Article
Reference
Pretransplant HLA mistyping in diagnostic samples of acute myeloid leukemia patients due to acquired uniparental disomy
DUBOIS, V, et al.
Abstract
Although acquired uniparental disomy (aUPD) has been reported in relapse acute myeloid leukemia (AML), pretransplant aUPD involving chromosome 6 is poorly documented. Such events could be of interest because loss of heterozygosity (LOH) resulting from aUPD in leukemic cells may lead to erroneous results if HLA typing for hematopoietic stem cell donor searches is performed on blood samples drawn during blastic crisis. We report here six AML patients whose HLA typing was performed on DNA extracted from peripheral blood obtained at diagnosis. We observed LOH involving the entire HLA region (three patients), HLA-A, B, C (two patients) and HLA-A only (one patient). An array-comparative genomic hybridization showed that copy number was neutral for all loci, thus revealing partial aUPD of chromosome 6p21. When HLA typing was performed on remission blood samples both haplotypes were detected. A 3-4% LOH incidence was estimated in AML patients with high blast counts. Based on DNA mixing experiments, we determined by PCR sequence-specific oligonucleotide hybridization on microbeads arrays a detection threshold for HLA-A, B, DRB1 [...]
DUBOIS, V, et al . Pretransplant HLA mistyping in diagnostic samples of acute myeloid leukemia patients due to acquired uniparental disomy. Leukemia , 2012, vol. 26, no. 9, p. 2079-85
DOI : 10.1038/leu.2012.68 PMID : 22488219
Available at:
http://archive-ouverte.unige.ch/unige:29052
Disclaimer: layout of this document may differ from the published version.
1 / 1
ORIGINAL ARTICLE
Pretransplant HLA mistyping in diagnostic samples of acute myeloid leukemia patients due to acquired uniparental disomy
V Dubois1, F Sloan-Be´ na2, A Cesbron3, BG Hepkema4, K Gagne3,5, S Gimelli2, D Heim6, A Tichelli6, J Delaunay7, M Drouet8, S Jendly9, J Villard9and J-M Tiercy9
Although acquired uniparental disomy (aUPD) has been reported in relapse acute myeloid leukemia (AML), pretransplant aUPD involving chromosome 6 is poorly documented. Such events could be of interest because loss of heterozygosity (LOH) resulting from aUPD in leukemic cells may lead to erroneous results if HLA typing for hematopoietic stem cell donor searches is
performed on blood samples drawn during blastic crisis. We report here six AML patients whose HLA typing was performed on DNA extracted from peripheral blood obtained at diagnosis. We observed LOH involving the entire HLA region (three patients), HLA-A, B, C (two patients) and HLA-A only (one patient). An array-comparative genomic hybridization showed that copy number was neutral for all loci, thus revealing partial aUPD of chromosome 6p21. When HLA typing was performed on remission blood samples both haplotypes were detected. A 3 -- 4% LOH incidence was estimated in AML patients with high blast counts. Based on DNA mixing experiments, we determined by PCR sequence-specific oligonucleotide hybridization on microbeads arrays a detection threshold for HLA-A, B, DRB1 heterozygosity in blood samples witho80% blasts. Because aUPD may be partial, any homozygous HLA result should be confirmed by a second typing performed on buccal swabs or on blood samples from the patient in remission.
Leukemia(2012)26,2079--2085; doi:10.1038/leu.2012.68
Keywords: acquired uniparental disomy; loss of haplotype; HLA typing; array-comparative genomic hybridization; luminex microbeads array
INTRODUCTION
Loss or downregulation of HLA class I antigens may allow tumor cells to escape T-cell immunosurveillance. HLA loss of hetero- zygosity (LOH), or haplotype loss, is a frequent event in a variety of solid tumors1,2but seems to be rare in hematological cancers. In a study on 25 acute myeloid leukemia (AML) patients at diagnosis, HLA class I expression was found to be preserved in all cases.3 The potential of adoptive T-cell immunotherapy by donor- lymphocyte infusions for treating leukemic patients has stimu- lated interest in HLA class I expression on leukemic cells that could impact on the efficacy of donor-lymphocyte infusions. An HLA and microsatellite genotyping analysis of 44 samples of freshly isolated leukemic blasts did not reveal any case of haplotype loss, although a downregulation of HLA class I expression was observed in 9.1%
of the patients samples.4Recently, a loss of mismatched HLA has been demonstrated in a significant proportion of AML patients relapsing after donor-lymphocyte infusions following transplanta- tion with hematopoietic stem cells from haploidentical donors.5 -- 7 As revealed by single-nucleotide polymorphism array analysis, HLA loss was copy-number neutral and thus resulted from acquired uniparental disomy (aUPD),5a chromosomal aberration that occurs when both copies of a chromosome or a chromosomal segment originate from one parent (reviewed Tuna et al.8).
Chromosomal segregation or non-disjunction error in mitosis, or mitotic recombination in the S/G2 phase of the cell cycle can
account for the occurrence of aUPD.8 The role of copy number mutants in tumor-suppressor genes and oncogenes in leukemia outcome has stimulated detailed analyses of aUPD as a mechanism underlying molecular aberrations affecting disease outcome. Although cancers such as retinoblastoma, rectal, basal cell or MUTYH-associated polyposis carcinomas exhibit high percentages of aUPD (50 -- 85%), the rates are much lower in AMLs.8 Following smaller scale studies,9 -- 12 a single-nucleotide polymorphism microarray study on 454 AML patients has disclosed aUPD in 17% of the samples.13In this study, the authors reported a non-random distribution of aUPD more specifically affecting chromosomes arms 13q, 11p and 11q. In addition, losses accounted for 15% and gains for 12% of the AML samples. A more recent genome-wide single-nucleotide polymorphism analysis of 157 cytogenetically normal AML adult patients also reported a 17% aUPD incidence.14
We report herein six patients whose HLA typing was performed on peripheral blood lymphocytes obtained from AML patients in acute phase at diagnosis and led to an apparent LOH as determined by PCR sequence-specific oligonucleotide (SSO)/
sequence-based typing (SBT) using locus-specific primers. Com- parative genomic hybridization (CGH) on microarrays did not reveal any copy number variation of HLA loci at 6p21 that could account for the loss of haplotypes, and thus allowed to confirm that the HLA homozygosity was due to aUPD. In addition, through
Received 18 November 2011; revised 31 January 2012; accepted 7 March 2012; accepted article preview online 14 March 2012; advance online publication, 10 April 2012
1HLA Laboratory, EFS Rhoˆne Alpes, Lyon, France;2Service of Medicine Genetics, University Hospital Geneva, Geneva, Switzerland;3HLA Laboratory, EFS Pays de la Loire, Nantes, France;4Department of Laboratory Medicine, University Medical Center, University of Groningen, Groningen, The Netherlands;5EA4271, EFS Pays de la Loire, Nantes, France;
6Division of Hematology, University Hospital Basel, Basel, Switzerland;7Service of Clinical Hematology, CHU Hotel Dieu Nantes, Nantes, France;8Histocompatibility Laboratory, CHU Limoges, Limoges, France and9National Reference Laboratory for Histocompatibility, Transplantation Immunology Unit, University Hospital Geneva, Geneva, Switzerland.
Correspondence: Dr J-M Tiercy, National Reference Laboratory for Histocompatibility, Transplantation Immunology Unit, Department of Genetics and Laboratory Medicine, Department of Medical Specialties, Geneva University Hospitals and University of Geneva, 4 rue Gabrielle-Perret-Gentil, 1211 Geneva 14, Switzerland.
E-mail: jean-marie.tiercy@unige.ch
www.nature.com/leu
in-vitroDNA mixing experiments, we have determined a threshold for the detection of HLA alleles by the reverse hybridization on microbeads arrays (luminex technology).
PATIENTS AND METHODS
The six patients reported in this study (Table 1) have been HLA typed in the laboratories of five transplant centers. Patient ELHM is a 26-year-old female with AML M1 without cytogenetic anomaly and mutations in FLT3, WT1 and NPM1. Patient’s marrow and blood samples at diagnosis had 96%
and, respectively, 95% blasts. Patient DMF is a 54-year-old female AML M5b patient with FLT3 and WT1 mutations. Patient’s blood sample at diagnosis had 96% blasts. Patient ND is a 44-year-old female with AML M5a, with a weakly positive FLT3 mutation and NPM1 mutation. Patient’s blood sample at diagnosis had 85% blasts. Patient PA is a 57-year-old female with AML M1 with an FLT3 mutation; the blast count was not available for this patient. Patient JH is a 47-year male with AML M1, FLT3-ITD positive.
Patient’s blood sample at diagnosis had 96% blasts. Patient CJ is a 62-year- old male with AML type M4, with a Y chromosome loss, an FLT3 duplication and an insertion-type NPM1 mutation. Patient’s blood sample at diagnosis had 27% blasts and 57% monocytes (29.8 G/L) at diagnosis.
Patients ELHM, FDM, JH, and CJ were initially typed in 2010 -- 2011, and patients ND and PA in 2003 -- 2005.
Genomic DNA was extracted using the GenoM6 automat (Qiagen cartridge, Qiagen GmbH, Hilden, Germany), or by spin columns on the QIAcube (Qiagen GmbH), or by a manual salting out method.
HLA typing was performed by PCR-SSO hybridization on microbeads arrays (One Lambda HD kits, Canoga Park, CA, USA; or Gene-Probe kits, San Diego, CA, USA), by PCR sequence-specific primers (SSPs) (Genovision, West Chester, PA, USA) and by biallelic PCR-SBT (Abbott Molecular Park, Chicago, IL, USA).
Array-CGH (aCGH) was performed on the probands using the Human Genome CGH Microarray Kit 180K (Agilent Technologies, Palo Alto, CA, USA) with 13 kb overall median probe spacing giving an average resolution of 30 kb.
Briefly, 1mg of patient and a sex-matched pooled reference DNAs (Promega Corporation, Madison, WI, USA) were processed according to manufacturer’s protocol. Fluorescence was scanned in a dual-laser scanner (Agilent DNA microarray scanner G2565CA; Agilent Technologies) and the images were extracted and analyzed with Agilent Feature Extraction software (v9.5.3.1) and CGH Analytics software (v3.5.14; Agilent Technologies, Santa Clara, CA, USA), respectively. Changes in test DNA copy number at a specific locus are observed as the deviation of the log-ratio value from a modal value of 0.
The normal range is indicated by black dots (Figure 2), whereas a loss is indicated for probes with a log-ratio value of1 (green dots in Figure 2), and a gain is indicated by probes with a log-ratio value40.6 (red dots in Figure 2). At leastX2 contiguous copy number alterations are considered as significant. We reported herein only copy number variations located in the HLA region at 6p21. Genomic content analysis was performed by aCGH for female and male reference DNA samples to identify potential imbalances in the HLA region of interest. Both reference DNA samples were shown to be positive for the HLA-DRB5 locus by PCR-SSP genotyping.
DNA mixing experiments were set up by mixing two DNA samples homozygous for the HLA-A, B, DRB1 loci. DNA#1 (A*01:01; B*08:01;
DRB1*03:01) and DNA#2 (A*32:01; B*35:01; DRB1*11:01) were diluted at
20 ng/ml and then mixed at different ratios: 100 -- 0%, 90 -- 10%, 85 -- 15%, 80 -- 20%, 70 -- 30% and 50 -- 50%. In all, 2-ml samples (40 ng) of each mix were amplified with locus-specific primers as described by the supplier and typed by the microbeads arrays (luminex) in parallel to routine clinical samples (using One Lambda LabType RSSO1A, RSSO1B, RSSO2B1 reagents). For each HLA-A,B,DRB1 allele that could not be assigned all allele-specific negative probes were listed and the mean fluorescence intensity values compared with the cutoff values provided by the manufacturer. Excluding beads that are positive with both alleles, a total of 53 allele-specific beads were analyzed: beads 27, 56, 61, 67, 74, 75, 95 for HLA-A*01:01; beads 1, 3, 6, 17, 26, 33, 39, 52, 54, 62, 72, 76 for HLA-A*32:01;
beads 24, 40, 56, 78, 81, 99, 521, 530, 545, 546, 553, 570, 574, 599 for HLA- B*08:01; beads 71, 87, 94, 514, 529, 544, 567, 578, 590 for HLA-B*35:01;
beads 16, 38, 46, 58, 82 for HLA-DRB1*03:01; and beads 28, 39, 57, 64, 69, 73 HLA-DRB1*11:01.
In addition, a DNA mixing experiment was performed by using homozygous DNA#1 (A*01:01; B*08:01; DRB1*03:01) and heterozygous DNA#3 (A*01:01, A*11:01; B*08:01, B*35:01; DRB1*03:01, DRB1*11:04).
Detection of the A*11:01, B*35:01 and DRB1*11:04 alleles present at a 10, 15, 20, 30 and 50% ratio was determined by the evaluation of 29 allele-specific beads using the luminex typing assay. Beads 27, 31, 36, 63, 64, 77, 87, 525, 552, 554, 583 were specific for A*11:01; beads 71, 87, 514, 529, 544, 567, 568, 590, 594 were specific for B*35:01; and beads 1, 28, 29, 57, 64, 69, 73, 83, 89 were specific for DRB1*11:04. Because the interpretation software of the luminex typing system can assign only two HLA alleles, mixing experiments could only be performed with between AA and BB or AB haplotypes.
RESULTS HLA typing data
DNA typing of three AML patients at diagnosis (ELHM, FDM and JH) revealed a complete HLA-A, B, C, DRB1, DQB1 homozygosity (Table 2, rows 1, 3 and 9), as determined by reverse PCR-SSO hybridization on microbeads arrays (luminex technology). Biallelic PCR-SBT revealed HLA class I homozygosity in two additional patients (PA and ND) (Table 2, rows 5 and 7), whereas they were heterozygous for HLA-DRB1 as shown by PCR-SSO hybridization on microbeads arrays. A sixth patient tested at diagnosis revealed homozygosity at HLA-A with a rare A*02:05 allele, but hetero- zygosity at all other loci (Table 2, row 11). Patients showed a percentage of blasts ranging from 85 to 95% (Table 2) when blood samples were drawn for HLA typing in view of a hematopoietic stem cell donor search. For patients DMF, PA and JH, the family pedigree immediately suggested an apparent LOH. A microsatellite analysis using four informative microsatellites markers in the HLA region independently confirmed LOH in patient ELHM (data not shown).
When retyping was subsequently performed for all six patients on blood samples drawn in remission, complete HLA class I and II heterozygosity was disclosed (Table 2, rows 2, 4, 6, 8, 10 and 12).
When HLA typing was performed at diagnosis by the PCR-SSP method the second allele could be readily detected, although the primer mixes specific for the minor allele resulted in systematically weaker amplifications, as shown for HLA-DPB1 typing of patient
Table 1. Clinical features of the six AML patients with LOH
Patient Diagnostic Age (years) Cytogenetics Molecular mutations
ELHM AML M1 26 Normal FLT3, WT1, NPM1
DMF AML M5b 54 Normal FLT3, WT1
ND AML M5a 44 Normal FLT3 weak, NPM1
PA AML M1 57 Normal FLT3
JH AML M1 47 Normal FLT3
CJ AML M4 62 Y chromosome loss FLT3, NPM1
Abbreviations: AML, acute myeloid leukemia; LOH, loss of heterozygosity.
Pretransplant HLA mistyping in diagnostic samples of AML patients V Duboiset al 2080
Leukemia (2012) 2079 -- 2085 &2012 Macmillan Publishers Limited
DMF (Figure 1). For patient PA monoallelic sequencing, as opposed to biallelic sequencing, revealed complete HLA class I heterozygosity (data not shown). Interestingly in the case of patient CJ even monoallelic SBT did not allow to detect the A*02:01 allele when performed on the DNA extracted from patient’s blood sample obtained at diagnosis, presumably because the heterozygosity concerned two alleles within the same amplification group (A*02).
Array-CGH
To detect putative deletions in the major histocompatibility complex region, the six patients’ DNA samples were analyzed by aCGH as described in Materials and Methods. No evidence of a homozygous deletion at any of theHLA-A, B, C, DRB1, DQB1, DPB1 loci could be detected (Figures 2a and b), showing that HLA loss at 6p21 was copy-number neutral and can only be interpreted as a result of aUPD. Interestingly, patients DMF, ND (Figure 2a) and CJ (Figure 2b) showed a small homozygous deletion of a DNA segment (6p21.32, rs32480027 -- 32521929) corresponding to the DRB5/DRB6 loci according to NCBI build 37 (see HLA typing in Table 2), as revealed by 2 -- 3 contiguous probes showing log-ratio values between1.5 and2. The absence ofDRB5was expected since DMF blast cells were homozygous for DRB1*14 and ND blast cells heterozygous DRB1*03,*11. For patients ELHM (Figure 2a) and JH (Figure 2b), whose blast cells would also be expected to lack theDRB5 locus, a signal was obtained with one probe only and thus not considered as significant by the software. Patient PA blasts that were typed as DRB1*08,*15 (Table 2) are expected to lack the DRB5 locus on one haplotype only. Indeed, aCGH
indicates a signal corresponding to a heterozygous deletion (circle in Figure 2b).
In addition, a few small microdeletions were observed at 6p21 in four of the six patients. Patient DMF (Figure 2a) has a heterozygous deletion at theHLA-Blocus (6p21.33, rs31287351 -- 31326021), which may result from some mosaicism in the blast population, according to ratio values around 0.8. Patient PA (Figure 2b) has a homozygous deletion at 6p22.1 centromeric to HLA-A (rs29854870 -- 29873933). Patient JH (Figure 2b) has two deletions at 6p22.1 very close to the HLA-A gene, one homozygous deletion (rs29854870 -- 29873933) encompassing the HCG2P7 and HLA-H genes and a heterozygous deletion (rs29887268 -- 29896657) comprising theHCG4P6gene. An aCGH test performed on the mother of patient JH (data not shown) allowed to conclude that these microdeletions were not inherited from the mother. Patient CJ has a short duplication (rs29842826 -- 29918779) close toHLA-A(Figure 2b). Thus, in addition to aUPD which lead to complete or partial HLA homozygosity, the blast cells of these patients showed a DNA content with microdeletions or duplications, either of parental origin, or occurringde novo15 at the blastic level.
DNA mixing experiments
To define more precisely a cutoff limit for the detection of a heterozygous DNA present in a fraction of the cells by the luminex technology, the following mixing experiments were performed.
First, two HLA-A, B, DRB1 homozygous DNA samples (DNA#1: HLA- A*01:01; B*08:01; DRB1*03:01 and DNA#2: HLA-A*32:01; B*35:01;
DRB1*11:01) were mixed at a 90 -- 10%, 85 -- 15%, 80 -- 20%, 70 -- 30% and 50 -- 50% ratio. Each mix was then tested by the luminex assay and mean fluorescence intensity values analyzed for each allele-specific bead. The results for all negative beads are shown in Table 3 and are representative of three experiments. For example, of the seven A*01:01-specific probes, three were positive at all dilutions (beads 74, 75, 95), whereas four probes (beads 27, 56, 61, 67) were negative at a 10/90% DNA ratio, but positive at ratiosX15%. Altogether, of the 53 beads that were specific for the 6 HLA-A,B,DRB1 alleles, 23 (41.5%) were negative at a 10/90% DNA ratio. The negative probes were on average 20% below the cutoff (rangeo1 -- 40%) with somewhat larger differences for the HLA-B- specific probes. At a 15% ratio the luminex assay allowed to assign both HLA alleles for each locus, although for three combinations the heterozygous result was assigned by the HLA Fusion software (One Lambda) with a single false negative bead (5.6% of the specific beads): bead 17 for A*32:01, bead 521 for B*08:01 and Table 2. HLA typing of six AML patients as performed on blood samples drawn at diagnosis and after treatment
Patient % Blasts A* B* C* DRB1* DQB1* DPB1*
ELHM 95% 03, --- 35, --- 04, --- 13, --- 06, --- NT
ELHM --- 03:02, 02:01 35:01, 15:01 04:01, --- 13:01, 01:01 06:02, 05:01 NT
DMF 91% 30, --- 13, --- 06, --- 14, --- NT 02:02, 04:01a
DMF --- 30:01, 29:02 13:02, 15:01 06:02, 03:04 14:54, 04:01 NT 02:02, 04:01
ND 85% 11, --- 55, --- 03, --- 03, 11 NT NT
ND --- 11:01, 24:02 55:01, 40:01 03:03, 03:04 03:01, 11:01 02:01, 03:01 01:01, 04:01
PA n/a 24, --- 39, --- 07, --- 08, 15 NT NT
PA --- 24:02, 02:01 39:06, 07:02 07:02, --- 08:01, 15:01 04:02, 06:02 04:01, 10:01
JH 96% 24, --- 18, --- 07, --- 04, --- 03, --- NT
JH --- 24:02, 02:01 18:01, 07:02 07:01, 07:02 04:01, 15:01 03:02, 06:02 02:01, 04:01
CJ 27%b 02:05, --- 13, 50 06, 07 03, 13 02, 06 NT
CJ --- 02:01, 02:05 13:02, 50:01 06:02, 07:02 03:01, 13:02 02:01, 06:09 0401, ---
Abbreviations: AML, acute myeloid leukemia; n/a, not available; NT, not tested; SSO, sequence-specific oligonucleotide; SSP, sequence-specific primer. Typing was performed by reverse PCR-SSO hybridization on microbeads (luminex technology) or direct sequencing after locus-specific amplification. High resolution typing on patients blood samples obtained after treatment was done by reverse PCR-SSO on microbeads, PCR-SSP and/or direct sequencing.aResult shown in Figure 1.bWith a monocytosis of 57% (29.8 G/L).
HLA-DPB1 SSP (lot N°47G)
3 4 11 22
HLA-DPB1*02:02: mix 8,9,10,13,14,15,47 HLA-DPB1*04:01: mix 3,4,11,22
Figure 1. HLA-DPB1 PCR-SSP typing (Genovision, lot 47G) of patient DMF at diagnosis. The four mixes (nos. 3, 4, 11, 22) specific for the DPB1*04:01 allele present in o10% of the peripheral blood lymphocytes appear as faint bands due to low DNA template amounts.
2081
bead 514 for B*35:01 (Table 3). For the assignment of B*35 at a 15% ratio, three probes were negative but very close to the cutoff level (20%, 8% and 13% below, respectively). These results show that the detection level of the luminex microbeads assay for heterozygous combinations is 15% in mixing experiments of two homozygous DNA samples. Taking into account the double amount
of the minor DNA template due to homozygosity we calculate that the HLA misassignment threshold of 10% would correspond to a 20% ratio of normal heterozygous cells over 80% leukemic blasts with LOH. This result was then confirmed by a mixing experiment using the homozygous HLA-A*01:01; B*08:01; DRB1*03:01 DNA#1 and a heterozygous HLA-A*01:01,A*11:01; B*08:01,B*35:01;
ELHM
DMF
ND
HLA-A
HLA-B HLA-C
HLA-C HLA-B
HLA-DRB5
HLA-DRB5
PA
JH
CJ
HLA-A HLA-G
HLA-J
HLA-E
HLA-A
HLA HLA-DRB5 p21.33
p21.32
Figure 2. aCGH assays using oligonucleotide aCGH. The profiles of a 7-Mb window at 6p21p22 are depicted for the six patients analyzed in this study. Regions showing for each patient the details of deleted segments are shown (arrows). For patient ELHM, the regions of the class I loci are enlarged (top right) to illustrate the lack of deletion involving the HLA genes. For patients DMF, ND (a), and PA, JH, CJ (b), the regions with unambiguous deletions/duplications (with signals onX2 contiguous probes) are enlarged on the right part of the figure. For patients ELHM, PA and JH, the homozygous or heterozygous absence of DRB5 is marked by a circle.
Pretransplant HLA mistyping in diagnostic samples of AML patients V Duboiset al 2082
Leukemia (2012) 2079 -- 2085 &2012 Macmillan Publishers Limited
DRB1*03:01,DRB1*11:04 DNA#3 at the same 90--10%, 85--15%, 80-- 20%, 70--30% and 50--50% ratios. Figure 3 shows the hybridization patterns of five representative A*11:01-specific beads. For the detection of the A*11:01 allele, seven beads were negative at a
15% ratio (for example, beads 27,63,64,554 in Figure 3), four at a 20%
ratio (for example, bead 554 in Figure 3) and only one at a 30% ratio (data not shown). The combined luminex analysis of the 29 A*11:01-, B*35:01- and DRB1*11:04-specific Table 3. Reactivity (MFI) of negative beads specific for the HLA allele present at a 10/90% and 15/85% ratio in mixing experiments with homozygous DNA#1 and DNA#2
Bead nb Cutoff A*01 A*32 B*08 B*35 DRB1*03 DRB1*11
10%a 15%b 10% 15% 10% 15% 10% 15% 10% 15% 10% 15%
27 45 37 pos
56 45 42 pos
61 45 40 pos
67 70 65 pos
3 45 43 pos
17 25 18 23
26 35 31 pos
54 15 o15 pos
24 25 15 24
99 25 24 pos
521 20 14 o20
530 15 11 pos
545 25 17 pos
570 40 29 pos
71 45 41 pos
94 15 9 12
514 70 56 64
529 15 10 13
567 30 27 pos
46 40 o40 pos
58 20 13 18
39 25 22 pos
73 35 29 pos
Abbreviation: MFI, mean fluorescence intensity. At a 10/90% ratio, none of the six HLA-A, B, DRB1 alleles could not be assigned because of 23 negative beads of a total of 53 allele-specific beads. At a 15/85% ratio, the two alleles at each locus could be assigned by modifying the cutoff of a single probe except for B*35 (three negative probes but close to the cutoff value). In the second column, the cutoff values for bead positivity are represented as indicated by the supplier.
aAt a 10/90% ratio, the MFI values for each bead when they were below the cutoff value are shown.bAt a 15/85% ratio, the MFI values for each bead when they were below the cutoff value are shown, all MFI values above the cutoff are indicated as pos.
27
10%
15%
20%
30%
64
100 90 80 70 60 50 40 30 2010
100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 50 40 30 20 10
100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 50 4030 20 10 10090 80 70 60 50 40 30 20 10
100 90 8070 60 50 40 30 20 10 100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 50 40 30 20 10 100 90 80 7060 50 40 3020 10
100 90 80 70 60 50 40 30 20 10 100 90 80 70 60 5040 30 20 10
100 90 80 70 60 50 4030 20 10 10090 80 70 60 50 40 30 20 10
100 90 80 70 60 50 4030 20 10 100 90 80 70 60 50 40 30 20 10 100 90 8070 60 5040 30 20 10 100 90 80 70 60 50 40 30 20 10
63 554 583
Figure 3. HLA microbeads array genotyping after mixing experiments of homozygous DNA#1 (HLA-A*01:01; B*08:01; DRB1*03:01) with decreasing amounts of heterozygous DNA#3 (HLA-A*01:01,*11:01; B*08:01,*35:01; DRB1*03:01,*11:04), with ratios of 50, 30, 20, 25 and 10%.
Screen prints (HLA Fusion software, One Lambda) of five representative HLA-A*11:01-specific beads after analyzing samples containing 10, 15, 20 and 30% A*11:01-positive DNA template are shown. Of the 11 A*11-specific beads, 9 were negative at a 10% ratio, 7 at a 15% ratio, 4 at a 20% ratio and 1 at a 30% ratio. HLA-A heterozygosity could only be unambiguously assessed in samples with 420% DNA#3. Red bars correspond to the samples tested in the mixing experiments. Horizontal line represents the cutoff value as defined by the manufacturer.
yaxis: MFI. Other clinical samples have been analyzed in parallel (green bars).
2083
beads showed that the detection level of the heterozygous DNA was in the range of 20 -- 30% with 19 (66%) positive beads at a 20% ratio and 25 (86%) positive beads at a 30% ratio. Therefore, loss of a given HLA allele would go undetected by the luminex technique if tested on a blood sample containingX80% blasts.
DISCUSSION
We describe here six AML patients with LOH in the HLA region due to aUPD as shown by aCGH assays performed on the blood samples collected at diagnosis. Although the patients originated from different centers, the results reported in this study show that copy-neutral LOH involving chromosome 6 may not be an exceptional event. Furthermore, if aUPD concerns only part of the HLA loci, such as observed in three patients, its frequency may also be underappreciated by the HLA laboratories. As a minimal frequency estimation, we can rely on two studies. In the largest study performed so far on 454 AML patients, only one aUPD event involving 6p21 was reported (Supplementary Figure 2F, Gupta et al.13), which would represent 0.22% of all AML patients. In the same study, two gains of the entire chromosome 6 (0.44%) were observed, but such alterations are not expected to affect the HLA types. In the second study on 157 cytogenetically normal AML patients, aUPD on 6p21 was observed in two (1.27%) patients (Figure 1a, Bullinger et al.14). Very recently three cases of LOH leading to HLA homozygosity have been described representing about 4% of patients suffering of myeloid malignancies tested in this center.16Although mechanisms contributing to the LOH were not addressed in this study, we predict that aUPD may account for these LOH cases.
In our study, the actual frequency of LOH among AML patients is difficult to assess precisely, because the six cases have been detected in five different laboratories with four of them in the last year and two in previous years. Also, the algorithm of requesting patient’s HLA typing straight at diagnosis or in remission varies depending on the transplant centers connected to the five HLA laboratories. We could estimate that 150 -- 200 AML patients, whose blood samples were drawn at diagnosis, have been tested in the timeframe in which the six cases were detected. Therefore, a 3 -- 4% LOH incidence can be calculated as a minimal estimate, which is very close to the recently reported rate.16aUPD involving 6p21 in pretransplant samples may thus be more frequent than expected. The risk of HLA allele misassignment would even be higher if aUPD does not involve the entire major histocompat- ibility complex thereby leading to only partial HLA homozygosity, such as observed in patient CJ. An apparent LOH will be readily detected by the segregation analysis upon family genotyping, as observed in three of the patients described in this study. In case no family member is available, the second control typing, as is mandatory for any patient by the laboratory standards for histocompatibility testing,17 should be performed on a buccal swab, or on a blood sample from the patient at remission.
Alternatively, any homozygous HLA typing result obtained in blastic cells should be verified by PCR-SSP.
The HLA typing results (Table 2) obtained by PCR-SSO reverse hybridization on microbeads arrays and on biallelic SBT show that when locus-specific PCR is used for HLA typing, the second haplotype will be missed ifX80% cells with aUPD are present in the blood sample. Generic primers did not allow an adequate amplification of the second allele when present in o20% of peripheral blood lymphocytes. Only when applying monoallelic SBT and PCR-SSP the second allele could be detected, although at a much lower level (Figure 1). Because the PCR-SSO reverse hybridization on microbeads arrays (luminex technology) is an increasingly used method for routine HLA typing of clinical samples we sought to define a threshold level for the detection of two alleles at each of the HLA-A, B, DRB1 loci. Based on DNA mixing experiments, we have estimated that the detection
threshold of HLA heterozygotes was achieved when blood samples hado80% blasts. These results imply that upon typing AML patients in blastic phase one must be cautious in interpreting any result showing homozygosity in the whole HLA region or even at a single HLAlocus. In the absence of aUPD, a microdeletion such as the one involving the HLA-B in patient DMF may also lead to an apparent homozygosity that would go undetected if it occurs early during the blastic transformation and if no family genotyping is performed to assess the segregation of the HLA haplotypes. In chronic myeloid leukemia patients, specific chromosomal alterations, losses and gains were unique to cases with clinically manifested or suspected accelerated/blast stage.18
To our knowledge, this is the first study showing an aCGH pattern for patients with aUPD involving part of or the entire HLA region. The technique also revealed constitutive or de-novo microdeletions or gains at 6p21 in four of the six patients.
It would be interesting to evaluate a larger set of AML patients by aCGH searching small-scale copy number variations that could possibly involve any of the HLA or non-HLA genes. HLA homozygosity could provide a selective advantage to neoplastic cells by restricting the potential of presentation of tumor-specific epitopes by HLA antigens. Interestingly, a recent study19 has reported a slightly increased homozygosity rate at HLA-A, B, C, DRB1 loci in a large cohort of chronic lymphocytic leukemia patients compared with HLA allele distribution in the population.
Copy-neutral LOH at 6p arms has recently been observed in 13%
aplastic anemia patients, all involving the HLA region. A proposed mechanism for escape hematopoiesis in these patients with 6p LOH postulated that CTL autoimmunity would be biased by reducing the presentation of still unknown autoantigens by the missing HLA molecules.20
If aUPD in AML blasts involving part of, or the entire HLA region leads to decreases T-cell immunosurveillance, then the possibility exists that haploidentical or mismatched unrelated hematopoietic stem cell transplantation donors or cord blood units could be selected on the basis of the lost HLA haplotype or HLA single antigen as detected in the pretransplant AML blasts with aUPD involving 6p21.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
We are grateful to Dr A Devys, P Herry (EFS Nantes) and I Mollet (EFS Lyon) for their technical support. This study has been supported by the Swiss National Science Foundation (Grant 320030_130483).
REFERENCES
1 Garrido F, Ruiz-Cabello F, Cabrera T, Pe´rez-Villar JJ, Lopez-Botet M, Duggan-Keen Met al.Implications for immunosurveillance of altered HLA class I phenotyped in human tumors.Immunol Today1997;18: 89 -- 95.
2 Marincola FM, Jaffee EM, Hicklin DJ, Ferrone S. Escape of human solid tumors from T-cell recognition: molecular mechanisms and functional significance.
Adv Immunol2000;74: 181 -- 273.
3 Wetzler M, Baer MR, Stewart SJ, Donahue K, Ford L, Stewart CCet al.HLA class I antigen cell surface expression is preserved on acute myeloid leukemia blasts at diagnosis and at relapse.Leukemia2001;15: 128 -- 133.
4 Masuda K, Hiraki A, Fujii N, Watanabe T, Tanaka M, Matsue K et al.Loss or down-regulation of HLA class I expression in freshly isolated leukemic blasts.
Cancer Sci2007;98: 102 -- 108.
5 Vago L, Perna SR, Zanussi M, Mazzi B, Barlassina C, Lupo MTet al. Loss of mismatched HLA in leukemia after stem-cell transplantation.New Engl J Med 2009;361: 478 -- 488.
6 Villalobos IB, Takahashi Y, Akatsuka Y, Muramatsu H, Nishio N, Hama Aet al.
Relapse of leukemia with loss of mismatched HLA resulting from uniparental disomy after haploidentical hematopoietic stem cell transplantation.Blood2010;
115: 3158 -- 3161.
Pretransplant HLA mistyping in diagnostic samples of AML patients V Duboiset al 2084
Leukemia (2012) 2079 -- 2085 &2012 Macmillan Publishers Limited
7 Waterhouse M, Pfeifer D, Pantic M, Emmerich F, Bertz H, Finke J. Genome-wide profiling in AML patients relapsing after allogeneic hematopoietic cell transplantation.Biol Blood Marrow Transpl2011;17: 1450 -- 1457.
8 Tuna M, Knuutila S, Mills GB. Uniparental disomy in cancer.Trend Mol Med2009;
15: 120 -- 128.
9 Raghavan M, Lillington DM, Skoulakis S, Debernardi S, Chaplin T, Foot NJet al.
Genome-wide single nucleotide polymorphism analysis reveals frequent uniparental disomy due to somatic recombination in acute myeloid leukemias.
Cancer Res2005;65: 375 -- 378.
10 Gorletta T, Gasparini P, D’Elios MM, Trubia M, Pelicci PG, Di Fiore PP. Frequent loss of heterozygosity without loss of genetic material in acute myeloid leukemia with a normal karyotype.Genes Chromosomes Cancer2005;44: 334 -- 337.
11 Serrano E, Carnicer MJ, Orantes V, Estivill C, Lasa A, Brunet S et al.
Uniparental disomy may be associated with microsatellite instability in acute myeloid leukemia (AML) with normal karyotype.Leuk Lymphoma2008;
49: 1178 -- 1183.
12 Raghavan M, Smith LL, Lillington DM, Chaplin T, Kakkas I, Molloy G et al.
Segmental uniparental disomy is a commonly acquired genetic event in relapsed acute myeloid leukemia.Blood2008;112: 814 -- 821.
13 Gupta M, Raghavan M, Gale RE, Chelata C, Allen C, Molloy Get al.Novel regions of acquired uniparental disomy discovered in acute myloid leukemia. Genes Chromosomes Cancer2008;47: 729 -- 739.
14 Bullinger L, Kro¨nke J, Scho¨n C, Radtke I, Urlbauer K, Botzenhardt U et al.
Identification of acquired copy number alterations and uniparental disomies in cytogenetically normal acute myeloid leukemia using high-resolution single-nucleotide polymorphism analysis.Leukemia2010;24: 438 -- 449.
15 Itsara A, Wu H, Smith JD, Nickerson DA, Romieu I, London SJet al. De novorates and selection of large copy number variation.Genome Res2010;20: 1469 -- 1481.
16 Pereira S, Vayntrub T, Hiraki DD, Cherry AM, Arai S, Dvorak CCet al.Short tandem repeat and human leukocyte antigen mutations or losses confound engraftment and typing analysis in hematopoietic stem cell transplants.Human Immunol2011;
72: 503 -- 509.
17 http://www.efiweb.eu/index.php?id¼129. EFI Standards for histocompatibility testing, 2009.
18 Brazma D, Grace C, Howard J, Melo JV, Holoyke T, Apperly JFet al.Genomic profile of chronic myeloenous leukemia: imbalances associated with disease progression.
Genes Chromosomes Cancer2007;46: 1039 -- 1050.
19 Shah N, Decker WK, Lapushin R, Xing D, Robinson SN, Yang H et al. HLA homozygosity and haplotype bias among patients with chronic lymphocytic leukemia: implications for disease control by physiological immune surveillance.
Leukemia2011;25: 1036 -- 1039.
20 Katagiri T, Sato-Otsubo A, Kashiwase K, Morishima S, Sato Y, Mori Yet al.Frequent loss of HLA alleles associated with copy number-neutral 6pLOH in acquired aplastic anemia.Blood2011;118: 6601 -- 6609.
2085
