Over the past decade, xenografting human leukemia cells into mice with different levels of immunodeficiency, with or without preconditioning, has provided an important tool to study various aspects of leukemia biology and to identify distinct clinical risk groups for evaluation of novel therapeutic strategies, as well as the possibility of amplifying human leukemia cells in vivo. Interestingly, these models using human acute lymphoblastic leukemia and acute myeloid leukemia cells as xenografts recapitulate many clinical features of the disease. Similar to the human environment (for example, in the bone marrow), transplanted leukemia cells in the murine setting are exposed to both favorable and unfavorable conditions for engraftment that may exert a distinct pressure for selection of subclones. Thus, results obtained in these models may vary depending on the experimental setup. The impact of in vivo growth of human leukemia cells on the background of a more or less hostile murine environment for leukemia biology and the course of the disease in patients are discussed in the context of the diversity of xenograft models. Cancer Res; 71(23); 7141–4. ©2011 AACR.

More than 25 years ago, a mutation leading to a severe combined immunodeficiency (SCID) in mice was described (1), allowing the transplantation of human hematopoietic cells into these recipients without graft rejection. Leukemia cell lines and primary patient cells have been engrafted into SCID mice (2, 3), leading to leukemia manifestation similar to the patient's disease. However, the residual immunity of SCID mice limits xenografting in these recipients. Backcrossing of SCID mice onto a nonobese diabetic (NOD/Lt) background resulted in animals with profound immunodeficiency designated as NOD/SCID [NOD/LtSz-scid/scid], providing superior engraftment capacity (4). Transplantation of primary acute lymphoblastic leukemia (ALL) cells onto NOD/SCID recipients leads to a leukemia model that resembles the distribution and course of the human disease (5, 6). The development of mouse strains with impaired innate immunity and absent natural killer (NK) cell activity due to additional defects of the interleukin-2 receptor gamma chain (IL-2Rγ) provided recipient mice [NOD-scid IL2Rnull (NSG; ref. 7); NOD/SCID/cnull (NOG; ref. 8)] with up to 100% engraftment and in vivo differentiation of transplanted hematopoietic stem cells, because of almost complete absence of a murine immune system. However, the impact of xenograft models with different permissiveness on the biology of xenografted leukemia is still unclear.

Recently, we characterized an engraftment phenotype of patient leukemia cells transplanted into NOD/SCID mice that is indicative for early patient relapse (9). A series of 50 primary B-cell precursor (BCP) ALL samples directly obtained from patients at diagnosis were transplanted into unconditioned, female NOD/SCID mice, and time to leukemia (TTL) was analyzed as the time from transplantation until onset of leukemia-related morbidity of the recipients for each sample transplanted. The TTLshort phenotype, which was found to be a strong prognostic factor for early patient relapse, is characterized by a gene expression signature pointing to involvement of molecules regulating cell proliferation and apoptosis. When applied to an independent set of gene profiles obtained from 197 BCP-ALL patients at diagnosis, the TTL signature identified a high-risk cluster of patients with early relapse, showing the biologic impact of leukemia engraftment observed in this xenotransplantation model.

Engraftment and in vivo proliferation of patient leukemia cells transplanted into immunodeficient mice have been described to be associated with patient outcome, although with conflicting results. In 2 studies, small numbers of ALL samples obtained at relapse were transplanted into irradiated SCID or NOD/SCID mice, showing an association of engraftment rate and extent of organ infiltration with patient remission duration; however, this association was not seen for samples obtained at patient diagnosis (5, 10). Upon transplantation of diagnostic samples of high-risk stratified BCP-ALL patients, engraftment in SCID mice varied, with a higher relapse rate for patients with SCID leukemia (11). However, in a large series of 681 BCP-ALL samples, including all risk groups, the previous finding could not be confirmed (12). In contrast, we found a significant correlation of TTL and remission duration independent of risk groups and only for low-risk stratified patients. This finding may be explained by the distinct models used: different recipients, such as SCID (showing low overall engraftment of 15% in this study; ref. 12) versus NOD/SCID (9); a relatively short observation time (12 weeks) of the SCID mice in contrast to 20 weeks' follow-up in our study; and the use of cell bank versus fresh samples. However, when cryopreserved cell bank samples were transplanted in our model, the distinct engraftment phenotypes of primary leukemias associated with early relapse were still maintained. We monitored some leukemias for longer observation times and detected leukemia manifestation up to 10 months after transplantation, suggesting that, ultimately, almost all samples would engraft and show in vivo disease manifestation. In line with our findings, superior patient survival has been reported if acute myeloid leukemia (AML) cells could not repopulate NOD/SCID mice, in contrast to in vivo proliferation associated with inferior outcome (13). Interestingly, a distinct engraftment phenotype upon leukemia xenotransplantation was also recently observed in different, albeit much more permissive, models.

Cryopreserved ALL samples transplanted intraosseously showed rapid or prolonged engraftment, although no significant association with patient outcome was found in NK cell–depleted, irradiated NOD/SCID mice (14). However, using NSG mice, an association between engraftment and risk stratification of patients was observed (15). Further evidence that engraftment and proliferation are determined by the aggressiveness of the leukemia itself has been provided by in vivo treatment studies. Time to reappearance of ALL after chemotherapy of leukemia-bearing mice was associated with remission duration and patient outcome (5). However, because we observed significantly different engraftment phenotypes of leukemia cells isolated at diagnosis from so-far untreated patients, this difference does not seem to be sufficiently explained by distinct sensitivity or resistance to treatment. Moreover, when transplanting different doses of TTLshort and TTLlong leukemia cells, engraftment properties are retained independently of the number of cells transplanted and, thus, are an intrinsic feature of the leukemia cells. Similarly, the engraftment capability of AML cells into NOD/SCID mice was shown to be intrinsic to the leukemia cells independent of the cell numbers transplanted (13).

The biology and clinical course of leukemia in the xenograft model is not only characterized by the recipient mouse strain used but also determined by a number of other factors contributing to successful transplantation and overall engraftment capacity of the respective model: (i) source and processing of leukemia cells, (ii) transplantation procedure and conditioning of the recipient, (iii) the recipient itself, and (iv) definition of leukemia manifestation (Fig. 1). Due to the known high rate of spontaneous apoptosis in ALL cells upon removal from bone marrow, the shipping, cryopreservation, storage, and preprocessing of cells are of crucial relevance. Leukemia cells that have surface-bound antibodies after sorting are prevented from engraftment if transplanted onto NOD/SCID mice (16). Upon transplantation of frozen cell bank samples in comparison to freshly isolated cells, we and others observed a longer latency to leukemia (9, 17). Furthermore, the route of injection determines engraftment. Direct transfer of cells into the bone marrow niche, circumventing the homing process, leads to higher engraftment frequencies, in contrast to systemic i.v. injection (18). Preconditioning regimens using sublethal irradiation modulate the murine environment and improve engraftment. However, successful ALL engraftment has been shown without conditioning prior to transplantation (6), and strikingly, irradiation has been shown to have a negative impact on homing of ALL cells (19). Residual murine immunity may affect engraftment efficacy not only on the level of immune responses against human leukemia cells but also on the level of cytokine production upon encounter of human ALL cells with the murine environment. Additionally, cross-species reactivity of homing molecules and growth factors (murine on human and vice versa) influences xenograft growth (20), and humanized models transgenic for human cytokines are developed to increase engraftment and differentiation of hematopoietic cells (21). Accordingly, recipients treated with or transgenic for human growth factors show increased AML engraftment (22–24); however, whether leukemia subtypes or subclones are selectively stimulated is unclear. Comparing recipients with different grades of immunodeficiency by parallel transplantation, aggressive ALL with rapid leukemia onset in permissive (NK cell–depleted NOD/SCID or NSG) mice also engrafted in less permissive NOD/SCID mice, whereas samples with prolonged NSG engraftment did not engraft in NOD/SCID recipients within 24 weeks (14). Similarly, transplantation of high-risk or relapsed ALL samples revealed accelerated leukemia onset in NSG versus NOD/SCID mice (25). Preconditioning treatment as used in these studies will attenuate the residual immunity of the recipients, leading to a diminished selection pressure on leukemia cells independent of the mouse strain used. However, despite preconditioning, AML samples that failed to engraft in NOD/SCID mice also did not engraft in more immunocompromised recipients (13). On the other hand, each patient ALL sample that engrafts in NOD/SCID mice will engraft in more permissive NSG mice, although it remains to be investigated whether ALL samples that failed to engraft in NOD/SCID mice will succeed in a less hostile microenvironment and engraft in NSG recipients. It can be expected that samples with failed or prolonged engraftment in low permissive mice (e.g., 10 months observed in our NOD/SCID model) will be engrafted at higher rates or shorter times in NSG mice. Furthermore, the age and gender of the recipient affect xenografting. Young age is associated with superior engraftment (26), and newborn mice have been used to avoid the effect of innate immunity (27). Most interestingly, the gender of the recipient has significant impact with more than 10 times superior engraftment in female versus male mice (28). Finally, the endpoint of the observation time (engraftment in a given time versus leukemia manifestation) and estimation of leukemia onset will be pivotal for the outcome of the experiment.

Figure 1.

Diversity of human leukemia xenograft mouse models. Xenograft mouse models in which primary patient cells are transplanted into immunodeficient recipients are increasingly used to study leukemia biology. Distinct proliferation of individual leukemia samples reflects disease progression in the patient and shows prognostic impact. A variety of distinct factors and experimental conditions define engraftment and in vivo leukemia growth of xenotransplanted leukemia cells and, thereby, determine the outcome of the experiment. However, preferable conditions for analyzing leukemia biology by means of in vivo xenograft growth mimicking the situation of the patient are (i) unprocessed patient cells (ii) systemically injected into (iii) animals with remaining immunity and then (iv) assessment of full-blown leukemia. ab, antibody; hu, human; i.f., intrafemoral; i.p., intraperitonial; i.t., intratibial.

Figure 1.

Diversity of human leukemia xenograft mouse models. Xenograft mouse models in which primary patient cells are transplanted into immunodeficient recipients are increasingly used to study leukemia biology. Distinct proliferation of individual leukemia samples reflects disease progression in the patient and shows prognostic impact. A variety of distinct factors and experimental conditions define engraftment and in vivo leukemia growth of xenotransplanted leukemia cells and, thereby, determine the outcome of the experiment. However, preferable conditions for analyzing leukemia biology by means of in vivo xenograft growth mimicking the situation of the patient are (i) unprocessed patient cells (ii) systemically injected into (iii) animals with remaining immunity and then (iv) assessment of full-blown leukemia. ab, antibody; hu, human; i.f., intrafemoral; i.p., intraperitonial; i.t., intratibial.

Close modal

Depending on the aim of the experiment, different xenograft models may be suitable for different purposes. To achieve high engraftment rates and cell amplification (and for cells with limited viability; e.g., shipped and/or frozen samples), highly permissive models supporting as unrestricted xenograft growth as possible will be preferred. In contrast, if features of leukemia biology are characterized by means of in vivo growth of xenografted cells mimicking the situation of the patient, careful adaptation of an appropriate model is required. Unmodified cells with the highest functional integrity and viability possible (i.e., directly from the patient) should be favored. Although the topical origin of leukemia is largely unknown, precursor or initiating cells are hematogenously spread during leukemogenisis and home into the supportive (bone marrow) environment, subsequently leading to full-blown disease. This homing process with xenograft–host interaction, adhesion, and migration will be maintained by systemic i.v. injection, but it is bypassed if cells are transplanted intraosseously directly into their niches. In the patient, leukemia cells originate and develop in a microenvironment that is, with its cytokines and immune cells, both supportive and hostile and most certainly exerts selection pressure. Thus, an in vivo model in which leukemia cells have to deal with and ultimately overcome residual immunity in a xenograft setting will be advantageous in resembling human leukemogenesis and disease progression compared with a model of widely unrestricted proliferation under highly permissive conditions. In patients, leukemia is generally diagnosed at the stage of fully developed disease. In a xenotransplantation setting, assessment of leukemia engraftment identifying low percentages of xenografted cells in the recipient might not mirror the composition and biology of the different populations present at the patient's diagnosis, which, in contrast, will be reflected at onset of full-blown leukemia in the recipient mouse. However, for this strategy, careful daily investigation of the recipient and immediate termination of the experiment upon signs of leukemia-related morbidity should be ensured. Thus, to analyze leukemia biology reflecting the situation of patients, the following conditions seem appropriate: (i) direct transplantation of unmanipulated cells by (ii) systemic injection into (iii) recipients bearing residual immunity, with (iv) onset of overt leukemia as the end point.

The interplay between leukemia aggressiveness on one side and hostility of the model on the other side determines engraftment and in vivo proliferation and has a profound impact on our understanding of leukemia biology. Thus, the repopulating activity of xenografted tumor cells is a characteristic of initiating or stem cells, and different models may affect the frequency of these cells. A hierarchical concept of only a few immature cells possessing stem cell properties at the apex of a differentiation tree has been challenged by the proposal of a stochastic model in which a high proportion of cancer cells, including cells with more mature phenotypes, harbors stem cell activity. It is striking that the paradigmatic change of the ALL stem cell definition from rarely present, immature cells to more frequent, mature cells may depend on the permissiveness of the xenograft model employed. Although ALL stem cells have been characterized as infrequently present and phenotypically immature cells using classical NOD/SCID mice (29, 30), the use of recipients with increased immunodeficiency (like NSG mice) has identified more mature cells at higher frequencies to have leukemogenic potential (27, 31, 32). Even more evident, in malignant melanoma with only rare cancer stem cells as assayed in NOD/SCID mice, as many as 1 stem cell in 4 melanoma cells was identified using NSG mice (33). In this context, leukemia progression and evolution of ALL subclones analyzed in xenograft models will also depend on the model used. Recent studies reported clonal diversity of BCP-ALL cells when serially transplanted intraosseally into NSG mice (14, 15, 34), suggesting a lack of selection pressure supporting proliferation of different subclones. Thus, in addition to the intrinsic oncogenic potential of transplanted leukemia cells, the selection pressure put on human leukemia cells by transplantation conditions is expected to have a profound impact on the “true frequency” of leukemia-initiating cells, on clonal evolution, and on other functional features.

Taken together, xenograft models are diverse, not only characterized by recipient permissiveness but also by several additional factors that, in summary, will determine engraftment outcome. The model used is of critical relevance if aspects of leukemia biology are analyzed based on xenograft in vivo growth and clinically relevant conclusions are drawn from these experiments. The question of which xenograft model will most precisely reflect the patient situation ultimately may depend on the focus of the experiment. Thus, highly permissive xenograft models providing as much unrestricted engraftment and in vivo proliferation as possible might be superior or inferior compared with less permissive models, which probably more closely reflect the patient setting.

No potential conflicts of interest were disclosed.

1.
Bosma
GC
,
Custer
RP
,
Bosma
MJ
. 
A severe combined immunodeficiency mutation in the mouse
.
Nature
1983
;
301
:
527
30
.
2.
Kamel-Reid
S
,
Letarte
M
,
Sirard
C
,
Doedens
M
,
Grunberger
T
,
Fulop
G
, et al
A model of human acute lymphoblastic leukemia in immune-deficient SCID mice
.
Science
1989
;
246
:
1597
600
.
3.
Lücking-Famira
KM
,
Daniel
PT
,
Möller
P
,
Krammer
PH
,
Debatin
KM
. 
APO-1 (CD95) mediated apoptosis in human T-ALL engrafted in SCID mice
.
Leukemia
1994
;
8
:
1825
33
.
4.
Shultz
LD
,
Schweitzer
PA
,
Christianson
SW
,
Gott
B
,
Schweitzer
IB
,
Tennent
B
, et al
Multiple defects in innate and adaptive immunologic function in NOD/LtSz-scid mice
.
J Immunol
1995
;
154
:
180
91
.
5.
Lock
RB
,
Liem
N
,
Farnsworth
ML
,
Milross
CG
,
Xue
C
,
Tajbakhsh
M
, et al
The nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model of childhood acute lymphoblastic leukemia reveals intrinsic differences in biologic characteristics at diagnosis and relapse
.
Blood
2002
;
99
:
4100
8
.
6.
Baersch
G
,
Möllers
T
,
Hötte
A
,
Dockhorn-Dworniczak
B
,
Rübe
C
,
Ritter
J
, et al
Good engraftment of B-cell precursor ALL in NOD-SCID mice
.
Klin Padiatr
1997
;
209
:
178
85
.
7.
Shultz
LD
,
Lyons
BL
,
Burzenski
LM
,
Gott
B
,
Chen
X
,
Chaleff
S
, et al
Human lymphoid and myeloid cell development in NOD/LtSz-scid IL2R gamma null mice engrafted with mobilized human hemopoietic stem cells
.
J Immunol
2005
;
174
:
6477
89
.
8.
Ito
M
,
Hiramatsu
H
,
Kobayashi
K
,
Suzue
K
,
Kawahata
M
,
Hioki
K
, et al
NOD/SCID/gamma(c)(null) mouse: an excellent recipient mouse model for engraftment of human cells
.
Blood
2002
;
100
:
3175
82
.
9.
Meyer
LH
,
Eckhoff
SM
,
Queudeville
M
,
Kraus
JM
,
Giordan
M
,
Stursberg
J
, et al
Early relapse in all is identified by time to leukemia in NOD/SCID mice and is characterized by a gene signature involving survival pathways
.
Cancer Cell
2011
;
19
:
206
17
.
10.
Kamel-Reid
S
,
Letarte
M
,
Doedens
M
,
Greaves
A
,
Murdoch
B
,
Grunberger
T
, et al
Bone marrow from children in relapse with pre-B acute lymphoblastic leukemia proliferates and disseminates rapidly in scid mice
.
Blood
1991
;
78
:
2973
81
.
11.
Uckun
FM
,
Sather
H
,
Reaman
G
,
Shuster
J
,
Land
V
,
Trigg
M
, et al
Leukemic cell growth in SCID mice as a predictor of relapse in high-risk B-lineage acute lymphoblastic leukemia
.
Blood
1995
;
85
:
873
8
.
12.
Uckun
FM
,
Sather
HN
,
Waurzyniak
BJ
,
Sensel
MG
,
Chelstrom
L
,
Ek
O
, et al
Prognostic significance of B-lineage leukemic cell growth in SCID mice: a Children's Cancer Group Study
.
Leuk Lymphoma
1998
;
30
:
503
14
.
13.
Pearce
DJ
,
Taussig
D
,
Zibara
K
,
Smith
LL
,
Ridler
CM
,
Preudhomme
C
, et al
AML engraftment in the NOD/SCID assay reflects the outcome of AML: implications for our understanding of the heterogeneity of AML
.
Blood
2006
;
107
:
1166
73
.
14.
Notta
F
,
Mullighan
CG
,
Wang
JC
,
Poeppl
A
,
Doulatov
S
,
Phillips
LA
, et al
Evolution of human BCR-ABL1 lymphoblastic leukaemia-initiating cells
.
Nature
2011
;
469
:
362
7
.
15.
Schmitz
M
,
Breithaupt
P
,
Scheidegger
N
,
Cario
G
,
Bonapace
L
,
Meissner
B
, et al
Xenografts of highly resistant leukemia recapitulate the clonal composition of the leukemogenic compartment
.
Blood
2011
;
118
:
1854
64
.
16.
Taussig
DC
,
Miraki-Moud
F
,
Anjos-Afonso
F
,
Pearce
DJ
,
Allen
K
,
Ridler
C
, et al
Anti-CD38 antibody-mediated clearance of human repopulating cells masks the heterogeneity of leukemia-initiating cells
.
Blood
2008
;
112
:
568
75
.
17.
Malaise
M
,
Neumeier
M
,
Botteron
C
,
Dohner
K
,
Reinhardt
D
,
Schlegelberger
B
, et al
Stable and reproducible engraftment of primary adult and pediatric acute myeloid leukemia in NSG mice
.
Leukemia
2011
;
25
:
1635
9
.
18.
Mazurier
F
,
Doedens
M
,
Gan
OI
,
Dick
JE
. 
Rapid myeloerythroid repopulation after intrafemoral transplantation of NOD-SCID mice reveals a new class of human stem cells
.
Nat Med
2003
;
9
:
959
63
.
19.
Spiegel
A
,
Kollet
O
,
Peled
A
,
Abel
L
,
Nagler
A
,
Bielorai
B
, et al
Unique SDF-1-induced activation of human precursor-B ALL cells as a result of altered CXCR4 expression and signaling
.
Blood
2004
;
103
:
2900
7
.
20.
Manz
MG
. 
Human-hemato-lymphoid-system mice: opportunities and challenges
.
Immunity
2007
;
26
:
537
41
.
21.
Rongvaux
A
,
Willinger
T
,
Takizawa
H
,
Rathinam
C
,
Auerbach
W
,
Murphy
AJ
, et al
Human thrombopoietin knockin mice efficiently support human hematopoiesis in vivo
.
Proc Natl Acad Sci U S A
2011
;
108
:
2378
83
.
22.
Wunderlich
M
,
Chou
FS
,
Link
KA
,
Mizukawa
B
,
Perry
RL
,
Carroll
M
, et al
AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF, GM-CSF and IL-3
.
Leukemia
2010
;
24
:
1785
8
.
23.
Bonnet
D
,
Bhatia
M
,
Wang
JC
,
Kapp
U
,
Dick
JE
. 
Cytokine treatment or accessory cells are required to initiate engraftment of purified primitive human hematopoietic cells transplanted at limiting doses into NOD/SCID mice
.
Bone Marrow Transplant
1999
;
23
:
203
9
.
24.
Feuring-Buske
M
,
Gerhard
B
,
Cashman
J
,
Humphries
RK
,
Eaves
CJ
,
Hogge
DE
. 
Improved engraftment of human acute myeloid leukemia progenitor cells in beta 2-microglobulin-deficient NOD/SCID mice and in NOD/SCID mice transgenic for human growth factors
.
Leukemia
2003
;
17
:
760
3
.
25.
Morisot
S
,
Wayne
AS
,
Bohana-Kashtan
O
,
Kaplan
IM
,
Gocke
CD
,
Hildreth
R
, et al
High frequencies of leukemia stem cells in poor-outcome childhood precursor-B acute lymphoblastic leukemias
.
Leukemia
2010
;
24
:
1859
66
.
26.
Ballen
KK
,
Valinski
H
,
Greiner
D
,
Shultz
LD
,
Becker
PS
,
Hsieh
CC
, et al
Variables to predict engraftment of umbilical cord blood into immunodeficient mice: usefulness of the non-obese diabetic—severe combined immunodeficient assay
.
Br J Haematol
2001
;
114
:
211
8
.
27.
Kong
Y
,
Yoshida
S
,
Saito
Y
,
Doi
T
,
Nagatoshi
Y
,
Fukata
M
, et al
CD34+CD38+CD19 +as well as CD34+CD38-CD19+ cells are leukemia-initiating cells with self-renewal capacity in human B-precursor ALL
.
Leukemia
2008
;
22
:
1207
13
.
28.
Notta
F
,
Doulatov
S
,
Dick
JE
. 
Engraftment of human hematopoietic stem cells is more efficient in female NOD/SCID/IL-2Rgc-null recipients
.
Blood
2010
;
115
:
3704
7
.
29.
Cox
CV
,
Evely
RS
,
Oakhill
A
,
Pamphilon
DH
,
Goulden
NJ
,
Blair
A
. 
Characterization of acute lymphoblastic leukemia progenitor cells
.
Blood
2004
;
104
:
2919
25
.
30.
Cobaleda
C
,
Gutiérrez-Cianca
N
,
Pérez-Losada
J
,
Flores
T
,
García-Sanz
R
,
González
M
, et al
A primitive hematopoietic cell is the target for the leukemic transformation in human philadelphia-positive acute lymphoblastic leukemia
.
Blood
2000
;
95
:
1007
13
.
31.
le Viseur
C
,
Hotfilder
M
,
Bomken
S
,
Wilson
K
,
Röttgers
S
,
Schrauder
A
, et al
In childhood acute lymphoblastic leukemia, blasts at different stages of immunophenotypic maturation have stem cell properties
.
Cancer Cell
2008
;
14
:
47
58
.
32.
Heidenreich
O
,
Vormoor
J
. 
Malignant stem cells in childhood ALL: the debate continues!
Blood
2009
;
113
:
4476
7
,
author reply 4477
.
33.
Quintana
E
,
Shackleton
M
,
Sabel
MS
,
Fullen
DR
,
Johnson
TM
,
Morrison
SJ
. 
Efficient tumour formation by single human melanoma cells
.
Nature
2008
;
456
:
593
8
.
34.
Anderson
K
,
Lutz
C
,
van Delft
FW
,
Bateman
CM
,
Guo
Y
,
Colman
SM
, et al
Genetic variegation of clonal architecture and propagating cells in leukaemia
.
Nature
2011
;
469
:
356
61
.