Recipient CTLA-4*CT60-AA genotype is a prognostic factor for acute graft-versus-host disease in hematopoietic stem cell transplantation for thalassemia.

Polymorphisms of the cytotoxic T-lymphocyte antigen-4 gene (CTLA-4) have been associated with autoimmune diseases and it has recently been reported that donor genotypes correlate with the outcome of allogeneic hematopoietic stem cell transplantation in leukemia patients. With the aim of confirming this finding in thalassemia patients, we investigated the influence of genotype distribution of 3 CTLA-4 gene polymorphisms in 72 thalassemia patients and their unrelated donors. A significant association was observed for recipient CT60-AA genotype and onset of grade II-IV (63.2% vs 24.5%; p = 0.001) and grade III-IV (36.4% vs 7.6%; p = 0.005) acute graft-versus-host disease (aGVHD). The same association was observed for the 88-base-pair allele of the CTLA-4 (AT)n polymorphism, which was determined to be in complete linkage disequilibrium with the CT60 A allele. Multinomial Cox regression demonstrated that this association was independent of CT60 donor genotypes or other risk factors (p = 0.016; hazard ratio = 2.8). Our data confirm that the genetic variability in CTLA-4 is an important prognostic factor for aGVHD and suggest that some of the risk factors for this complication are generated by recipient cells that persist after the myeloablative conditioning regimen. Copyright Â

Introduction

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is currently one of the most effective methods of treatment for both hematological malignancies and nonmalignant disorders; it also represents a promising therapeutic approach in some solid tumors [1,2]. Unfortunately, allo-HSCT is not exempt from serious complications, such as acute graft-versus-host disease (aGVHD), an immune-mediated condition caused by the attack of donor T lymphocytes against recipient tissues, which represents the primary cause of treatment-related failure [3]. By contrast, alloimmune responses directed against malignant recipient cells have a graft-versus-tumor effect, which plays a major role in permitting the achievement and/or persistence of complete remission of the neoplastic disease. Hence, the recognition of factors involved in the regulation of alloimmune response is an essential step toward a broader and more successful application of allo-HSCT.

As mentioned above, aGVHD is largely the result of donor T-cell alloreactivity against disparate major or minor histocompatibility antigens of the recipient. This process is strongly influenced by a plethora of membrane-bound receptors and soluble molecules capable of either up- or downregulating the duration and entity of the immune response. Among these, the cytotoxic T lymphocyte–associated antigen-4 (CTLA-4) cell surface molecule plays a prominent role [4]. In humans, the CTLA-4 protein is encoded by the homonymous gene located on chromosome 2p22. The CTLA-4 molecule exists as 2 isoforms: a full-length membrane-bound isoform (flCTLA-4) and a soluble form (sCTLA-4) lacking the entire transmembrane domain. The alternative splicing that generates these isoforms seems to be regulated by genetic polymorphisms within or in close proximity to the CTLA-4 gene [5] It was initially believed that sCTLA-4 was basally expressed in healthy individuals [5] and, therefore, a decrease in protein expression correlated with susceptibility to autoimmune diseases [6,7].

However, several studies have demonstrated elevated serum sCTLA-4 levels in patients with autoimmune disorders compared with healthy individuals [8-11]. Consequently, sCTLA-4 expression seems to be associated with lymphocyte activation rather than quiescence [12,13].

A recent report has uncovered the importance of CTLA-4 for regulatory T cell (Treg) function by demonstrating that antibody blockade of CTLA-4, particularly at the intestinal level, completely abrogates Treg function [14], thus confirming the key role of CTLA-4 in the determination of immune tolerance. The role of Tregs in tolerance to antigens has been extensively described in a recent review [15].

In recent reports, CTLA-4 polymorphisms have been correlated with the clinical outcome of allo-HSCT. Pérez-Garcia et al. [16] reported that the risk for aGVHD was significantly increased in leukemia patients receiving a graft from a related donor homozygous for a polymorphism located downstream of the CTLA-4 gene (CT60-AA). When patients were transplanted from donors with this genotype, there was a significant improvement in overall survival. This improvement was interpreted by the authors as being the result of an increment in the graft-versus-tumor effect, thus preventing malignancy recurrence [16].

However, the most recent studies on the role of CTLA-4 polymorphisms in the outcome of allo-HSCT reported different conclusions [17-20]. The discrepancies observed in these previously published studies may be explained by patient heterogeneity and differences in conditioning regimens and/or study design.

In an attempt to obtain further information on this debated issue, we studied the role of CTLA-4 gene polymorphisms in the development of aGVHD in thalassemia patients given HSCT from an unrelated donor. Thalassemia patients have a competent immune system and constitute a homogeneous cohort in terms of disease, stem cell source, conditioning regimen, and GVHD prophylaxis. This clinical setting facilitates the evaluation of immunogenetic factors and their role in determining the outcome of transplantation.

Subjects and methods

Patients

We retrospectively investigated the influence of +49AG (rs231775), CT60 (rs3087243), and the polymorphic microsatellite marker CTLA-4(AT)n on the outcome of unrelated HSCT in 72 patients affected by thalassemia major.

The study was conducted on β-thalassemia major patients transplanted in 3 different Italian transplant centers. Approval was obtained from the local institutional review board of each participating center; informed consent was obtained from all patients or from their parents or legal guardians. The mean age of patients (41 males and 31 females) was 12.9 ± 7.9 years. The mean age of donors (39 males and 33 females) was 33.2 ± 8.0 years. High-resolution molecular typing of donor and recipient pairs was performed for the human leukocyte antigen (HLA)-A, -B, and -C loci and the HLA-DRB1, -DQB1, and -DPB1 loci, as previously reported [21]. Molecular typing confirmed that all pairs were completely identical for the HLA-A, -B, -C, -DRB1, and -DQB1 loci. In 33 allo-HSCT (46%) both the donor and the recipient were seropositive for cytomegalovirus. Eight donors (11%) had a mismatch with the recipient for an HLA-DPB1 allele. All patients received a myeloablative conditioning regimen. In 43 cases, the conditioning regimen was based on the combination of either busulfan or treosulfan (16 mg/kg and 14 g/m2 administered over 4 and 3 days, respectively) with thiotepa (8 mg/kg) and fludarabine (160 mg/m2). The remaining 29 patients were given a modified conditioning regimen with busulfan (14 mg/kg), thiotepa (10 mg/kg), and cyclophosphamide (120 mg/kg). Cyclosporin A and short-term methotrexate were used for GVHD prophylaxis in all patients. aGVHD was diagnosed and graded according to the Seattle criteria [21]. All patients were given unmanipulated bone marrow cells. Twenty-five patients developed grade II–IV aGVHD within 9 to 30 days after allo-HSCT. Ten of these patients had grade III–IV aGVHD. Seven patients rejected, 4 of whom lost the graft within 100 days of transplantation. One patient died within 100 days of transplantation without any evidence of GVHD.

CTLA-4 genotyping

Genotypes for +49AG and CT60 polymorphisms were determined by direct cycle sequencing. Briefly, for each sample, 20 to 50 ng of DNA was amplified by polymerase chain reaction (PCR) using specific primers. Five microliters of PCR products was purified by ExoSap and sequenced using the DYEnamic ET dye terminator cycle sequencing kit (GE Healthcare Biosciences, Piscataway, NJ). After a second purification carried out using Millipore Montage SEQ (Millipore, Co, Billerica, MA), the reactions were loaded onto a MegaBace 1000 capillary sequencer (GE Healthcare Biosciences). Genotypes at +49AG and CT60 were assigned by manual inspection of electrofluorograms.

Genotypes of the CTLA-4(AT)n microsatellite polymorphism were assigned by fluorescent fragment analysis. The PCR products, obtained using primers Fluo-GCC AGT GAT GCT AAA GGT TG-3′ and 5′-AAC ATA CGT GGC TCT ATG CA-3′, were loaded onto an automated capillary sequencer for electrophoretic separation and registration of electropherograms. The alleles were defined by comparisons with molecular markers.

Statistical analysis

Clinical outcome was analyzed after a median follow-up period of 69 months (range 3–155 months).

Comparison of cumulative incidence curves in the presence of competing risks were used to assess the statistical distribution of the genotypes of CTLA-4 polymorphisms in patients with and without GVHD. Death without evidence of aGVHD or rejection was considered a competing risk. Analysis was performed using the publicly available software R, cmprsk package (version 2.8.1, http://www.r-project.org). The CumIncidence function was used according to a previously reported procedure [22].

Departures from Hardy–Weinberg equilibrium and haplotype frequencies were calculated by a Markov chain simulation method and the expectation–maximization algorithm, respectively, as implemented in ARLEQUIN software package v. 3.0 [23]. A Windows version of the software is freely available at http://cmpg.unibe.ch/software/arlequin3.

Multinomial Cox regression with software package SPSS v. 13 (SPSS, Inc., Chicago, IL) was used to test for independence with respect to factors putatively involved in transplantation outcome. To adjust for the effects of each of the factors analyzed, these were simultaneously incorporated into the model. Multiple logistic regression was used to discriminate between associations of single polymorphisms and haplotype associations. A p value of ≤0.05 was considered statistically significant.

Results

We genotyped 2 single-nucleotide polymorphisms (49AG, CT60) and 1 microsatellite marker (CTLA-4(AT)n) within the CTLA-4 gene in 72 thalassemia patients and their respective unrelated donors. Initially, we tested both recipients and donors for independence and the eventual presence of hidden sampling stratifications. As expected, the 2 groups exhibited independence for allele and genotype distribution. The CTLA-4 genotypes of both donor and recipient groups were in Hardy–Weinberg equilibrium (p > 0.1).

We then compared the genotype distribution for each polymorphism versus either the presence or the absence of aGVHD. Two types of comparisons were performed: grade II–IV GVHD versus no GVHD–grade I GVHD and grade III–IV GVHD versus no GVHD–grade II GVHD. We determined that recipients with the CT60-AA genotype had a significantly higher risk of developing grade II–IV (63.2% vs 24.5%; p = 0.001; Fig. 1) or grade III–IV aGVHD (36.4% vs 7.6%; p = 0.005; Fig. 2). Identical association values were obtained for the microsatellite CTLA-4(AT)n (88-base-pair [bp] allele). As discussed afterward, this finding can be interpreted in the light of the complete linkage disequilibrium between the A allele of CT60 and the 88-bp allele of CTLA-4(AT)n. A weak but statistically significant association was observed for the risk of grade III–IV aGVHD in recipients with the 49AG-AA genotype (23.6% vs 5.6%; p = 0.04).

Fig. 1

Cumulative incidence of grade II–IV acute graft-versus-host disease (aGVHD) according to recipient CT60 genotype estimated 100 days after allogeneic hematopoietic stem cell transplantation (allo-HSCT).

Fig. 2

Cumulative incidence of grade III–IV acute graft-versus-host disease (aGVHD) according to recipient CT60 genotype estimated 100 days after allogeneic hematopoietic stem cell transplantation (allo-HSCT).

Multinomial Cox regression was used to test whether the association observed between recipient CT60-AA genotype and susceptibility to aGVHD was influenced by other factors. Both recipient and donor genotypes were included in the analysis to exclude any eventual interaction. Cox regression analysis demonstrated that the associations observed for CTLA-4 polymorphisms were independent of other putative or known risk factors for aGVHD, such as donor age, the female donor/male recipient combination, human cytomegalovirus occurrence, and HLA-DPB1 mismatch (see Table 1 for additional details).

Table 1

Multivariate Cox regression analysis to evaluate the impact of CTLA-4 genotypes on acute graft-versus-host disease (aGVHD) risk

Grade II–IV aGVHD risk variables	p	HR (95% CI)
Recipient CTLA-4 CT60 AA	0.016	2.8 (1.2–6.6)
Donor CTLA-4 CT60 AA	0.589	1.3 (0.5–3.0)
Gender mismatch	0.062	2.3 (1.0–5.4)
Donor age (>35 years)	0.464	1.3 (0.6–3.4)
Positive human cytomegalovirus in patients and their donors	0.089	2.0 (0.9–4.6)
DPB1 mismatch	0.850	1.2 (0.2–5.5)

HR, hazard ratio; 95% CI, 95% confidence interval; CTLA-4, cytotoxic T-lymphocyte antigen-4 gene.

Given the association with more than 1 polymorphism reported in this study, we analyzed the distribution of CTLA-4 haplotypes in recipients and tested the correlation of the latter with the onset of moderate or severe aGVHD. Eleven haplotypes with a frequency higher than 0.02 were present in our cohort of donors and recipients. To distinguish the role of the single alleles encoded by these haplotypes, we applied logistic regression analysis using 1, 2, or 3 loci as dichotomous variables. The results indicate that susceptibility to moderate or severe aGVHD is associated with the haplotype carrying the CT60-A and CTLA-4(AT)n 88-bp alleles. These 2 alleles were reported to be in complete linkage disequilibrium (Ds = 1; p = 1 × 10−8); thus, any further attempts to dissect the role of these 2 polymorphisms were abandoned. Moreover, each of these 2 alleles can completely explain the association identified in our study cohort, which means that 1 of these 2 alleles is more likely to render the host susceptible to aGVHD than the haplotype.

Discussion

CTLA-4 is a member of the immunoglobulin superfamily, encoding a protein that transmits an inhibitory signal to T cells that downregulates their activation. The role played by CTLA-4 gene polymorphisms in determining the clinical outcome of patients given allo-HSCT remains unclear. In particular, 2 recent studies [18,19] did not confirm the association of CTLA-4 polymorphisms with the risk of aGVHD originally reported by Pérez-Garcia et al. [16]. Heterogeneity among studies, low statistical power, and the relatively low risk conferred by the susceptibility alleles (odds ratio < 2) were considered to justify the discrepancies between the different studies.

From the results of the present study, it would seem that genetic variability of CTLA-4 is implicated in the development of aGVHD following allo-HSCT. Acute GVHD was significantly more frequent in patients homozygous for the CT60-A allele than in patients with the CT60-G allele (genotype AG+GG). Our findings are consistent with those obtained in the study performed by Pérez-Garcia et al. [16]. Even the fact that the association observed in our study is specific for the homozygous genotype (AA) agrees with the dominant effect of the G-allele observed by these authors [16]. By contrast, our data provide evidence that only recipient CTLA-4 genotype confers susceptibility to aGVHD following allo-HSCT. This finding is supported by the results of logistic regression analysis, which excluded the possibility of association with the different genotype combinations among donor and recipient pairs.

Our search for loci that influence the outcome of allo-HSCT was performed on a cohort of patients and their unrelated donors, which means that, except for the loci of the major histocompatibility complex, recipients and donors form 2 genetically independent groups. This offers a significant advantage over genetic studies of patients transplanted from first-degree family donors in whom the genotypes are expected to be consistent with mendelian probability.

For loci not linked to the major histocompatibility complex region, the probability of genotype identity between an unrelated donor and the recipient is given as the square of the genotype frequency within the population, whereas the probability of genotype identity for a related donor is directly dependent upon the degree of kinship. For example, given the frequency of the CT60-AA genotype within our study cohort, the probability of identity with the patient for this genotype would be 1:3 for sibling donors but only 1:14 for unrelated donors. This may have caused bias in previous studies and may also explain why we observed a higher risk for aGVHD compared with that observed by Pérez-Garcia et al., who only considered related donor genotypes [16].

Some of the controversy in the literature may also result from the variability of conditioning regimens and the different disorders studied (thalassemia vs leukemia). It is also possible that the genetic variability of CTLA-4 has a stronger impact on the outcome of allo-HSCT for thalassemia than for leukemia and lymphoma. Indeed, the immune system of thalassemia patients undergoing allo-HSCT is less compromised than that of patients undergoing transplantation for hematologic malignancies. In the latter patients, it is likely that the outcome is influenced by a multitude of genetic and environmental factors and that, in the end, each of these factors contributes to diluting the effect of the others. Although no study can be completely free of bias, we tried to reduce bias to a minimum by recruiting a homogeneous cohort of thalassemia patients, taking care to eliminate any environmental or other factors that could potentially limit the statistical power of the results. The association of CT60-AA genotype with better overall survival reported by Pérez-Garcia et al. [16] could not be evaluated in the present study. The differences between thalassemia and leukemia, as well as the factors influencing the respective HSCT outcomes, make it impossible to compare overall survival rates in these 2 patient groups.

Seen from a functional viewpoint, our data support the hypothesis that the G-allele determines protection against aGVHD by reducing the expression of sCTLA-4 [16]. The significantly increased expression of sCTLA-4 associated with the CT60-AA genotype seems to determine susceptibility to aGVHD by impeding or reducing the possibility of binding between B7 and membrane-bound CTLA-4 expressed on activated T lymphocytes [11-14]. As in autoimmune disorders, this would heighten the reactivity of effector T cells, which fail to efficiently transcribe the inhibitory signals leading to their inactivation [11].

By contrast, our data indicate that the risk factors for aGVHD induced by CTLA-4 are expressed by cells of recipient origin that persist after myeloablative conditioning treatment. Donor T lymphocytes, although being the effector component of aGVHD, do not seem to determine genetic susceptibility to aGVHD.

Our results are more in line with an alloimmune model, wherein the risk factor for aGVHD acts as a single short-term rather than long-term event capable of triggering or repressing aGVHD during the conditioning or early phases of the transplantation procedure.

Given the present state of our knowledge, it is difficult to explain the mechanisms underlying such an event. It could be that, under inflammatory conditions, residual immunocompetent recipient cells of the intestinal tract trigger the onset of aGVHD by expressing high levels of sCTLA-4. CTLA-4 is a fundamental receptor for the activity of Treg [14], which, in turn, play a pivotal role in the induction of immune tolerance during allo-HSCT [24,25].

Functional studies on animal models have provided evidence that specialized populations of regulatory T cells play an important role in the control of intestinal inflammation [26]. A ratio of expression (sCTLA-4:flCTLA-4) increased for an isoform at the expense of another could have a significant impact on Treg, enhancing or reducing their capacity to migrate, expand, and induce tolerance.

Understanding the role of genetic factors capable of influencing the outcome is fundamental to the increasingly safer application of HSCT. Genetic factors conferring protection or susceptibility to aGVHD are particularly important, considering that this complication is the main cause of morbidity and mortality following HSCT for thalassemia, but, conversely, can have a beneficial effect on the survival of patients transplanted for leukemias or lymphomas. Within this context, our data suggest that the determination of CTLA-4 genotypes in patients is a useful tool for evaluating the pretransplantation risk that, combined with the other currently known risk factors, can form the basis for appropriate tailoring of strategies aimed at preventing the onset of aGVHD.