The Role of Donor Selection for a Second Allogeneic Stem Cell Transplantation in Patients with AML Relapsing after a First Transplant; A Study on Behalf of the Acute Leukemia Working Party of EBMT

Introduction. Recurrent disease is the major cause of treatment failure after allogeneic stem cell transplantation (SCT) in patients with AML. Second SCT (SCT2) is a valid treatment option in this setting but outcome is relatively poor. Haplo-identical (haplo) SCT is increasingly used over the last decade due to the introduction of non T-depleted methods. Prior studies have shown similar outcome when using the same or different HLA-matched donor for SCT2. However, there is relatively limited data on the use of haplo-donors.

Methods and Results. The study included 556 patients with AML relapsing after a first allogeneic SCT (SCT1) given in CR1 from an HLA-matched sibling (sib, n= 294) or a matched unrelated donor (MUD, n=262) and given SCT2 during the years 2006-2016. The median age at SCT2 was 46 years (20-73). 247 patients were in CR2 (44%) and 309 had active leukemia (55%) at the time of SCT2. The conditioning regimen was myeloablative (MAC, 66%) or reduced-intensity (RIC, 34%) for SCT1, and 41% and 59%, respectively for SCT2. 19% of all patients had acute GVHD grade II-IV and 20% had chronic GVHD after SCT1 and before relapse. Patients were divided into 3 groups based on the donor selected for SCT2; 1) same donor (n=163, sib/sib-112, MUD/MUD-51), 2) different HLA-matched donor (n=305, sib/different sib-44, sib/MUD-93, MUD/ different MUD- 168), 3) haplo-donor (n=88, sib/haplo-45, MUD/haplo-43). All haploSCT were non T-depleted. There were some differences between the 3 groups in the timing of relapse and SCT2. The median time from SCT1 to relapse was similar; 10.6, 12.5 and 9.3 months, respectively (P=0.14). However, the median time from relapse to SCT2 was shorter for the same donor group; 2.8, 3.7 and 3.5 months, respectively (P<0.001) and the median time between SCT1 and SCT2 was longer for the different donor group; 14.3, 17.5 and 13.8 months, respectively (P=0.03). There were no difference between the groups in patient age, gender, disease status at SCT2 or conditioning regimen intensity for SCT1 or SCT2. The 2-year leukemia-free survival (LFS) after SCT2 was 23.5%, 23.7% and 21.8%, respectively (unadjusted P=0.30). Multivariate analysis of factors predicting relapse after SCT2 showed no effect of the second donor type, hazard ratio (HR) 0.96 (P=0.83) and 1.20 (P=0.47) for different matched donor and haplo-donor compared to the same donor, respectively. MUD donor in SCT1, CR2 compared to active disease and chronic GVHD after SCT1 were associated with reduced relapse risk after SCT2, HR 0.70 (P=0.02), 0.60 (P=0.001) and 0.66 (P=0.03), respectively. Age, gender, conditioning regimen used for SCT1 or SCT2 and time to first relapse or to SCT2 did not predict relapse rate after SCT2. The second donor type did predict for non-relapse mortality (NRM) after SCT2; HR 1.26 (P=0.41) and 2.18 (P=0.02) for different matched donor and haplo-donor compared to same donor, respectively. Advanced age and MAC in SCT1 also predicted for NRM, HR 1.40 (P<0.001) and 0.61 (P=0.04), respectively. The second donor also predicted for LFS after SCT2; HR 1.05 (P=0.77) and 1.55 (P=0.03), respectively. Advanced age and SCT2 in CR2 also predicted for LFS; HR 1.11 (P=0.06) and 0.66 (P=0.002), respectively. In all, there were no differences between same or different matched donors in SCT2 outcomes, but haploSCT2 was associated with higher NRM and lower LFS. Significant interaction was detected between second donor type and conditioning for SCT1. The inferior outcome after SCT2 with a haplo-donor was limited to patients given MAC in SCT1. In this setting it was associated with higher relapse and NRM rates and lower LFS, HR 1.86 (P=0.05), 3.40 (P=0.005) and 2.25 (P=0.001), respectively. However, there was no difference in any of these outcomes in patients given RIC in SCT1. Unadjusted analysis showed that in patients with no chronic GVHD after SCT1, haploSCT2 was associated with lower LFS, due to higher NRM. However, LFS was similar in patients with prior chronic GVHD. Multivariate analysis was not feasible due to low patient numbers.

Conclusions. Second SCT with the same donor or different matched donor is associated with similar outcomes in patients with relapsed AML after a first SCT. However, SCT2 with a haplo-donor is associated with higher NRM and lower LFS, mostly in patients given MAC in SCT1. Prior chronic GVHD after SCT1 is associated with lower relapse rate after SCT2. The role of prior chronic GVHD in donor selection should be further investigated.