Arsenic trioxide and ascorbic acid: synergy with potential implications for the treatment of acute myeloid leukaemia?


Univ. Prof. Dr Heinz Gisslinger, University of Vienna, Department of Internal Medicine I, Division of Haematology and Blood Coagulation, Währinger Gürtel 18–20, A-1090 Vienna, Austria. E-mail:


Arsenic trioxide (As2O3) induces remission in a high proportion of patients with acute promyelocytic leukaemia (APL) via induction of apoptosis. Preliminary reports suggest that the apoptotic effect of As2O3 is not specific for APL but can also be observed in non-APL acute myeloid leukaemia (AML) cells, although these are less sensitive than APL cells. Ascorbic acid has recently been demonstrated to enhance the apoptotic effect of As2O3. We have therefore evaluated combined As2O3/ascorbic acid treatment in various clinical samples of AML. Our results indicate a significant synergistic effect of As2O3 and ascorbic acid, suggesting a possible future role of As2O3/ascorbic acid combination therapy in patients with AML.

Arsenic trioxide (As2O3) has recently been identified as a potent anti-leukaemic agent in patients with refractory acute promyelocytic leukaemia (APL) (Soignet et al, 1998). At clinically achievable concentrations of 1–2 μmol/l, As2O3 induces apoptosis in the t(15;17) APL cell line NB4, as well as APL cells in vitro and, to some extent, in patients with APL without significant myelosuppression (Chen et al, 1996, 1997; Soignet et al, 1998). Recent preliminary reports suggest that the apoptotic effect of As2O3 is not specific for APL but can also be observed in non-APL acute myeloid leukaemia (AML) cells, although these may display an up to 10-fold lower sensitivity to As2O3 than APL cells (Kitamura et al, 1997; Huang et al, 1999). This has led to a search for agents that enhance the apoptotic effect of As2O3 in cells intrinsically less sensitive to As2O3. Ascorbic acid has been found to be a promising agent in this regard as it is able to synergize with the growth-inhibitory and apoptotic effects of As2O3 both in vitro and in vivo (Dai et al, 1999). In order to investigate whether ascorbic acid is suitable for enhancing the apoptotic effect of As2O3 in different kinds of AML, we screened a panel of freshly isolated leukaemic cells from patients with various subtypes of AML (AML-M1, -M2, -M3, -M4 and -M5a) with a combination of As2O3 and ascorbic acid. Our results indicate a significant synergistic effect of As2O3 and ascorbic acid on apoptosis in clinical samples of AML and, thus, suggest a possible future role of this treatment approach in patients with the disease.

Patients and methods

Compounds A stock solution of As2O3 (Sigma Chemical, St Louis, MO, USA) was made at a concentration of 10 mmol/l and diluted to the working concentration of 1 μmol/l before use. Ascorbic acid was purchased from Sigma Chemical and diluted to the working concentration of 125 μmol/l separately for each experiment.

Patient samples After informed consent, heparinized peripheral blood was obtained from AML patients with at least 50% circulating blast cells. For isolation of leukaemic cells, blood samples were layered onto Lymphoprep (Nycomed, Oslo, Norway) and centrifuged at 450 g for 30 min at room temperature.

Cell culture Freshly isolated leukaemic cells were cultured by seeding 5 × 105 cells/ml of Roswell Park Memorial Institute (RPMI)-1640 medium containing 25 mmol/l HEPES buffer and L-glutamine (Gibco-BRL, Grand Island, NY, USA), supplemented with 10% heat-inactivated fetal calf serum (FCS; Gibco-BRL), with or without addition of the compounds described above, in a 5% CO2 humidified atmosphere at 37°C.

Cell viability and morphological observation Cell viability was assessed on d 3 using trypan-blue dye exclusion. For morphological observation, cells were centrifuged onto slides using a cytocentrifuge (Shandon, 500 r.p.m., 5 min) and subsequently stained with May–Grünwald–Giemsa stain. Apoptotic cells were identified according to the criteria introduced by Kerr et al (1972).

TUNEL assay The terminal deoxynucleotidyl-transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick end-labelling (TUNEL) assay was performed according to the recommendations of the manufacturer (Boehringer, Mannheim, Germany). For quantification of biotinylated dUTP incorporated in nucleotide polymers, cells were analysed using flow cytometry (Becton Dickinson, San Jose, CA, USA).

Statistics Differences between values obtained in leukaemic cells treated with different experimental conditions were determined using two-sided Student's t-test analyses. A P-value < 0·05 was considered statistically significant.


Eleven patients with an established diagnosis of AML were included in the investigations. Patient characteristics are given in Table I.

Table I.  Patient characteristics and apoptotic responses of isolated AML samples.




(× 109/l)

% Blast cells

% Apoptosis
SpontaneousAs2O3As2O3+ AA
  • *

    Some patients received HU or ATRA shortly prior to isolation of leukaemic cells. However, we did not detect any statistically significant differences in apoptotic responses between samples derived from patients who had received HU or ATRA and samples derived from patients who had remained untreated.

  •   †Numbers in parentheses represent treatment-specific apoptosis (cumulative treatment-induced apoptosis minus spontaneous apoptosis).

  •   ‡PML-RARα-positive.

  •   As2O3,1 μmol/l arsenic trioxide; AA, 125 μmol/l ascorbic acid; [R], relapse; HU, hydroxyurea; ATRA, all-trans retinoic acid.

174M1 [R]6094HU45·483·7 (38·3)85·0 (39·6)
248M21466836·583·4 (46·9)84·0 (47·5)
369M1158046·571·7 (25·2)66·1 (19·6)
444M35376ATRA4·211·4 (7·2)45·1 (40·9)
557M4104503·318·2 (14·9)36·6 (33·3)
640M113289HU20·932·2 (11·3)52·9 (32·0)
782M214085HU1·715·0 (13·3)22·5 (20·8)
876M42018011·913·1 (1·2)22·5 (10·6)
972M419883HU15·016·3 (1·3)22·2 (7·2)
1060M1 [R]2169HU2·15·5 (3·4)6·0 (3·9)
1176M5a [R]34509·013·3 (4·3)13·6 (4·6)

First, we addressed the spontaneous apoptotic responses of the AML samples using morphological observation and the TUNEL assay. Consistent with previous observations (Banker et al, 1997), AML cells showed widely variable spontaneous apoptosis after 72 h in vitro. Morphological observation of cytospin preparations revealed different degrees of typical features of apoptosis, including chromatin condensation and fragmentation of the nuclei, as well as formation of apoptotic bodies. In accordance with morphological findings, the TUNEL assay detected 1·7% to 46·5% apoptotic cells in the investigated AML samples (Table I), with median spontaneous apoptosis amounting to 11·9% (Fig 1).

Figure 1.

Apoptotic responses in clinical samples from 11 patients with AML. Leukaemic cells were treated with 1 μmol/l As2O3 and 125 μmol/l ascorbic acid (AA) alone or in combination. Apoptotic cells were detected using the TUNEL method as described in Patients and Methods. The data are shown in a box-chart plot, with markings that indicate the 0th, 25th, 50th, 75th and 100th percentiles. P-values: P = 0·36 (AA vs. control), P = 0·008 (As2O3 vs. control), P = 0·036 (As2O3+ AA vs. As2O3).

When ascorbic acid was added to the cultures, no substantial apoptotic effect was observed: the median percentage of TUNEL-positive cells in cultures treated with ascorbic acid amounted to 14·8% (range 1·6% to 48·8%), which was slightly increased as compared with untreated controls, although it did not reach statistical significance (P = 0·36) (Fig 1).

Next, we addressed the apoptotic effect of As2O3. Samples with high levels (> 35%) of spontaneous apoptosis (derived from patients 1–3) responded moderately to well to As2O3, with As2O3-specific apoptosis (cumulative apoptosis upon treatment with 1 μmol/l As2O3 minus spontaneous apoptosis) ranging from 25·2% to 46·9%. However, the remaining eight samples [noticeably, also that of patient 4 (AML-M3)] displayed considerably lower sensitivity to As2O3, with As2O3-specific apoptosis values of less than 15% (Table I). The median percentage of TUNEL-positive cells in cultures treated with As2O3 was relatively low (16·3%, range 5·5% to 83·7%), albeit significantly increased as compared with untreated controls (P = 0·008) (Fig 1).

We finally addressed the effects of ascorbic acid on As2O3-mediated induction of apoptosis. Noticeably, ascorbic acid led to substantial enhancement of As2O3-specific apoptosis in six out of eight (75%) samples with low intrinsic sensitivity to As2O3 (Table I): in five samples (patients 4–6, 8 and 9), As2O3-specific apoptosis could be at least doubled, and in another sample (patient 7), it could be increased by more than 50%, indicating a distinct synergism of As2O3 and ascorbic acid in the respective cells. The most prominent synergistic effect was observed in the sample from patient 4 (AML-M3), in which As2O3-specific apoptosis could be increased almost sixfold from 7·2% (As2O3) to 40·9% (As2O3 and ascorbic acid). Overall, combined treatment with ascorbic acid resulted in a significant increase in the median percentage of TUNEL-positive cells from 16·3% in As2O3-treated cultures to 36·6% (range 6·0–85·0%) in cultures treated with As2O3 and ascorbic acid (P = 0·036) (Fig 1).


A growing body of evidence suggests that As2O3-induced apoptosis is mediated through generation of reactive oxygen species, particularly H2O2 (Wang et al, 1996; Jing et al, 1999). Moreover, it has been found that the cellular glutathione redox system modulates the growth-inhibitory and apoptotic effects of As2O3, with increasing levels of reduced glutathione (GSH) conferring resistance to As2O3 (Dai et al, 1999). Lowering cellular GSH as well as increasing intracellular H2O2 levels are therefore suitable approaches to enhance As2O3-induced apoptosis in malignant cells that are otherwise refractory to 1–2 μmol/l As2O3. Indeed, agents such as buthionine sulphoximine, which depletes GSH, as well as mercaptosuccinic acid and aminotriazol, which increase H2O2 levels by inhibiting glutathione peroxidase and catalase, have been successfully used to augment As2O3-induced apoptosis in cell lines intrinsically less sensitive to As2O3 (Dai et al, 1999; Jing et al, 1999).

Ascorbic acid has recently been shown to equally enhance As2O3-induced apoptosis owing to its capacity to undergo auto-oxidation resulting in elevated intracellular H2O2 levels (Dai et al, 1999). Importantly, ascorbic acid does not potentiate the effects of As2O3 on the colony formation capacity of normal haematopoietic cells, suggesting that the combination of As2O3 and ascorbic acid may be selectively toxic to malignant cells without causing severe toxicity in normal tissues (Dai et al, 1999). To date, As2O3 and ascorbic acid have been found to synergize in As2O3-resistant HL-60 cells and primary cultures of chronic lymphocytic leukaemia cells, as well as in a mouse lymphoma model in which combined treatment substantially increased the survival time of mice injected with P388D1-lymphoma cells, whereas single agent treatment had no influence on survival (Dai et al, 1999). Steady-state ascorbic acid plasma levels of 125 μmol/l, the concentration used during our experiments, can be easily achieved and maintained using daily intravenous infusion of approximately 1·3 g of ascorbic acid, with minimal toxicity (Riordan et al, 1995). Parenteral administration is necessary as plasma levels tend to approach an upper limit (90 μmol/l) when ascorbic acid is given orally, even in doses exceeding 2·5 g (Blanchard et al, 1997).

To our knowledge, the present report is the first to provide evidence that As2O3 and ascorbic acid synergize to induce apoptosis in clinical samples of AML: co-administration of ascorbic acid significantly enhanced As2O3-induced apoptosis in the AML samples tested, predominantly owing to a substantial synergistic effect of ascorbic acid in most samples with low sensitivity to As2O3. Noticeably, the effect mediated by ascorbic acid was not restricted to a particular subtype of AML, as evidenced by the fact that pronounced increases in apoptosis could be achieved in a variety of subtypes including AML-M1, -M2, -M3 and -M4. The most impressive synergism was noted in a sample of PML-RARα-positive APL (AML-M3, patient 4), suggesting that APL cells that are intrinsically less responsive to As2O3 may be particularly sensitive to combination treatment with As2O3 and ascorbic acid.

In conclusion, our study suggests that the use of As2O3 and ascorbic acid may represent a promising treatment strategy in patients with AML. Adding ascorbic acid to As2O3 treatment may prove particularly useful to treat AML patients who are intrinsically resistant to As2O3, as well as to overcome acquired As2O3 resistance in patients with APL. Therefore, further studies on the combination of As2O3 and ascorbic acid are clearly warranted.


This work has been supported by grants from the University of Vienna and the Hochschuljubiläumsstiftung der Stadt Wien. We thank Dr Christine Brostjan and Dr Alexander Gornikiewicz for excellent technical assistance.