Missing HLA C Group 1 Ligand In Patients With AML and MDS Is Associated With Reduced Risk Of Relapse After Allogeneic Stem Cell Transplantation With Fludarabine and Treosulfan Reduced Toxicity Conditioning

Allogeneic stem cell transplantation (SCT) is a curative therapy for patients with AML and MDS. Reduced toxicity conditioning with fludarabine and treosulfan (FT) has been shown to be a dose intensive regimen with enhanced anti-leukemia effect and acceptable toxicity. However, relapse after SCT remains the main obstacle to cure. Natural killer (NK) cell alloreactivity has a documented role in reducing relapse risk after haplotype mismatched SCT in AML. The role of NK alloreactivity in HLA matched SCT is more controversial. The missing ligand theory suggests that missing ligands in the recipient for donor inhibitor Killer immunoglobulin-like receptors (KIR) may drive NK alloreactivity after SCT in the absence of HLA mismatch. In this analysis we defined the prognostic factors for relapse and overall survival (OS) in patients with AML and MDS given FT conditioning. In particular we tested the role of recipient HLA ligands for NK KIRs in controlling relapse. The study included 203 patients with AML (n=129, 29 of them secondary to MDS or prior chemotherapy) and MDS (n=74) given SCT in 2 institutions. The median age was 58 years (range, 21–76), 126 male, 77 female. The donor was an HLA-matched sibling (n=97) or matched unrelated (n=106). Disease status was CR1 (n=65), CR2/later CR (n=24), no CR (n=44) or previously untreated MDS (n=70). Eighty patients (39%) had poor-risk cytogenetic or molecular markers. FT included treosulfan 12 gr/m² x3 days in one institution (FT12, n=123) or 14 gr/m² x3 in the second (FT14, n=80). 135 patients with an unrelated donor or no prior therapy were also given ATG. High-resolution HLA typing revealed that 67% expressed at least one Bw4 antigen, 81% expressed group 1 C alleles and 66% group 2 alleles. With a median follow-up of 48 months (range, 6-108 months) 86 patients are alive, 51 died of non-relapse causes (NRM) and 66 died of relapse. Seven additional patients relapsed but are currently alive. The 5-year OS and leukemia-free survival (LFS) rates were 39% (95%CI, 32-47)and 36% (95%CI, 28-43), respectively. The 5-year cumulative incidence of relapse and NRM was 38% (95%CI, 31-45) and 27% (95%CI, 31-45), respectively. The most significant predictor of relapse was the status of disease at SCT. 5-year relapse rates were 33% (95%CI, 26-42) and 55% (95%CI, 42-71) for patients in CR/untreated or refractory disease, respectively (p=0.001). Patients expressing 1-2 HLA C group 1alleles had a relapse rate of 45% (95%CI, 37-56) compared to 26% (95%CI, 13-49)in patients with missing HLA C group 1 ligand (p=0.03). Missing HLA C group 2 and/or Bw4 ligands had no effect on relapse. Multivariate analysis identified no CR at SCT (HR 3.6, p=0.001), missing HLA C group 1 ligand (HR 2.6, p=0.03), sibling donor (HR 1.8, p=0.04), poor cytogenetics (HR 1.7, p=0.05) and female donor to male recipient (HR 0.5, p=0.06) as independent factors predicting relapse. The reduced relapse rate associated with missing HLA C group 1 ligand was more pronounced in patients with MDS/secondary AML, 39% Vs 7%, respectively (p=0.02) and in patients not in CR or in untreated disease at SCT, 46% Vs. 8%, respectively (p=0.02). Missing HLA ligands were not associated with GVHD or NRM. The predicting factors for NRM were SCT beyond CR1 (HR 8.6, p=0.0002), Sibling donor (HR 0.6, p=0.08) and SCT comorbidity score > 2 (HR 1.8, p=0.08). The reduced relapse rate combined with no effect on NRM of missing HLA C group1 ligand translated into improved LFS, been 46% (95%CI, 27-65) compared to 30% (95%CI, 21-40) in patients expressing the ligand (p=0.07). Multivariate analysis identified SCT not in CR (HR 2.8, p=0.0007), SCT comorbidity score > 2 (HR 1.5, p=0.06) and missing HLA C group 1 ligand (HR 1.9, p=0.02) as independent predicting factor for LFS. In conclusion, missing HLA C group 1 ligand in SCT recipients with AML and MDS may be associated with reduced relapse risk, similar NRM and improved LFS. This effect is more pronounced in patients with MDS and secondary AML and in patients with advanced disease. The speculated mechanism is enhanced NK alloreactivity in this setting according to the missing ligand theory. These observations merit further study in larger cohorts and in patients given other conditioning regimens. They may also impact planning of immune therapies after SCT.