GvHD prophylaxis modulates the influence of HLA mismatches on the unshared haplotype on the outcomes after T-cell repleted haploidentical stem cell transplantation.
F. Lorentino (Milan, IT)
Haploidentical hematopoietic stem cell transplantation (haplo-HSCT) with T-repleted grafts is an option for high risk acute leukemia patients (pts) who lack a matched donor, with GvHD prophylaxis mainly based on anti-thymocyte globulin (ATG) or post-transplant cyclophosphamide (PTCy). We aimed to evaluate if HaploSCT outcomes could be shaped by HLA mismatches (mm) on the unshared haplotype (UH) between pts and donors, possibly refining current criteria for donor selection. We included 509 pairs with HLA-A, -B, -C and -DRB1 typing at low (66%) or high (34%) resolution. A validated high-resolution imputation algorithm, developed by the German National Donor Registry, was used to impute allele-level matching in 418 pairs. More than half of pts were in complete remission, and most transplants were performed with reduced intensity conditioning (52%). Stem cell source was peripheral blood (PB) for 61% of pts, bone marrow for 39%. ATG was used in 39% of cases, PTCy in 61%. In multivariate analysis (including center effect, pts age, donor gender, diagnosis, disease status, conditioning intensity, stem cells source) an antigenic but not an allelic mm at the DRB1 locus was an independent risk factor for aGvHD ≥2 in PTCy (HR 2.0; 95% CI 1.2-4.0, p=0.02) but not in ATG regimens. Neither mm at any HLA locus nor predicted NK alloreactivity based on KIR ligand mm in the GvL vector influenced transplant-related mortality, relapse and disease-free survival. The influence of clinical variables on aGvHD was modulated by the adopted GvHD prophylaxis: PB as stem cell source, myeloablative conditioning and female donors were independently associated with the hazards of aGvHD in PTCy (HR 2.2, 95% CI 1.4-3, p<0.01; HR 1.7, 95% CI 1.1-2.5, p=0.04; HR 1.8, 95% CI 1-3.2, p=0.05, respectively) but not in ATG regimens. Our data suggest that HLA mm on the UH do not appear to be of prominent importance for haplo-donor selection. However, its role and other factors influencing alloreactivity might be modulated by the adopted GvHD prophylaxis, calling for further investigations in this field.
Screening for a biomarker panel for prediction of graft-versus-host disease in humans
H. Budde (Göttingen, DE)
Biomarkers are important utilities to allow personalized therapy. Up to now no biomarker for GvHD occurrence after hematopoietic stem cell transplantation (HSCT) is in clinical practice. As potential biomarkers we analyzed CD3+ T cells, CD3+ CD4+ T-helper cells, CD3+ CD8+ cytotoxic T cells, CD4+ CD25+ FoxP3+ Tregs and CD19+ CD21- precursor B cells. Furthermore we quantified the serum levels of soluble IL-2 receptor, soluble TNF receptor and hepatocyte growth factor. We included 28 patients with hematopoietic malignancies in the study. Biomarker levels were determined at the time point of hematopoietic engraftment around two to three weeks after HSCT. During follow-up for 12 months occurrence and clinical severity of GvHD was monitored. We found that the CD4/CD8 T cell ratio was significantly increased in the patients who later developed GvHD. Another potential biomarker with significant differences between the GvHD and non-GvHD group was sIL2-R with lower concentrations in GvHD group. Highly significant differences were found by combination of a panel of the 5 biomarkers CD4+ cells, CD8+ T cells, CD4/CD8 T cell ratio, CD19- CD21+ precursor B cells and sIL-2R. For each of these markers we set a cut-off value. Reaching this value was defined with 1 point resulting in a GvHD score with a maximum of 5 points for every patient. Patients with a score of at least 4 points developed GvHD with a sensitivity of 77% and a specificity of 89% with an area under curve of 0.90 in the receiver operating curve. In summary our data provide information about potential cellular and cytokine biomarkers for GvHD prognosis. In particular, a panel of five combined biomarkers seems to be promising to predict future GvHD occurrence already at the date of the hematopoietic engraftment. A multi-center study is necessary to evaluate our results for clinical use.
A rapid and sensitive molecular tool for the diagnosis of HLA loss relapses after partially-incompatible allogeneic HSCT
M. Ahci (Essen, DE)
Genomic loss of the mismatched HLA haplotype represents a frequent and clinically-relevant mechanism of leukemia immune evasion and relapse after partially-incompatible allogeneic HSCT. Since donor lymphocyte infusions are expectedly inefficacious against HLA loss variants, reliable diagnostic assays are needed to inform the choice of relapse treatment in all centers performing partially HLA-mismatched HSCT. Here we designed an innovative methodology to detect HLA loss relapses, based on the combination of qPCR-based chimerism for "outside HLA" host markers with ad hoc designed qPCR reactions targeting HLA allele groups ("inside HLA" markers). Concordance or discrepancy between inside and outside HLA host markers identifies "classical" and HLA loss relapses, respectively. We designed and validated a total of 10 "inside HLA" reactions specific for the most frequent HLA-A, -C and -DPB1 allele groups, covering two thirds of a representative series of 165 consecutive haploidentical HSCT. Each reaction was tested against a panel of HLA-typed cell lines (n>20) to confirm specificity and absence of cross-reactivity. All reactions displayed over 80% efficiency, with superimposable performance in water and in target-negative DNA (R=0.99, p<0.0004), sensitivity of at least 0.2%, and high concordance between the expected and experimentally-determined chimerism (R= 0.99, p<0.0001). The clinical utility of this system was shown in 9 relapses after partially HLA-mismatched HSCT. In all cases we detected host-specific chimerism by outside HLA markers and, as expected, inside HLA markers resulted positive in case of classical relapses (5/5) and negative for HLA loss relapses (4/4). In conclusion, we developed a highly sensitive, reliable, and easy-to-implement molecular tool to unequivocally discriminate between classical and HLA loss relapses, facilitating further studies and the implementation of this analysis in the therapeutic algorithm for post-transplantation relapses.
Investigating HLA-B/C and -DRB1/DQB1 linkage disequilibrium to the third and fourth field of resolution using Pacific Biosciences Single Molecule Real Time (SMRT®) sequencing.
D. Merlo (London, GB)
Current criteria for the selection of unrelated donors for haematopoietic stem cell transplantation include matching for the alleles of the most polymorphic HLA loci within the major histocompatibility complex. The HLA loci are located on chromosome 6 and HLA-B/C and -DRB1/DQB1 are in tight linkage-disequilibrium (LD): this means that they are non-randomly associated with each other due to their physical proximity. In January 2016 our centre introduced SMRT® sequencing for HLA typing patients and donors routinely in a clinical environment at allele-level resolution. The aim of this study is to provide an elucidation of HLA-B/C and -DRB1/DQB1 linkage to the third and fourth field of resolution. For example, we have observed B*07:02:01 associated with C*07:02:01:03 5861 times versus 165 with C*07:02:01:01. Also, DRB1*01:01:01 has been typed in association with DQB1*05:01:01:03 3129 times, and only 29 with DQB1*05:01:01:01. Furthermore, we observed a large difference in LD between intronic variants of the same allele. HLA-B*08:01:01:01, frequently observed in the British-Irish population, has been typed in association with C*07:01:01:01 2152 times versus 17 with C*07:02:01:01 whereas B*08:01:01:02, commonly observed in Indian/Pakistani population, has been seen in association with C*07:02:01:01 78 times and only twice with C*07:01:01:01. As described by Petersdorf et al in 2007, haplotype mismatches between patient and donor can induce graft-versus-host responses. We hypothesise that synonymous and intronic differences in HLA genes leading to mismatches at the third and fourth field of resolution may serve as proxies for identifying haplotype mismatches that could potentially trigger graft-versus-leukaemia effect, associated with a lower relapse risk. This assumption might support the clinicians in the unrelated donor selection process, as donors mismatched at third or fourth field of resolution could be preferentially chosen for patients suffering from malignant diseases associated with high risk of relapse. Alternatively, a fully matched donor could be selected for patients with genetic or metabolic blood disorders, or malignant conditions with a lower relapse risk. For the laboratory routine, more in-depth LD information will support the HLA typist, increasing confidence in HLA genotyping results.
MICA and NKG2D polymorphisms have a significant impact on acute graft versus host disease after HLA-matched hematopoïetic stem cell transplantation.
N. GUILLAUME (AMIENS, FR)
MICA (MHC class I polypeptide-related sequence A) is a highly polymorphic gene closely linked to the HLA-B locus. It encodes a cell stress inducible glycoprotein, which mediates an activatory signal towards the NKG2D receptor expressed on NK-cells, CD8+ T-cells and NKT-cells. MICA polymorphisms have been shown to influence NKG2D signaling. Indeed, a methionine to valine change at position 129 in exon 3 categorized the MICA alleles into strong (MICA-129 met) and weak (MICA-129 val) binders of NKG2D receptor. Five repetitions of GCT with 1 additional nucleotide insertion (G) in exon 5 designed the MICA A5.1 alleles with a premature stop codon. Moreover, NKG2D polymorphisms identified alleles associated with a low (NKC3 C/C and NKC4 C/C) or high cytotoxic activity (NKC3 G/G and NKC4 T/T).
Here, we evaluated whether recipient MICA and donor NKG2D polymorphisms (respectively MICA-129, MICA A5.1 and NKC3, NKC4) could influence the incidence of graft versus-host disease (GvH), overall survival (OS) and relapse free survival (RFS) on 124 patients undergoing allogenic hematopoietic stem cell transplantation using an HLA-matched donor (10/10).
In an univariate model, Recipient MICA A5.1 heterozygosity (p=0.030), donor NKC4 C/C polymorphism (p=0.013) are associated with the increase of incidence of acute GvH (grade I to IV). Recipient MICA A5.1 heterozygosity is also associated with chronic GVH (p=0.04) while Recipient MICA-129 val/val tends to be a risk factor of chronic GVH without being statistically significant. These polymorphisms have no significant impact on OS and RFS in our study. Our data suggest that a MICA or NKG2D low activity status can be related to an increase of acute GVH according to a mechanism that remains to be elucidated.
A new NGS approach for total (classical and non-classical) HLA genotyping
M. Alizadeh (Rennes, FR)
In the recent years we developed an approach using NGS sequencing from the long range PCR for HLA typing. There are some particularities in our approach. First, HLA-E, -F and -G are typed at the same time; second, amplification of these loci and classical class I loci takes place in the same vials and third, we also amplify HLA-J, -H and -K. The 9 loci, HLA-A, -B, -C, -E, -F, -G, -J, -H and -K are amplified in two vials. This number of genes allows a better haplotype determination. All class I genes are amplified from 5’ to 3’ UTR, the same as for DQB1, while DPB1 is amplified from intron 2 to 3’ UTR and DRB1,3, 5 are amplified in two vials one from 5’ UTR to intron 1 and the other from intron 1 to 3’ UTR. We also developed new software (Profiler) that allows analysis of all loci in a unique row without any influence of all class I genes between each other. In this work, 60 fully HLA typed samples (SBT or SSO plus SSP) were retyped by our strategy and analyzed blindly for all classical HLA (except DRB4) alleles plus HLA-E, -F and -G. There are a few differences between the results. In two cases we identified new polymorphism in exon 7 for HLA-C, one for C*16:01 and the second for C*12:03 (which is known today as *12:143). We also observed differences in the DPB1 locus, that concerned DPB1*03:01 and *104:01 with a polymorphism located in exon 4. In this study a total of 31 new alleles were described for HLA-B (1), -C (1), -DRB1 (14), -DQB1 (5) and -DPB1 (10). Thirty six (36) samples have a DRB3 allele with 8 new alleles among them and 5 samples have one DRB5 without any new alleles. Only one new allele was observed with an exonic region (HLA-C*16:01:01G, exon 7 1087 G > A, Thr replaces Ala). All other novel SNPs were located in intronic regions. Considering HLA-E, -F and -G, 4 new alleles were observed with one mutation located in exon 3 (synonymous) for HLA-F. Interestingly some of these novelties have been found more than once, i.e. DRB1*14:54:01 which has a difference with reference sequence (intron 2 1578 G > A) in all 6 cases studied. Another (intron 5 301 G > C) might be associated with a specific haplotype A*02:01:01:01, C*03:03:01, B*15:01:01:01, DRB1*13:01:01:NEW, DQB1*06:03:01 (2 cases).
Association between multiple mismatches at the HLA-DPB1 and DRB3/4/5 genes and adverse outcomes in HLA-A, -B, -C, -DRB1 and -DQB1 identical hematopoietic stem cell transplantation
S. Ducreux (LYON, FR)
Matching for HLA-A, -B, -C, –DRB1 loci (8/8 match) is associated with the highest overall survival (OS) rates after unrelated donor (URD) hematopoietic stem cell transplantation (HSCT). A detrimental role of additional HLA-DQB1, HLA-DPB1 and HLA-DRB3/4/5 mismatches (MM) has been recently identified on OS. We investigated the impact of HLA-DPB1 and HLA-DRB3/4/5 MM on outcomes in a large cohort of 10/10 matched URD HSCTs. 2,393 patients who received an initial HSCT from a 10/10 matched URD in 35 French centers were included between January 2000 and October 2012. High-resolution typing was performed for HLA-A, -B, -C, -DRB1, -DQB1, -DPB1 and -DRB3/4/5 loci for all donor/recipient pairs. Clinical data were obtained through ProMISe. Patients were classified into 5 different groups according to their global matching level for the 7 considered loci (10/14 to 14/14). HLA-DPB1 MM were classified as permissive or non-permissive. The primary composite endpoint for the analysis was GvH disease (GvHD)-free and relapse-free survival (GRFS). Acute GvHD (aGvHD), chronic GvHD, relapse and OS were also studied. The median follow-up was 59 months. Compared to 14/14 pairs, a significantly lower early GRFS (within 3 months after HSCT) was observed for patients who received a 10-11/14 (Hazard Ratio (HR) 2.0, 95% CI 1.4 to 2.9, p=0.0003) or a 12/14 URD HSCT (HR 1.4, 95% CI 1.1 to 1.8, p=0.01). This was related to an increased risk of grade III-IV aGvHD associated with 10-11/14 (HR 2.3, 95% CI 1.5 to 3.7, p=0.0004) and 12/14 pairs (HR 1.7, 95% CI 1.2 to 2.4, p=0.003). 10-11/14 pairs were also associated with a higher risk of death at 3 months (HR 2, 95% CI 1.1 to 3.6, p=0.024). In patients matched for HLA-DRB3/4/5 but MM for HLA-DPB1 (n=1846, 77.1%), two HLA-DPB1 MM and non-permissive HLA-DPB1 MM were associated with a lower early GRFS (HR 1.4, 95% CI 1.1 to 1.9, p=0.01 and HR 1.3, 95% CI 1.0 to 1.7, p=0.02 respectively) due to an increased risk of aGvHD (HR 1.7, 95% CI 1.2 to 2.5, p=0.002 and HR 1.50, 95% CI 1.08 to 2.08, p= 0.017 respectively). Only 19 pairs (0.8%) were DPB1 matched and DRB3/4/5 mismatched; HLA-DRB3/4/5 MM could thus not be analyzed. Multiple HLA-DPB1 and HLA-DRB3/4/5 MM have an early impact after 10/10 matched URDs HSCT. Prospective evaluation of matching for HLA-DPB1 and HLA-DRB3/4/5 is warranted to reduce early post-HSCT toxicity in donor-recipient 10/10 matched pairs.
HLA null alleles defined using NGS recruitment typing in unrelated donors on the National Marrow Donor Program’s Be The Match Registry
M. Bauer (Minneapolis, Minnesota, US)
Null alleles are relevant for HLA matching in the HSCT setting. This study describes the occurrence of null alleles, which is poorly understood, in an unrelated donor (URD) population. In total, 495,262 URDs recently typed using targeted exon or whole gene, long range NGS methods and added to the Be The Match Registry® were evaluated. Associations were compared to ASHI Ad Hoc Committee’s 2012 report and the ZKRD URD population described in 2016 (Eberhard et al.). Our observed results showed that the expected allele associations were most frequent for HLA-A*24:09N, -B*51:11N, -C*04:09N, -DRB4*01:03N and -DRB5*01:08N. However, additional associations were observed for all these alleles aside from DRB5*01:08N. Interestingly, while A*24:11N, part of the A*24:02:01G group, occurred commonly with the anticipated C*05:01 in URDs self-identified as white, the same was not shown in URDs self-identified as Asian or black, where it primarily associated with either C*04:01 or C*06:02:01G. Eleven null alleles not included in the CWD 2.0.0 catalogue were observed in at least 5 URDs or at least 3 URDs in conjunction with a conserved haplotype. These may be candidates for a future CWD catalogue. Of particular interest for matching is A*31:14N, part of the A*31:01:02G group, due to the polymorphism outside the ARS in exon 4. It was detected in 5 URDs, all self-identified as white, with 3 of 5 carrying the B*40-C*03-DRB1*04-DQB1*03 extended haplotype. As expected, C*04:09N was almost always observed in association with B*44:03 (99%). In 3 URDs (1%), it was detected in association with B*44:02. Unexpectedly, it was confirmed outside of the B*44 association in one URD, appearing to associate with B*37:01, likely within the A*23:01-B*37:01-DRB1*10:01-DQB1*05:01 extended haplotype. To our knowledge, this is the first published description of C*04:09N not in the presence of B*44. The evolution of NGS HLA typing methodologies, which are increasingly available and cost effective, are likely to broaden our understanding of null allele frequencies and associations in populations. Our results provide evidence for labs to consider this level of testing as a new standard for best patient care.