BackgroundAlthough most pediatric acute myeloid leukemia (AML) protocols no longer prescribe glucocorticoids (GCs) as part of antileukemic treatment, several low-income countries may still rely on them. Furthermore these agents are globally being used in the prevention or treatment of chemotherapy-induced nausea, or in the treatment of acute respiratory distress syndrome in children with leukemia. Therefore, to study potential effects of GCs on AML cells remains important.The aim of this study was to evaluate the in vitro responsiveness of pediatric AML cells to GCs in relation to cell biological features.MethodsIn vitro drug resistance or potential proliferation upon GCs (prednisone) was determined using a methyl-thiazol-tetrazolium salt (MTT)-assay. Samples with more than 70% blasts at day 4 in the control wells and a control optical density (OD) >0.050 units were used to calculate the LC50-value, i.e. the PRD concentration lethal to 50% of the AML blasts. Samples were classified as highly sensitive (LC50 PRD ≤0.1 µg/ml), intermediately sensitive (LC50 PRD >0.1-150 µg/ml) or resistant (LC50 PRD ≥150 µg/ml). PRD-induced leukemic cell proliferation was defined as a leukemic cell survival (LCS=(OD drug-exposed well)/(mean OD control wells)) of >105% plus 2 times the standard deviation of the control cell survival in the PRD containing wells.Potential glucocorticoid (GC)-induced differentiation was evaluated using basic morphology. To validate the overall effects of GCs, a subset of randomly selected samples was tested for both prednisone and dexamethasone (DXM).Samples of patients treated according to (at that time) GC-containing protocols, provided by the BFM-Germany and the DCOG, were used to evaluate associations between in vitro effects and clinical outcome, both at an univariate and multivariate level.Within these protocols GCs were only administered during consolidation.ResultsApproximately 70% (n=117) of the 167 available pediatric AML samples could be evaluated for GC resistance and GC-induced proliferation. None of the samples were highly sensitive and only 15% (n=17) were intermediately sensitive to the lytic action of GCs. FAB type M1 samples were significantly more often intermediately sensitive than other FAB types (43% of the M1 samples versus 0-13% in FAB type M0, and M2-M6 samples) (P=0.006). FAB type M7 samples were more intermediately sensitive as well (67%, n=2), but sample numbers were small.GC-induced differentiation was not observed after evaluating 10 unselected samples.Contrary, 27% (n=32) of the samples showed proliferation upon GC-exposure. GC-induced proliferation occurred independently from spontaneous proliferation.Samples with FAB type M5 or activating FLT3 mutations were significantly more prone to this phenomenon: 63% (n=12) of the M5 samples compared to 0-33% of the other samples (P<0.001), and 60% (n=6) of the FLT3itd-positive samples versus 23% (n=22) of the negative samples (P=0.011) showed GC-induced proliferation. There were no significant differences in GC-induced proliferation between samples with favorable or other cytogenetics.The results for the subset of samples tested for both PRD and DXM were similar with regard to GC-resistance, GC-induced proliferation, and GC-differentiation.Probabilities of relapse free survival (pRFS) were lower for patients with GC-induced proliferation at a borderline significant level (P=0.045, univariately). However, in a multivariable COX model, including white blood cell count, cytogenetics, and FLT3 status an independent association between inferior pRFS and GC-induced resistance or GC-induced proliferation could not be confirmed.ConclusionAssuming that the proliferation induced in vitro is paralleled in vivo, it is tempting to speculate that GCs given during consolidation treatment may have an adverse effect on any present minimal residual leukemic cells, and that such exposure may increase the risk of relapse in a significant subset of patients.Although we did not study this in a prospective or in vivo design, given the lack of GC-induced differentiation effects, and the potential of GC-induced proliferation in this study - with at best unknown clinical consequences - the use of GCs in pediatric AML patients should be discouraged in general, and especially in subgroups with FAB type M5 or activating FLT3 mutations. DisclosuresKaspers:Galen Pharmaceuticals: Consultancy; Boehringer Ingelheim: Consultancy; Celgene: Consultancy.