Interleukin-10 and tumor necrosis factor alpha region haplotypes predict transplant-related mortality after unrelated donor stem cell transplantation

Tumor necrosis factor alpha (TNFα) is a member of the TNF receptor superfamily of ligand receptors and, as a T-helper cell 1 (Th1) cytokine, is an important component of the proinflammatory immune response. Intensive chemotherapy and irradiation given prior to stem cell transplantation (SCT) is associated with increased serum levels of TNFα.^1-3  High levels of serum TNFα measured before transplantation are associated with increased probability of early transplant-related mortality (TRM) and acute graft-versus-host disease (aGVHD).^2,4 

Interleukin-10 (IL-10) is a member of an expanding family of cytokines known as the interferon receptor superfamily.⁵  It is an important anti-inflammatory cytokine in a number of immune pathways. A reduction in TRM after SCT has been associated with increased recipient levels of IL-10 immediately before and after transplantation; however, this observation remains controversial.^4,6,7 

The genes encoding TNFα (TNFSF2) and IL-10 (IL10) are both highly polymorphic. The TNFSF2 gene maps to chromosome 6 (6p21.3) and contains a number of single nucleotide polymorphisms (SNPs), predominantly in the 5′ promoter region.⁸ TNFSF2 gene is surrounded by a number of other genes with immunologic functions. A number of well-studied microsatellites have been described within genes flanking TNFSF2.⁸  TNFa, TNFb, and TNFc microsatellites are found in or near the lymphotoxin alpha gene, TNFSF1. TNFd and TNFe microsatellites are located in intron 3 of the leukocyte-specific transcript gene, LST1. The IL10 gene maps to chromosome 1 (1q31-32). The 5′ promoter region of this gene contains a number of proximal⁹  and distal¹⁰  SNPs as well as 2 microsatellites termed IL-10R and IL-10G.^11,12  In both TNFSF2 and IL10 genes investigators have sought polymorphic profiles that can be defined as haplotypes.^11,13,14 

Polymorphisms in the TNFSF2 region and IL10 genes have been associated with clinical outcome in several identical sibling stem cell transplantation (ID-SCT) studies.^12,15,16  However for unrelated donors (UD-SCT), very few studies have been undertaken.¹⁷  Therefore, we investigated the impact of TNFSF2 and IL10 polymorphisms on TRM after UD-SCT by analyzing individual polymorphisms and statistically inferred haplotypes. We have identified polymorphic haplotypes in the TNFSF2 region and in IL10 that may facilitate the selection of patients and donors for UD-SCT by identifying individuals at higher risk of TRM.

One hundred eighty-two patients undergoing UD-SCT at the University of Minnesota Hospital between 1990 and 2000 were studied. Forty-six percent of patients were younger than 18 years of age and 32% received transplants for nonmalignant disease. Sixty-nine percent of those with hematologic malignancy had early disease, acute leukemia in first or second complete remission, or chronic myelogenous leukemia (CML) in chronic phase at the time of transplantation. All donors were older than 18 years of age and the median donor age was 37 years. Sixty-five percent of transplantations were performed after 1995 and the median follow-up on survivors was 4.8 years. Eighty-nine percent of patients received total body irradiation-based pretransplantation protocols and 66% received T-cell-replete marrow transplants and received cyclosporine and methotrexate after transplantation. In 34% of cases T cells were removed by elutriation. Pretransplantation serologic typing indicated all donor-recipient pairs were matched at human leukocyte antigen A (HLA-A) and B loci. HLA-C locus typing was not performed. Sequence-specific oligonucleotide probes were used to genotype for HLA-DRB1, which indicated that 7% of pairs were HLA-DRB1 mismatched. Approval was obtained from the University of Minnesota institutional review board, and informed consent was provided by all individuals according to the Declaration of Helsinki.

Genotyping was performed on stored DNA or crude cell lysate. The quantity of DNA did not allow all pairs to be analyzed at all loci. Cytokine SNPs and HLA class I alleles were genotyped using induced heteroduplex generator (IHG) technology.¹⁸  Analysis of the TNFSF2 region and IL10 microsatellites was performed using standard polymerase chain reaction (PCR) methodology.

For IL10 haplotype analysis, 2 previously described methods were used to produce a physically linked 4-locus haplotype (10G, -1082, -819, -592).^12,19  A PHASE-based statistical approach was used to generate haplotypes including the IL-10R microsatellite.²⁰  For the TNFSF2 region, the PHASE-based approach was used alone because of large distances between individual polymorphic loci. Cumulative incidence rates and their 95% confidence intervals (CIs) were used to estimate treatment-related mortality, treating relapse-related deaths as a competing risk.²¹  Univariate comparisons of the major end points were completed by using the log-rank statistic. In preparation for inclusion into the multiple regression models, the proportional hazards assumption was tested by the Cox regression model using the Wald chi-square test and a time-dependent covariate of the logarithm of time. Cox proportional hazard models were then used to evaluate the independent effect of cytokine polymorphism.²²  In addition to all significant TNFSF2 region and IL10 genotype factors, other potential confounders were included in the Cox regression models. These factors included recipient age, donor age, sex, sex mismatch, diagnosis, disease status, T-cell depletion, pretransplantation conditioning, and HLA disparity.

Relevant allele frequencies for both recipients and donors are summarized in Table 1. No significant differences were found between allele frequencies for TNFSF2 region and IL10 in the recipient and donor cohorts. In contrast to ID-SCT studies there was a considerable level of mismatch of TNFSF2 region genotypes between recipients and donors. At the TNFSF2 -238 locus, 8.8% of pairs were mismatched for one or more alleles and at the TNFSF2 -308 locus 13.3% of pairs were mismatched. As the TNFSF2 region is in linkage disequilibrium with flanking HLA genes, this level of mismatch is consistent with the levels of HLA allele disparity found using the IHG-based HLA typing technique.¹⁸ 

The association between donor and recipient TNFSF2 region and IL10 genotypes and the cumulative incidence of TRM at one year was investigated (Table 1). The presence of the TNFd4 allele in conjunction with the TNFSF2 -1031C allele in recipients or donors significantly increased the cumulative incidence of TRM (P < .01; Figure 1A). Cox proportional hazards analysis showed that the TNFd4/TNFSF2 -1031 haplotype increased the risk of TRM at one year independent of all other factors. Relative risk (RR) = 2.2, CI 1.2-4.3, P = .02; and TNFa5 RR = 2.8, CI 1.2-6.2, P = .01. The other TNFSF2 region genotypes investigated were not associated with TRM.

Individual IL10 polymorphic loci were not associated with TRM. However haplotype analysis revealed significant associations. The cumulative incidence of TRM at one year was higher in the presence of R2-G-C-C haplotype¹⁰  (P = .01; Figure 1B), whereas the R3-G-C-C haplotype was shown to be protective (P = .01; Figure 1C).

Cox proportional hazards analysis supported the association of the R3-G-C-C haplotype with TRM (RR = 0.4, CI 0.2-0.8, P = .03), although results could not be verified for the R2-G-C-C haplotype as hazards were not proportional. Cavet et al¹²  reported that ID-SC transplant recipients possessing longer 10G alleles were more likely to develop grades II to IV aGVHD. We have shown R2-G-C-C, a haplotype with predominantly long 10G alleles, is a risk factor for TRM after UD-SCT. The patients included in this study were a heterogeneous group. Furthermore, it must be emphasized that these analyses involved multiple comparisons and therefore P values between .05 and .01 should be interpreted with caution.

This is the first study to describe an association of cytokine polymorphism with TRM after UD-SCT, although no independent associations with aGVHD were identified. The most important finding is that recipient or donor TNFd4/TNFSF2 -1031C haplotype is a risk factor for TRM (Table 1). The same TNFd4 allele has recently been implicated as a risk factor for moderate to severe aGVHD in ID-SCT.¹⁶  It is likely that the TNFd4 allele is part of an extended chromosome 6 haplotype comprising immune HLA and a complement gene. We have shown significant associations between polymorphisms of TNFSF2 region and IL10 with TRM after UD-SCT. Confirmation of our results in a multicenter context would rule out any possibility of a center effect being responsible for our findings. Cytokine gene polymorphism analysis could become a useful pretransplantation tool to identify high-risk donors and recipients.