Incidence, Risk Factors, Treatment and Outcome of Isolated Extramedullary Relapse after Allogeneic Stem Cell Transplantation for Acute Leukemia and Myelodysplastic Syndrome

Allogeneic stem cell transplantation (allo-SCT) is an effective treatment with a curative potential for acute myeloid leukemia (AML), myelodysplastic syndrome (MDS) and acute lymphoblastic leukemia (ALL). The major cause of death after allo-SCT remains disease relapse, with relapse rate of 35-45%. Isolated extramedullary relapse (iEMR) after allo-SCT is considered relatively rare and there is limited data on this type of relapse.

We performed a retrospective analysis in 685 consecutive patients with AML (n=446), MDS (n=119) and ALL (n=120) given allo-SCT from sibling or unrelated donor over a 13-year period in a single institution to study the incidence, risk factors, treatment options and outcome of patients with iEMR after allo-SCT. With a median follow-up of 57 months (range, 1-166), 271 patients are alive 158 died of treatment related causes and 292 patients relapsed. The 5-year overall and disease-free survival rates were 38.8% (95CI, 34.9-42.7) and 33.9% (95CI, 30.1-37.6) respectively. The cumulative incidence of relapse was 43.9% (95CI, 40.2-47.9). Among 292 patients who experienced relapse post-transplant, 260 had bone marrow relapse (BMR) (5 year cumulative incidence 39% (95CI, 35.4-42.9)) and 32 had iEMR (5 year cumulative incidence 4.9% (95CI, 3.5-7), 10.9% of all 1^st relapses). Seventeen patients had combined BM and EMR and were included in the final analysis with the BM group since their disease course was similar to that of patients with BMR. Sixty four patients had EMR at some point of disease course thus the 5-year cumulative incidence for any EMR was 10.1% (95CI, 7.9-12.8). iEMR occurred significantly later than BMR, 11.1 (range, 0.9-67.2) vs 3.8 (range, 0.4-101.7) months post-transplant respectively (p<0.001). Univariable analysis identified factors that conferred higher cumulative risk of iEMR: younger age (<55y) (CI=6.8% vs 2.5%, p=0.03), diagnosis of ALL vs AML (CI=12.3% vs 4.0%, p<0.001), myeloablative vs reduced intensity or reduced toxicity conditioning (CI=8.0%, 5.0% and 2.6% respectively, p=0.02) and prior extra medullary disease (EMD) (CI=20.6% vs 3.0%, p<0.001). Acute and chronic GVHD reduced the risk of BMR but did not protect against iEMR. Risk factors that remained significant with multivariable analysis were diagnosis of ALL vs MDS (HR=14.05, p=0.03), poor cytogenetics (HR=2.29, p=0.04) and prior EMD (HR=3.80, p=0.002). When excluding patients with MDS the risk for iEMR was higher in ALL than in AML patients (HR for AML=0.4, p=0.05) and the other above mentioned risk factors remained significant.

Most of the patients with iEMR received systemic treatment combined with local radiation and DLI if feasible. With a median follow-up of 32.6 (1.6-140.4) months after 1^st relapse, 37 patients are alive and 255 have died. The 1-year and 3-year OS were 21.8% (16.9-26.6) and 10.8% (6.9-14.7). The 3-year OS was 8.5% (4.8-12.2) and 29.1% (12.0-46.3) after BMR and iEMR, respectively (p=0.003). Female gender (HR=0.7, p=0.01), diagnosis of MDS vs AML or ALL (HR=0.57, p=0.01) and 1^st iEMR (HR=0.56, p=0.02) were factors that independently predicted better 3-year survival after 1^st relapse in multivariable analysis while advanced status at diagnosis (HR=1.45, p=0.01) and poor cytogenetics (HR=1.37, p=0.03) were independent predictors of inferior survival. One hundred and one patients achieved 2^nd CR, 21 of them died of treatment related causes, 25 are alive and 56 had second relapse, 30 in the BM and 26 EM. The 3-year cumulative incidence of second iEMR was high and quite similar to BMR - 29.0 vs 32.3%. Among 32 patients who relapsed initially EM 13 relapsed again, the minority in BM (3 patients) while most EM (10 patients) (CI 16.7% (5.9-46.8) and 55.6% (36.8-84.0) respectively). The 3-year OS after 2^nd relapse was 13.8% with significantly better survival after iEMR (23.8 vs 4.0% for BMR, p=0.03).

In conclusion, the incidence of iEMR of acute leukemia and MDS after SCT is 4.9%, with higher incidence in patients with ALL or prior extra medullary disease. It occurs later than BMR and more commonly in patients with chronic GVHD, suggesting less effective protection of graft versus leukemia on extra medullary sites. When treated aggressively with a combination of chemotherapy, radiation and immunotherapy prognosis is better with iEMR than with BMR. Patients with iEMR tend to relapse again extra medullary, but long-term survival is feasible in these patients.

*AN and AS equally contributed