Incidence, Risk Factors and Clinical Outcome Of Leukemia Relapses Due To Loss Of The Mismatched HLA Haplotype After Partially-Incompatible Hematopoietic Stem Cell Transplantation

Allogeneic Hematopoietic Stem Cell Transplantation (HSCT) represents the best curative option for many patients with high-risk myeloid malignancies, mainly due to its potent immuno-mediated antileukemic effect. Still, post-transplantation relapse remains an unsolved issue. We and others described genomic loss of the mismatched HLA haplotype as a mechanism by which leukemic cells evade donor T cell-mediated immune pressure and cause clinical relapse after partially HLA-incompatible HSCT (Vago et al, N Engl JMed, 2009), but the actual incidence and risk factors of this phenomenon are to date largely unknown.

We analyzed retrospectively 224 consecutive partially HLA-mismatched HSCTs performed in our Institute in the last ten years (Unrelated Donor, UD: 60; Mismatched Related Donor, MMRD: 164) in patients affected by myeloid malignancies (Acute Myeloid Leukemia, AML: 173; Myelodisplastic Syndrome, MDS: 27, Myeloproliferative Neoplasms: 17; others: 7). All patients received myeloablative conditioning and infusion of donor T cells, either as part of the graft or as an add-back. Patients’ follow-up included bone marrow genomic HLA typing to identify HLA loss relapses. In selected cases of HLA loss relapse cryopreserved serial serum samples harvested after HSCT were analyzed for the eventual presence of anti HLA Class I or Class II antibodies.

We documented 77 cases of relapse: 66 after MMRD and 11 after UD HSCT. Out of 77 relapses 21 (27%) were due to genomic loss of the mismatched HLA in leukemic cells. HLA loss occurred in 19 patients with AML, one with MDS and one with myelofibrosis. All the 21 cases of HLA loss occurred after MMRD HSCT (32%), so the analysis for putative risk factors were limited to this subgroup of transplants (n=164), comparing the frequencies of putative risk factors between patients with HLA loss and “classical” relapses (n=21 and 45, respectively). HLA loss relapses occurred significantly later than their classical counterparts (median time to relapse 307 vs 86 days, p<0.0001) in this very high-risk population, suggesting that outgrowth of the mutant variants require a considerable lapse of time. None of the disease-related factors we addressed (amongst which disease subtype, cytogenetics, molecular profile and disease status at HSCT) correlated significantly with eventual HLA loss. Use of an unmanipulated T cell-repleted graft resulted to be a risk factor for HLA loss relapses (Chi2=6.36; p=0.01). Both acute (HR:4.67, CI 95%: 1.53-14.22; p=0.007) and chronic (HR: 1.71; CI 95%: 0.68-4.28; p=0.01) Graft-versus-Host Disease (GvHD) occurred more frequently in patients with HLA loss relapses. Intriguingly, HLA-C*04 was more frequent in the mismatched haplotype of patients with HLA loss as compared to those with classical relapse (Chi2= 8.07; p=0.04), possibly suggesting an higher immunogenicity of the allele, hinted also by a similarly higher frequency in patients who did not relapse (Chi2=2.77; p=0.096). In our series, predicted NK alloreactivity had no apparent impact on eventual HLA loss. In none of the five patients studied to date we could evidence circulating anti-HLA antibodies, suggesting that humoral immunity does not play a major role in this phenomenon. Since lymphocyte infusions from the original donor are expected to be inefficacious to treat HLA loss relapses, whenever fit these patients were candidate to re-transplantation from alternative donors, HLA-mismatched and putatively alloreactive against the relapsed leukemia. Still, outcome was poor, with 5 of 8 re-transplanted patients dying of transplant-related mortality and only one alive in complete remission at a follow-up of 18 months.

Genomic loss of the mismatched HLA haplotype is an extremely frequent mechanism of leukemia immune evasion and relapse after MMRD HSCT. It appears to be prompted by selective immune pressure mediated by donor-derived T cells, and accordingly occurs more frequently upon T cell-repleted transplants and in the presence of acute and chronic GvHD, clinical hallmarks of T cell alloreactivity. Conversely the role of NK and B cells in HLA loss needs further investigation, but appears to date less pronounced. Given the poor outcome of re-transplantation, mainly due to toxicity, novel diagnostic and therapeutic approaches are needed to anticipate the detection and improve the treatment of these frequent variants of leukemia relapse.

Bordignon:MolMed SpA: Employment. Bonini:MolMed SpA: Consultancy.