Refined Graft-Versus-Host Disease-Free, Relapse-Free Survival after Allogeneic Hematopoietic Stem Cell Transplantation in Patients with Intermediate and Unfavorable Prognosis Acute Myeloid Leukemia Ttransplanted in First Complete Remission from HLA-Identical Related or Unrelated Donors: A Retrospective Study on Behalf of the ALWP of the EBMT

The refined endpoint of graft-versus-host disease (GVHD)-free, relapse-free survival (GRFS) represents an important measure of outcomes of allogeneic hematopoietic stem cell transplant (HSCT) including both clinical results and quality of life without ongoing morbidity.

We analyzed the refined GRFS in 5059 patients with de novo acute myeloid leukemia (AML) with intermediate and unfavorable cytogenetics in first complete remission (CR1), undergoing HSCT from a matched sibling (MSD, n=3731) or unrelated donor (UD, n=1328). HSCT were performed between 2000 and 2015 and reported to EBMT. Median age for the whole population was 49 years (range 18-76)

There were statistical differences between the 2 groups: compared to MSD, UD recipients were younger (60 vs 32 years, p<0.01), had more often a male donor (54% vs 69%, p<0.01), had a longer interval from diagnosis to HSCT (140 vs 168 days, p<0.01) and were more often transplanted from a CMV negative donor (62% vs 41%, p<0.001). UD recipients received more often reduced intensity conditioning regimen (59% vs 47%, p<0.01), in vivo T-cell depletion (TCD) (72% vs 26%, p<0.01) and peripheral blood stem cells (PBSC) (79% vs 74%, p<0.01).

Median follow up was longer for MSD recipients (60 vs 32 months, p<0.01)

In univariate analysis, cumulative incidence (CI) of relapse (RI) was 25% in both groups (p=0.89); UD recipients experienced a higher CI of grade II-IV acute GVHD (aGVHD, 26% vs 23%, p<0.01), of extensive chronic GVHD (cGVHD, 23% vs 19%, p<0.01), non-relapse mortality (NRM, 12% vs 15%, p<0.01) and had a lower probability of leukemia-free survival (LFS, 60% vs 63%, p<0.01) and of overall survival (OS, 65% vs 69%, p<0.01).

Probability of GRFS at 4 years was not different, being 44% in both groups (p=0.9).

Regardless the donor type, probability of 4-year GRFS was higher in patients receiving in-vivo TCD (48% vs 41%, p<0.01), with the use of myeloablative conditioning (45% vs 40%, p<0.01), in patients aged less than median age (47% vs 40%, p<0.01), in patients with intermediate risk cytogenetics (47% vs 33%, p<0.01), when interval from diagnosis to HSCT was higher than 5 months (45% vs 41%, p<0.03), when bone marrow was used as stem cell source (49% vs 41%, p<0.01), in female recipients (46% vs 41%, p<0.01).

In a multivariate analysis adjusted for the differences between the 2 groups, UD was associated with lower GRFS (HR 1.19, CI 1.07-1.31, p<0.01), higher risk of grade II-IV aGVHD (HR 1.79, CI 1.53-2.09, p<0.01) and extensive cGVHD (HR 1.42, CI 1.19-1.69, p<0.01), higher NRM (HR 1.64, CI 1.34-1.99, p<0.01) and OS (HR 1.13, CI 1.00-1.28, p<0.01).

The other factors independently associated with GRFS were unfavorable cytogenetics (HR 1.42, p<0.01), time from diagnosis to HSCT (HR 0.96, p<0.01), male recipients from female donors (HR 1.23, p<0.01), PBSC as stem cell source (HR 1.22, p<0.01), use of in-vivo TCD (HR 0.73, p<0.01).

Unfavorable cytogenetics was associated with lower LFS (HR 1.64, CI 1.50-1.81, p<0.01) and OS (HR 1.68, CI 1.53-1.86, p<0.01) and higher RI (HR 1.94, CI 1.73-2.17, p<0.01).

When performing a subgroup analysis according to the use (n=1955) or not (n=3104) of in-vivo TCD, GRFS was not different in the two groups (47% vs 48%, p=0.54). GRFS was lower in UD recipients not receiving in-vivo TCD (37% vs 41%, p<0.02).

In multivariate analysis, a lower GRFS was observed in UD versus MSD when in-vivo TCD was not used (HR 1.23, CI 1.05-1.43, p<0.01), while no significant differences were found for those receiving in-vivo TCD. Unfavorable cytogenetics was associated to a worse GRFS in both groups.

In conclusion, UD transplant is associated with lower GRFS, probably due to higher incidence of GVHD and NRM. Importantly, in-vivo TCD might improve outcomes of UD recipients, leading to a GRFS comparable to MSD recipients. Moreover, in AML in CR1, unfavorable cytogenetics remains a strong predictor of worse outcomes either in MSD and UD recipients as compared to intermediate cytogenetics.