Bone Marrow Transplantation (2015) 50, 540–544
© 2015 Macmillan Publishers Limited All rights reserved 0268-3369/15
www.nature.com/bmt
ORIGINAL ARTICLE
Physiochemical disparity of mismatched HLA class I
alloantigens and risk of acute GVHD following HSCT
V Kosmoliaptsis1,2, MM Jöris3,4, DH Mallon1, AC Lankester5, PA von dem Borne6, J Kuball7, M Bierings8, JJ Cornelissen9,
ME Groenendijk–Sijnke10, B van der Holt10, JA Bradley1, M Oudshoorn3,4, JJ van Rood3,4, CJ Taylor2 and FHJ Claas4
We determined whether assessment of the immunogenicity of individual donor–recipient HLA mismatches based on differences in
their amino-acid sequence and physiochemical properties predicts clinical outcome following haematopoietic SCT (HSCT). We
examined patients transplanted with 9/10 single HLA class I-mismatched grafts (n = 171) and 10/10 HLA-A-, -B-, -C-, -DRB1- and
-DQB1-matched grafts (n = 168). A computer algorithm was used to determine the physiochemical disparity (electrostatic mismatch
score (EMS) and hydrophobic mismatch score (HMS)) of mismatched HLA class I specificities in the graft-versus-host direction.
Patients transplanted with HLA-mismatched grafts with high EMS/HMS had increased incidence of ⩾ grade II acute GVHD (aGVHD)
compared with patients transplanted with low EMS/HMS grafts; patients transplanted with low and medium EMS/HMS grafts had
similar incidence of aGVHD to patients transplanted with 10/10 HLA-matched grafts. Mortality was higher following single HLAmismatched HSCT but was not correlated with HLA physiochemical disparity. Assessment of donor–recipient HLA incompatibility
based on physiochemical HLA disparity may enable better selection of HLA-mismatched donors in HSCT.
Bone Marrow Transplantation (2015) 50, 540–544; doi:10.1038/bmt.2014.305; published online 26 January 2015
INTRODUCTION
Haematopoietic SCT (HSCT) represents a potentially curative
treatment option for life-threatening haematological disorders such
as leukaemias and bone marrow failure syndromes. However, the
diversity of the human MHC system and the limited number of stem
cell donors necessitates the selection of HSCT donor–recipient pairs
who are not fully HLA matched. Incompletely HLA-matched grafts
are associated with increased morbidity, including acute GVHD
(aGHVD), and mortality following transplantation.1 The ability,
therefore, to select HLA-mismatched donor grafts that are of low
immunogenicity for a given recipient is of great interest in HSCT.
It has been suggested that alloimmune responses in the context
of transplantation may be predicted based on information derived
from the amino-acid (AA) sequences of HLA molecules.2 However,
previously developed tools for assessing HLA immunogenicity were shown to be unsuitable for predicting cellular
alloreactivity.3,4 We have recently proposed an algorithm to assess
HLA compatibility in HSCT based on the number, position and
nature of AA polymorphisms in the α1/α2 domains of mismatched
donor–recipient HLA class I molecules.5 Although this proved
useful in predicting donor cytotoxic T lymphocyte precursor
reactivity against mismatched recipient HLA molecules in vitro, the
algorithm was not predictive of clinical outcomes, such as aGVHD
and mortality, following single HLA mismatch HSCT.6 In the
context of solid-organ transplantation, the immunogenicity of HLA
class I and class II alloantigens relates not only to the number of
AA polymorphisms between mismatched HLA molecules but also
to their physiochemical differences.7,8 In this report, we describe
an analysis of fully matched and single HLA class I-mismatched
grafts and show that the physiochemical disparity between
mismatched HLA alloantigens is predictive of the risk of
developing aGVHD following HSCT.
MATERIALS AND METHODS
Donor–recipient pairs
The study cohort consisted of 171 consecutive patients who received a
9/10 single HLA-A (n = 60), HLA-B (n = 27) or HLA-C (n = 84) mismatched
HSCT graft between 1988 and 2008 in three centres in the Netherlands
(Erasmus MC Rotterdam, University Medical Centre Utrecht and Leiden
University Medical Centre). As controls we selected 168 10/10 HLA-A, -B, -C,
-DRB1 and -DQB1 allele-matched donor–recipient pairs, matched for HSCT
year, donor type, patient age and disease diagnosis to the single HLA class
I-mismatched cohort (total cohort: n = 339). The study population included
pediatric (n = 112) and adult patients (n = 227) with malignant (n = 296) and
nonmalignant (n = 43) diseases. Donors originated from national or
international volunteer bone marrow donor registries, and 20 were family
donors. Clinical patient information was made available by the HOVON
Data Centre through the European group for Blood and Marrow
Transplantation patient database.
HLA genotyping, AA sequencing and physicochemical properties
All HSCT donors and patients were HLA typed at allele level for
HLA-A, -B, -C, -DRB1 and -DQB1, as described previously.9 The AA sequence
comparisons between mismatched HLA class I specificities at the
graft-versus-host direction were performed using a previously
described computer program7 (available on request). The program allows
identification of the position and nature of all disparate AAs in the entire
1
Department of Surgery, University of Cambridge, and NIHR Cambridge Biomedical Research Centre, Cambridge, UK; 2Tissue Typing Laboratory, Cambridge University
Hospitals NHS Foundation Trust, Addenbrooke’s Hospital, Cambridge, UK; 3Europdonor Foundation, Leiden, The Netherlands; 4Department of Immunohematology and Blood
Transfusion, LUMC, Leiden, The Netherlands; 5Department of Pediatrics, LUMC, Leiden, The Netherlands; 6Department of Hematology, LUMC, Leiden, The Netherlands;
7
Department of Hematology and Immunology, UMCU, Utrecht, The Netherlands; 8Wilhelmina Children’s Hospital, UMCU, Utrecht, The Netherlands; 9Department of Hematology,
Erasmus MC Cancer Institute, Rotterdam, The Netherlands and 10HOVON Data Centre, Erasmus MC Cancer Institute-Clinical Trial Center, Rotterdam, The Netherlands.
Correspondence: Dr V Kosmoliaptsis, Academic Department of Surgery, Box 202, Addenbrooke’s Hospital, Cambridge CB2 0QQ, UK.
E-mail: [email protected]
Received 27 May 2014; revised 9 November 2014; accepted 26 November 2014; published online 26 January 2015
Physiochemical HLA disparity and acute GVHD
V Kosmoliaptsis et al
541
α1 and α2 domains. For the purposes of this study, focussing on cellular
allorecognition, the program was adjusted to enable separate consideration of residues that are important in TCR recognition and/or peptide
binding. Accordingly, AA at positions 49–85 and 137–180 in the α-helices
were considered important for TCR contact and residues at positions 3–13,
21–29, 31–38, 45–48, 93–104, 108–119, 121–127 and 132–136 in the
β-sheets were considered important for peptide binding.10–13 Donor–
recipient AA mismatches at each sequence position were scored according
to their hydrophobic (using the Hopp–Woods hydrophobicity scale14) and
isoelectric point disparities. For each donor–recipient HLA class I mismatch
combination, the individual AA physiochemical disparity scores were
summed to give a total hydrophobicity mismatch score (HMS) and a total
electrostatic mismatch score (EMS).7
End points and competing risks
Primary end point for the HSCT analysis was incidence of ⩾ grade II aGVHD.
Peak severity of aGVHD after HSCT defined according to grade (no, I, II, III
or IV) was available for 317 patients. Patients were censored at date of
donor leukocyte infusion, relapse or second HSCT within 100 days after
HSCT and at 100 days after HSCT. Death without ⩾ grade II aGVHD within
100 days after HSCT was treated as a competing risk. Secondary outcomes
in the HSCT analysis were incidence of relapse and mortality. Patients were
censored at date of donor leukocyte infusion or second HSCT and those
who survived were censored at time of last contact. For analysis of relapse,
death without relapse was defined as competing risk.
Statistical analysis
For the single 9/10 HLA class I-mismatched grafts, the number of AA
differences, EMS and HMS variables were divided into three groups based
on their distribution (overall, EMS ranged from 0 to 58.9 whereas EMS for
α-helices and β-sheets ranged from 0 to 47.3 and from 0 to 18.2,
respectively; similarly, HMS ranged from 0 to 66.2 whereas HMS for
α-helices and β-sheets ranged from 0 to 45.5 and from 0 to 20.3,
respectively). The lowest tertile was set as reference category. Tertiles were
used to ensure objectivity and even-sized categories.
Association between EMS/HMS of single 9/10 HLA class I mismatches and
HSCT outcome was investigated relative to 10/10 HLA-matched HSCT.
Differences in patient, donor and HSCT characteristics between low, medium
and high EMS/HMS were examined with Pearson’s χ2 test or Fishers’ exact
test for discrete variables and Kruskal–Wallis test for continuous variables.
Hazard ratios, 95% confidence intervals and P-values were obtained using
the proportional hazards model for the subdistribution of competing risks of
Table 1a.
Crude association between donor–recipient HLA EMS and
outcomes following HSCT
N
Univariate analysis
HR (95% CI)
P-value
⩾ Grade II aGVHD
Low EMS
Medium EMS
High EMS
10/10 HLA matched
317
55
49
54
159
1.00
1.28 (0.55–2.99)
1.90 (0.87–4.17)
0.91 (0.44–1.88)
Reference
0.560
0.110
0.790
Relapse
Low EMS
Medium EMS
High EMS
10/10 HLA matched
296b
54
47
47
148
1.00
1.41 (0.70–2.82)
0.87 (0.42–1.81)
1.03 (0.58–1.84)
Reference
0.330
0.710
0.930
Mortality
Low EMS
Medium EMS
High EMS
10/10 HLA matched
339
60
54
57
168
a
1.00
1.10 (0.68–1.78)
0.93 (0.58–1.50)
0.64 (0.43–0.96)
Reference
0.710
0.770
0.030
Abbreviations: aGVHD = acute GVHD; CI = confidence interval; EMS = electrostatic mismatch score; HR = hazard ratio; HSCT = haematopoietic SCT.
a
Information on aGVHD was not available for 22 patients (13 had single
9/10 HLA-mismatched grafts and 9 had 10/10 HLA-matched grafts). bOnly
patients with malignant diseases were considered (N = 296).
© 2015 Macmillan Publishers Limited
Table 1b. Multiple variable analysis: influence of donor and recipient
factors on outcomes following HSCT
Multivariate analysis
HR (95% CI)
P-value
1.00
1.71 (0.68–4.27)
2.82 (1.22–6.53)
0.91 (0.31–2.7)
Reference
0.25
0.016
0.86
Recipient age
Pediatric
Adult (418 years)
1.00
3.83 (1.93–7.6)
Reference
o 0.001
ATG prophylaxis
No
Yes
1.00
1.27 (0.75–2.17)
Reference
0.38
Campath use
No
Yes
1.00
0.61 (0.28–1.32)
Reference
0.21
Calcineurin inhibitor
No
Yes
1.00
1.22 (0.85–1.74)
Reference
0.29
HLA locus
HLA-A or -B
HLA-C
1.00
0.94 (0.66–1.36)
Reference
0.75
T-cell depletion
No
Yes
1.00
0.73 (0.34–1.57)
Reference
0.42
Donor type
Unrelated
Related
1.00
1.15 (0.3–4.43)
Reference
0.84
Relapse
Low EMS
Medium EMS
High EMS
10/10 HLA matched
1.00
1.38 (0.51–3.77)
2.06 (0.75–5.65)
1.20 (0.49–2.95)
Reference
0.52
0.16
0.69
Recipient age
Pediatric
Adult (418 years)
1.00
1.29 (0.75–2.23)
Reference
0.35
Myeloablative conditioning
Yes
No
1.00
1.48 (0.82–2.65)
Reference
0.19
HLA locus
HLA-A or -B
HLA-C
1.00
0.89 (0.64 - 1.24)
Reference
0.48
Graft type
Bone marrow
Peripheral blood
1.00
0.69 (0.38–1.24)
Reference
0.22
1.00
(0.66–1.82)
(0.48–1.38)
(0.29–0.98)
(1.01–1.05)
Reference
0.72
0.45
0.04
0.01
HLA locus
HLA-A or -B
HLA-C
1.00
0.92 (0.73–1.15)
Reference
0.45
Myeloablative conditioning
Yes
No
1.00
0.98 (0.69–1.38)
Reference
0.89
Primary disease
Nonmalignant
Malignant
1.00
1.68 (0.86–3.28)
Reference
0.13
⩾ Grade II aGVHD
Low EMS
Medium EMS
High EMS
10/10 HLA matched
Mortality
Low EMS
Medium EMS
High EMS
10/10 HLA matched
Recipient age at transplant (per year)
1.10
0.82
0.54
1.03
Abbreviations: aGVHD = acute GVHD; ATG = antithymocyte globulin;
CI = confidence interval; EMS = electrostatic mismatch score; HR = hazard
ratio; HSCT = haematopoietic SCT. P-values in bold denote statistical
significance.
Bone Marrow Transplantation (2015) 540 – 544
Physiochemical HLA disparity and acute GVHD
V Kosmoliaptsis et al
542
Fine and Gray15 to quantify the association between EMS/HMS and
cumulative incidence of ⩾ grade II aGVHD, relapse and mortality.
Patient, donor and HSCT characteristics that were associated with EMS or
HMS and HSCT end points were considered as potential confounding
variables in our data set. They were evaluated for each end point separately
and identified as confounding variables when including them in the model
(one by one); the crude association (β estimate) changed by 410%.
For aGVHD, relapse and mortality models, the following variables were
considered: HSCT centre (Erasmus MC, University Medical Centre Utrecht
or Leiden University Medical Centre), mismatched locus (HLA-A, -B or -C),
HSCT year (continuous), patient age at HSCT (continuous), donor age at
HSCT (continuous), cyclosporine use (y/n), campath use (y/n), T-cell
depletion of the graft (y/n), TBI part of the conditioning (y/n) and
conditioning type (myeloablative/reduced intensity). All identified confounding variables, and variables considered clinically important, were
included in the multivariate models. Two-sided P-values of ⩽ 0.050 were
considered statistically significant. Statistical analysis was performed using
SPSS 19.0 for Windows (IBM Corp., Armonk, NY, USA) and the CMPRSK
package in R (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
We have analysed clinical HSCT outcomes in a large cohort of fully
HLA-matched and single HLA class I-mismatched grafts (n = 339).
b
1.0
Low EMS
Medium EMS
High EMS
Matched
0.8
Cumulative incidence of replase
Cumulative incidence of aGVHD ≥ Grade 2
a
There was no association between the number of AA mismatches
in single HLA-A-, -B- or -C-mismatched grafts and the incidence of
aGVHD, relapse or mortality (data not shown). Similarly, separate
consideration of AA mismatches in sequence positions important
for TCR recognition or peptide binding showed no association
with primary or secondary outcomes. However, donor–recipient
pairs with a single HLA mismatch and a high EMS showed a
statistically significant increase in the cumulative incidence of
⩾ grade II aGVHD compared with pairs with a single HLA
mismatch and low EMS (hazard ratio 2.82, 95% confidence
interval 1.22–6.53, P = 0.016). Of patients transplanted with 9/10
single HLA-mismatched grafts, those with low and medium EMS
had similar risk of aGVHD compared with patients transplanted
with 10/10 HLA-matched grafts (Tables 1a and b and Figure 1a).
On multivariable analysis, only high EMS and recipient age
(pediatric versus adult) had an independent effect on risk of
⩾ grade II aGVHD (Table 1b). In this study cohort, the incidence of
severe aGVHD (⩾ grade III aGVHD) was relatively low
(Supplementary Figure 1); although there was a trend towards
increased risk of ⩾ grade III aGVHD in patients with single HLAmismatched grafts and high EMS, this did not reach statistical
significance (Supplementary Table 1).
0.6
0.4
0.2
0.0
1.0
Low EMS
Medium EMS
High EMS
Matched
0.8
0.6
0.4
0.2
0.0
0
20
40
60
80
0
100
500
Survival time (Days)
Cumulative incidence of death
c
1000
1500
Survival time (Days)
1.0
Low EMS
Medium EMS
High EMS
Matched
0.8
0.6
0.4
0.2
0.0
0
1000
2000
3000
Survival time (Days)
Figure 1. Association between donor–recipient HLA EMS and outcomes following HSCT. The cumulative incidence of grade ⩾ II aGVHD (a),
relapse (b) and mortality (c) is shown for single 9/10 HLA class I-mismatched grafts (n = 171), according to the HLA EMS, and for 10/10
HLA-matched grafts (n = 168). The EMS was calculated for the entire HLA α1/α2 domains at the graft-versus-host direction. Analysis based on
HMS showed similar results (Supplementary Figure 2). Information on aGVHD was not available for 22 patients (total cohort n = 317). Only
patients with malignant diseases were considered for analysis of relapse (n = 296).
Bone Marrow Transplantation (2015) 540 – 544
© 2015 Macmillan Publishers Limited
Physiochemical HLA disparity and acute GVHD
V Kosmoliaptsis et al
543
Patients transplanted with a 10/10 HLA-matched graft had
superior survival compared with patients transplanted with a 9/10
single HLA-mismatched graft but, within the latter group of
patients, no difference was observed in cumulative mortality
between groups with different levels (low, medium and high) of
EMS (Tables 1a and b and Figure 1c). Finally, the cumulative
incidence of disease relapse was similar for patients transplanted
with HLA-matched and HLA-mismatched grafts, independent of
physiochemical disparity scores (Tables 1a and b and Figure 1b).
Patients with low, medium and high EMS scores were similar with
respect to all patient, donor and HSCT characteristics examined,
with the exception of the frequency of HLA-A, -B and -C
mismatches (Table 2). However, the mismatched HLA locus had
no effect on HSCT outcomes. Moreover, we observed no
difference in HLA-DPB1 mismatches between donor–recipient
pairs with high, medium and low EMS.
Table 2.
When the impact of HMS was considered, similar results were
obtained (aGVHD hazard ratio 2.96, 95% confidence interval 1.09–
8.05, P = 0.033 for patients with high EMS grafts; Supplementary
Figures 1 and 2) with no independent effect of HMS over EMS
(EMS and HMS values were highly correlated). Similarly, when the
physiochemical disparity of single HLA-mismatched grafts was
calculated based on AA differences located in TCR or peptide
contact positions, the relationship with incidence of aGVHD
remained significant but the strength of the association did not
improve (data not shown).
DISCUSSION
In the absence of a fully HLA-matched graft, selection of a suitable
donor for a given recipient represents a major challenge in HSCT.
In this pilot study, we have shown that HSCT recipients of single
Characteristics of patients transplanted with a single HLA class I-mismatched graft
Low EMS, n = 60
Patient age at HSCT,
32 (1–65)
median (range)
Donor age at HSCT,
35 (20–55)
median (range)
Year of HSCT, median
2002 (1988–2008)
(range)
Unrelated donor, n (%)
59 (98)
HSCT centre, n (%)
Leiden
31 (52)
Rotterdam
8 (13)
Utrecht
21 (35)
DPB1 mismatch GVH,
39 (65)
n (%)
HLA-A mismatch GVH,
19 (32)
n (%)
HLA-B mismatch GVH,
23 (38)
n (%)
HLA-C mismatch GVH,
18 (30)
n (%)
Female (donor) to male
13 (22)
(recipient), n (%)
ABO mismatch, n (%)
33 (55)
Donor CMV negative to
14 (23)
CMV positive recipient,
n (%)
Diagnosis, n (%)
Acute leukaemia
27 (45)
Chronic leukaemia
6 (10)
Plasma cell disorder
5 (8)
MDS/MPS
12 (20)
Bone marrow failure
3 (5)
Inherited disorder
3 (5)
Lymphoma
4 (7)
Disease type, n (%)
Myeloid malignancy
34 (58)
Lymphoid
20 (32)
malignancy
Nonmalignant
6 (10)
ATG, n (%)
27 (45)
Myeloablative
38 (66)
conditioning, n (%)
Cyclosporin, n (%)
55 (96)
TCD, n (%)
28 (45)
Campath, n (%)
13 (23)
TBI, n (%)
42 (71)
Bone marrow graft,
33 (55)
n (%)
Medium EMS, n = 54
High EMS, n = 57
P-value
Total, n = 171
Controls, n = 168
P-value
29 (1–66)
29 (1–62)
0.626
29 (1–66)
29 (1–64)
0.724
37 (23 –63)
37 (8–67)
0.431
37 (8–67)
33 (2–55)
0.039
2002 (1989–2008)
2003 (1992–2008)
51 (94)
50 (88)
35
4
15
33
35
8
14
39
(65)
(7)
(27)
(61)
(61)
(14)
(25)
(68)
0.685
0.062
0.507
0.722
2003 (1988–2008) 2004 (1987–2009)
160 (94)
159 (95)
101
20
50
111
76
41
51
125
(59)
(12)
(29)
(65)
0.0173
60 (35)
2 (4)
o0.0001
27 (16)
39 (72)
27 (47)
o0.0001
84 (49)
13 (24)
11 (25)
0.830
37 (23)
35 (65)
13 (24)
31 (54)
9 (16)
0.459
0.488
99 (58)
36 (21)
24
11
2
6
2
4
5
23
8
3
8
7
2
6
13 (24)
28 (49)
2 (4)
(45)
(24)
(30)
(74)
23 (14 )
108 (64)
20 (12)
0.642
(44)
(20)
(4)
(11)
(4)
(7)
(9)
(40)
(14)
(5)
(14)
(12)
(4)
(11)
29 (52)
18 (32)
8 (15)
19 (35)
39 (74)
9 (16)
25 (44)
41 (72)
44
29
17
40
29
49
34
22
43
31
(88)
(53)
(33)
(74)
(54)
(92)
(60)
(41)
(75)
(54)
0.846
0.005
0.057
0.055
0.228
0.023
0.997
74
25
10
26
12
9
15
(43)
(15)
(6)
(15)
(7)
(5)
(9)
75
24
12
25
12
8
12
(45)
(14)
(7)
(15)
(7)
(5)
(7)
0.892
30 (57)
17 (32)
0.723
0.301
93 (54)
55 (32)
80 (48)
68 (40)
0.462
0.613
23 (13)
71 (42)
118 (69)
20 (12)
114 (68)
116 (69)
0.251
0.370
0.143
0.868
0.662
148
91
52
125
93
134
79
51
123
79
(87)
(53)
(30)
(73)
(54)
(81)
(47)
(31)
(73)
(47)
o 0.0001
0.971
0.003
0.302
1.000
1.000
0.839
Abbreviations: ATG = antithymocyte globulin; EMS = electrostatic mismatch score; HSCT = haematopoietic SCT; MDS = myelodysplastic syndrome;
MPS = myeloproliferative syndrome; TCD = T-cell depletion. Analyses with hydrophobic mismatch score (HMS) showed similar results (data not shown).
P-values in bold denote statistical significance.
© 2015 Macmillan Publishers Limited
Bone Marrow Transplantation (2015) 540 – 544
Physiochemical HLA disparity and acute GVHD
V Kosmoliaptsis et al
544
HLA-mismatched grafts with high physiochemical disparity (EMS)
were more likely to have aGVHD compared with recipients of
grafts with low EMS; the risk of aGVHD for patients transplanted
with low and medium EMS grafts was similar to that of HSCT
recipients of fully (10/10) HLA-matched grafts. This association was
independent of the mismatched HLA locus (HLA-A, -B or -C) and of
the number of AA polymorphisms between donor and recipient
HLA and was evident when either the α-helices and β-sheets only
or the entire α1/α2 domains were considered. We were not able to
identify an independent effect of AA substitutions at specific
peptide binding positions on the risk of aGVHD,16,17 although we
recognise that our study may have lacked the necessary power to
fully discern such an effect. The increased incidence of aGVHD
observed for high EMS grafts did not translate into higher
mortality rates and, in this cohort, physiochemical disparities or AA
sequence differences between donor–recipient HLA were not
predictive of patient mortality. This is perhaps not surprising given
that mortality following HSCT is a composite outcome and to a
significant extent the result of infections in an immunocompromised host. On the other hand, a larger cohort of HSCT patients
may have to be analysed to better delineate this relationship.
A potential limitation of our study is the analysis of a relatively
small and heterogeneous patient cohort. To avoid confounding,
given than mismatching for a single HLA-DRB1 is associated with
increased incidence of aGVHD and mismatching at HLA-DQB1 in
the presence of mismatches at other loci (HLA-A, -B, -C or -DRB1)
may have a detrimental effect on mortality,18 all donor–recipient
pairs in this analysis were matched for HLA-DRB1 and HLA-DQB1.
Mismatching for HLA-DPB1 is also associated with increased risk of
aGVHD.19 Additional matching of our donor–recipient pairs for
HLA-DPB1 was not possible. Only 15–20% of 10/10 HLA-matched
pairs are also matched at the HLA-DPB1 locus because of weak
linkage disequilibrium between HLA-DPB1 and the rest of the
MHC. However, we observed no difference in HLA-DPB1
mismatches between donor–recipient pairs with high, medium
and low EMS. Factors that were found to be associated with
aGVHD, relapse and/or mortality were included in the multivariate
model to correct for confounding. The change in the estimates
was minor and therefore confounding was unlikely to have played
a major role in our analysis. A major strength of our survival
analysis was the utilisation of competing risk analysis in both
unrelated donor search and transplantation analysis preventing
overestimated estimates.
In conclusion, this report shows for the first time that high
physiochemical disparity between mismatched HLA class I
alloantigens correlates with increased risk of aGVHD following
HSCT. Assessment of donor–recipient HLA compatibility based on
physiochemical disparities may provide valuable information
when fully HLA-compatible HSCT is not an option. Further studies
in a larger patient cohort are needed to determine the clinical
applicability of both HLA class I and class II physiochemical
disparity scores in HSCT.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
VK, CJT and JAB were funded by the NIHR Cambridge Biomedical Research Centre.
We thank the staff of the section Immunogenetics and Transplantation Immunology
of the Leiden University Medical Centre for HLA typing all donors and patients.
REFERENCES
1 Furst D, Muller C, Vucinic V, Bunjes D, Herr W, Gramatzki M et al. High resolution
HLA-matching in hematopoietic stem cell transplantation: a retrospective collaborative analysis. Blood 2013; 122: 3220–3229.
2 Claas FH, Roelen DL, Mulder A, Doxiadis II, Oudshoorn M, Heemskerk M. Differential immunogenicity of HLA class I alloantigens for the humoral versus the
cellular immune response: "towards tailor-made HLA mismatching". Hum Immunol 2006; 67: 424–429.
3 Dankers MK, Heemskerk MB, Duquesnoy RJ, Doxiadis II, Oudshoorn M, Roelen DL
et al. HLAMatchmaker algorithm is not a suitable tool to predict the alloreactive
cytotoxic T-lymphocyte response in vitro. Transplantation 2004; 78: 165–167.
4 Heemskerk MB, Doxiadis II, Roelen DL, Claas FH, Oudshoorn M. The HistoCheck
algorithm does not predict T-cell alloreactivity in vitro. Bone Marrow Transplant
2005; 36: 927–928.
5 Joris MM, van Rood JJ, Roelen DL, Oudshoorn M, Claas FH. A proposed algorithm
predictive for cytotoxic T cell alloreactivity. J Immunol 2012; 188: 1868–1873.
6 Joris MM, Lankester AC, von dem Borne PA, Kuball J, Bierings M, Cornelissen JJ
et al. Translating in vitro prediction of cytotoxic T cell alloreactivity to hematopoietic stem cell transplantation outcome. Transpl Immunol 2014; 30: 59–64.
7 Kosmoliaptsis V, Chaudhry AN, Sharples LD, Halsall DJ, Dafforn TR, Bradley JA et al.
Predicting HLA class I alloantigen immunogenicity from the number and physiochemical properties of amino acid polymorphisms. Transplantation 2009; 88:
791–798.
8 Kosmoliaptsis V, Sharples LD, Chaudhry AN, Halsall DJ, Bradley JA, Taylor CJ.
Predicting HLA class II alloantigen immunogenicity from the number and physiochemical properties of amino acid polymorphisms. Transplantation 2011; 91:
183–190.
9 Oudshoorn M, Doxiadis II, van den Berg-Loonen PM, Voorter CE, Verduyn W,
Claas FH. Functional versus structural matching: can the CTLp test be replaced by
HLA allele typing? Hum Immunol 2002; 63: 176–184.
10 Heemskerk M, Roelen D, Dankers M, van Rood J, Claas F, Doxiadis I et al. Allogeneic MHC class I molecules with numerous sequence differences do not elicit a
CTL response. Hum Immunol 2005; 66: 969–976.
11 Bjorkman PJ, Saper MA, Samraoui B, Bennett WS, Strominger JL, Wiley DC. The
foreign antigen binding site and T cell recognition regions of class I histocompatibility antigens. Nature 1987; 329: 512–518.
12 Garboczi DN, Ghosh P, Utz U, Fan QR, Biddison WE, Wiley DC. Structure of the
complex between human T-cell receptor, viral peptide and HLA-A2. Nature 1996;
384: 134–141.
13 Saper MA, Bjorkman PJ, Wiley DC. Refined structure of the human histocompatibility antigen HLA-A2 at 2.6 A resolution. J Mol Biol 1991; 219: 277–319.
14 Hopp TP, Woods KR. Prediction of protein antigenic determinants from amino
acid sequences. Proc Natl Acad Sci USA 1981; 78: 3824–3828.
15 Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a
competing risk. J Am Stat Assoc 1999; 94: 496–509.
16 Pidala J, Wang T, Haagenson M, Spellman SR, Askar M, Battiwalla M et al. Amino
acid substitution at peptide-binding pockets of HLA class I molecules increases
risk of severe acute GVHD and mortality. Blood 2013; 122: 3651–3658.
17 Kawase T, Morishima Y, Matsuo K, Kashiwase K, Inoko H, Saji H et al. High-risk HLA
allele mismatch combinations responsible for severe acute graft-versus-host
disease and implication for its molecular mechanism. Blood 2007; 110:
2235–2241.
18 Lee SJ, Klein J, Haagenson M, Baxter-Lowe LA, Confer DL, Eapen M et al. Highresolution donor-recipient HLA matching contributes to the success of unrelated
donor marrow transplantation. Blood 2007; 110: 4576–4583.
19 Shaw BE, Gooley TA, Malkki M, Madrigal JA, Begovich AB, Horowitz MM et al. The
importance of HLA-DPB1 in unrelated donor hematopoietic cell transplantation.
Blood 2007; 110: 4560–4566.
Supplementary Information accompanies this paper on Bone Marrow Transplantation website (http://www.nature.com/bmt)
Bone Marrow Transplantation (2015) 540 – 544
© 2015 Macmillan Publishers Limited