Clinical characteristics and outcome of refractory/relapsed myeloid leukemia in children with Down syndrome

Down syndrome (DS) is one of the most common congenital disorders and is associated with an increased risk of acute leukemia.¹  Acute myeloid leukemia (AML) in patients with DS is categorized as myeloid leukemia associated with DS (ML-DS) in the 4th edition of the World Health Organization classification. Clinical and biologic features of ML-DS in children are quite different from those of AML in children without DS and include: younger age at onset, lower white blood cell (WBC) count at diagnosis, and greater incidence of acute megakaryoblastic leukemia.^2,3  ML-DS is known to exhibit good sensitivity against cytotoxic agents, especially cytarabine (Ara-C), and outcomes in recent clinical trials are favorable: long-term event-free survival has been reported in approximately 80% of patients.^4-11  However, little attention has focused on refractory or relapsed (R/R) cases because most treatment failures are the result of toxicities rather than to resistant or recurrent leukemia. We present is a nationwide retrospective analysis of patients with R/R ML-DS in Japan.

The present retrospective study was conducted on 120 institutions that belong to the Japanese Pediatric Leukemia/Lymphoma Study Group (JPLSG), a Japanese nationwide collaborative study group for childhood hematologic malignancies, and data on 29 patients with R/R ML-DS treated in 26 hospitals were collected. Patients were either enrolled on one of the AML clinical trials or registered on patient database of the collaborative study group at initial diagnosis of ML-DS, of which the protocols and registrations were approved by the institutional review boards of each participating center with informed consent obtained in accordance with the Declaration of Helsinki. The present retrospective study was approved by the JPLSG steering committee and institutional review boards of Shiga University of Medical Science for all aspects of this investigation.

Inclusion criteria on current analyses were ML-DS patients initially treated with curative intent between January 1, 2000, and December 31, 2010, and age younger than 18 years at the onset of ML-DS. Patients with myelodysplastic syndrome with DS also were included because it is currently recognized that there are no biologic and clinical differences between myelodysplastic syndrome and overt AML in patients with DS. Clinical data at initial diagnosis of ML-DS, including sex, age, WBC count, extramedullary disease, French-American-British (FAB) morphology, therapy protocol given, and duration from initial diagnosis to relapse, were collected. In addition, treatment data, including achievement of further remission, HSCT, secondary cancer, outcome, and cause of death after diagnosis of R/R ML-DS, also were collected.

Descriptive statistical analyses to assess baseline characteristics and the clinical course of patients diagnosed with R/R ML-DS were performed by use of the χ² tests for categorical variables and Wilcoxon rank-sum tests for continuous variables. Overall survival (OS) was defined as the length of time from the diagnosis of R/R ML-DS to death from any cause. OS percentages and standard errors were calculated with the Kaplan-Meier method, and log-rank tests were used for group comparisons. A Cox proportional hazards regression model was used to investigate risk factors that were associated with survival after diagnosis of R/R ML-DS. Variables including sex (female vs male), age at initial diagnosis (≤ 2 years vs > 2 years), WBC at initial diagnosis (≥ 10 000/μL vs < 10 000/μL), FAB morphology at initial diagnosis (M7 vs others), disease status of R/R ML-DS (induction failure or relapse ≤ 6 months vs relapse > 6 months), achievement of further remission (yes vs no), and treatment of HSCT (yes vs no) that were significantly associated with survival in the univariate analyses were considered for inclusion in the model. Significant variables associated with survival were then identified. No statistical adjustment was made for performing multiple tests, but 2-sided P values greater than .05 were interpreted with caution. All data analyses were performed by the use of SAS Version 9.1.3 statistical software (SAS Institute). Follow-up data were actualized as of December 31, 2011.

Relevant clinical data of the 29 patients at initial diagnosis are shown in Table 1; a slight predominance of male patients existed (male/female ratio was 18/11), median age at initial diagnosis for ML-DS was 2 years (range, 7 months to 16 years) at which 23 of 29 patients were younger than 4 years of age, median WBC count was 5600/μL (range, 900-143 600/μL), and only 1 patient had an extramedullary disease (at skin). Morphologically, 22 (75.8%) patients showed FAB M7 blasts. Karyotype analysis at initial diagnosis showed monosomy 7 in 2 patients, monosomy 7 associated with a ring or marker chromosome in 5 patients,¹²  t(8;21)(q22;q22) with FAB M2 morphology in 1 patient, other various cytogenetic abnormalities in 13 patients, and 6 patients with normal karyotype and sole constitutional trisomy 21. Full karyotypes are listed in supplemental Table 1 (available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

All patients initially were treated with one of the protocols specifically designed for ML-DS. Twenty patients were treated according to the AML99 Down protocol of the Japanese Childhood AML Cooperative Study⁶ ; 7 patients were officially enrolled in the study, and the rest were treated according to the institutional choice. Among the remaining 9 patients, 8 patients were treated with the JPLSG AML-D05 protocol (registered at http://www.umin.ac.jp/ctr/ as UMIN000000989) and 1 patient with the Japanese Children's Cancer and Leukemia Study Group AML9805 Down protocol.⁷  The AML99 and AML-D05 protocol, with which 28 of 29 patients were treated, consists of 5 courses of pirarubicin (25 mg/m², 1-hour intravenous infusion on days 1 and 2), intermediate-dose Ara-C (100 mg/m², 1-hour intravenous infusion on days 1-7), with or without etoposide (150 mg/m², 2-hour intravenous infusion on days 3-5). Because of few incidences of CNS leukemia and CNS relapse among patients with ML-DS, cerebrospinal fluid was not routinely examined, and no CNS prophylaxis was delivered throughout the treatment on these patients. None of the patients received HSCT before the diagnosis of R/R ML-DS.

Relevant clinical data of the patients at induction failure or at first relapse are shown in Table 2. There were 3 induction failures and 26 relapsed cases. Among the 26 relapsed cases, duration from initial diagnosis to relapse was 2.4 -71.8 months (median, 8.6 months). All patients who relapsed within 6 months (n = 8) were on chemotherapy for ML-DS. Twenty-four patients (92%) relapsed within 2 years after the initial chemotherapy for ML-DS. Twenty-five patients relapsed in the BM, and 1 relapsed with an isolated extramedullary mass in a skin lesion. The WBC count at relapse was between 1700 and 25 700/μL (median, 4100/μL). Twenty-three (88.5%) showed FAB M7 morphology. Ten patients had chromosomal abnormalities that were the same as at the initial diagnosis, and 12 patients had additional abnormalities.

Nine patients were examined for GATA1 mutation of the leukemic blasts either at initial diagnosis of ML-DS or at relapse; 8 of them were confirmed to have the mutation (Table 3).

The clinical data and outcome of all the 29 patients in this study are described in Table 3. Twenty-six of the 29 patients received various salvage chemotherapies with curative intent. Six patients were treated with an ML-DS–oriented induction regimen as previously reported^6,7 ; 12 patients were treated with etoposide, mitoxantrone, and intermediate dose of Ara-C with continuous intravenous infusion¹³ ; and the other 8 patients were treated with various chemotherapy regimen, such as FLAG (fludarabine, high-dose Ara-C, and G-CSF), AVC (pirarubicin, vincristine, and Ara-C with continuous intravenous infusion), low-dose Ara-C + etoposide, and vincristine + asparaginase. No deaths because of toxicities were observed during reinduction therapy. Two patients only received palliative therapy and eventually died of disease progression (Table 3, no. 19 and 28). The details of the postrelapse clinical course could not be identified in one patient (no. 26), and this patient died of unknown cause after undergoing HSCT.

Among the 26 patients who were treated with curative intent, 13 patients (50%) achieved complete remission (CR). Eight of the 13 patients who achieved CR subsequently received allogeneic HSCT, and 2 survived without leukemia. The remaining 5 patients who achieved subsequent CR were treated with chemotherapy alone, and 4 were alive without any evidence of leukemia (no. 15, 16, 17, and 23). The 3-year OS rate of the patients who achieved CR was 57.7% ± 14.7%. All 13 patients who did not achieve CR eventually died because of disease progression. Six of those patients received allogeneic HSCT without attaining CR: 4 died because of disease progression, and 2 died because of transplantation-related toxicities. No secondary cancer was observed.

Preconditioning regimen varied among the 14 patients who received HSCT. We therefore categorized the conditioning regimen into 4 groups¹⁴ : busulfan (BU)–based myeloablative conditioning (MAC) regimen (BU-MAC), when > 8 mg/kg of BU combination was used; BU-based reduced intensity conditioning (RIC) regimen (BU-RIC), when a lower dose of BU combination was used; total body irradiation (TBI)–based MAC (TBI-MAC), when ≥ 8 Gy of fractionated TBI combination was used; and TBI-based RIC (TBI-RIC), when a lower dose of the TBI combination was used. A total of 5 patients received BU-MAC, 3 received BU-RIC, 4 received TBI-MAC, and 2 received TBI-RIC. We did not find any difference in survival and toxic death among these 4 conditioning regimen (data not shown).

Finally, the 3-year OS rate of all patients was 25.9% ± 8.5% (Figure 1). The median follow-up period for all 29 patients was 0.9 years (range, 0.2-7.0 years). Among the 8 patients with GATA1 mutation, 7 patients, including 1 patient with palliative therapy (no. 19), did not achieve subsequent remission and died of disease progression. In contrast, 11 of the 20 patients whose GATA1 status was not examined attained subsequent remission, and 6 of 11 patients are alive without disease. The patient with wild-type GATA1 (no. 10) died of transplantation-related toxicity although receiving HSCT in second CR.

Several predictive factors for OS were evaluated by univariate and multivariate analyses (Table 4). The unadjusted 3-year OS was found to be better for patients with a longer duration from initial diagnosis to relapse (patients with the duration from diagnosis to relapse > 12 months, 66.7% ± 27.2%; ≤ 12 to > 6 months, 23.1% ± 11.7%; and ≤ 6 months, 0%). In Cox regression analysis, the adjusted hazard ratios of patients who relapsed more than 6 months after the initial diagnosis were significantly better compared with those who did not achieve CR or who relapsed within 6 months after the diagnosis (hazard ratio 3.14; 95% confidence interval 1.28-7.67; P = .012), and this finding was still detected after we controlled for reinduction regimens (etoposide, mitoxantrone, and intermediate dose of Ara-C with continuous intravenous infusion vs others), HSCT (yes vs no), and biologic factors, including chromosomal abnormalities and GATA1 status (hazard ratio 4.56; 95% confidence interval 1.07-19.41; P = .04). Two clinical factors were found to be associated with the duration from initial diagnosis to relapse. First, patients older than 2 years of age at initial diagnosis were more likely to relapse in a short period from initial diagnosis (≤ 6 months) compared with younger patients (trend test, P = .02). Second, patients who relapsed earlier were less likely to achieve CR when they received second-line chemotherapy (P = .001). Other factors, including sex, WBC at initial diagnosis, FAB morphologies (M7 vs others), chromosomal abnormalities, and GATA1 status, were not significantly associated with survival. HSCT did not influence the prognosis even if performed after achieving further remission.

Treatment strategy for ML-DS is based on reducing the intensity of chemotherapy protocol designed for non-DS AML patients, considering both the potential risk of treatment-related toxicities and greater sensitivity to cytotoxic agents. With this strategy, there have been successful reports on treating patients with ML-DS from several collaborative groups.^4-11  However, it is assumed that salvage of patients with R/R ML-DS is quite difficult because the OS and event-free survival rate are almost the same in these reports. Because little attention has been focused on these cases so far, a nationwide retrospective study was conducted. The present study clearly showed that patients with R/R ML-DS are resistant to second-line chemotherapy and that the disease course rarely is salvaged by allogeneic HSCT.

It is well recognized that the outcome of ML-DS is much better than that of AML in non-DS patients; however, the OS rate of R/R ML-DS cases in the present study (26%) was no better than that of the reported survival rate of non-DS AML patients, which is 23%-33%.^15,16  Usually, non-DS AML patients are heavily pretreated before relapse (eg, HSCT), and this explains one aspect of therapeutic resistance to second-line treatment that could lead to the low salvage rate in these patients. It is notable that the salvage rate of patients with ML-DS in the R/R setting also was very poor, even though these patients had initially received the low-intensive ML-DS–oriented chemotherapy. The initial therapies given for the present cases are even less intensive than the other ML-DS protocols used in developed countries and, of course, than that of the non-DS AML protocols.

Only 50% of patients with R/R ML-DS in the present study had achieved further remission with attempts of various reinduction therapies. It has been reported that the reinduction rate for non-DS patients with AML is 65%-77%, and achievement of subsequent CR was uniformly a good prognostic factor.^15,16  When we consider the results of the present study, which indicate that the achievement of further remission is a good prognostic factor, improvement of the reinduction rate for R/R ML-DS would be mandatory for a better prognosis.

The reported 5-year OS rate of patients with non-DS AML who attained second CR and subsequently received HSCT were 58%-62%.^17-19  In the present study, 8 of the 13 patients subsequently received allogeneic HSCT, but only 2 of those 8 patients (25%) survived. It is well known that the transplantation-related mortality of patients with ML-DS is greater than in patients with non-DS AML²⁰ ; however, the main cause of transplantation failure was disease progression, not transplantation-related complications, where only one death in remission was documented. Our result was consistent with the report by Meissner et al in which they described that relapse was the major cause of treatment failure in children with DS treated by HSCT for acute leukemia.²¹ 

However, 5 of the 13 R/R ML-DS patients with further remission subsequently were treated with chemotherapy only, and 4 of the 5 patients survived (duration after relapse, 22-45 months). Three of those 4 patients had relapsed more than 12 months from the initial diagnosis, which was found to be good prognostic factor in this study. Moreover, most of these patients were treated with continuous and/or high-dose Ara-C, which might be a key component of a salvage regimen for R/R ML-DS. Allogeneic HSCT might not be essential for these patients, especially for “late” relapsed patients, as for the non-DS cases^15,16 

The duration from initial diagnosis to R/R was shown as the strongest prognostic factor, but biologic factors (including chromosomal abnormalities and GATA1 status) were not relevant in this R/R ML-DS study. Although more than 80% of patients with ML-DS could be cured with low-intensive chemotherapy, methods that can be used to identify the remaining poor subgroups with a poor prognosis and a treatment strategy distinct from “usual” ML-DS are urgently needed. This necessity is because of the fact that they are rarely salvageable once they experience morphologic induction failure or relapse, as was indicated in this retrospective study. Future treatment protocols in this patient population could include adherence to a very low-intensity chemotherapy for the majority of patients with ML-DS, identification of the subgroup with a poor prognosis using minimal residual disease, and stratification of these patients to receive a more intensive chemotherapy containing high-dose and/or continuous infusion of intermediate-dose Ara-C.

The authors are grateful to Dr I. Iguchi (Hokkaido University), Dr A. Sato (Miyagi Children Hospital), Dr T. Watanabe (Niigata Cancer Center), Dr M. Sotomatsu (Gunma Children Hospital), Dr K. Ida (Tokyo University), Dr M. Nagasawa (Tokyo Medical and Dental University), Dr T. Kaneko (Tokyo Metropolitan Children's Medical Center, Dr R. Ooyama (St Marianna Medical College), Dr M. Akiyama (Jikei Medical College), Dr M. Yabe (Tokai University), Dr D. Toyama (Showa University), Dr Y. Taneyama (Chiba Children's Hospital), Dr K. Matsumoto (Nagoya First Red Cross Hospital), Dr M. Inoue (Osaka Medical Center and Maternal and Child Research Institute), Dr D. Hasegawa (Hyogo Children's Hospital), Dr I. Usami (Kobe City Hospital), Dr Y. Kataoka (Matsuyama Red Cross Hospital), Dr S. Kawakami (Ehime University), Dr T. Anan (Kumamoto University), and the 120 institutions and all of the doctors in the Japanese Pediatric Leukemia/Lymphoma Study Group for their invaluable contributions to data collection. They also thank Dr T. Toki and Dr R. Wang (Hirosaki University Graduate School of Medicine), for performing GATA1 mutation analysis.

This study was supported in part by a Grant-in-Aid for Cancer Research from the Ministry of Health and Welfare of Japan.

Contribution: T. Taga and D.T. designed, organized, and performed research, analyzed data, and wrote the paper; A.M.S. designed research and collected and analyzed clinical data; K. Kudo, H.M., A.K., S.I., H.N., H.T., A.T., and A.S. designed, organized, and performed research and analyzed data; K.T. performed mutation screening and designed, organized, performed research and analyzed data; T. Taki designed, organized, and performed research and analyzed chromosomal data; K. Koh and H.K. provided clinical samples and data; and S.A. provided clinical data, designed and organized research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Takashi Taga, Department of Pediatrics, Shiga University of Medical Science, Seta-tsukinowa, Otsu, Shiga 520-2192, Japan; e-mail: ttaga@belle.shiga-med.ac.jp.