Adult umbilical cord blood transplantation: a comprehensive review.

Abstract

Recent registry studies have established umbilical cord blood (UCB) transplantation as a safe and feasible alternative to bone marrow transplantation in adults when no sibling donor is available. There is, however, no gold standard to guide optimal treatment choices. We review here factors leading to the choice of the 'best available donor' and 'best available unit' in the case of UCB. For instance, it is clear that higher cell dose may partially overcome the negative impact of certain histocompatibility leukocyte antigen (HLA) disparities in UCB transplantation, leading us to choose the more closely HLA-matched unit with a cell dose >2.5 x 10(7)/kg. New approaches in adult UCB transplantation are systematically covered, with a quantitative appreciation of the evidence available to date. Reduced intensity conditioning, for example, broadens the range of potential recipients by reducing transplant-related mortality, but suffers from unproven risks and benefits long term. Potential advantages of multiple units over single unit transplants are discussed, with a particular emphasis on confounding factors that impact interpretation. The limited clinical results of ex vivo UCB expansion, the possible benefits of co-infusion of haploidentical cells and controversial issues (e.g. killer immunoglobulin-like receptor matching and alternative graft sources) are also addressed with a debate on the future of UCB transplantation.

Full text

Page 1

REVIEW
Adult umbilical cord blood transplantation: a comprehensive review
H Schoemans1,2, K Theunissen1, J Maertens1, M Boogaerts1,3, C Verfaillie2,4and J Wagner4,5
1Department of Hematology – Gasthuisberg University Hospital Leuven, Leuven, Belgium;2Stem Cell Institute Leuven (SCIL),
Leuven, Belgium;3Leuven Cancer Institute, Leuven, Belgium;4Stem Cell Institute, University of Minnesota, McGuire Translational
Research Facility, Minneapolis, MN, USA and5MMC 366, University of Minnesota, Minneapolis, MN, USA
Recent registry studies have established umbilical cord
blood (UCB) transplantation as a safe and feasible
alternative to bone marrow transplantion in adults when
no sibling donor is available. There is, however, no gold
standard to guide optimal treatment choices. We review
here factors leading to the choice of the ‘best available
donor’ and ‘best available unit’ in the case of UCB. For
instance, it is clear that higher cell dose may partially
overcome the negative impact of certain histocompatibility
leukocyte antigen (HLA) disparities in UCB transplant-
ation, leading us to choose the more closely HLA-matched
unit with a cell dose 42.5?107/kg. New approaches in
adult UCB transplantation are systematically covered,
with a quantitative appreciation of the evidence available
to date. Reduced intensity conditioning, for example,
broadens the range of potential recipients by reducing
transplant-related mortality, but suffers from unproven
risks and benefits long term. Potential advantages of
multiple units over single unit transplants are discussed,
with a particular emphasis on confounding factors that
impact interpretation. The limited clinical results of
ex vivo UCB expansion, the possible benefits of co-infusion
of haploidentical cells and controversial issues (e.g. killer
immunoglobulin-like receptor matching and alternative
graft sources) are also addressed with a debate on the
future of UCB transplantation.
BoneMarrow Transplantation
doi:10.1038/sj.bmt.1705403; published online 5 June 2006
Keywords:
umbilical cord blood; allogeneic transplant-
ation; stem cells; adults
(2006)38,83–93.
Evidence supporting the efficacy of umbilical cord blood
(UCB) transplantation in adults has significantly increased
over the past years, as it now becomes a standard
alternative to bone marrow transplantation (BMT) in some
hematopoietic stem cell transplantation (HSCT) centers. As
of August 2005, there were 3724 UCB transplantations
recorded in the NETCORD inventory, almost half of
which performed in adults (n¼1405). This article sum-
marizes the evidence available to date supporting the
efficacy of umbilical cord blood transplant (UCBT) in
adults based on published UCB transplant clinical trials
and puts into perspective the various approaches currently
under investigation to improve these results.
We have reviewed the current literature on UCB trans-
plantation in adults, from January 2001 until December
2005. Clinical cohorts containing less than 10 patients,
reviews lacking original results or abstracts with incomplete
or non-concordant information have been excluded.
Contact with authors has been made to limit the risk of
duplication between reports from individual institutions or
multi-institutional studies.
Comparing UCBT to BMT
The three most distinctive features distinguishing unrelated
donor UCB transplantation from peripheral blood stem
cells (PBSC) or BMT (Table 1) are: (1) the number of stem
cells available for transplantation, (2) the speed of their
availability and (3) the histocompatibility leukocyte antigen
(HLA) matching requirements.
Cell dose
It is clear that transplantation outcome after UCB
transplantation is correlated with the cell dose infused: a
threshold must be reached to get consistent engraftment
and lower incidence of transplant-related events.1–3Cell
dose also directly correlates with rate of neutrophil and
platelet recovery1–3such that recipients of higher cell doses
have significantly more rapid recovery as compared to
those with lower cell doses. The current empirically
accepted threshold limits are 1.7?105infused CD34þ
cells/kg3or 2.5?107cyropreserved nucleated cells/kg.4
UCB transplantation requires consequently about one log
less cells than BMT, possibly because of the high
proliferation index of the infused UCB stem cells.
Unfortunately, UCB stem and progenitor cells are intrinsi-
cally limited by the amount that can be collected from a
placenta, thereby restricting the choice of the recipient
as body weight becomes a limiting factor. Additional cell
Received 7 February 2006; revised 27 April 2006; accepted 28 April 2006;
published online 5 June 2006
Correspondence: Dr H Schoemans, Stem Cell Institute Leuven (SCIL),
O.& N., Herestraat 49 – bus 804, BE-3000 Leuven, Belgium.
E-mail: helene.schoemans@med.kuleuven.be
Bone Marrow Transplantation (2006) 38, 83–93
& 2006 Nature Publishing Group All rights reserved 0268-3369/06 $30.00
www.nature.com/bmt

Page 2

infusions from the same donor are also rendered impos-
sible, since the whole UCB unit needs to be used for the
primary transplantation.
Graft availability
UCB units are almost immediately available for transplant
as they are fully HLA-typed before storage without risk of
donor morbidity or attrition. According to Barker et al.,5
the median time to donor identification is 13.5 days (range:
2–387 days) for a UCB unit in contrasts to 49 days (range:
32–393 days) for bone marrow. With the expansion of the
UCB banks, the search of a unit now takes now about 1
day for University of Minnesota searches, whereas an
unrelated PBSCT/BMT donor search will take an average
of 3–4 months.
Such rapid availability can be particularly useful for
patients with high-risk malignancy or rapidly progressive
non-malignant diseases. Unfortunately, the clinical signifi-
cance of this potential advantage is as yet unknown since
most studies published to date first ruled out the possibility
of an unrelated BMT before UCB transplantation was
considered (Hamza et al.6showed that the time from
diagnosis to transplant was 21.7 months for cord blood,
and 8.2 months for bone marrow), except if the practitioner
considered the recipient was too sick to wait any longer.
This reflects the fact that UCB has historically been a
second choice and that only high-risk patients were offered
this stem cell source.
HLA matching
UCB is less restricted with regards to HLA matching
requirements relative to bone marrow stem cells from adult
donors.7–9In fact, Ko ¨ gler et al. have recently reported that
HLA matching was only predictive for survival after UCB
transplantation if performed at the serological level for
HLA-A and HLA-B, and high resolution allelic typing for
HLA-DRB1. Any other attempts to better define compat-
ibility through high resolution typing failed to correlate
with survival, in sharp contrast to unrelated BMT where
any single serological mismatch or multiple high resolution
mismatch are considered as risk factors.10
This permissive HLA mismatching consequently in-
creases the number of potential units available per patient,
thereby improving the odds of finding a suitable unit for
patients with uncommon tissue types,9without increased
risk of engraftment failure or graft-versus-host reactions. It
was recently shown that patients had a 99% chance of
finding a 4/6 HLA matched UCB unit, and a 70% chance
of finding a well-matched UCB unit (defined as 5/6 or 6/6
HLA match, without mismatch in the graft-versus-host
disease (GVHD) direction) within the New York Blood
Center (NYBC) inventory, with further limitations being
rather linked to the cell content of the unit.11
This apparent immune-tolerance may be attributed to
the immaturity of the cord blood lymphocytes, as suggested
by the observation that: (1) activated UCB mononuclear
cells (MNC) have a different mRNA and protein expres-
sion proper for several growth factors and interleukins
important in immune cells as compared to normal adult
peripheral blood MNC, (2) UCB cells generate decreased
T-helper-1(Th1)-type cytotoxic cellular responses to mito-
genic stimulation, (3) the phenotype of UCB B-cells is more
immature than in normal adults, (4) there is a lower
number of CD4þ/CD45þ T-cells in UCB, (5) the activity
of the UCB CD4þ/CD45þ T-cells and natural killer-cells
(NK-cells) is relatively low and (6) most UCB dendritic
cells tend to have a lymphoid and immature phenotype,
with a decreased capacity to produce IL-12 (thus further
impairing Th1 response).12
Beyond HLA-matching, killer immunoglobulin-like re-
ceptor (KIR)-ligand mismatch is an area of intense
investigation, based on the search for ‘perfect mismatch’
between donor NK-cells possessing an inhibitory KIR
receptor and recipient cells lacking the corresponding
HLA, theoretically leading to selective killing of residual
host leukemia cells and antigen presenting cells. Conse-
quently, better survival through increased GVL and
decreased GVHD might be expected. Clinical results with
PBSC or BMT, however, show relatively inconsistent
results, potentially modulated by a variability of the
Table 1
Comparison of the main features of UCBT and BMT (adapted from Grewal et al.57)
UCB transplantPBSC* – BMT**
Number of available donors (worldwide), as of
10/2005 (www.bmdw.org)58
208000 units
208000 (100%) ABDR typed,
189000 (91%) DNA class II typed
and 113000 (54%) DNA class I
typed.
Fixed unit cell content
B2.5?107/kg
9840000 donors
6650000 (68%) ABDR typed,
5770000 (59%) DNA class II typed
and 3880000 (39%) DNA class I typed
Major limiting factor
Minimum number of total nucleated cells (TNC)
needed for transplantation
Second graft or DLI
Median Speed of donor availability
Donor morbidity
HLA match and donor attrition
B2.0?108/kg
Impossible
B1 day
None
Possible
B3–4 months
(50–68%)* – (80–85%)** fatigue, local
pain and/or lower-back pain59–61
Median time to return to normal activity:
7 days59
Possible
None
Mostly 8/8
EBV/CMV transmission to recipient
Risk for transmission of congenital disease
Standard HLA match requirements
Negligible
Theoretically possible
Minimum 4/6
Adult UCB transplantation: a comprehensive review
H Schoemans et al
84
Bone Marrow Transplantation

Page 3

transcriptional KIR repertoire and the impact of viral
reactivations and infections.13
et al.14recently showed that KIR-ligand mismatch in
UCB transplant recipients seemed to have no effect
on overall survival, relapse graft failure or GVHD, but
were predictive of increased transplant-related mortality
(TRM).
Interestingly, Brunstein
Choosing the ‘best’ unit
Choosing ‘the best unit’ for transplantation remains a
challenge, as no study has formally demonstrated to date
which parameter has higher priority (cell dose or HLA
match). While higher cell dose may partially overcome the
negative impact of HLA disparity,3the best matched unit
with a cell dose 42.5?107cryo-preserved nucleated cells
per kilogram should be used.4Further studies, however, are
needed to more precisely define this algorithm. This also
calls for rationalization of the methods of determining both
compatibility (serological versus high resolution allelic
typing) and cell dose (total nucleated cells versus CD34þ
fraction).
Studies comparing UCB transplantation and
unrelated BMT
Prospective comparative studies exploring the differences in
outcomes of UCB transplant versus unrelated BMT have
yet to be performed. The most recent reports concerning
adult patients are from the International Bone Marrow
Transplant Registry (IBMTR)/NYBC,15the European
Bone Marrow transplant (EBMT) consortium16and a
single center retrospective Japanese study17(Table 2).
Although retrospective, they demonstrate the feasibility
of UCB transplantation in adults and highlight major
differences between the two stem cell sources. Several other
US,18,19European20–22and Japanese23–28authors have
published earlier smaller scale reports concerning myelo-
ablative, single unit UCB transplantation in adults, but
these are not described here to avoid overlap with the
larger reports.
As detailed in Table 2, the most frequent indications for
transplantation were acute and chronic leukemia. The UCB
grafts contained a median of 2.2–2.5?107total nucleated
cells (TNC) per kilogram, zero to three HLA mismatches
and were administered after a myeloablative conditioning.
Engraftment was slower in the UCB cohort, with a time to
neutrophil recovery ranging from 22 to 27 days, and poorer
engraftment as reported by the EBMT (75% at day 60). In
spite of the high level of HLA disparity in the UCB group,
similar rates of acute GVHD (aGVHD) were observed as
compared to the HLA-matched unrelated BMT in the US
study. Rocha et al. demonstrated less aGVHD and
Takahashi et al. demonstrated less high-grade aGVHD in
recipients of HLA-mismatched unrelated UCB transplant-
ation as compared to HLA-matched unrelated BMT;
chronic GVHD (cGVHD), however, was found to be
similar. In the US study, a higher risk of cGVHD in the
UCB cohort was observed, but with a smaller fraction of
extensive disease.
Although follow-up remains limited (25–40 months), we
note similar relapse rates in all groups (range: 16–23% for
UCB transplantation versus 23–25% for unrelated BMT at
2 years), and similar overall survival (range: 26–74% versus
35–44% at minimum 2 years for UCB transplantation and
unrelated BMT, respectively). Notably, Takahashi et al.
showed a survival of 74% at 2 years in the UCB transplant
cohort, which is markedly higher than that observed in
recipients of unrelated BMT.
In summary, Laughlin et al. concluded that UCBT was
safe, considering its results were comparable to those
observed in recipients of one-HLA-mismatch unrelated
BMT, but inferior to a fully HLA-matched unrelated
BMT. The EBMT report considered both transplantation
modalities (HLA-mismatched UCB transplantation versus
HLA-matched unrelated BMT) to be equivalent and
Takahashi et al. concluded that UCB transplantation was
superior to unrelated BMT in the light of their survival
results.
Discrepancies in results between those three reports have
been commented on at length after their publication. The
difference in results between the IBMTR/NYBC and
EBMT reports could be explained by several factors: lower
mean weight of the European population, difference in
observation periods (which included the pioneering period
of UCB transplantation in the US study) and differences in
HLA disparity (greater HLA mismatch in the IBMTR/
NYBC cohort).29Furthermore, the impressive results of
the Japanese study might be accounted for by specific
characteristics such as: an extensive experience with
UCB in more than 555 adults to date (Takahashi 2004,
personal communication), lower median weight of Japanese
patients, more homologous HLA genotype on the island,24
prolonged hospitalization of patients after transplant, the
high proportion of limited cGVHD and possibly differ-
ences in conditioning therapy.25
In addition to these comparative studies, the Cord Blood
Transplantation (COBLT) Study reported the first pro-
spective multi-center UCB transplantation study in 32
adults.30This study is remarkable by its unexpectedly poor
engraftment (median 31 days for neutrophil recovery, with
75% engraftment at day 42) and very poor 1-year survival
estimate (17% at 1 year) all attributed to the high-risk
profile of the transplant population.
Improving transplantation outcome by reducing TRM
Even if the greatest body of evidence available to date
undeniably concerns myeloablative single unit UCBT,
several approaches have been proposed to circumvent the
slow engraftment kinetics and high TRM of conventional
myeloablative UCB transplantation (Table 3). The use of
reduced intensity conditioning, co-infusion of multiple
UCB units, expansion of the UCB progenitor cells and
co-infusion of PBSC have been tested with varying degrees
of popularity. Many other tactics are also currently being
explored but remain at the preclinical or case-report level,
for instance: improvement of bone marrow homing via
inhibition of the truncation of stromal cell-derived factor-1
(SDF-1)/CXCL12,31UCB-derived somatic stem cells32or
in vivo injection of SCF and filgrastim in adult recipients of
unrelated UCBT.33
Adult UCB transplantation: a comprehensive review
H Schoemans et al
85
Bone Marrow Transplantation

Page 4

Table 2
Studies comparing single unit Unrelated UCB Transplantation and Matched Unrelated BMT in Adults (2001–2005)
Laughlin et al.15
Multicentric registry based (IBMTR+NYBC)
1996–2001
Rocha et al.16
Multicentric registry based (Eurocord and EBMT)
1998–2002
Takahashi et al.17
Single center (retrospective)
1996–2003
UCB TransplantationUnrelated BMT UCB Transplantation Unrelated BMT UCB TransplantationUnrelated BMT
Number of patients
Age (years)
Weight (kg)
150 36798584 6845
31 (16–58)
68 (44–133)
37 (16–60)
76 (40–156)
24 (15–55)
58 (38–92)
32 (15–59)
68 (40–108)
36 (16–53)
55 (36–76)
26 (16–50)
60 (37–85)
Diagnosis
AML
ALL
CML
Other
68 (45%)
45 (30%)
37 (25%)
0
140
82
145
0
45 (46%)
53 (54%)
0
0
317 (54%)
267 (46%)
0
0
39 (57%)
15 (22%)
5 (7%)
10 (14%)
15 (33%)
8 (18%)
18 (40%)
4 (9%)
Donor Recipient match
0 Mismatch
1 Mismatch
2 Mismatch
X3 Mismatch
0 367 (100%)
0
0
0
6 (6%)
48 (51%)
37 (39%)
4 (4%)
584 (100%)
0
0
0
039 (87%)
6 (13%)
0
0
34 (23%)
116 (77%)
0
14 (21%)
37 (54%)
17 (25%)
Cell dose median
TNC (?107/kg weight) 2.2 (1.0–6.5)y
24 (0.2–170) 2.3 (0.9–6.0)y
29 (10–90)2.5 (1.1–5.3)y
33 (6.6–50)
Hematopoietic recovery
Time to ANC engraftment
(days)
Cum. Incid. of neutrophil
recovery
Graft failure/secondary
graft rejection
27 (25–29)*y
18 (18–19) 26 (14–80)y
19 (5–72)22 (16–41)y
18 (12–33)
Not Av.Not Av.75% (66–84)* at d60y
89% (87–91)* at d60 92% (85–99)* at d42100% at d42
Not Av.Not Av. 20 (20%)y
43 (7%) 5 (8%)y
0
GVHD
Acute GVHD II–IV
(% Cum. Incid.)
Chronic GVHD
(% Cum. Incid.)
61/150 pt 176/367 pt26% (14–38)* at d100y
39% (31–47)* at d10030/60 pt30/45 pt
35/69 pty
86/243 pt 30% (20–40)* at 2 years46% (44–48)* at 2 years 42/54 pt 26/35 pt
Statistical analysis
Median follow-up
(months)
TRM (% Cum. Incid.)
40 (12–85)48 (12–78)27 (3–66)24 (1–76) 26 (4–68)14 (1–100)
95/150ypt169/367 pt 44% at 2 years38% at 2 years 9% (2–16)* at 1–2yearsy
29% (15–42)* at
1–2years
44% (30–59)* at 2 years
25% (12–37)* at 2 years
Median overall survival %
Relapse (% Cum. Incid.)
26% (19–32)* at 3 yearsy
Not Av.
35% (30–39)* at 3 years
Not Av.
36% at 2 years
23% at 2 years
42% at 2 years
23% at 2 years
74% (63–85)* at 2 yearsy
16% (7–25)* at 2 years
All values are given as median (% or range), unless otherwise specified.
Abbreviations: AML¼acute myeloid leukemia; ALL¼acute lymphocytic leukemia; ANC¼absolute neutrophil count; CML¼chronic myeloid leukemia; Cum. Incid.¼cumulative incidence; GVHD¼graft-
versus-host disease; Not Av.¼not available; Pt¼patients; TNC¼total nucleated cells; TRM¼transplant-related mortality.
A superscript ‘y’ indicates a significant difference between UCB transplantation and unrelated BMT cohorts.
*95% confidence interval.
Adult UCB transplantation: a comprehensive review
H Schoemans et al
86Bone Marrow Transplantation

Page 5

Reduced intensity conditioning in the setting of UCBT
From 2001 to 2005, the early results of about 330 reduced
intensity conditioning (RIC) UCBT have been published
(Table 4 shows the largest studies published by each
group34–41), with a follow-up of 7.5–23 months. As
expected, patients were 10–20 years older than those
included in myeloablative studies and transplant indica-
tions crossed a wide range of hematological diseases. The
number of days to neutrophil recovery were highly variable
(range: 12–26), with relatively high rates of graft failure or
secondary graft rejection (range: 14–67%). Interpretation
of this data is difficult as some of these studies were carried
out with multiple UCB units and conditioning regimens
varied between centers. Recent analysis also suggested a
positive correlation between engraftment and the timing of
chemotherapy received until transplantation,42,43a para-
meter which is not frequently mentioned in studies, and
thus hardly comparable retrospectively. GVHD was
roughly similar to that observed after myeloablation
(aGVHD: 20–61 and 26% for RIC and myeloablative
conditioning, respectively; cGVHD: 21–26 versus 30%,
respectively), with one notable exception: in a direct
comparison of myeloablative versus RIC in a single
center study, significantly more grade III–IV aGVHD was
observed in the RIC group.35Finally, in these high-risk,
heavily pre-treated patients with significant comorbidities
at the time of UCBT, survival was comparable to that
observed in younger, fitter patients transplanted after a
myeloablative therapy (31–80% Kaplan Meier survival at
1 year), probably relating to higher TRM and/or relapse
rates. Unfortunately, this comparison suffers lack of
adjustment for disease stage at time of UCB transplant,
as it is not systematically reported. Reassuring disease-free
survival results (64–79% at 1 year), also seem to indicate
that GVL effect is preserved despite the low T-cell dose,40
although this needs confirmed once longer follow-up is
achieved.
Multiple unit UCB transplantation
Results of double unit UCB transplantation in 142 patients
have been reported (Table 5), the majority having taken
place at the University of Minnesota.34,40,41,44,45These
studies are recent and offer follow-up periods ranging from
7 to 23 months. We note a wide range of neutrophil
engraftment, partially explained by the different condition-
ings used (12–26 days), but impressively low frequencies of
graft failure (0–22%). Interestingly, only one of the two
units infused predominated over time (by day 100, most
patients have chimerism derived from one of the two
units), with no factor predicting which unit would prevail
(Wagner, personal communication 2005). Incidence of
grade II–IV aGVHD appeared to be slightly higher and
cGVHD seemed to lie in the same range as that previously
described (44–65% for aGVHD and 21–25% for cGVHD).
TRM rates remained low (14–48%) and 1-year overall
survival ranged from 31 to 79%. UCB transplantation
with 2 units appears therefore safe and feasible both in
the myeloablative and non-myeloablative setting; extending
the application of this treatment to almost all adults
for whom a single cord blood unit would have been
insufficient.
Some recent studies compare single unit to double unit
UCB transplantation. The retrospective comparative ana-
lysis of the University of Minnesota data set demonstrated
high engraftment and less relapse, but increased grade II–
IV aGVHD, with possibly improved survival with double
unit UCB transplantation.46Although this might suggest
an improved GVL effect, several other factors differed
between the two groups. As double UCB unit transplants
were initiated, the standard conditioning regimen also
shifted from ATG to fludarabine and GVHD prophylaxis
was changed from methylprednisone/cyclosporine to myco-
phenolate mofetyl/cyclosporine. Analysis of the NYBC
data set suggested shorter time to engraftment and better
overall survival in patients treated with fludarabine-
containing regimen and double UCB transplantation.47
Creer et al.48also showed better time to engraftment in the
double unit UCB transplant group (17 versus 20 days)
and better survival (14 versus 80%), essentially because of
the absence of deaths due to opportunistic infections in the
double UCB transplant cohort. These results highlight the
success of using double UCB units but the underlying
mechanisms are still unclear. Prospective trials are now
Table 3
Selection of major studies relative to UCB transplantation in adults published between 2001 and 2005
Total Myeloablative Nonmyeloablative
Standard single UCB
unit
573 Patientsa
350 Patients: 150 patients,1598 patients,16
68 patients,1734 patients30
223 Patients: 18 patients,3721 patients,39
13 patients,36129 patients,3813 patients,35
17/95 patients received single unitsa,34
12/21 patients received single unitsa 41
Multiple UCB units 142 Patientsa
34 Patients: 23 patients,4411 patients45
108 Patients: 21 patients,4078/95 patients
received double unitsa,349/21 patients
received double unitsa 41
Expansion of UCB units64 Patients
(27 adults)
64 patients (27 adults): 37 patients
(25 adults),4927 patients (2 adults)50
None
Co-infusion of peripheral
blood CD34+cells
28 Patients28 patients: 28 patients53
None
Total: 439 adult patients Total: 331 adult patients
aThere is some overlap between reports, since some studies report both single and double unit cord blood transplants and some reports include registry data.
Adult UCB transplantation: a comprehensive review
H Schoemans et al
87
Bone Marrow Transplantation