Fifty-eight adult patients with relapsed AML treated at the Division of Hematology, Severance Hospital, Yonsei University College of Medicine, between January 2005 and December 2010 were included in this study. Written informed consent was obtained from each patient. Diagnosis of AML was established on the basis of bone marrow (BM) aspiration findings. AML was classified according to the morphological French-American-British (FAB) criteria, and the diagnosis was confirmed on the basis of cytochemical and immunological analyses findings [6]. At diagnosis and relapse, samples of BM or blood were examined for cytogenetic abnormalities using standard banding techniques and were classified according to the International System for Human Cytogenetic Nomenclature [7]. Cytogenetic analysis was successfully performed for 54 of 58 patients (93%) at diagnosis and for 48 of 58 patients (82%) at relapse. Karyotypes were classified as favorable (t(8;21), chromosome 16 abnormalities), adverse (complex karyotype, -7, 3q abnormality, del(5q), -5, 11q23), or intermediate (no abnormality, +8, +21, del(7q), del(9q), +22, with other numerical and structural abnormalities) according to the MRC AML 10 study [8]. Induction therapy with idarubicin (IDA) (12 mg/m²/d intravenous [IV] over 30 min on days 1.3) in combination with cytosine arabinoside (Ara-C) (100 mg/m²/d IV continuous infusion on days 1.7) was administered to all patients, except for 1, who received mitoxantrone, etoposide, and cytarabine (MEC)-based chemotherapy (mitoxantrone, 12 mg/m²/day on days 1.4; etoposide, 100 mg/m²/day on days 1.3; and Ara-C, 1,000 mg/m²/12 hr on days 1.3). As salvage therapy, 40 patients received MEC-based chemotherapy with or without gemtuzumab ozogamicin (GO) (90 mg/m² on day 4), 13 patients received other intensive combination chemotherapy (Ara-C, topotecan, and IDA, N=5; fludarabine, IDA, and Ara-C: N=4; IDA and Ara-C: N=2; and GO: N=1), and the remaining 5 patients received allogeneic hematopoietic stem cell transplantation (HCT). To determine the impact of salvage therapy on relapsed AML, patients were classified into prognosis subsets according to the new European Leukemic Net (ELN) recommendations, recently proposed by an expert panel [1].

CR was defined as normal cellular BM with normal erythroid and myeloid elements and less than 5% myeloblasts and normalization of peripheral blood counts (neutrophils >1×10⁹/L and platelets >100×10⁹/L). CR with incomplete platelet recovery (CRp) was defined according to the same criteria for CR but with additional platelet transfusion independence and platelet counts of <100×10⁹/L. CR and CRp cases were considered in the computation of the overall response (OR). All other types of responses were considered as treatment failure. Relapse was defined as the detection of at least 5% leukemic blasts in BM aspirates or new extramedullary (EM) leukemia in patients with previously documented CR [9].

Descriptive statistics are reported as frequencies or medians and ranges. OR, overall survival (OS), and event-free survival (EFS) were the primary endpoints. The χ² test was used to test for factors associated with OR. Survival was calculated from the date of the documentation of relapse until death or last follow-up for OS or until failure, relapse, death, or last follow-up for EFS. OS and EFS were estimated using the Kaplan-Meier product limit method. The following factors were analyzed for their association with OR, OS, and EFS: age at relapse, type of AML (de novo vs. secondary), EM involvement at diagnosis, cytogenetic findings at diagnosis and relapse, number of induction cycles of chemotherapy required to achieve the first CR, previous HCT, relapse-free interval (RFI), type of salvage regimen, and karyotypic changes between diagnosis and relapse. The relationships between patient and disease characteristics and clinical outcomes were evaluated using multivariate analysis. OR was analyzed using multivariate logistic regression analysis, and OS and EFS were analyzed using Cox's proportional hazard model. P values were two-tailed and considered statistically significant if <0.05. All factors associated with a P value of <0.20 in univariate analysis were entered into multivariate analyses. The final analysis was performed using SPSS Inc., 12.0 (Chicago, IL, USA).

Karyotype records at both diagnosis and relapse were available for 45 patients (77%). A change in karyotype was observed in 17 out of 45 cases (37%, group 1) and no change was noted in 28 out of 45 cases (62%, group 2). Karyotypic changes between diagnosis and relapse in group 1 are shown in Table 1. An intermediate karyotype at diagnosis changed to a complex karyotype at relapse in 8 cases (47%). Favorable karyotypes (2 cases of t(8;21) and 1 of inv(16)) at diagnosis changed to normal karyotypes at relapse in 2 cases and were the same aberrant karyotypes as at initial diagnosis but with additional aberrations at relapse in 1 case. Adverse karyotypes at diagnosis changed to complex karyotypes in 3 cases, and a translocation involving 11q23 changed to a normal karyotype in 1 case.

We enrolled 31 male and 27 female patients, with a median age at diagnosis of 49 years (range, 17.76 years) and a median follow-up duration from relapse of 14.3 months (range, 0.2-149.0 months). Patient characteristics of groups 1 and 2 are summarized in Table 2. In group 1, all patients had de novo AML, and AML was significantly more prevalent in group 1 than in group 2 (71.4%, P=0.016). The frequencies of cytogenetic changes at relapse were significantly different between the 2 groups (P<0.001), although both groups were comparable in terms of the karyotype classification, or the cytogenetic risk, at diagnosis. In group 1 at relapse, 1 patient (5.9%) had a favorable karyotype, 5 patients (29.4%) had an intermediate karyotype, and 11 patients (64.7%) had an unfavorable karyotype. In group 2, 3 patients (10.7%) had a favorable karyotype, 22 patients (78.5%) had an intermediate karyotype, and 3 patients (10.8%) had an unfavorable karyotype. In group 2, 23 (82%) patients received HCT before relapse, compared to only 5 patients (52.9%) in group 1 (P=0.036). Cut-off points for age and RFI were chosen on the basis of the mean values in all patients. The 2 groups did not show statistical differences in age (<49 vs. ≥49 years), sex, FAB classification, white blood cell count at diagnosis (<100 vs. ≥100×10⁹/L), EM involvement at diagnosis, type of salvage therapy, number of induction cycles required to reach first CR and relapse, or RFI (<7.5 vs. ≥7.5 months).

Clinical, biological, and treatment-related data were analyzed to establish whether they had prognostic value with respect to OR, OS, and EFS. The OR (CR and mCR) rate was significantly lower in group 1 than in group 2 (23% vs. 38%; P=0.016). Univariate analysis showed that the cytogenetic risk at relapse (P=0.030) and the type of salvage therapy (P=0.039) were associated with OR (Table 3).

Fig. 1 shows Kaplan-Meier curves for OS and EFS. Patients in group 1 had a shorter median OS (14.2 vs. 21.1 months, P=0.004) and shorter median EFS (4.1 vs. 6.0 months, P=0.010) than those in group 2. Age at relapse (P=0.033) and type of salvage therapy (P=0.044) were significantly associated with OS, and the type of AML (de novo vs. secondary) (P=0.025) and cytogenetic risk group at relapse (P=0.028) were associated with EFS (Table 3). We also applied multiple multivariate Cox regression analyses to identify independent prognostic factors for OR, OS, and EFS. Karyotypic change between diagnosis and relapse was the only independent predictor of OS (P=0.023; response rate [RR]=2.655) and EFS (P=0.033; RR=0.033) that retained significance (Table 4).