Bone Marrow Transplantation (2014) 49, 902–906

© 2014 Macmillan Publishers Limited All rights reserved 0268-3369/14

www.nature.com/bmt

ORIGINAL ARTICLE
Allogeneic haematopoietic SCT for natural killer/T-cell
lymphoma: a multicentre analysis from the Asia Lymphoma
Study Group
E Tse1,5, TSY Chan1,5, L-P Koh2, W-J Chng2, W-S Kim3, T Tang4, S-T Lim4, AKW Lie1 and Y-L Kwong1

Eighteen patients (men = 14; women = 4) with natural killer (NK)/T-cell lymphomas (CR1, N = 9; CR2, N = 7; PR, N = 1; progressive
disease, N = 1) undergoing allogeneic haematopoietic SCT (HSCT) (myeloablative, N = 14; reduced intensity, N = 4) were analyzed.
With a median follow-up of 20.5 months, the 5-year OS was 57% and 5-year EFS was 51%. The use of the SMILE regimen pre-HSCT
was the most important positive prognostic indicator, resulting in significantly superior OS and EFS (P o 0.01). Acute GVHD had a
significant negative impact on OS (P = 0.03). CR1 and CR2 patients had similar survivals, but all patients who were not transplanted
in remission died. Chronic GVHD, International Prognostic Index, disease stage, primary sites of involvement, conditioning regimen
and source of HSC did not affect survival. Although allogeneic HSCT leads to reasonable survival for NK/T-cell lymphoma patients,
results need to be compared with those in patients receiving L-asparaginase-containing regimens. Novel prognostic models
incorporating biomarkers, such as circulating EBV DNA, are needed to identify high-risk patients who may benefit from
allogeneic HSCT.

Bone Marrow Transplantation (2014) 49, 902–906; doi:10.1038/bmt.2014.65; published online 28 April 2014

INTRODUCTION
Natural killer (NK)/T-cell lymphoma is a very rare disease.1
Occurring worldwide, it has a predilection for Asian and South
American populations.2,3 Most of the cases are derived putatively
from NK cells, without T-cell receptor (TCR) gene rearrangement.
A minority may be of bona fide T-cell lineage, with clonal TCR gene
rearrangement.1 There is an invariable association with EBV
infection.3 Clinically, it is classified into ‘nasal’ or ‘non-nasal’ types.
In the nasal type, the primary sites include the nose or
nasopharynx, although the paranasal sinuses, tonsils, Waldeyer’s
ring and oropharynx may also be affected.2 In about 20% of
patients, the tumours may arise from the skin, gastrointestinal
tract or testis without clinically evident nasal involvement.2,3
However, with sensitive radiological imaging such as positron
emission tomography computed tomography, most ‘non-nasal’
cases can in fact be shown to have occult nasal involvement.3,4
For early stage (I/II) disease, conventional chemotherapeutic
regimens designed for aggressive B-cell lymphomas, such as
CHOP (CY, adriamycin, vincristine, prednisone) or CHOP-like
regimens, give unsatisfactory results.3,5 Recent studies of
L-asparaginase-containing regimens combined with radiotherapy
have reported overall response rates of about 90%, with CR in
70–80% of cases.6–8 The 2-year PFS and OS were 80–86% and
86–89%, respectively.6–8 However, in advanced-stage or relapsed/
refractory diseases, the treatment outcome is still unfavourable. In
a phase II study of SMILE (dexamethasone, methotrexate,
ifosfamide, L-asparaginase, etoposide) in stage IV and relapsed/
refractory diseases, the overall response rate was 79%, with CR in
45% of cases.9 The 1 year PFS and OS were 53% and 55%,
respectively.9
High-dose chemotherapy with haematopoietic SCT (HSCT) has
been used to improve the outcome of NK/T-cell lymphoma.
However, autologous HSCT for early-stage NK/T-cell lymphoma
does not confer additional benefit compared with combined
chemotherapy and radiotherapy.10 In advanced-stage disease, the
role of autologous HSCT is debatable, because its results are either
comparable with or in fact inferior to L-asparaginase-containing
regimens.6,10 Owing to a putative graft-versus-lymphoma effect,
allogeneic HSCT has also been explored, especially in poor-risk
patients with advanced-stage disease.11-13 In an earlier study of
allogeneic HSCT in 28 patients with CD56-positive neoplasms,
which included 22 patients with NK/T-cell lymphoma, the 2-year
OS and PFS were 40% and 34%, with a TRM of 20–30% depending
on the conditioning regimen.11 Unfortunately, results of patients
with NK/T-cell lymphomas were not separately described, and the
survivals were not spectacular in comparison with L-asparaginase-
containing regimens. In another study of 12 patients with
NK/T-cell lymphomas treated with allogeneic HSCT, the 3-year
OS and PFS were 55 and 53%, with a TRM of 8%.13 These survivals
appeared to be only comparable with those from the use of
L-asparginase-containing regimens.
With the advent of L-asparaginase-containing regimens as
treatment for NK/T-cell lymphoma, the role of allogeneic HSCT in
these patients should be reappraised. To address this issue, we
conducted a retrospective analysis to examine the use and
outcome of allogeneic HSCT in NK/T-cell lymphoma patients.

1
Department of Medicine, Queen Mary Hospital, Hong Kong, China; 2Department of Haematology-Oncology, National University Cancer Institute, Singapore; 3Division of Hemato-
Oncology, Samsung Medical Center, Seoul, Korea and 4Department of Medical Oncology, National Cancer Centre, Singapore. Correspondence: Dr Y-L Kwong, Department of
Medicine, Professorial Block, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
E-mail: ylkwong@hkucc.hk.hk
5
These athours contributed equally to this work.
Received 11 October 2013; revised 18 December 2013; accepted 19 December 2013; published online 28 April 2014
 Allogeneic HSCT in NK/T-cell lymphoma
E Tse et al
903
MATERIALS AND METHODS
Table 1. Clinicopathologic features of 18 natural killer/T-cell
Patients lymphoma patients undergoing allogeneic haematopoietic SCT
Four tertiary referral centres specialized in the management of NK/T-cell
malignancies participated in this retrospective analysis. According to the Clinicopathologic features Number (%)
Asia Lymphoma Study Group proposed algorithm, allogeneic HSCT was
considered for patients with NK/T-cell lymphoma if they satisfied the Gender
following criteria: age o60 years, stage IV disease with BM involvement Male 14 (78)
at first CR1 or chemotherapy-sensitive relapsed/refractory disease and Female 4 (22)
suitable HSC donors available.
Median age 40.5 (17–57) years
Eastern Cooperative Oncology Group performance
Conditioning regimens score
Both myeloablative and reduced-intensity conditioning regimens were 0 11 (61)
used. For myeloablative conditioning, they were based primarily on high- 1 4 (22)
dose BU, TBI or combination of CY, BCNU and etoposide. Combination of 2 3 (17)
fludarabine and melphalan was used in patients receiving reduced-
intensity conditioning HSCT. Lactate dehydrogenase
Normal 3 (17)
Increased 15 (83)
Statistical analysis
Differences between categorical variables were analyzed by Χ2-test. Stage at diagnosis
Continuous variables were analyzed by independent two sample t-test. I 3 (17)
Analysis of OS and EFS was conducted by the Kaplan-Meier method. II 2 (11)
Two-tailed P-values of less than 0.05 were taken as significant. All tests III 1 (6)
were performed with the SPSS 15.0 software package (Chicago, IL, USA). IV 12 (66)

International Prognostic Index

RESULTS Low (0–1) 5 (28)
Patients Low intermediate (2) 5 (28)
High intermediate (3) 7 (39)
Eighteen patients (14 men; 4 women) were studied (Table 1). The High (4–5) 1 (6)
median age was 40.5 years. Most of the patients had advanced-
stage (stage III/IV) disease (74%). Most patients (74%) had received Disease status
SMILE for induction of remission before HSCT. First CR 9 (50)
The majority of patients (89%) were in CR (CR1, N = 9; CR2, N = 7) Second CR 7 (39)
before HSCT. For patients transplanted at CR1, the initial PR 1 (6)
chemotherapeutic protocols were CHOP-like regimens (N = 5) Progressive disease 1 (6)
and SMILE (N = 4). For patients transplanted at CR2, the initial
Chemotherapy before transplantation
treatment protocols were CHOP-like regimens (N = 5), radio- SMILE 14 (78)
therapy alone (N = 1) and SMILE (N = 1). At first relapse, all cases CHOP 2 (11)
were treated with SMILE and achieved CR2. Conditioning ICE 1 (6)
regimens were myeloablative in 78% of cases. CALGB 19 802 1 (6)

Conditioning regimen
HSCT outcome Myeloablative
Acute GVHD developed in nine patients (grade 1-2: N = 7; grade CY/TBI 8 (44)
3-4: N = 2), which responded to treatment in all cases. Four CY/TBI/anti-thymocyte globulin 1 (6)
patients developed chronic mucocutaneous GVHD (mild: N = 3; Etoposide/TBI 1 (6)
severe: N = 1), which was controlled with immunosuppressants. At TBI 1 (6)
a median follow-up of 20.5 (1.6–65) months, relapse occurred in CY/BCNU/etoposide 2 (11)
BU/CY 1 (6)
five patients, with two of them dying from disease progression,
Reduced intensity
and three salvaged with chemotherapy (two were alive at Fludarabine/melphalan 4 (22)
follow-up 1 and 31 months after salvage therapy, but one died
of pneumonia at 2 months). Another four patients died of Type of donor
infection and one patient committed suicide. HLA-matched sibling donor 12 (67)
Matched unrelated donor 6 (33)
Survivals Source of haematopoietic stem cell
The 5-year OS was 57%, with the median survival not reached. The BM 7 (39)
5-year EFS was 51%. The survival curves appeared to have Peripheral blood 11 (61)
plateaued after 20 months (Figures 1a and b).
Abbreviations: CALGB = daunorubicin, CY, vincristine, dexamethasone,
methotrexate; CHOP = CY, doxorubicin, vincristine, prednisolone;
Prognostic indicators ICE = ifosfamide, carboplatin, etoposide; SMILE = dexamethasone,
methotrexate, ifosfamide, L-asparaginase, etoposide.
Univariate analysis showed that patients who had received SMILE
before HSCT had significantly better OS and EFS (median OS and
EFS not reached) than patients treated with other regimens
(median OS and EFS were 6.7 and 2.8 months, respectively; survival (Supplementary Figure A). Other clinicopathologic para-
P o0.01) (Figures 1c and d). The status of the remission (CR1 vs meters, including chronic GVHD (Supplementary Figures B and C),
CR2) had no impact on survivals (Figure 1e). The development of International Prognostic Index (Supplementary Figures D and E),
acute GVHD had a significantly negative impact on OS (Figure 1f) stage of disease (stage I/II vs stage III/IV), primary site
but not EFS, whereas chronic GVHD had no significant impact on of involvement (nasal vs non-nasal), conditioning regimen

© 2014 Macmillan Publishers Limited Bone Marrow Transplantation (2014) 902 – 906
 Allogeneic HSCT in NK/T-cell lymphoma
E Tse et al
904
100 100

Cumulative survival (%)

Cumulative survival (%)

80 80

60 60

40 40

20 20

0 0
0 20 40 60 0 20 40 60
Overall survival (months) Event free survival (months)

100 100
P < 0.01 P < 0.01
Cumulative survival (%)

Cumulative survival (%)

80 80
SMILE
60 60
SMILE
40 40

20 20
Non-SMILE Non-SMILE
0 0
0 20 40 60 0 20 40 60
Overall survival (months) Event free survival (months)

100 100
P = 0.43
Cumulative survival (%)

Cumulative survival (%)

80 CR2 80 No aGVHD

60 60 P = 0.03
CR1
40 40

20 20 aGVHD

0 0
0 20 40 60 0 20 40 60
Overall survival (months) Overall survival (months)
Figure 1. Survival of 18 patients with natural killer/T-cell lymphoma after allogeneic haematopoietic SCT (HSCT). (a) OS for the entire cohort.
(b) EFS for the entire cohort. (c) Patients treated with SMILE pre-HSCT had superior OS compared with those treated with non-SMILE regimens
pre-HSCT. (d) Patients treated with SMILE pre-HSCT had superior EFS compared with those treated with non-SMILE regimens pre-HSCT. (e) No
difference in OS for patients transplanted in first or second CR. (f) Patients without acute GVHD (aGVHD) had superior OS.

(myeloablative vs non-myeloablative) and source of HSC (BM vs
peripheral blood) had no significant impact on survivals. On
multivariate analysis, the use of SMILE chemotherapy pre-HSCT
still remained the only significant factor impacting on EFS
(P = 0.03), and showed a trend toward impacting on OS
(P = 0.07). Acute GVHD was no longer a significant factor for
OS (P = 0.25) and ESF (P = 0.8).
DISCUSSIONS
To date, this report is the largest study that focused exclusively on
allogeneic HSCT for extranodal NK/T-cell lymphoma. More
importantly, it is also the first study that examined HSCT in
patients treated with SMILE, now considered a standard regimen
for advanced-stage or relapsed/refractory NK/T-cell lymphoma.3
The outcome in our patients was encouraging (5-year EFS and OS
at 51 and 57% respectively). Results from other studies involving
smaller numbers of NK/T-cell lymphoma patients treated with
allogeneic HSCT are summarized in Table 2. Two studies were
conducted before L-asparaginse-containing regimens were
used,11,14 whereas two studies contained some patients treated
with L-asparaginase-containing regimens.12,13 Owing to the
heterogeneity and small numbers of patients involved, most
studies only stressed the feasibility of allogeneic HSCT.
There were several interesting observations from this analysis.
First, patients who received SMILE as induction or salvage
chemotherapy prior to allogeneic HSCT had a significantly better
outcome than those receiving non-L-asparaginase-containing
regimen. In our analysis, the number of patients not receiving
SMILE pre-HSCT was relatively small compared with those
receiving SMILE. This limitation was due to SMILE being
considered a standard regimen for patients with NK/T-cell

Bone Marrow Transplantation (2014) 902 – 906 © 2014 Macmillan Publishers Limited
 Allogeneic HSCT in NK/T-cell lymphoma
E Tse et al
905
Table 2. Selected studies on allogeneic HSCT in patients with natural killer/T-cell lymphomas

Study No. Age Stage Status Chemotherapy Conditioning OS PFS TRM (%)

Tse et al. (present study) 18 40.5 I/II = 5 CR = 16 L-asp = 14 MAC = 14 5 years, 5-year 22
III/IV = 13 PR = 1 Others = 4 RIC = 4 57% 51%
PD = 1
Ennishi et al.13 12 28 I/II = 4 CR = 4 L-asp = 5 MAC = 9 3 years, 55% 3-year 8.3
III/IV = 8 PR = 4 Others = 7 RIC = 3 53%
PD = 4
Yokoyama et al.12 5 30 I/II = 3 CR = 3 L-asp = 4 MAC = 5 ND ND 0
III/IV = 2 PR = 0 Others = 1 RIC = 0
PD = 2
Susuki et al.14 6 28 I/II = 2 CR = 1 L-asp = ND MAC = 6 ND ND 16.6
III/IV = 4 PR = 0 Others = ND RIC = 0
PD = 5
Murashige et al.11 22 38a I/II = 1a CR = 8a L-asp = ND MAC = 17 2 yearsa, 2 yearsa, 28.5a
III/IV = 19a PR = ND Others = ND RIC = 5 40% 34%
PD = ND
Abbreviations: Age = median age; Chemo = chemotherapy regimen prior to HSCT; Conditioning = conditioning regimen for HSCT; HSCT = haematopoietic SCT;
L-asp = L-asparaginase-containing regimen; MAC = myeloablative; ND = no data or not reported; No. = number of patients; PD = progressive disease;
RIC = reduced-intensity conditioning; Stage = stage of disease; Status = disease status at HSCT. aIndicates that the data included patients with other natural
killer cell neoplasms in the study.

lymphoma in most centres, so that fewer and fewer patients
would receive non-SMILE induction protocols. To further explore
this point, a comparison of our results with two previously
published studies of NK/T-cell lymphoma patients undergoing
allogeneic HSCT, where survival data were provided, might be
informative (Table 2). In the study by Murashige et al.,11 in which
patients did not receive L-asparaginase-containing regimens, the
2-year PFS and OS were only 34 and 40%. However, in the study
by Ennishi et al.,13 in which 42% of patients had received
L-asparaginase-containing regimens before HSCT, the 3-year PFS
and OS were 53 and 55%, which were comparable with our
results.These observations together with ours might reflect a
better quality, or deeper, remission and disease control achieved
by SMILE or L-asparaginase-containing protocols as compared
with other regimens. It is also consistent with the notion that the
presence of residual disease at the time of HSCT is associated with
a poor outcome.
Occurrence of the lymphoma primarily in non-nasal sites has
been considered to be a poor prognostic indicator when
conventional chemotherapy regimens were used.2 With the SMILE
regimen, it has been shown that the primary site of lymphoma
no longer has any impact on outcome.6 In this study, because the
majority of patients had received SMILE before HSCT, the primary
site also did not appear to impact on survival. This observation
again suggested that with appropriate cytoreduction, the primary
presentation site in NK/T-cell lymphoma had no significant clinical
implication.
Attaining a remission has been found to be a significant factor
impacting on the outcome after allogeneic HSCT in NK/T-cell
lymphoma patients.10 In this study, the majority of patients were
transplanted in CR1 or CR2. Interestingly, CR1 and CR2 patients
had comparable survivals. Only two patients were transplanted
while not in remission, and both died. Hence, obtaining a
remission is important before transplantation. Whether the quality
or depth of remission, as determined by sensitive tumour
biomarkers such as circulating EBV DNA,15 may impact on survival
will need to be determined.
Allogeneic HSCT remains a potentially curative option for
patients with advanced-stage or relapsed/refractory NK/T-cell
lymphoma. However, its use in patients treated with
L-asparaginase-containing regimens should be carefully evalu-
ated; in view of the short- and long-term toxicities of allogeneic
HSCT, as well as the very good results achieved by SMILE or similar
regimens. Given that the CR rate of using SMILE in relapsed/
refractory patients is very high,6,9 and that CR1 and CR2 patients
fare similarly after allogeneic HSCT, frontline allogeneic HSCT for
patients in CR1 does not seem justifiable. Furthermore, patients in
CR1 induced by SMILE had been shown to have an estimated
4-year disease-free survival of 60%.6 For relapsed patients
achieving CR2 induced by SMILE, a 4-year disease-free survival
of 68.2% was achieved.6 Hence, HSCT may not be necessary for
more than half of the patients in CR1 and CR2 induced by SMILE.
Therefore, allogeneic HSCT should conceivably be reserved for
CR1/CR2 patients who are at high risks of relapse.
In conclusion, with evolving strategies for more effective
treatment of NK/T-cell lymphomas, the role of allogeneic HSCT
in NK/T-cell lymphoma must be strictly evaluated. Recently, it has
been shown that failure to achieve undetectable circulating EBV
DNA after one to two courses of SMILE portends a very poor
prognosis.16 This test, together with the use of more conventional
risk stratifiers such as the International Prognostic Index, should be
evaluated as indicators for allogeneic HSCT in high-risk patients in
remission.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
REFERENCES
1 Chan JK, Quintanilla-Martinez L, Ferry JA, Peh SC. Extrannodal NK/T-cell
lymphoma, nasal type. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA,
Stein H et al. (eds). WHO Classification of Tumours of Haematopoietic and
Lymphoid Tissues: IRAC, Lyon 2008, pp 285–288.
2 Kwong YL. Natural killer-cell malignancies: diagnosis and treatment. Leukemia
2005; 19: 2186–2194.
3 Tse E, Kwong YL. How I treat NK/T-cell lymphomas. Blood 2013; 121: 4997–5005.
4 Tse E, Leung R, Khong PL, Lau WH, Kwong YL. Non-nasal natural killer cell
lymphoma: not non-nasal after all. Ann Hematol 2009; 88: 185–187.
5 Kim SJ, Kim WS. Treatment of localized extranodal NK/T cell lymphoma,
nasal type. Int J Hematol 2010; 92: 690–696.
6 Kwong YL, Kim WS, Lim ST, Kim SJ, Tang T, Tse E et al. SMILE for natural killer/T-cell
lymphoma: analysis of safety and efficacy from the Asia Lymphoma Study Group.
Blood 2012; 120: 2973–2980.
7 Jiang M, Zhang H, Jiang Y, Yang Q, Xie L, Liu W et al. Phase 2 trial of ‘sandwich’
L-asparaginase, vincristine, and prednisone chemotherapy with radiotherapy
in newly diagnosed, stage IE to IIE, nasal type, extranodal natural killer/T-cell
lymphoma. Cancer 2012; 118: 3294–3301.
8 Wang L, Wang ZH, Chen XQ, Li YJ, Wang KF, Xia YF et al. First-line combination of
gemcitabine, oxaliplatin, and L-asparaginase (GELOX) followed by involved-field

© 2014 Macmillan Publishers Limited Bone Marrow Transplantation (2014) 902 – 906
 Allogeneic HSCT in NK/T-cell lymphoma
E Tse et al
906
radiation therapy for patients with stage IE/IIE extranodal natural killer/T-cell
lymphoma. Cancer 2013; 119: 348–355.
9 Yamaguchi M, Kwong YL, Kim WS, Maeda Y, Hashimoto C, Suh C et al. Phase II
study of SMILE chemotherapy for newly diagnosed stage IV, relapsed, or refrac-
tory extranodal natural killer (NK)/T-cell lymphoma, nasal type: the NK-Cell Tumor
Study Group study. J Clin Oncol 2011; 29: 4410–4416.
10 Kwong YL. High-dose chemotherapy and hematopoietic SCT in the
management of natural killer-cell malignancies. Bone Marrow Transplant 2009; 44:
709–714.
11 Murashige N, Kami M, Kishi Y, Kim SW, Takeuchi M, Matsue K et al. Allogeneic
haematopoietic stem cell transplantation as a promising treatment for natural
killer-cell neoplasms. Br J Haematol 2005; 130: 561–567.
12 Yokoyama H, Yamamoto J, Tohmiya Y, Yamada MF, Ohguchi H, Ohnishi Y et al.
Allogeneic hematopoietic stem cell transplant following chemotherapy contain-
ing l-asparaginase as a promising treatment for patients with relapsed or
Supplementary Information accompanies this paper on Bone Marrow Transplantation website (http://www.nature.com/bmt)
Bone Marrow Transplantation (2014) 902 – 906
refractory extranodal natural killer/T cell lymphoma, nasal type. Leuk Lymphoma
2010; 51: 1509–1512.
13 Ennishi D, Maeda Y, Fujii N, Kondo E, Shinagawa K, Ikeda K et al. Allogeneic
hematopoietic stem cell transplantation for advanced extranodal natural
killer/T-cell lymphoma, nasal type. Leuk Lymphoma 2011; 52: 1255–1261.
14 Suszuki R, Suzumiya J, Nakamura S, Kagami Y, Kameoka JI, Sakai C et al.
Hematopoietic stem cell transplantation for natural killer-cell lineage neoplasms.
Bone Marrow Transplant 2006; 37: 425–431.
15 Au WY, Pang A, Choy C, Chim CS, Kwong YL. Quantification of circulating
Epstein-Barr virus (EBV) DNA in the diagnosis and monitoring of natural killer cell
and EBV-positive lymphomas in immunocompetent patients. Blood 2004; 104:
243–249.
16 Kwong Y-L, Pang AWK, Leung AYH, Chim C-S, Tse E. Quantification of circulating
Epstein-Barr virus DNA in NK/T-cell lymphoma treated with the SMILE protocol:
diagnostic and prognostic significance. Leukemia 2014; 28: 865–870.
© 2014 Macmillan Publishers Limited