Addition of hydroxyurea (hydroxycarbamide) enhances the efficacy of fludarabine/cytarabine-based salvage regimens against acute myeloid leukaemia

Abstract

Despite intensive treatment, patients with relapsed or refractory acute myeloid leukaemia (AML) have a dismal prognosis. A cornerstone therapy for relapsed/refractory AML is a combination of fludarabine (F-ara-A) and cytarabine (ara-C), so-called FLA. As the enzyme SAMHD1 mediates resistance to both ara-C and F-ara-A, we investigated whether SAMHD1 inhibition via hydroxyurea (hydroxycarbamide; HU) could improve FLA efficacy. Here, we show that HU synergistically enhanced ara-C-, F-ara-A- and FLA-induced cytotoxicity in an SAMHD1-dependent manner in AML cell lines, primary AML cells and an immunocompetent AML mouse model. Mechanistically, HU significantly increased the active metabolite triphosphates of ara-C and F-ara-A in FLA combinations. Furthermore, leukaemic SAMHD1 protein expression negatively correlated with overall survival in a cohort of FLA-treated refractory AML patients. Our findings suggest that the addition of HU improves the efficacy of FLA-based regimens and warrant clinical trials to test the safety and efficacy of this combination in patients with relapsed/refractory AML.

Keywords: AML; SAMHD1; relapse/refractory.

FIGURE 1

Hydroxycarbamide (HU) sensitizes cells to ara‐C, F‐ara‐A and the FLA combination and leads to increased apoptosis and checkpoint activation in SAMHD1‐positive acute myeloid leukaemia (AML) cells. Proliferation inhibition assay combining HU with ara‐C (A), F‐ara‐A (B) or the FLA (ara‐C + F‐ara‐A) combination (C) in SAMHD1‐proficient THP‐1 (left panels), MV4‐11 (middle panels) and OCI‐AML‐3 (right panels) cells. For (C), ara‐C treatment was combined with 2.9 μM F‐ara‐A (FLA), 60 μM HU (ara‐C + HU) or a combination of both (FLA + HU). Error bars indicate mean and standard deviation from replicates from a representative out of three independent experiments. (D) EC₅₀ values plotted as a function of HU concentration for ara‐C (solid line) and F‐ara‐A (dashed line), for cell lines for experiments shown in (A) and (B). Cells were treated for 72 h. Statistical testing was performed by comparing the logEC₅₀ values using an extra‐sum‐of‐squares F‐test: (A) THP‐1: F = 21.3, MV4‐11: F = 15.82, OCI‐AML‐3: F = 26.69 with DFn = 4, DFd = 60 for all; (B) THP‐1: F = 6.861, MV4‐11: F = 14.82, OCI‐AML‐3: F = 41.73 with DFn = 3, DFd = 48 for all; (C) THP‐1: F = 13.51, MV4‐11: F = 13.26, OCI‐AML‐3: F = 15.81 with DFn = 3, DFd = 40 for all. ****p < 0.0001. (E) Percentage of Annexin‐V‐positive cells in SAMHD1‐proficient THP‐1 cells. Cells were treated for 48 h with 500 nM ara‐C, 300 nM F‐ara‐A, 60 μM HU as indicated. Error bars represent mean and standard deviation from triplicate assays from a representative out of three independent experiments. (F) Fraction of pCHK1‐positive cells detected in SAMHD1‐proficient THP‐1 cells by immunofluorescence for indicated treatments. Error bars represent mean and standard error from two independent experiments. Cells were treated for 24 h with the same concentrations as given in (E). Analyses were performed using unpaired two‐tailed t‐tests. *p < 0.05. Representative images with pCHK1 in red and DAPI (4',6‐diamidino‐2‐phenylindole; to stain for the DNA) in blue. Scale bars, 10 μm.

FIGURE 2

Hydroxyurea (HU) increases levels of active ara‐C and F‐ara‐A metabolites in an SAMHD1‐dependent manner. SAMHD1‐proficient THP‐1, OCI‐AML‐3, MV4‐11 and SAMHD1‐deficient THP‐1 cells were treated as indicated, and deoxyribonucleotide triphosphate (dNTP) and metabolite levels were determined using Liquid Chromatography‐Tandem Mass Spectrometry (LC‐MS/MS). Concentrations used: HU—60 μM, ara‐C—500 nM, F‐ara‐A—150 nM. Levels of ara‐CTP (A) and F‐ara‐ATP (B) in pmol/10⁶ cells after 24 h. (C, D) Metabolite levels over time as indicated in cells treated with FLA or FLA + HU in THP‐1 SAMHD1 ^+/+ (left) and SAMHD1 ^−/− (right) for ara‐CTP (C) and F‐ara‐ATP (D) in pmol/10⁶ cells. AUC, area under the curve. (E) Ratio of dCTP‐to‐dATP for indicated treatments in THP‐1 SAMHD1 ^+/+ (left) and SAMHD1 ^−/− (right) cells. Points, columns and error bars correspond to individual values, means and SEM of three independent experiments. Statistical analyses were done using unpaired two‐tailed t‐tests: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 3

Addition of Hydroxycarbamide (HU) to FLA is effective in a murine acute myeloid leukaemia (AML) model and in AML primary patient samples. (A) Kaplan–Meier survival analysis of CD45.2 C57BL/6J mice intravenously injected with mouse MLL‐AF9‐transformed AML blasts (day 0) and treated with PBS, FLA or FLA + HU days 20–24. n = 6 per treatment group. Statistical analysis was done using the Mantel–Cox log‐rank test; p = 0.03 for FLA versus PBS, p = 0.008 for FLA + HU versus PBS. (B) Primary paediatric AML sample at diagnosis (A4326) with HU added to ara‐C (left panel), F‐ara‐A (middle panel) and FLA (right panel). FLA with and without HU in an AML paediatric sample at relapse (A3793) (C), an adult patient sample at diagnosis (ALG20_049) (D) and an adult patient sample at relapse (ALG20_018) (E). (F) Dose–response curves for an adult patient at diagnosis (ALG2011_23). Points represent the mean and standard deviation from replicate values after treating cells for 72 h. Concentrations used: 0.7 μM F‐ara‐A, 60 μM HU. Absolute ara‐CTP (G) and F‐ara‐ATP (H) levels in pmol/10⁶ cells 24 h after indicated treatments as determined by LC‐MS/MS in the same patient as shown in (F); statistical analyses were done using unpaired two‐tailed t‐tests. For metabolite measurements, 500 nM ara‐C, 150 nM F‐ara‐A and 60 μM HU were used. Points, columns and error bars correspond to individual values, means and SD of two independent experiments. Statistical testing for dose–response curves in (B–F) was performed by comparing the logEC₅₀ values using an extra‐sum‐of‐squares F test: (B) left: F = 56.51, DFn = 1, DFd = 24; (B) middle: F = 35.24, DFn = 1, DFd = 24; (B) right: F = 91.64, DFn = 1, DFd = 24; (C) F = 98, DFn = 1, DFd = 24; (D) F = 12.83, DFn = 1, DFd = 24; (E) F = 15.98, DFn = 1, DFd = 20; (F) ara‐C only versus ara‐C + HU: F = 8.18, DFn = 1, DFd = 20; (F) FLA versus FLA + HU: F = 54.59, DFn = 1, DFd = 20. (I) High and low SAMHD1 groups as determined from blots shown in Figure S8A plotted against the EC₅₀(log₁₀M) of ara‐C for FLA (ara‐C with 0.7 μM F‐ara‐A) and FLA + HU (ara‐C + 0.7 μM F‐ara‐A + 60 μM HU). Individual points correspond to patient samples as colour‐coded in Figure S8A; horizontal lines depict median. Statistical analysis in (I) was performed using paired non‐parametric t‐test (Wilcoxon test). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 4

SAMHD1 levels correlate with outcome in FLA‐treated relapsed/refractory acute myeloid leukaemia (AML) patients. (A) Representative image of a patient with high SAMHD1 expression (top) and a patient with low SAMHD1 expression (bottom). Left panels show haematoxylin–eosin (H&E) staining; right panels show SAMHD1 (brown)/CD68 (magenta) staining. SAMHD1 is mainly expressed in the nucleus of AML blasts, while CD68⁺ macrophages are strongly SAMHD1‐positive and serve as an internal positive control. Scale bars = 100 μm. (B) Kaplan–Meier survival curves showing overall survival in patients treated with FLA‐based regimens in the subgroup of refractory disease only. *p < 0.05.

FIGURE 5

Model for the action of SAMHD1 on ara‐C and F‐ara‐A metabolism. Ara‐C and F‐ara‐A are substrates of the rate‐limiting enzyme deoxycytidine kinase (dCK) in the process of their phosphorylation and activation. SAMHD1 limits the efficacy of these agents by converting the cytotoxic triphosphate metabolites of ara‐C and F‐ara‐A to their inactive precursors. Hydroxycarbamide (HU) inhibits SAMHD1, preventing it from metabolizing ara‐CTP and F‐ara‐ATP, and allowing accumulation of cytotoxic products that lead to apoptosis of leukaemic cells. In addition, HU can activate dCK, leading to further accumulation of active ara‐C and F‐ara‐A in the presence of HU.