriglyceridemia in 21 percent, and elevated transaminases in 12 percent of patients; there was one treatment-related death.

Less-fit patients — Less-fit status is based on performance status and medical comorbidities, not age, per se, as discussed above. (See 'Medical fitness' above.)

Choices for less-fit patients with localized disease — There is no consensus regarding optimal management of the less-fit patient with localized disease. Less-fit patients may be able to tolerate RT with or without chemotherapy. Treatment should be individualized, with the choice of therapy informed by medical comorbidities, adverse effects, and patient values and preferences.

We generally offer sequential CMT, as we consider CCRT too toxic in this setting. Some patients are better suited for lower-intensity chemotherapy regimens or RT alone.

We consider the following to be acceptable for less-fit patients with localized disease:

●RT followed by dose reduced DeVIC (dexamethasone, etoposide, ifosfamide, carboplatin) [22,32,33].

●RT followed by VIDL (etoposide, ifosfamide, dexamethasone, L-asparaginase) [34].

●Two or three cycles of m-SMILE (dexamethasone, methotrexate, ifosfamide, pegaspargase, etoposide) followed by RT [39].

●Lower-intensity chemotherapy alone. (See 'Less fit with advanced disease' below.)

●RT alone. (See 'Radiation therapy alone' below.)

No controlled prospective studies have directly compared treatments in this setting and no single approach is clearly superior. Options for less-fit patients are presented in the following sections.

Details of the chemotherapy regimens used for CMT are presented above. (See 'Preferred CMT protocols' above.)

Chemotherapy alone or with RT — Some less-fit patients may be able to tolerate chemotherapy, either alone or in sequence with RT. The choice of treatment in this setting must be individualized, based on adverse effects, comorbid illness, and patient values and preferences. No prospective controlled studies have directly compared various regimens in this setting and no single approach is clearly superior.

Some less-fit patients may be able to tolerate dose-reduced DeVIC, VIDL, m-SMILE, or P-GEMOX alone, or in sequence with RT. For other less-fit patients, lower-intensity regimens, such as GDP (gemcitabine, dexamethasone, cisplatin) or AspaMetDex (L-asparaginase, methotrexate, dexamethasone), may be more suitable and better tolerated. Details of chemotherapy regimens, outcomes, and toxicity are presented separately. (See 'Choice of CMT protocol' above and 'Less fit with advanced disease' below.)

Importantly, we consider anthracycline-based regimens (eg, CHOP and related regimens) inadequate for treatment of ENKL, as described below. (See 'Less fit with advanced disease' below.)

Radiation therapy alone — RT alone is acceptable for patients with localized ENKL who are not medically fit to receive chemotherapy. RT can also provide symptomatic relief for some less-fit or frail patients with advanced disease.

For RT alone, we treat with ISRT 50 to 55 Gy to the gross tumor volume with a sufficient margin, using methods to limit treatment volumes to the region of involvement [8,44]. The clinical target volume should encompass the entire nasal cavity, nasopharynx, and the entire involved region (as defined by magnetic resonance imaging [MRI] and computed tomography [CT]), with expansion to sinuses that were partially involved, all adjacent paranasal sinuses, and 0.5 to 1 cm of surrounding soft tissue [44]. Contemporary techniques, such as intensity-modulated RT, are commonly used, but there has been no evidence demonstrating their superiority to three-dimensional conformal radiation [32,34].

RT to localized disease generally achieves 60 to 80 percent OR, 40 to 80 percent CR, and 40 to 50 percent five-year OS [6,45]. Approximately half of patients relapse; in-field failures are primarily due to inadequate dose (ie, <50 Gy) and/or inadequate imaging to fully define the extent of disease [46]. Systemic failure occurs in 25 to 50 percent, indicating that some patients with apparently localized disease probably had more advanced disease [40,45].

No well-controlled studies have directly compared various RT techniques and doses, RT alone versus chemotherapy alone, or RT versus CMT. Uncontrolled retrospective studies have reported that RT alone achieved superior outcomes compared with chemotherapy alone, but these studies generally utilized anthracycline-based regimens, which are now considered suboptimal therapy [1,9,30,45,47-50]. As an example, for patients with localized ENKL, compared with anthracycline-based chemotherapy, RT achieved superior three-year OS (57 versus 30 percent) [1].

A retrospective study of 1273 patients with localized ENKL reported that outcomes with RT alone were inferior for patients with one or more of the following risk factors: age >60 years, Eastern Cooperative Oncology Group (ECOG) performance status >1, stage II disease, elevated serum lactate dehydrogenase (LDH), and primary disease extending into neighboring structures and/or organs [51]. Estimated five-year OS rates for patients with adverse features versus those without adverse features were 60 versus 89 percent, respectively. Importantly, nearly half of patients who present with ENKL are >60 years of age and would therefore be classified as high risk [22,52].

RT as a component of CMT is described above. (See 'Medically fit patients' above.)

ADVANCED DISEASE — Advanced disease comprises stage III or stage IV nasal type ENKL and all stages of extranasal ENKL, except stage I cutaneous extranasal ENKL. (See 'Staging' above.)

Management should take place in a center that has expertise with ENKL, and we encourage participation in a clinical trial, when possible. Treatment should begin promptly, but it is important to first stabilize a patient's condition and control complications. Examples of such complications include hemophagocytic syndrome, uncontrolled infection, or significant bleeding. (See 'Medical emergencies' above.)

Treatment of advanced disease is informed by medical fitness, as described in the following sections.

Medically fit with advanced disease — For medically fit patients with advanced disease, we suggest treatment with an intensive asparaginase-based chemotherapy regimen rather than other chemotherapy approaches. Compared with less intensive chemotherapy regimens or non-asparaginase-based approaches (eg, anthracycline-based), treatment with intensive asparaginase-based chemotherapy achieves superior outcomes with acceptable levels of toxicity.

No specific intensive asparaginase-based chemotherapy regimen provides clearly superior outcomes for advanced-stage ENKL, and all are associated with substantial adverse effects. However, compared with anthracycline-based regimens, non-anthracycline-based regimens (eg, asparaginase-, platinum-, or gemcitabine-based regimens) are associated with superior outcomes, as discussed below.

Preferred regimens vary by institution:

●SMILE or modified SMILE – SMILE and modified SMILE (m-SMILE) use the same core chemotherapy protocol but are distinguished by use of L-asparaginase versus pegaspargase (pegylated asparaginase), respectively; m-SMILE is more convenient to administer and generally better tolerated because of the use of pegaspargase. Importantly, SMILE is generally not suitable for patients ≥70 years of age and/or lymphocyte count <500/microL due to the risk of severe bone marrow suppression and infection; m-SMILE is acceptable for these categories of patients.

The core SMILE regimen consists of 28-day cycles of methotrexate 2 g/m² intravenously (IV) over 6 hours on day 1, leucovorin 15 mg x 4 IV or orally on days 2 to 4, ifosfamide 1500 mg/m² IV on days 2 to 4, mesna 300 mg/m² x 3 IV on days 2 to 4, dexamethasone 40 mg/day IV or orally on days 2 to 4, etoposide 100 mg/m² IV on days 2 to 4, and granulocyte-colony stimulating factor (G-CSF) beginning on day 6 and continuing until recovery beyond the neutrophil nadir [41]. SMILE includes seven doses of L-asparaginase (6000 units [U]/m² IV days 8, 10, 12, 14, 16, 18, 20) per cycle, whereas m-SMILE includes pegaspargase 2500 U/m² intramuscularly (IM) on day 8 of each cycle and is repeated every three weeks.

•Treatment with SMILE achieved 79 percent overall response (OR) and 45 percent complete response (CR) in 28 patients with advanced ENKL; one-year progression-free survival (PFS) and overall survival (OS) were 53 and 55 percent [41]. Toxicity included grade ≥3 neutropenia in 100 percent of patients (despite use of G-CSF) and anemia and thrombocytopenia in 50 and 64 percent, respectively. Grade ≥3 nonhematologic adverse events included elevated transaminases or hyponatremia in approximately one-third of patients, and 61 percent grade ≥3 infections, including two fatal infections.

In a study of 87 patients with newly-diagnosed or relapsed ENKL treated with SMILE, rates of OR and CR were 81 and 66 percent, and five-year OS was 50 percent [42].

•In a single-institution retrospective study of 43 patients, compared with CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone)-like regimens, m-SMILE achieved superior rates of CR (80 versus 30 percent) [39].

●P-GEMOX – P-GEMOX consists of gemcitabine 1250 mg/m² IV, oxaliplatin 85 mg/m² IV, and pegaspargase 2500 international units (IU)/m², all on day 1 and repeated every two weeks; an alternative P-GEMOX protocol is gemcitabine 1000 mg/m² IV on days 1 and 8, oxaliplatin 130 mg/m² IV, and pegaspargase 2500 IU/m² IM on day 1 and repeated every three weeks [43,53].

A retrospective report of 117 patients treated with P-GEMOX for newly diagnosed or relapsed/refractory ENKL reported 89 percent OR; with median follow-up of 17 months, three-year OS and PFS were 73 and 58 percent, respectively [53]. Severe (grade ≥3) toxicity includes severe anemia, thrombocytopenia, and neutropenia in 26 to 41 percent of patients, elevated transaminases, and hypertriglyceridemia.

●DDGP – The DDGP regimen consists of 21-day cycles of pegaspargase 2500 IU/m² IM on day 1, gemcitabine 800 mg/m² IV on days 1 and 8, cisplatin 20 mg/m² IV on days 1 to 4, and dexamethasone 15 mg/m² IV on days 1 to 5 [54].

•Treatment with DDGP achieved 95 percent OR and 71 percent CR in 21 patients with advanced stage ENKL; with median follow-up of 14 months, rates of one-year OS and PFS were 90 and 86 percent, respectively, and two-year OS was 86 percent [54]. Grade ≥3 anemia, thrombocytopenia, and neutropenia were reported in 10 to 15 percent each and 5 percent had grade ≥3 nausea and vomiting.

Preliminary data presented in abstract form only, with 42 month follow-up of 40 patients, reported that DDGP achieved 90 percent OR, 68 percent CR, 57 percent three-year PFS, and 74 percent five-year OS [55]. Details of the study design are unclear, but a comparison of DDGP versus SMILE in this study is invalidated by an unconventional staging system and important differences in administration of SMILE that exaggerated the toxicity of that regimen.

Importantly, CMT regimens such as those described above are associated with improved survival compared with anthracycline-based regimens, such as CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) or related protocols. In an uncontrolled retrospective study that included 334 patients with advanced disease, compared with anthracycline-based regimens, non-anthracycline-based chemotherapy was associated with superior OS (HR 0.70; 95% CI 0.51-0.92) and PFS (HR 0.67; 95% CI 0.50-0.86) [29]. The relative resistance to CHOP-like regimens may be attributable, at least in part, to high level expression by NK cells of the multidrug resistant (MDR) P-glycoprotein, which efficiently exports anthracyclines [56].

Less fit with advanced disease — Less-fit patients generally cannot tolerate intensive chemotherapy regimens. Importantly, less-fit status is based on functional status and is not based on age, per se, as discussed above. (See 'Medical fitness' above.)

Because of the adverse prognosis and limited therapeutic options in this setting, treatment is unlikely to achieve long-term disease control. Management should be individualized, based on the severity and nature of symptoms, medical comorbidities, adverse effects of therapy, and patient values and preferences. Treatment should aim to relieve symptoms and possibly extend survival, while minimizing treatment-related adverse effects.

Treatment options in this setting include lower-intensity asparaginase-based chemotherapy or non-asparaginase-based regimens. We consider the following to be acceptable options for less-fit patients with advanced ENKL:

●AspaMetDex – The AspaMetDex regimen consists of L-asparaginase 6000 IU/m² IM on days 2, 4, 6, 8; methotrexate (3 g/m² IV on day 1); and oral dexamethasone (40 mg day 1 to 4) in 21-day cycles [57]. For patients >70 years, methotrexate and dexamethasone doses should be reduced to 2 g/m² and 20 mg on days 1 to 4, respectively. AspaMetDex treatment of 19 patients with either relapsed or refractory ENKL achieved 78 percent OR and 61 percent CR, with a 12-month median duration of response. The most common grade ≥3 adverse events were neutropenia (44 percent), anemia (22 percent), and hepatitis (17 percent).

●GDP – The GDP regimen consists of gemcitabine 1000 mg/m² IV over 30 minutes on days 1 and 8, dexamethasone 20 mg/day orally on days 1 to 4 and days 11 to 14, and cisplatin 25 mg/m² IV over 60 minutes on days 1 to 3. Treatment of 41 patients with newly-diagnosed advanced disease or relapsed/refractory disease was associated with 83 percent OR; after median follow-up of 16 months, one-year PFS and OS were 73 and 55 percent, respectively [58]. Toxicity included grade ≥3 neutropenia (34 percent) and thrombocytopenia (20 percent); there was no grade ≥3 nonhematologic toxicity.

We suggest not using CHOP or related anthracycline-based regimens because they are less efficacious than non-anthracycline-based chemotherapy (eg, asparaginase-, platinum-, or gemcitabine-based) regimens for ENKL. In an uncontrolled retrospective study that included 334 patients with advanced disease, compared with anthracycline-based regimens, non-anthracycline-based chemotherapy was associated with superior OS (HR 0.70; 95% CI 0.51-0.92) and PFS (HR 0.67; 95% CI 0.50-0.86) [29]. (See 'Chemotherapy alone or with RT' above.)

Some less-fit and frail patients with advanced disease may benefit from palliative RT to symptomatic sites. Treatment with RT is discussed above. (See 'Radiation therapy alone' above.)

CENTRAL NERVOUS SYSTEM MANAGEMENT — The risk of central nervous system (CNS) recurrence can be assessed with the CNS-PINK model. (See 'Prognostic scoring systems' above.)

There is no clear evidence that prophylaxis can reduce the rate of CNS recurrence in ENKL. We suggest not administering CNS prophylaxis for patients treated with methotrexate-containing regimens (eg, SMILE, m-SMILE) [15,16].

A study of 399 patients treated with non-anthracycline-based chemotherapy reported that risk for CNS relapse was increased in association with intermediate or high risk prognostic index of NK lymphoma (PINK) scores and extranodal involvement at ≥2 sites [27]. SMILE-like regimens with intermediate-dose methotrexate (eg, ≥2 g/m²) were associated with a lower incidence rate of CNS relapse than other regimens that did not include intermediate-dose methotrexate.

For patients treated with radiation alone, we suggest establishing a sufficient margin of radiation therapy field to the side of the CNS to lessen the risk of direct invasion of the brain.

Evaluation of the patient with neurologic findings and management of CNS involvement is discussed separately. (See "Secondary central nervous system lymphoma: Clinical features and diagnosis" and "Secondary central nervous system lymphoma: Treatment and prognosis".)

POST-INDUCTION MANAGEMENT — Following completion of induction therapy, the response to treatment should be assessed and a decision made regarding whether to proceed with post-induction consolidation therapy.

Monitoring after completion of therapy is discussed below. (See 'Monitoring' below.)

Response assessment — The response to initial therapy should be evaluated to determine if the patient has achieved complete response (CR), partial response (PR), or remains in no response (NR).

Approximately one month after the completion of planned therapy (or sooner if the outcome is unfavorable), the response to treatment should be documented by history, physical examination, flexible nasal endoscopic examination, laboratory studies, and imaging:

●Laboratory studies – Complete blood count (CBC), lactate dehydrogenase (LDH), and biochemical profile. Quantitative polymerase chain reaction (PCR) for Epstein-Barr virus (EBV) DNA should be repeated [17,19].

●Flexible nasal endoscopy – Flexible nasal endoscopy should include biopsies of any suspicious lesions and biopsies of random, normal-appearing mucosa.

●Imaging

•Computed tomography (CT) and/or magnetic resonance imaging (MRI) should be repeated for previously involved sites.

•Positron emission tomography (PET) should be obtained 6 to 8 weeks after completion of chemotherapy and 12 weeks after the completion of radiation therapy [59]. PET provides information on the size and activity of residual masses and allows for the distinction between active disease and fibrosis or chronic sinusitis.

It should be recognized that Waldeyer's ring and certain extranodal sites have high physiologic uptake; as a result, the PET response should be compared with adjacent normal tissue, rather than compared with liver. For patients with Deauville 4 to 5 score, we suggest biopsy of fluorodeoxyglucose (FDG)-avid sites to distinguish persistent/progressive disease from inflammatory conditions.

Based on all of the above studies and using the Lugano response criteria for non-Hodgkin lymphoma (table 4), we categorize the response to initial treatment as:

●Complete response (CR) – No evidence of residual disease based on clinical evaluation, laboratory studies, nasal endoscopy, and imaging, including PET Deauville score 1 to 3.

●Partial response (PR) – Response ≥50 percent in all measurable disease or PET Deauville score 4 to 5 (but with reduced FDG uptake compared with baseline).

●No response (NR) – Response <50 percent in measurable disease or PET score 4 to 5 (with no significant decrease in FDG uptake compared with baseline).

For patients who achieve CR or PR, post-induction management is described below. (See 'Post-induction consolidation therapy' below.)

For patients who did not achieve at least a PR, management of refractory disease is described below. (See 'Relapsed or refractory disease' below.)

Post-induction consolidation therapy — Decisions regarding post-induction management are informed by the initial disease stage, response to induction therapy, medical fitness, and individual values and preferences.

For patients in whom hematopoietic cell transplantation (HCT) is a consideration, we arrange early referral to a transplant specialist to discuss eligibility, seek potential donors (if allogeneic HCT is selected), and consider the potential benefits and risks of HCT. (See "Determining eligibility for allogeneic hematopoietic cell transplantation" and "Determining eligibility for autologous hematopoietic cell transplantation".)

There is no firm consensus regarding optimal post-induction management of ENKL and no well-controlled prospective studies or randomized trials to guide decision-making. Our approach to post-induction management follows:

●Localized disease in CR – For patients with localized disease who achieve CR, we suggest observation rather than HCT, consolidation chemotherapy, or immunotherapy. Long-term benefits for patients with ENKL have not been proven for consolidation chemotherapy or immunotherapy, and the favorable outcomes in this setting outweigh adverse effects of HCT. Outcomes for treatment of localized disease are presented above. (See 'Preferred CMT protocols' above.)

●Localized disease with PR

•Medically fit – For patients who achieve PR after localized disease and are medically fit for transplantation, we suggest HCT rather than consolidation chemotherapy, immunotherapy, or observation alone. HCT is the only post-induction treatment in this setting that has been shown to achieve long-term disease control and this benefit outweighs transplant-related toxicity; long-term benefits of other approaches for consolidation have not been demonstrated.

•Less fit – We encourage participation in a clinical trial for patients who are not eligible for HCT. Outside of a trial, there is no consensus regarding a preferred approach, and we offer chemotherapy, immunotherapy, or observation according to individual values and preferences. Potential treatment options are discussed below. (See 'Relapsed or refractory disease' below.)

●Advanced disease – For patients with advanced disease who achieve either CR or PR and are medically fit for transplantation, we suggest HCT rather than further chemotherapy, immunotherapy, or observation alone. The potential for HCT to achieve long-term disease control outweighs transplant-related toxicity in this adverse prognosis setting; long-term benefits of the other approaches are not well-defined.

●Refractory disease – For patients with localized or advanced disease that was refractory to induction therapy, we encourage participation in a clinical trial. For others, the decision to treat refractory disease versus provide supportive care to palliate symptoms should be individualized, based on medical fitness and individual values and preferences. (See 'Relapsed or refractory disease' below.)

Monitoring of the patient after completion of treatment is described below. (See 'Monitoring' below.)

HEMATOPOIETIC CELL TRANSPLANTATION — Hematopoietic cell transplantation (HCT) is used as post-induction consolidation therapy for selected patients with ENKL and as treatment for refractory or relapsed disease:

●Selection of patients for HCT as post-induction consolidation therapy is discussed above. (See 'Post-induction consolidation therapy' above.)

●HCT for treatment of relapsed or refractory disease is discussed separately. (See 'Post-induction consolidation therapy' above and 'Relapsed or refractory disease' below.)

The choice of autologous versus allogeneic HCT varies by institution and there are no clear data to guide this decision. This choice may be informed by comorbid conditions, availability of an acceptable allogeneic donor, performance status, available social and medical support, and patient values and preferences.

Autologous HCT is an acceptable option for ENKL because marrow involvement is uncommon. For allogeneic HCT, there is no consensus regarding a preferred conditioning regimen or choice of graft source (ie, peripheral blood versus bone marrow). The American Society for Blood and Marrow Transplantation has published suggestions for the use of HCT in ENKL [60].

Studies that reported outcomes with HCT for ENKL include:

●Allogeneic HCT – Most reports of allogeneic HCT for ENKL are from uncontrolled retrospective studies:

•In a Center for International Blood and Marrow Transplant Research (CIBMTR) report of 82 patients with ENKL, 30 percent received upfront allogeneic HCT and 60 percent underwent HCT at a later time point [61]. Rates of three-year progression-free survival (PFS) and overall survival (OS) were 28 and 34 percent, respectively. There were no differences in PFS, OS, or recurrence regarding the timing of HCT (upfront versus later), disease status (ie, complete response [CR] versus partial response [PR] versus no response [NR]), disease risk, or race (Asian compared with White populations).

•In a retrospective analysis of 18 adults (<60 years) with suitable donors who underwent allogeneic HCT, five-year estimated OS and PFS were 57 and 51 percent, respectively; at the time of transplantation, nine patients were in first CR, seven in second CR, one with PR, and one had progressive disease [62].

•Toxicity associated with allogeneic HCT for ENKL is considerable. In a study of 40 patients who underwent allogeneic HCT, transplant-related mortality was observed in one-quarter and another quarter had disease progression [63].

•A retrospective report of 12 patients with refractory ENKL treated with allogeneic HCT reported that outcomes were associated with disease status at the time of transplantation [64]. With median follow-up of 13 months, seven of the eight patients with a CR or PR at the time of transplant were alive without disease recurrence, whereas all patients with progressive disease at the time of transplant died of disease progression.

●Autologous HCT – No well-controlled prospective studies have directly compared post-induction consolidation HCT versus observation for ENKL.

•A survival benefit for HCT was suggested by a retrospective analysis that compared 47 patients who underwent autologous HCT (using various conditioning regimens) versus 107 historical controls matched for prognostic factors and disease status at the time of HCT [65]. With nine-year follow-up, the median survival had not been reached for patients who had undergone HCT but was 43.5 months for the historical controls. For patients who achieved CR, patients who underwent HCT had higher five-year disease-specific survival compared with controls (87 versus 68 percent). Among patients not in CR, HCT was associated with a trend towards improved one-year disease-specific survival (67 versus 29 percent). Treatment-related mortality occurred in 9 percent of transplanted patients; one patient died with sepsis, one died with pneumonia, and two causes were unspecified. Conclusions from this study are limited because the patient groups were not well-matched for chemotherapeutic regimens or timing of radiotherapy.

•A study that included 36 patients with localized disease and 21 with advanced disease reported no apparent difference in long-term outcomes between patients who underwent autologous HCT versus historical controls who received no post-induction treatment [66,67].

MONITORING — Surveillance for relapse and complications of treatment should continue lifelong.

Some patients may require reconstructive surgery after completing treatment. Bony structures should be evaluated to determine whether reconstructive surgery is needed [68]. Examples include a collapsed orbital floor leading to visual impairment and a perforated hard palate resulting in difficulties in eating. (See "Mandibular and palatal reconstruction in patients with head and neck cancer".)

Patients should be seen periodically to monitor for treatment complications and possible relapse. We generally schedule visits every two months during the first two years, every six months starting two years after complete response, and annually after year 5. However, the follow-up schedule and protocol should be individualized based on clinical status and the comfort of both the patient and clinician. No controlled studies have compared schedules and protocols for follow-up.

At each visit we perform the following:

●History and physical examination – Relapsed or progressive disease can present with B symptoms (ie, fever, night sweats, weight loss), cytopenias, an extranodal mass, or as enlargement of the lymph nodes, liver, or spleen.

●Laboratory studies – Complete blood count (CBC), serum chemistries and lactate dehydrogenase (LDH), and plasma Epstein-Barr virus (EBV) DNA by quantitative polymerase chain reaction (PCR).

●Bone marrow examination should be performed only if there is concern raised by cytopenias or other hematologic complications.

●Imaging – We obtain magnetic resonance imaging (MRI) and/or computed tomography (CT) of the head every six months for the first two years, then once a year at year 4 and year 5. We only obtain additional positron emission tomography/computed tomography (PET/CT) scans to evaluate concerning findings from CT or MRI. When imaging studies suggest possible relapse, a biopsy should be performed to confirm the diagnosis before proceeding to salvage therapy.

We do not obtain routine imaging after year 5, unless clinical or laboratory findings suggest the possibility of relapse. When planning the post-treatment surveillance strategy, care should be taken to limit the number of CT scans, particularly in younger individuals, given concerns about radiation exposure and the risk for second malignancies. (See "Radiation-related risks of imaging".)

RELAPSED OR REFRACTORY DISEASE — We encourage participation in a clinical trial for patients with relapsed or refractory ENKL, whenever possible.

There is no preferred treatment for patients who had an inadequate response to induction therapy (ie, refractory disease) or disease that relapses after an initial response. Approximately one-quarter of patients with localized disease and more than three-quarters of patients with advanced disease will either not achieve remission or will relapse.

Outside of a trial, our approach is informed by medical fitness:

Eligible for HCT — Hematopoietic cell transplantation (HCT) is the only approach that has been proven to provide long-term disease control in patients with relapsed or refractory ENKL.

For patients who are medically fit for transplantation, we generally treat with salvage chemotherapy, followed by HCT in those who achieve a complete response (CR) or partial response (PR).

The preferred salvage chemotherapy regimen varies by institution. Our approach follows:

●We favor asparaginase-based chemotherapy for patients who did not receive such treatment as induction therapy; examples are discussed above. (See 'Medically fit patients' above.)

●For other patients, acceptable treatments include:

•Combination chemotherapy regimens, such as DHAP (dexamethasone, cisplatin, cytarabine), ESHAP (etoposide, methylprednisolone, cytarabine, cisplatin), GDP (gemcitabine, dexamethasone, cisplatin), GemOx (gemcitabine, oxaliplatin), and ICE (ifosfamide, carboplatin, etoposide). (See "Treatment of relapsed or refractory peripheral T cell lymphoma", section on 'Traditional regimens'.)

•Single agents, such as pralatrexate and brentuximab vedotin (for CD30+ disease) [69].

Suitable salvage regimens are discussed separately. (See "Treatment of relapsed or refractory peripheral T cell lymphoma", section on 'Chemotherapy'.)

●Immune checkpoint inhibitors (ICIs) have activity against relapsed or refractory ENKL. It should be noted that for some other types of lymphoma, treatment with pembrolizumab prior to allogeneic HCT was associated with high rates of transplant-related complications, but the risk in patients with ENKL is presently unknown. (See "Toxicities associated with immune checkpoint inhibitors", section on 'Combining or sequencing immunotherapy with other therapies'.)

Not eligible for HCT — For patients who are not medically fit for transplantation, do not achieve at least a PR to second-line chemotherapy regimens, or relapse after HCT, the goal of further management is palliation. We favor therapies that will provide symptom relief while minimizing treatment-related adverse effects.

Outside of a clinical trial, we generally treat patients with relapsed or refractory ENKL using pegaspargase-based combination chemotherapy, if they did not receive such a regimen during induction therapy. For others, platinum-based chemotherapy or ICIs are active for relapsed or refractory disease. Examples of treatment for patients who are not eligible for HCT include:

●Chemotherapy

•AspaMetDex (L-asparaginase, methotrexate, dexamethasone) – Treatment with AspaMetDex in 19 patients with relapsed or refractory ENKL reported 78 percent overall response rate (ORR) and 61 percent CR [57]. Patients with localized disease received RT. With median follow-up of 26 months, median progression-free survival (PFS) and overall survival (OS) were both one year. The main toxicities were hepatitis, cytopenias, and allergy.

•GDP (gemcitabine, dexamethasone, cisplatin) – A retrospective study of GDP in 41 patients with newly diagnosed stage IV or relapsed/refractory ENKL reported 83 percent ORR and 42 percent CR, with 55 percent one-year PFS after median follow-up of 16 months [58]. The median number of cycles of GDP was six. Grade 3/4 neutropenia was documented in 34 percent of the patients.

•LVD (L-asparaginase, vincristine, dexamethasone) plus radiation therapy (RT) – A retrospective analysis of 45 patients treated with LVD RT (if not previously given) for ENKL that relapsed or was refractory to CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) reported 56 percent CR [70]. With median follow-up of 35 months, three-year OS was 67 percent and there appeared to be a plateau after 16 months. Major toxicities included myelosuppression, liver dysfunction, hyperglycemia, and allergic reaction.

●Immunotherapy – ICIs are effective against ENKL, but the possibility for excessive toxicity in patients who subsequently undergo allogeneic HCT must be considered, as discussed above. (See 'Eligible for HCT' above.)

•Pembrolizumab – A pilot study of single-agent pembrolizumab reported responses in seven patients with ENKL who had relapsed after receiving asparaginase-containing regimens and/or allogeneic HCT [71]. Two patients achieved PR and five achieved CR; all five patients with CR maintained the response after a median of seven cycles of pembrolizumab (range, 2 to 13) and a median follow-up of six months (range, 2 to 10). Grade 2 skin GVHD occurred in one patient with previous allogeneic HCT.

Another study of 14 patients treated with pembrolizumab for relapsed or refractory ENKL reported 43 percent ORR and 36 percent CR [72]. Another study reported 57 percent ORR and 29 percent CR in seven patients with relapsed or refractory ENKL who were treated with pembrolizumab [73].

•Nivolumab – Treatment of three patients with low-dose nivolumab (40 mg/day, two weeks interval) for ENKL that was refractory to SMILE (dexamethasone, methotrexate, ifosfamide, L-asparaginase, etoposide) and platinum reported CR in all three, but two died of infection [74].

●Radiation therapy can provide relief for pain or other symptoms, if needed. (See 'Radiation therapy alone' above.)

CLINICAL TRIALS — Often there is no better therapy to offer a patient than enrollment onto a well-designed, scientifically valid, peer-reviewed clinical trial. Additional information and instructions for referring a patient to an appropriate research center can be obtained from the United States National Institutes of Health (www.clinicaltrials.gov). (See "Treatment of relapsed or refractory peripheral T cell lymphoma", section on 'Clinical trials'.)

SUMMARY AND RECOMMENDATIONS

●Description – Extranodal natural killer (NK)/T cell lymphoma, nasal type (ENKL) is a rare category of non-Hodgkin lymphoma characterized by vascular damage and tissue destruction in the upper aerodigestive tract. Tumor cells are latently infected with Epstein-Barr virus (EBV), and most cases have an NK cell phenotype.

●Evaluation – History/physical examination, laboratory studies, imaging, bone marrow examination plus fiberoptic endoscopy. (See 'Pretreatment evaluation' above.)

●Medical fitness – Medical fitness is stratified by performance status (table 1) and comorbidities, as described above (see 'Medical fitness' above):

•Medically fit

•Less fit

•Frail

●Staging. (See 'Staging' above.)

•Localized – Includes:

-Stage I(E) – Single extranodal site, with or without contiguous involvement of adjacent structures

-Stage II(E) – Stage I(E) plus contiguous cervical lymph node involvement

•Advanced – Noncontiguous stage II(E) disease, stages III or IV, and all ENKL, non-nasal type

●Prognosis

•Prognostic factors – Outcomes are better with localized disease versus advanced disease; ENKL, nasal type versus ENKL, non-nasal type; and lower EBV load. (See 'Prognostic factors' above.)

•Scoring systems – Validated prognostic models include (see 'Prognostic scoring systems' above):

-Prognostic index of natural killer lymphoma (PINK)

-PINK-E (includes Epstein-Barr viral load)

-Central nervous system (CNS)-PINK

●Management of localized disease

•Medically fit – For medically fit patients with localized ENKL, we suggest combined modality therapy (CMT) rather than chemotherapy or radiation therapy (RT) alone (Grade 2C). (See 'Medically fit patients' above.)

-When RT can begin within two weeks, we suggest RT-first approach (either concurrent chemoradiation [CCRT] or RT-first sequential CMT) rather than chemotherapy-first CMT protocols (ie, either chemotherapy-first sequential or "sandwich therapy") (Grade 2C). (See 'Choice of CMT protocol' above.)

Examples of preferred CCRT and RT-first sequential CMT are presented above. (See 'Preferred CMT protocols' above.)

-When RT cannot begin within two weeks, chemotherapy-first sequential CMT or sandwich CMT is acceptable. (See 'Alternative CMT approaches' above.)

•Less fit – Treatment should be individualized, with therapy guided by comorbidities, adverse effects, and patient preference; options include:

-Lower-intensity chemotherapy, alone or in sequence with RT. (See 'Chemotherapy alone or with RT' above.)

-RT alone. (See 'Radiation therapy alone' above.)

●Advanced disease. (See 'Advanced disease' above.)

•Medically fit – For medically fit patients with advanced disease, we suggest asparaginase-based, intensive chemotherapy rather than other regimens (Grade 2C). (See 'Medically fit with advanced disease' above.)

•Less fit – For less-fit patients with advanced disease, treatment is individualized and may include lower-intensity asparaginase-based chemotherapy, non-asparaginase-based regimens, or RT, as described above. (See 'Less fit with advanced disease' above.)

●Post-induction management – Guided by disease stage and response to therapy. (See 'Post-induction management' above.)

•Localized disease

-Complete response (CR) – We suggest observation rather than hematopoietic cell transplantation (HCT), consolidation chemotherapy, or immunotherapy (Grade 2C). (See 'Post-induction consolidation therapy' above.)

-Partial response (PR) – We suggest HCT rather than consolidation chemotherapy, immunotherapy, or observation (Grade 2C).

•Advanced disease with CR or PR – We suggest HCT rather than consolidation chemotherapy, immunotherapy, or observation (Grade 2C).

Patients with stable or progressive disease are treated as refractory disease.

●Relapsed/refractory ENKL – We encourage enrollment in a clinical trial. Management is informed by prior therapy, eligibility for HCT, medical fitness, and patient preference. (See 'Relapsed or refractory disease' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Kensei Tobinai, MD, PhD, who contributed to earlier versions of this topic review.