Extended TNF Genotyping Identifies TNF -857T as a Risk Factor for Acute Graft-Versus-Host Disease Following Allogeneic Hematopoietic Stem Cell Transplantation.

Extensive evidence supports a role for non-HLA immunogenetic polymorphisms in determining risk and severity of acute GVHD following allogeneic hematopoietic stem cell transplantation (HCT). Tumor necrosis factor (TNF) polymorphisms have been frequently associated with GVHD risk. However, TNF is highly polymorphic and the TNF polymorphism(s) most strongly associated with GVHD remain to be determined. We previously studied three TNF single nucleotide polymorphisms (SNPs; −308, −238 and 488) in a retrospective study of 160 myeloablative HLA-matched related allo-HCTs and found a highly significant association between the intronic TNF 488A allele and risk and severity of acute GVHD and early mortality (Transplantation 2004;27:587). It was unclear if this association was primarily with TNF 488A or secondary to other functionally important TNF alleles in linkage disequilibrium. To clarify the importance of TNF variants in GVHD, we performed extended genotyping of TNF promoter SNPs in the previously described cohort, and in a prospective cohort of myeloablative (n=78) and non-myeloablative (n=56) sibling HLA-matched allo-HCT. Six TNF polymorphisms (−1030, −863, −857, −308, −238, 488) were genotyped in donors and recipients by PCR-SSP. TNF −308/−238/488 haplotypes were directly amplified by PCR, and full TNF haplotypes inferred using Haploview. The incidence of acute GVHD was 53% and 45.8% for myeloablative and non-myeloablative transplants, respectively. Nine TNF haplotypes were identified, and the TNF 488A allele was in strong, but not complete linkage disequilibrium with TNF −857T. TNF −857T was more strongly associated with acute GVHD (OR 11.1) and grade II–IV GVHD (OR 12.3) than 488A (acute GVHD OR 8.6 and grade II–IV GVHD OR 8.9) in the first cohort. These alleles were also associated with acute GVHD in the second cohort, but only in recipients receiving myeloablative conditioning (all acute GVHD, TNF 488A OR 2.9, −857T OR 3.4). Non-significant trends were observed with grade II–IV GVHD, possibly reflecting the reduced frequency of this complication in the later cohort (29% v. 48%). No association or trends were seen in non-myeloablative transplants. Moreover, no associations were identified with the extensively studied −308 or −238 SNPs in either cohort. Together, these data confirm the association between TNF polymorphisms and GVHD in myeloablative transplants, and suggest that −857T, which lies in closer proximity to TNF regulatory regions than 488A, may be the most useful marker for risk prediction. Our results also suggest that the immunogenetic determinants of GVHD vary according to the intensity of conditioning. Further work examining non-myeloablative transplants is required.