Allogeneic haematopoietic stem cell transplantation as a promising treatment for natural killer-cell neoplasms


Masahiro Kami, Haematopoietic Stem-Cell Transplantation Unit, National Cancer Centre Hospital, 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan.


The efficacy of allogeneic haematopoietic stem-cell transplantation (allo-HSCT) for natural killer (NK)-cell neoplasms is unknown. We investigated the results of allo-HSCT for NK-cell neoplasms between 1990 and 2003 through questionnaires. After reclassification by a haematopathologist, of 345 patients who underwent allo-HSCT for malignant lymphoma, 28 had NK-cell neoplasms (World Health Organization classification): extranodal NK/T-cell lymphoma (n = 22), blastic NK-cell lymphoma (n = 3), and aggressive NK-cell leukaemia (n = 3). Twelve were chemosensitive and 16 chemorefractory. Twenty-two had matched-related donors. Stem-cell source was bone marrow in eight and mobilised peripheral blood in 20. Conditioning regimens were myeloablative (n = 23) and non-myeloablative (n = 5). Grade 2–4 acute graft-versus-host disease (GVHD) and chronic GVHD developed in 12 and 8 respectively. Eight died of disease progression, three of infection, two of acute GVHD, one of veno-occlusive disease, one of interstitial pneumonitis, and one of thrombotic microangiopathy. Two-year progression-free and overall survivals were 34% and 40% respectively (median follow-up, 34 months). All patients who did not relapse/progress within 10 months achieved progression-free survival (PFS) during the follow-up. In multivariate analysis, stem cell source (BM versus peripheral blood; relative risk 3·03), age (≥40 years vs. <40 years; relative risk 2·85), and diagnoses (extranodal NK/T-cell lymphoma versus others; relative risk 3·94) significantly affected PFS. Allo-HSCT is a promising treatment for NK-cell neoplasms.

Neoplasms of natural-killer (NK)-lineage cells are rare. Although their clinical features and prognostic factors have been characterised in the latest decade, their optimal treatment strategies remain unclear (Jaffe et al, 1996; Chan et al, 1997; Kwong et al, 1997; Kim et al, 2002; Oshimi, 2003; Chim et al, 2004; Li et al, 2004). Notably, the prognosis of NK-cell neoplasms is generally poor except for stage I diseases; the median overall survival (OS) is reportedly 12–38 months for nasal NK/T-cell lymphoma (Kwong et al, 1997; Chim et al, 2004; Li et al, 2004) and 47–58 d for aggressive NK leukaemia (Song et al, 2002; Suzuki et al, 2004).

There is no standard treatment for advanced diseases. Response of nasal NK/T-cell lymphoma to combination chemotherapy, such as CHOP (cyclophosphamide, doxorubicin, vincristine and prednisone), is minimal or transient, and early progression is common. While radiation therapy is effective for this disease, its clinical application is restricted to those with localised diseases (Oshimi, 2003). A delivery of 40–50 Gy to the involved field is required to achieve complete remission (CR; Oshimi, 2003; Chim et al, 2004).

Patients with NK-cell neoplasms who attained initial CR with chemoradiotherapy had better 5-year OS than those who did not attain initial CR (60% vs. 3%; Li et al, 2004). The International Prognostic Index (IPI) was also associated with OS; IPI 0–1 was superior to IPI 2–4 for 20-year OS (57·4% vs. 27·6%)(Chim et al, 2004). However, the treatment effect of allogeneic haematopoietic stem cell transplantation (allo-HSCT) remains unknown because of the rarity of NK-cell neoplasms and the application of allo-HSCT to the disease.

A few reports on allo-HSCT for various haematological malignancies included some patients with NK-cell neoplasms and suggested an efficacy of the treatment. Izutsu et al (2004) included 12 patients with NK-cell neoplasms in their report of 124 patients with different histological subtypes of malignant lymphoma who underwent allo-HSCT from a human leucocyte antigen (HLA)-matched unrelated donor through the Japan Marrow Donor Programme, suggesting the feasibility of allo-HSCT. We conducted a nation-wide study which is, to our knowledge, the largest study on allo-HSCT for NK-cell neoplasms.


Patients and transplantation procedure

The data of Japanese patients who underwent allo-HSCT for lymphoma except adult T-cell leukaemia/lymphoma between 1990 and 2003 were collected by questionnaires. Of the 345 patients from 76 hospitals, 32 were reported to be NK-lineage neoplasms. After pathological review and restaining of the specimens, 28 evaluable patients were included in the analysis. The study was approved by the Institutional Review Board of the National Cancer Center Hospital.

The NK-cell neoplasms were chemosensitive in 12 patients and chemorefractory in 16 (Table I). Eight were in CR at the time of allo-HSCT. Allo-HSCT was performed according to the protocol at each centre and therefore the transplantation procedure varied (Table II). Pretransplant treatment regimens and conditioning regimens were decided at the discretion of the primary physicians at each centre. As pretransplant irradiation was used for local tumour control, it did not contraindicate total body irradiation (TBI) in conditioning regimens. The types of allo-HSCT were conventional myeloablative stem cell transplantation (CST) in 23 patients and reduced-intensity stem cell transplantation (RIST) in five patients. TBI was given to 20 patients. Graft-versus-host disease (GVHD) prophylaxis was ciclosporin and short-term methotrexate in 23 patients. All the patients had received a median of two courses (range, 1–10) of chemotherapy before allo-HSCT, including seven patients with history of high-dose therapy and autologous stem cell transplantation. Radiation was administered to 16 patients. The median time from diagnosis to allo-HSCT was 6·3 months (range, 1·6–53 months).

Table I.  Patient characteristics.
 Total (n = 28)Myeloablative (n = 23)Reduced-intensity (n = 5)
  1. Allo-HSCT, allogeneic haematopoietic stem cell transplantation.

Median age (years)38 (range, 19–54)35 (range, 19–53)52 (range, 40–54)
 Extranodal NK/T-cell lymphoma22175
 Blastic NK-cell lymphoma330
 Aggressive NK-cell leukaemia330
Stage at allo-HSCT
 Complete remission880
 Stage 1–2110
 Stage 3000
 Stage 419145
Chemosensitivity at allo-HSCT
Treatments before allo-HSCT
 Autologous HSCT725
 Median number of pretransplant chemotherapies2 (range, 1–10)2 (range, 1–5)5 (range, 4–10)
Table II.  Conditioning regimens and GVHD prophylaxis.
Conditioning regimensMyeloablative (n = 23)Reduced-intensity (n = 5)
  1. CY, cyclophosphamide; TBI, total body irradiation; AraC, cytarabine; VP-16, etoposide; BU, busulphan; Mel, melphalan; MCNU, ranimustine; CBDCA, carboplatin; TLI, total lymphoid irradiation; ATG, antithymocyte globulin; GVHD, graft-versus-host disease; CSP, ciclosporin; sMTX, short-term methotrexate; PSL, prednisolone; FK, tacrolimus; MMF, mycophenolate mofetil.

 CY/TBI (n = 8)Flu/Mel (n = 2)
 AraC/CY/TBI (n = 5)Flu/BU (n = 1)
 VP-16/CY/TBI (n = 3)Flu/BU/TBI(4 Gy) (n = 1)
 BU/Mel/TBI (n = 3)Flu/BU/ATG (n = 1)
 BU/CY (n = 2) 
 BU/CY/VP-16 (n = 1) 
 AraC/MCNU/CBDCA/CY/TLI (n = 1) 
GVHD prophylaxis

Matched-related donors were available for 22 patients, mismatched-related for four patients, matched-unrelated for one, and mismatched-unrelated for one patient. Stem cell source was bone marrow (BM) in eight and mobilised peripheral blood (PB) in 20 patients.


The pathological data at each hospital including CD3, CD4, CD5, CD8, CD20, CD56, terminal deoxynucleotidyl transferase, and Epstein–Barr virus small RNA (EBER) status were collected for the 32 patients with reported NK-cell neoplasms, and were reviewed by an expert haematopathologist. Twenty-five satisfied the criteria of the World Health Organization Classification of Tumors (Chan et al, 2001a,b, 2001c; Suzuki et al, 2005); two were excluded from the analysis; we collected unstained slides of five patients with insufficient data. Restaining for TIA1+, GranzymeB+, EBER, and other lacking markers excluded two more patients. The diagnoses of the excluded patients were two peripheral T-cell lymphoma, unspecified, one myeloid sarcoma, and one precursor T-cell lymphoblastic lymphoma. The remaining 28 patients were reclassified into the three categories of the World Health Organization Classification of Tumors (Table I): extranodal NK/T-cell lymphoma (n = 22), blastic NK-cell lymphoma (n = 3) and aggressive NK-cell leukaemia (n = 3).

Statistical analysis

The cumulative incidences of progression and progression-free mortality were evaluated using Gray's method, considering each other risk as a competing risk. OS and progression-free survival (PFS) were estimated using the Kaplan–Meier method. Potential confounding factors considered in the analysis were age, sex, donor type (HLA-matched-related donor or alternative donor), stem cell source (BM or peripheral blood stem cells), HLA-mismatch, disease status, conditioning regimen and the development of grade II-IV acute GVHD. To evaluate the influence of these factors on PFS, proportional hazard modelling was used, treating the development of acute GVHD as a time-dependent covariate. Factors associated with at least borderline significance (P < 0·15) in the univariate analyses were subjected to a multivariate analysis using proportional-hazard modelling. P < 0·05 were considered statistically significant.


Progression-free and overall survival

At a median follow-up duration of 34 months among survivors, 12 were alive in CR (n = 10) or with disease (n = 2). The 2-year PFS and OS were 34% [95% confidence interval (CI), 20–58%] and 40% (95%CI, 24–64%) respectively (Figs 1 and 2). Disease progression was observed in 18 patients and the median time from allo-HSCT to disease progression was 1·8 months (range, 0–9·8 months). All the patients who did not relapse/progress within 10 months remained free of progression and all those who survived 10 months remained alive during the follow-up.

Figure 1.

Progression-free survival was 34% at 2 years after allogeneic haematopoietic stem cell transplantation. Broken lines indicate 95% confidence intervals. The median follow-up was 34 months. All the patients who did not relapse/progress within 10 months remained free of progression during the follow-up.

Figure 2.

Overall survival was 40% at 2 years after allogeneic haematopoietic stem cell transplantation. Broken lines indicate 95% confidence intervals. The median follow-up was 34 months. All the patients who survived 10 months remained alive during the follow-up.

Graft-versus-host disease

Three patients who died without developing acute GVHD within 30 d of allo-HSCT were not eligible for the analysis of acute GVHD. Grade 2–4 acute GVHD developed in 12, of whom four had grade 3–4 acute GVHD. Among the 17 patients who survived longer than 100 d, limited and extensive chronic GVHD developed in five and three patients respectively. No patients received donor lymphocyte infusions.

Response to allo-HSCT

Overall response was observed in 15 of 22 (68%) evaluable patients. CR was achieved in 14 including the four patients who were in CR at allo-HSCT (Table III). The overall response rates in CST and RIST groups were 52% and 60% respectively.

Table III.  GVHD and response after allo-HSCT for NK-cell neoplasms.
GVHDTotal (n = 28)Myeloablative (n = 23)Reduced-intensity (n = 5)
  1. GVHD, graft-versus-host disease; allo-HSCT, allogeneic haematopoietic stem cell transplantation; CR, complete remission; PR, partial remission; SD, stable disease; PD, progressive disease.

Acute GVHD (n = 13)
 Grade 2–412102
 Grade 3–4440
Chronic GVHD (n = 9)
Response to allo-HSCT
 CR to CR440
Not evaluable660

Treatment-related mortality

The causes of death were primary disease progression in eight, systemic infection in three, acute GVHD complicated with systemic infection in two, interstitial pneumonitis in one, thrombotic microangiopathy in one, and veno-occlusive disease in one patient (Table IV). The numbers of primary disease progression (n = 8) and TRM (n = 8) were not significantly different. While TRM was 30% among the 23 CST patients, it was 20% among the five RIST patients.

Table IV.  Cause of death (n = 16).
Disease progression8
  1. TRM, treatment-related mortality; GVHD, graft-versus-host disease; TMA, thrombotic microangiopathy.

  2. TRM within 1 year includes TRM within 100 d.

 acute GVHD2
 Veno-occlusive disease1
 Interstitial pneumonitis1
Myeloablative (n = 23) 
 Disease progression8
 TRM within 100 d6
 TRM within 1 year7 (30%)
 Alive8 (35%)
Reduced-intensity (n = 5) 
 TRM within 100 d0
 TRM within 1 year1 (20%)
 Alive4 (80%)

Prognostic factors

In univariate analyses, relative risks for PFS of age (≥40 years vs. <40 years), diagnoses (extranodal NK/T-cell lymphoma versus others), and stem cell source (BM versus PB) were 2·17 (95% CI, 0·84–5·65; P = 0·11), 3·02 (95% CI, 0·69–13·3; P = 0·14), and 2·33 (95% CI, 0·90–6·25; P = 0·081) respectively. Sex, stage at allo-HSCT (stage 3–4 versus stage 1–2), chemosensitivity at allo-HSCT, disease status at allo-HSCT (CR versus non-CR), type of conditioning regimens (RIST versus CST), the use of TBI in conditioning regimens, donors (matched-related versus others), and acute GVHD (grade 2–4 versus grade 0–1) did not affect PFS.

In a multivariate analysis, relative risks for PFS of stem cell source (BM versus PB), age (≥40 years versus <40 years) and diagnoses (extranodal NK/T-cell lymphoma versus others) were 3·03 (95%CI, 1·10–8·33; P = 0·032), 2·85 (95%CI, 1·03–7·87; P = 0·044), and 3·94 (95%CI, 0·84–18·5; P = 0·083) respectively.


The prognosis of advanced NK-lineage neoplasm is invariably poor despite the use of multi-agent chemotherapy (Nakamura et al, 1997; Cheung et al, 1998; Chim et al, 2004). Allo-HSCT has been attempted for this rapidly progressive and incurable disease, although the number of patients has been limited because of the rarity of the disease (Imashuku et al, 1999; Takenaka et al, 2001; Izutsu et al, 2004). The present study with 28 patients showed durable PFS for patients with NK-cell neoplasms, and verified the promising results of previous studies with a small number of patients. Notably, many patients (19/28) had active stage IV disease at the time of allo-HSCT, but 34% of them achieved long-term PFS. The favourable outcome did not result from the selection bias of the patients with good disease status. Although stage at allo-HSCT, chemosensitivity at allo-HSCT or disease status at allo-HSCT did not affect PFS in our univariate analysis, the study might lack the power to show the association because of its small size.

One of the explanations for our successful results is the presence of a graft-versus-lymphoma (GVL) effect. A GVL effect is an immune-mediated anti-tumour effect against lymphoma after allo-HSCT, which has been discussed for almost 15 years (Jones et al, 1991). Ratanatharathorn et al (1994) showed that the probability of progression of lymphoma was significantly higher in those who underwent autologous stem cell transplantation than in allo-HSCT patients (P = 0·001), suggesting the existence of a GVL effect. Izutsu et al (2004) reported that disease progression for patients with grade 2–4 acute GVHD was significantly lower than those without acute GVHD with a landmark at day 60 (3-year cumulative incidences: 5·9% vs. 33·2%, P = 0·0053). In the present study of NK-cell lymphoma, no progression was observed after 10 months of allo-HSCT and 34% of the patients achieved long-term PFS. In contrast, patients who underwent autologous transplantation, except for those in the first CR, showed a high relapse rate (Au et al, 2003). Autologous transplantation evidently cures a fraction of patients with NK-cell lymphoma. Au et al (2003) reported a 3-year PFS of 71% in patients who underwent autologous transplantation in first CR. However, autologous transplantation is not sufficient to cure advanced disease. Although the reports on the presence of a GVL effect was limited to low-grade lymphoma (Bishop, 2003), it is also effective for NK-cell lymphoma and plays an important role in disease eradication.

The drawback of allo-HSCT is the morbidity and mortality. CST is associated with significant TRM, ranging from 28% to 41% (Bearman et al, 1988; Clift et al, 1990; Dominietto et al, 2002). TRM tends to be higher in patients with refractory or advanced diseases, who have been heavily treated, compared with those who have achieved remission before allo-HSCT (Bearman et al, 1988). Autologous transplantation with high-dose chemotherapy was followed by a high relapse rate (Au et al, 2003). As the intensification of preparative regimens results in an increased regimen-related toxicity and TRM, and do not sufficiently suppress relapse, conditioning regimens with high-dose chemotherapy and TBI may not improve survival. A new strategy of transplantation or RIST has been developed to reduce the regimen-related toxicity while preserving a GVL effect (Giralt et al, 1997; Slavin et al, 1998; Saito et al, 2002; Tanimoto et al, 2003). In the present study, TRM of RIST was lower than that of CST (20% vs. 30%), whereas response rate was equivalent (52% vs. 60%). Therefore, RIST can be a favourable treatment option for patients with NK-cell neoplasms.

Although the present study demonstrated that allo-HSCT is a feasible and promising treatment option for NK-cell neoplasms, the optimal timing for allo-HSCT remains undetermined. The clinical course of advanced NK-cell lymphoma is so aggressive that some patients miss the opportunity for allo-HSCT. In that context, our good results might be biased by patient selection, which is a limitation of a retrospective survey. Further prospective studies are warranted to determine the efficacy and the indication of allo-HSCT for NK-cell neoplasms.

Conflict of interest

None of the authors have any conflict of interest.


We are grateful to the following physicians for their contribution in data collection:

Tetsuo Nishiura: Department of Internal Medicine, Kure National Hospital; Akiyoshi Takami: Department of Internal Medicine (III) Kanazawa University School of Medicine; Katusji Shinagawa: Department of Medicine II, Okayama University Medical School; Yasunori Ueda: Transfusion and Haemapheresis Centre and Department of Internal Medicine, Kurashiki Central Hospital; Nobuo Maseki: Haematology Clinic, Saitama Cancer Centre Hospital; Hirofumi Teshima, Department of Internal Medicine, Osaka City General Hospital; Junichi Tsukada, First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health; Toshio Wakayama, Department of Haematology and Oncology, Shimane Prefectural Central Hospital; Shuichi Miyawaki, Division of Haematology, Saiseikai Maebashi Hospital; Tetsuya Eto, Hamanomachi General Hospital, Division of Haematology; Tomoyuki Endo, Department of Internal Medicine II, Hokkaido University School of Medicine; Kazuteru Ohashi, Department of Haematology, Tokyo Metropolitan Komagome Hospital; Kazuto Togitani, Department of Internal Medicine, Kochi Medical School; Toyonori Tsuzuki, The First Department of Internal Medicine, Nagoya Daini Red Cross Hospital.