Genomic loss of patient-specific HLA in acute myeloid leukemia relapse after well-matched unrelated donor HSCT

To the editor:

Allogeneic hematopoietic stem cell transplantation (HSCT) can grant long-term control and cure of acute myeloid leukemia (AML) thanks to the antitumor effect of the transplanted immune system. Still, relapse remains an open issue: in the haploidentical setting, we and others demonstrated that more than one-third of post- transplantation relapses are due to the de novo genomic loss of patient-specific HLA in leukemic blasts, which favors immune evasion from donor T cells.^1,2  Although highly relevant in the haploidentical context, HLA loss has been poorly assessed after matched unrelated donor (MUD) HSCT,³  which is 10 times more frequent in clinical practice.⁴ 

Here we report on a 36-year-old woman with normal karyotype AML, positive for the FLT3 internal tandem duplication (ITD) at diagnosis (clinical course in Figure 1A). The patient achieved complete remission after standard induction and 2 courses of consolidation chemotherapy, followed by myeloablative HSCT from a MUD (10/12 HLA matched, donor 1 in Figure 1B). Disease relapse occurred after 10 months, and was controlled for more than 2 years by salvage chemotherapy followed by serial donor lymphocyte infusions (DLIs). Thereafter, a second relapse occurred, requiring 2 cycles of chemotherapy and a second allogeneic HSCT from a different MUD (9/12 and 11/12 HLA matched with the patient and the first donor, respectively, donor 2 in Figure 1B). After 6 months from the second HSCT, a third relapse occurred (boxed in gray in Figure 1A). Interestingly, this time leukemia resulted negative for the FLT3-ITD mutation, prompting further investigation.⁵  HLA molecular typing and whole-genome single nucleotide polymorphism analysis were comparatively performed on leukemia at diagnosis and at third relapse, demonstrating de novo acquired uniparental disomy (UPD) of a large region of chromosome 6p. The genomic alteration spanned approximately 40 Mb and resulted in the loss of the HLA haplotype encompassing the C*02:02 and DPB1*04:02 mismatched alleles (Figure 1B and C), with the same mechanism described for relapses after haploidentical HSCT.^1,2  This observation suggests that a primeval leukemic clone, even though negative for FLT3-ITD (mapped to chromosome 13), may have persisted during the course of treatments and reappeared upon the immune advantage granted by genomic HLA loss. Because T cells from the current donor are ineffective against leukemic cells that have lost the mismatched HLA,¹  we proposed to our patient a third HSCT from her mother, sharing the haplotype that had been lost by UPD and therefore fully HLA-mismatched against the relapsing leukemia (donor 3 in Figure 1B). Unfortunately, the patient died in aplasia of transplant-related toxicity.

Our report demonstrates that immune escape by genomic HLA loss can occur not only in the haploidentical context, but also in leukemia relapses after well-matched unrelated donor HSCT. Further studies are warranted to address the frequency of this phenomenon; still, we would already encourage the use of HLA molecular typing in AML relapses also after MUD HSCT, to avoid the toxicity of inefficacious DLIs. Ultimately, novel diagnostic and therapeutic tools will be needed for early detection and targeted treatment of these peculiar variants of posttransplantation AML relapse.