Competing interest statement
Conflict of interest: the authors declare no potential conflict of interest.
Abstract
The outcome of patients underwent to allogeneic stem cell transplantation (allo- SCT)
is closely related to graft versus host disease (GvHD) and graft versus leukemia (GvL)
effects which can be mediated by mHAgs. 23 mHAgs have been identified and reported
to be differently correlated with GVHD or GVL and the aim of this work was develop
a method to genotype the mHAgs described so far. For this study we used MALDI-TOF
iPLEX Gold Mass Array technology. We tested 46 donor/recipient matched pairs that
underwent allo-SCT because of Philadelphia positive (Ph+) chronic myeloid leukemia
(n=29) or Ph+ acute lymphoblastic leukemia (n=17). Our data show that sibling pairs
had a lesser number of mHAgs mismatches compared to MUD pairs. Notably, donor/recipient
genomic mismatch on DPH1 was correlated with an increased risk of acute GvHD and LB-ADIR-1R
mismatch on graft versus host direction was correlated with a better RFS with no increase
of GvHD risk. Our work provides a simple, accurate and highly automatable method for
mHAgs genotyping and suggest the role of mHAgs in addressing the immune reaction between
donor and host.
Introduction
Allogeneic stem cell transplantation (allo-SCT) may be the only cure for patients
affected by acute myeloid or lymphoid leukemia, or other hematological diseases such
as lymphomas or multiple myeloma.
1
The curative effects of allo-SCT are closely related to graft versus leukemia (GvL).
However the severity of the graft versus host disease (GvHD) may override the GvL
benefit and worsen the outcome of allotransplanted patients.
1-3
Despite a full major HLA antigens (MHAgs) compatibility, minor histocompatibility
antigens (mHAgs) can also play a pivotal role in conditioning both GvL and GvHD response
in HLA full-matched allo-SCT. Evidence from experimental and clinical studies on HLA-identical
allo-SCT suggest that GvL and GvHD may be driven by donor T cell responses against
disparate mHAgs.
4-9
Indeed, mHAgs are polymorphic HLA-bound peptides derived from cellular proteins that
can induce powerful alloreactive T cell responses. The mHAgs recipient-donor disparity
may arise from a genomic variation in the coding region of the gene that leads to
differences in the amino acid sequence of the homologous protein and, in most cases,
it may depend on a nonsynonymous single nucleotide polymorphism (nsSNP) or on a deletion.
7,10,11
Recent advances in the molecular identification of mHAgs have significantly expanded
our knowledge to a total of 23 autosome-coded mHAgs and 10 Y-chromosome coded mHAgs,
leading to an increased interest in the clinical application of mHAgs typing. Although
several mHAgs, including Y-chromosome encoded mHAgs, are ubiquitously expressed, an
increasing number of autosomal-encoded mHAgs is being identified as expressed exclusively
by hematopoietic cells or by their malignant counterparts.
12-21
About this, ACC-1, ACC-2 and HA-2 have been correlated with the beneficial GvL effects,
while some mHAgs disparities, CD31, HA-5, HA-8 and UGT2B17, have been found to be
involved in the induction of GvHD.
8,22-28
The molecular identification of GvHDand GvL-associated mHAgs could allow the evaluation
of the clinical impact of mHAgs mismatches and their specific T cell responses triggered
by allo-SCT. Several studies in HLA-matched allo-SCT reported an association between
mHAgs mismatches and the clinical outcome,
29-33
but other studies have not confirmed these observations.
7,24
The heterogeneity of techniques suitable for mHAgs typing (SSP-PCR and Luminex) as
well as the complexity of integrating mHAgs typing data and clinical information are
likely the main reasons that do not facilitate the routinary evaluation of mHAgs in
clinics.
34-36
In our study, we set up a new method for mHAgs genotyping based on Matrix Assisted
Laser Desorption Ionization Time-of-Flight (MALDI-TOF) mass spectrometry (MS) and
we tested it in a training set of donor-recipient pairs with the aim to propose a
simple and standardizable methodology able to overcome the limits of the conventional
methods and to make mHAgs genotyping suitable for clinical application.
37-38
Materials and Methods
Patients and transplant procedures
For this study, we tested the MALDITOF iPLEX Gold method on a cohort of Ph+ CML and
Ph+ ALL patients who underwent allo-SCT at six Italian Centres from 1990 to 2011.
To this purpose, we retrospectively selected 46 donor-recipient pairs fully HLA compatible
for HLA-A, -B, -C, -DRB1 and -DQB1 alleles, according to SSP-PCR high resolution molecular
methods. Out of the 46 selected cases, 29 were Ph+ CML and 17 were Ph+ ALL patients
who underwent allo- SCT by sibling (29 cases, 63%) or MUD (17 cases, 37%).
GvHD effects, either acute or chronic, were defined according to the Glucksberg scale
and NHI criteria, respectively, and they were reported as cumulative incidence. Relapse
free survival (RFS) was calculated using Kaplan-Meier method and it was assumed as
an indicator of GVL effect.
39-40
All patients provided informed consent according to the policy of each participating
Centre. Patients and transplant features are reported in Table 1.
mHAgs’s biological characteristics and definitions
The HLA matched donor-recipient pairs evaluated for this study were genotyped for
a panel of 23 mHAgs (and causal SNPs). The biological characteristics of each mHAg
(gene, locus, SNP reference number, nucleotide switch and HLA restriction) are detailed
in Table 2. We specify that CD31 exists in two isoforms (CD31125 and CD31563) because
it results from two different SNPs (rs668 and rs12953, respectively). We genotyped
both SNPs, but we considered the two isoforms together during the analysis because
of the strong linkage between the two SNPs. On the contrary, the SNP rs2289702 determine
two different mHAgs, ACC-4 and ACC-5, according to the HLA molecule that present them.
For the purpose of this study, immunogenic mHAg difference was defined when within
a given donor/recipient pair, only one individual had an immunogenic phenotype of
a particular mHAg accompanied by the appropriate HLA restriction molecule. Genomic
mHAg difference was identified when mHAg genotypes in donor and recipient were different,
but phenotypically they were either the same or the mHAg immunogenic phenotype was
not accompanied by the appropriate HLA restriction molecule. Both genomic and immunogenic
mHAgs disparities were included in the analysis. This is due to an incomplete knowledge
of mHAgs because the epitope-prediction strategy often makes it hard to confirm the
immunogenicity of the predicted putative mHAgs and there is currently no controlled
way of isolating mHAgs-specific T cells directed against mHAgs.
mHAgs genotyping by MALDI-TOF iPLEX Gold technology
For the purpose of our study, the genomic DNA (gDNA) was extracted using QIAamp DNA
mini Kit (Qiagen) from peripheral blood mononuclear cells (PBMC) previously cryopreserved.
The PBMC collection was performed before allo-SCT for patients and before stem cells
harvest for donors. The purity of gDNA for each sample was determined by measuring
the absorbance at 260 and 280 nm, with the A260/A280 values being in the range of
1.5-1.9, and the concentration of the gDNA was adjusted to 12 ng/μL. A total of 30
ng of gDNA was used for genotyping all SNPs.
MS MALDI-TOF iPlex Gold is able to discriminate the two variants of an SNP in a very
efficient way, so it was considered suitable for the aim of the study. The MassARRAY
Assay Design software was used to design 3 different multiplex reactions to investigate
the 23 SNPs. Genotyping was performed using iPLEX Gold technology and MassARRAY high-throughput
DNA analysis with matrix-assisted laser desorption/ionization time-of-flight (MALDI□TOF)
MS [Agena Bioscience Inc., San Diego, CA], according to the manufacturer’s protocol.
41
Multiplex design and primer sequences are shown in Table 3.
The multiple-genotyping assay was validated using intra- and extra-run controls. Firstly,
a DNA sample (NA10859) from the CEPH (Centre d’Etude du Polymorphisme Humain CEPH,
Paris, France) panel was genotyped simultaneously in every single run. Six mHAgs (ACC-1,
ACC-2, ACC-6, HA-8, HB-1 and LB-ADIR-1R) were reported. Then, the genotype of each
polymorphism was validated in 10 randomly selected samples by amplification with PCR
and subsequent direct Sanger Sequencing (ABI Prism 3730, Applied Biosystems, Foster
City, CA) as gold standard.
Statistical analysis
For continuous factors, the median and ranges were calculated. The χ
2
-test was used to compare differences in percentage, and Mann-Whitney U test was used
to compare continuous values. The probability of GvHD (acute and chronic) was estimated
as cumulative incidence. In GvHD analysis, competing risks were relapse or death before
the onset of GvHD. Probabilities for RFS were calculated using the Kaplan-Meier method.
42
RFS was calculated from the date of allo-SCT until the date of relapse or death, whichever
occurred first. Death in remission was considered as a competing risk in the relapse
analysis. Differences in RFS were evaluated by log-rank testing in univariate analysis.
Multivariate analyses were performed using the Fine and Gray regression model. The
Cox proportional hazard regression model was used for multivariate analyses of variables
affecting RFS. The following patient- and transplantrelated variables were analyzed:
CML or ALL diagnosis and type of bcr-abl transcript, immunogenic/genomic mHAgs mismatches
between donor and recipient, patient age at SCT, type of donor, patient gender and
sex mismatch between donor and recipient, graft source, time from diagnosis to HSCT,
conditioning regimen, GvHD prophylaxis and development of GvHD. All P-values were
2-sided and P#x003C;0.05 was considered statistically significant. Each SNP was tested
for departures from the Hardy-Weinberg equilibrium (HWE).
Results
SNPs genotyping by MALDI-TOF iPLEX Gold technology
The MALDI-TOF iPLEX Gold technology method was used on a training group of 46 donor/recipient
pairs with the aim to evaluate the accuracy and reliability of the genotyping assay.
A total of 2116 genotypes resulted out of a predicted total number of 2116 (92 samples
for 23 SNPs) with a call rate of 100%.
In order to evaluate the accuracy and reliability of the genotyping assay, two different
approaches were adopted. Evaluation of method reproducibility was carried out by genotyping
of the DNA number NA10859 during the Sequenom run. This standard DNA is released the
genotype of only six (6 of 23, 26%; ACC-1, ACC-2, ACC-6, HA-8, HB-1 and LB-ADIR-1R)
mHAgs. The concordance between the released data and our genotyping was 100%. In the
second stage, we validated the set of 10 randomly selected samples using conventional
Sanger sequencing and also in this case we obtained a concordance of 100%. The Hardy-Weinberg
equilibrium (HWE) was satisfied for most SNPs on both populations (patients and donors).
rs12692566 (mHAgs LB-LY751K) was the only SNP showing a significant difference as
compared with the prediction under HWE assumptions. Since Hardy Weinberg disequilibrium
can indicate genotyping errors or population stratification, LBLY751K was excluded
from the statistical analysis (Table 4).
mHAgs mismatches, patients’ clinical features and correlation with GvHD/GvL effects
The analysis of immunogenic mismatches showed that sibling pairs had a lesser number
of mismatches compared to MUD pairs (median 1 vs. 3; t-test with P<0.003). The evaluation
of genomic mismatches point out that sibling pairs have higher identity than MUD pairs
(t-test, P<0.0001). In fact, the median number of genomic differences was 8 (range
0-15) in sibling pairs and 13 (range 11-17) in MUD pairs (t-test with P<0.05). Only
one sibling pair showed a perfect genomic mHAgs match.
We also tried to correlate if some mHAgs mismatches could be involved in GvHD development.
DPH1 genomic mismatch resulted to be correlated with the risk of grade ≤2 aGvHD development
(multivariate analysis HR 2.2, P=0.04, Table 5), while no mHAgs mismatches were found
to be correlated with an increased risk of cGvHD (Table 5).
By these evidences, we investigated any correlation between mHAgs mismatches and RFS
as a clinical surrogate of GvL effect. Despite some clinical factors affecting the
RFS (i.e. the underlying disease, b3a2 transcript isoform and chronic GvHD development),
in multivariate analysis we observed that only LB-ADIR-1R, with genomic mismatch on
graft versus host direction (HR 0.3, P=0.03, Table 5) was positively correlated with
a better RFS.
Discussion
The study aimed to set up a new laboratory assay for genotyping minor histocompatibility
antigens which are thought to play a key role in the allo-immune responses in fully
HLA-matched stem cell transplantations.
The MALDI-TOF iPLEX gold approach was used to overcome the limits of conventional
methods, such as SSP-PCR and Luminex, and to make mHAgs genotyping analysis suitable
for clinical application. PCR-SSP and Luminex are commonly used for HLA typing, but
both methods have several limitations. Complex primer design and identification of
the annealing temperature are critical for the PCR-SSP test; while biotinylated DNA
probes, beads and streptavidin-phycoerythrin binding are critical steps for Luminex.
34,35
MALDI-TOF was used effectively for KIR (killer-cell immunoglobulin-like receptor)
and platelet antigens genotyping and, due to the expected advantages in terms of rapidity,
simplicity and high throughput capability, it was identified as a potential new method
for mHAgs genotyping.
36,37
From a technical point of view, one of the main advantages of SNPs genotyping by MS
system consists in the direct measurement of the mass of the molecules of interest
without using any surrogate, such as fluorescence. MS genotyping has shown high accuracy;
moreover, this methodology is rapid and highly automated, with a genotyping throughput
of up to 128 matched pairs (256 samples) per run. The MS approach presents other advantages:
it requires only a small amount of DNA, it is highly reproducible, and, furthermore,
it works on multiplex and the design of each multiplex is made directly by the instrument
software. The only drawbacks of this method are that it does not allow the genotyping
of mHAgs resulting from deletions and can be used only if both the polymorphism and
the polymorphism’s flanking region are known.
36
The use of designed primers for SNPs of interest and the MS protocol in this training
set allowed us to genotype 100% of the SNPs (2116 genotypes of a predicted total number
of 2116) and mHAgs. Intra- and extra-run controls demonstrated the reliability of
this method. Analyzing the data obtained by genotyping the mHAgs of this set of donor/recipient
pairs with their clinical features, particularly GvHD development and RFS, some interesting
suggestions have emerged.
Sibling pairs have fewer mHAgs disparities despite the pairs with HLAmatched unrelated
donor (P<0.0001). This data may appear obvious, but from a biological point of view
no study has clearly shown that until now. This means that the genomic compatibility
of HLA full matched MUD pairs will never be greater than full HLA sibling pairs.
Established that HLA differences between donor and recipient are the major predictor
of GvHD, we investigated a possible role of mHAgs on GvHD development and relapse
incidence in a training set of Ph-positive CML and ALL allotransplanted patients.
These patients were chosen because representative of chronic and acute leukemias sharing
a unique cytogenetic alteration: t(9;22). The only observation is that genomic DPH1
mismatch appeared to be related to an increased risk of grade ≤2 aGvHD development.
This possible correlation between DPH1 and aGvHD is supported by the fact that DPH1
is expressed by a broad range of non-hematopoietic tissues. The role of DPH1 on extramedullary
toxicity has already been described by Warren, who pointed out that pulmonary toxicity
was observed with infusion of DPH1-specific T cells. On the contrary, leukemic blasts
were poorly recognized by DPH1-specific T cells.
43
Conversely, we found that genomic mismatch of LB-ADIR-1R on graft versus host direction
was related to a better RFS. Our findings on LB-ADIR-1R mismatch are consistent with
previous data from van Bergen, showing that LB-ADIR-1R specific T cells perform wide-reaching
antitumor activity with a limited recognition of nonactivated tissues. Indeed, LB-ADIR-1R
specific T cell recognize cell lines from haematological tumours, while generally
mesenchymal and biliary epithelial cells are recognized to be GvHD target tissues.
14
Conclusions
This work prove that MS may be a simple, effective, and accurate method for mHAgs
genotyping. The method requires a small amount of gDNA that can be easily extracted
also from cryopreserved cells. Furthermore, MS is able to genotype all mHAgs in a
single work session, thus saving a lot of time.
Data analysis of our patients training set lead us to say that despite the full major
HLA match, the minor-HLA genomic and immunogenic compatibility between a patient and
his unrelated donor is always lower compared to the genomic and immunogenic compatibility
of a patient and his sibling donor. In fact, sibling pairs had a lesser number of
mHAgs mismatches compared to MUD pairs (P=0.003). Of 23 mHAgs evaluated, only 2, DPH1
and LBADIR- 1R, proved to be correlated with the GvHD and GvL effect respectively,
and these results confirm the previous reports. Our study suggests that MS would be
used and useful for mHAgs genotyping. A larger and prospective trial would be warranted
to validate this method.