Stability of the Antimalarial Drug Dihydroartemisinin under Physiologically Relevant Conditions: Implications for Clinical Treatment and Pharmacokinetic and In Vitro Assays

Abstract

Artemisinins are peroxidic antimalarial drugs known to be very potent but highly chemically unstable; they degrade in the presence of ferrous iron, Fe(II)-heme, or biological reductants. Less documented is how this translates into chemical stability and antimalarial activity across a range of conditions applying to in vitro testing and clinical situations. Dihydroartemisinin (DHA) is studied here because it is an antimalarial drug on its own and the main metabolite of other artemisinins. The behaviors of DHA in phosphate-buffered saline, plasma, or erythrocyte lysate at different temperatures and pH ranges were examined. The antimalarial activity of the residual drug was evaluated using the chemosensitivity assay on Plasmodium falciparum, and the extent of decomposition of DHA was established through use of high-performance liquid chromatography with electrochemical detection analysis. The role of the Fe(II)-heme was investigated by blocking its reactivity using carbon monoxide (CO). A significant reduction in the antimalarial activity of DHA was seen after incubation in plasma and to a lesser extent in erythrocyte lysate. Activity was reduced by half after 3 h and almost completely abolished after 24 h. Serum-enriched media also affected DHA activity. Effects were temperature and pH dependent and paralleled the increased rate of decomposition of DHA from pH 7 upwards and in plasma. These results suggest that particular care should be taken in conducting and interpreting in vitro studies, prone as their results are to experimental and drug storage conditions. Disorders such as fever, hemolysis, or acidosis associated with malaria severity may contribute to artemisinin instability and reduce their clinical efficacy.

INTRODUCTION

Artemisinin derivatives have been the most potent antimalarials available to date; they are highly active against all Plasmodium species and parasite stages, including young gametocytes, and constitute the backbone of malaria case management for severe and uncomplicated malaria (as a component of artemisinin combination therapy [ACT]) (1). The emergence and spread of artemisinin resistance in Southeast Asia is, therefore, a serious threat to malaria control (2, 3).

All artemisinin derivatives share a distinctive 1,2,4-trioxane pharmacophore, which confers their antimalarial activity (the corresponding acyclic analogues lacking the endoperoxide bridge are inactive) (4, 5) but also makes these molecules particularly highly reactive and thus difficult to quantify in plasma or blood. Of the various artemisinin-type compounds in use, dihydroartemisinin (DHA), the reduced lactol derivative of artemisinin, is an antimalarial compound on its own (currently coformulated with piperaquine) and the main bioactive metabolite of artesunate and artemether. DHA itself is chemically fragile and displays a marked propensity to undergo ring opening of the lactol and rearrangement under neutral conditions, leading to a new, biologically active peroxide, which in turn rapidly decays to the inert end product deoxyartemisinin (6). Under aqueous conditions, artemisinins tend to react with labile ferrous iron and heme-Fe(II), and in certain organic solvents, such as dimethyl sulfoxide (DMSO), they degrade very quickly (7). Moreover, under physiological conditions at pH 7.4, artemisinins are reduced by biological reductants, such as flavin cofactors (8). Artemisinin-type compounds are short lived (depending on the route of administration, elimination half-lives (t_1/2) range from 0.67 to 20.2 h), but their antimalarial activity persists without measurable drug levels (9, 10).

While these characteristics are well-known, our understanding of how artemisinin derivatives behave in various conditions and how this relates to their antimalarial activity is incomplete. Here, we describe the effects on chemical decay and in vitro activity in cultures of exposing DHA to a range of conditions (pH, temperature, media, and blood products) likely to be relevant to the interpretation of in vitro assays and clinical use of DHA and which may affect its overall activity.

MATERIALS AND METHODS

In vitro cultures of Plasmodium falciparum.

The chloroquine (CQ)-sensitive (D10) and the CQ-resistant (W2) strains of P. falciparum were maintained at 5% hematocrit (human type A-positive red blood cells [RBCs]) in RPMI 1640 (EuroClone) medium supplemented with 1% AlbuMax II (lipid-rich bovine serum albumin) (Invitrogen), 0.01% hypoxanthine (Sigma), 20 mM HEPES (EuroClone), and 2 mM glutamine (EuroClone). All the cultures were maintained at 37°C in a standard gas mixture consisting of 1% O₂, 5% CO₂, and 94% N₂ (11).

Preparation of erythrocyte lysate.

Blood samples from human donors kept in citrate-phosphate-dextrose solution were obtained from AVIS Milano. Blood was centrifuged to remove plasma and leukocytes, and red blood cells (RBCs) were washed two times with RPMI medium.

One milliliter of 100% erythrocytes was diluted in 10 ml of phosphate-buffered saline (PBS) (pH 7.4). The suspension was then divided in two aliquots. In order to obtain carbon monoxide (CO) hemoglobin, 5 ml of PBS-diluted RBCs was flushed with a gas mixture containing 2% CO, 5% CO₂, and 93% nitrogen for 5 min. Normal and CO-flushed erythrocytes were then lysed by a cycle of freeze-thawing at −80°C; the stable complexation of Fe(II)-heme with CO was confirmed by UV-visible light spectrophotometry, which showed the shift in absorption from 414 nm of oxy-hemoglobin (HbO) to 420 nm of carboxy-hemoglobin (HbCO) (12, 13). Erythrocyte lysates were further diluted 1:5 in PBS to perform the experiments described below.

DHA preparation and incubation conditions.

DHA (a gift from R. Haynes) was weighed and dissolved in ethanol at a concentration of 40 mM. The drug was then diluted at 2 μM in PBS, plasma (inactivated at 56°C for 40 min), serum, plasma diluted in PBS, erythrocyte lysate, or CO-flushed erythrocyte lysate diluted 1:5 in PBS. Samples were then incubated at 37°C for different lengths of time (3, 6, or 18 h) in a CO₂ incubator. Some experiments were performed at 4°C, room temperature, or 40°C. DHA was also incubated in erythrocyte lysate or in plasma in the presence of ascorbic acid (Sigma-Aldrich) (0.5 or 1 mM) dissolved in PBS at 100 mM and then diluted to the desired concentrations directly in the incubation mixtures. Artesunate (a gift from R. Haynes) and chloroquine (Sigma-Aldrich) were dissolved at 40 mM in DMSO and water, respectively, and then diluted to 2 μM under the conditions described above.

For the experiments performed at different pHs, phosphate buffers at a pH of 7.2, 7.4, or 7.6 were prepared using different ratios of NaH₂PO₄ and Na₂HPO₄, and pHs were measured using a pH meter.

The activity of each DHA preparation was then determined by the semiautomated parasite lactate dehydrogenase (pLDH) assay (Freedom Evo 75; Tecan Italia, Srl). Briefly, samples containing DHA were diluted in complete medium at 100 nM DHA concentration, placed in 96-well flat-bottom microplates (Costar), and serially diluted. The control (fresh DHA, artesunate, and chloroquine-CQ) was dissolved in ethanol, DMSO, or water at 10 mg/ml and then diluted to the desired concentration in complete medium. Asynchronous cultures with parasitemia of 1% to 1.5% and 1% final hematocrit were aliquoted into the plates and incubated for 48 h at 37°C. Parasite growth was determined spectrophotometrically (optical density at 650 nm [OD₆₅₀]) by measuring the activity of the pLDH according to a modified version of Makler's method in control and drug-treated cultures (14). Antimalarial activity was determined as the concentration of drugs inducing 50% of growth inhibition (IC₅₀). Three independent experiments were performed, and each one was performed in duplicate.

Measurement of DHA half-life at different pHs and in plasma.

Chromatographic determinations were performed using high-performance liquid chromatography (HPLC)-grade acetonitrile by electrochemical detection (ECD) operating in the reductive mode at −1,000 mV in an oxygen-free environment (HPLC-ECD BAS 200A; Bioanalytical System, Inc., West Lafayette, IN, USA). Chromatographic separations were obtained on a Nucleosil C₁₈ stainless steel column (250-mm length by 4.6-mm inside diameter and 5-μm particle size) (Phenomenex, Torrance, CA, USA).

Buffer solutions of various species with pH ranges of 1.0 to 8.6 (hydrochloric acid buffer, pH <2; phthalate buffer, pH 2.0 to 4.5; phosphate buffer, pH 6.0 to 7.4; borate buffer, pH 8.6) were prepared at a constant total buffer species concentration of 0.2 M and ionic strength, μm = 0.5, by adding sodium chloride. Sample solutions of DHA at 20, 30, 40, and 50 mg/liter were prepared by using a specific volume of a 1.0 mg/ml DHA stock solution (in ethanol) in a final volume of 100 ml achieved by adding buffer (preheated to 37°C). The DHA stock solution was prepared in ethanol because of the poor solubility of DHA in buffer solutions. The resulting mixture was sealed to prevent water evaporation and maintained at the appropriate temperature in a thermostat water bath (water bath model W270; Memmert, Schwabach, Germany). The temperature was kept at 37 ± 0.1°C. At appropriate time intervals, 500-μl aliquots of the reaction mixture were removed from the reaction vessel, pipetted into test tubes, sealed, and immediately stored in a −70°C freezer and analyzed as quickly as possible. At the time of analysis, samples were removed from the freezer and thawed. One hundred microliters of the internal standard stock solution (40 mg/liter) prepared in ethanol/water (50:50 [vol/vol]) and 400 μl of ethanol/water (50:50 [vol/vol]) were added into the sample test-tube containing 500 μl sample and vortex mixed. A fixed volume (20 μl) of these samples was assayed chromatographically to determine the concentrations of DHA and artemisinin. This procedure was used for all pH values.

To determine the kinetics of decomposition of DHA in plasma, 40 μl of DHA (50 ng/μl) solution in ethanol/water (50:50 [vol/vol]) was spiked in a salinized glass culture tube containing 500 μl plasma and 460 μl of phosphate buffer (pH 7.4) to obtain a final concentration of 2,000 ng/0.5 ml plasma. The total reaction volume was 1 ml, and the incubation was carried out at 37 ± 0.1°C. At appropriate time intervals (0, 0.5, 1, 1.5, 2, 2.5, and 3 h), the samples were withdrawn and immediately stored in a −70°C freezer until analysis by HPLC-ECD. At the time of analysis, samples were removed from the freezer and thawed at 4°C. Twenty microliters of the internal standard stock solution (40 mg/liter) was added and extracted, according to the method of Navaratnam et al. (15), and subsequently analyzed by HPLC-ECD.

Statistical analysis.

Data were expressed as mean ± standard deviation (SD) and analyzed using two-tailed Student's t test with a level of significance of P < 0.05 or P < 0.01.

RESULTS

Blood components reduced DHA effectiveness against P. falciparum.

To investigate the effects of blood components on DHA activity against P. falciparum, DHA (2 μM) was incubated for 18 h at 37°C in PBS, erythrocyte lysate, or plasma. The role of Fe(II)-heme in DHA activity was investigated by flushing erythrocytes with CO, which firmly binds Fe(II)-heme (16), thus inhibiting its reactivity. Fresh DHA, weighed and dissolved immediately before the assay, was used as control. As shown in Fig. 1A, DHA activity was reduced after incubation in all the conditions used. In particular, DHA activity was almost completely lost after incubation in plasma and to a lesser extent in erythrocyte lysate; the presence of CO partially restored activity. When human serum was used instead of human plasma, the results were comparable, indicating that the reduction in drug activity may not be ascribed to anticoagulants. When DHA was incubated in plasma serially diluted in PBS, the reduction in DHA activity was inversely related to the dilution (Fig. 1B).

FIG 1.

DHA (2 μM) was incubated for 18 h in the presence of PBS, normal (HbO) and CO-flushed (HbCO) RBC lysate, and plasma (A) or in the presence of plasma diluted with PBS at different plasma/PBS ratios (B). P. falciparum (W2 strain) cultures were then exposed to the drug for 48 h. Fresh DHA dissolved immediately before the assay was used as the control. A representative experiment is shown.

To verify if the loss in drug activity was related to the peroxide moiety, artesunate (extensively converted to DHA [17]) and chloroquine (a 4-aminoquinoline) were incubated in PBS or in the blood components for 18 h before testing their activity on parasite. Table 1 shows the mean IC₅₀ of the drugs and the ratio between the IC₅₀ of the fresh drug and the IC₅₀ of the drug after incubation in blood components. Chloroquine was only marginally affected (the 7% to 20% drop in activity across the different conditions was not statistically significant). The effects of PBS and HbCO on artesunate were mild, but marked reductions were observed in plasma, though they were proportionally less than for DHA.

TABLE 1.

Antiplasmodial activities of DHA, artesunate, and chloroquine following exposure to different conditions

Condition	DHAa IC50 (nM)	Fresh DHA IC50/DHA IC50 after incubation	Artesunatea IC50 (nM)	Fresh artesunate IC50/artesunate IC50 after incubation	Chloroquinea IC50 (nM)	Fresh chloroquine IC50/chloroquine IC50 after incubation
Fresh drugb	2.6 ± 1.1		4.7 ± 1.2		436.2 ± 97
PBS	7.6 ± 4c	0.34	5.3 ± 1.5	0.87	548.8 ± 55	0.79
HbO	19.4 ± 8.5c	0.13	15.4 ± 6.4c	0.3	551.7 ± 31	0.79
HbCO	11.9 ± 5.3c	0.22	7.8 ± 0.9c	0.59	544.2 ± 35	0.8
Plasma	>200		34.9 ± 1.7d	0.13	464.5 ± 107	0.93

^aData are the mean IC₅₀ ± SD of three different experiments in duplicate.

^bDrugs (2 μM) were used as fresh preparations or incubated for 18 h at 37°C in PBS, normal or CO-flushed RBC lysate (HbO or HbCO), and plasma. P. falciparum (W2 strain) cultures were then exposed to the drugs for 48 h and the parasite viability was assessed by the pLDH method.

^cP < 0.05 versus fresh drug.

^dP < 0.01 versus fresh drug.

DHA activity was also reduced after 18 h of incubation in the serum-enriched medium usually used for in vitro cell or parasite cultures, namely, RPMI containing 10% fetal calf serum or 1% AlbuMax (Table 2). DHA activity was partially preserved at 4°C or at room temperature. These temperatures were chosen to reproduce common conditions for handling drugs in in vitro screening assays.

TABLE 2.

DHA activity after incubation in commonly used serum-enriched cell culture media at different temperatures

Medium	IC50 (nM)a at:
	4°C	Room temp	37°C
Fresh DHAb	2.73 ± 1.5
RPMI-AlbuMax	7.04 ± 1.7	9.73 ± 0.8c	118.1 ± 49c
RPMI-FBS	6.23 ± 1.7	10.6 ± 1.4c	108.6 ± 10d

^aData are the mean IC₅₀ ± SD of three different experiments in duplicate.

^bDHA (2 μM) was used as a fresh preparation or incubated for 18 h in a medium containing 10% fetal bovine serum (FBS) or 1% AlbuMax at three different temperatures. P. falciparum (W2 strain) cultures were then exposed to the drugs for 48 h and the parasite viability was assessed by the pLDH method.

^cP < 0.05 versus fresh DHA.

^dP < 0.01 versus fresh DHA.

DHA stability at different pHs and in plasma.

Table 3 provides the values of observed degradation rate constants (k_obs) and half-lives (t_1/2) at the various pH values and at 37°C. The pH rate profile curve for DHA given in Fig. 2 is U shaped, which is typical of reactions that are susceptible to specific acid-base catalysis. Assuming there is no appreciable catalytic effect of the buffer systems used in the study, the observed rate constant generally obeys the equation k_obs = k_H+ [H⁺]^m when the catalysis is largely by hydrogen ions and k_obs = k_OH− [OH]ⁿ when catalysis is mainly by hydroxyl ions, where k_H+ is the rate constant for hydrogen-ion-catalyzed reaction, k_OH− is the rate constant for the hydroxyl-ion-catalyzed reaction and m and n are constants. The values of k_H+, m, k_OH−, and n were obtained by plotting log k_obs versus pH in the low (pH <2.2) and high (pH >7) regions and were 1.62 × 10⁻³ Lmol⁻¹ s⁻¹, −1.19, 1.43 × 10⁻¹¹ Lmol⁻¹ s⁻¹, and 0.86, respectively.

TABLE 3.

Observed degradation rate constants and half-lives of DHA at various pHs at 37 ± 0.1°C

DHA pH	k (s−1)	t1/2 (min)	t1/2 (h)
1	1.02 × 10−4	113	1.9
1.2	6.22 × 10−5	186	3.1
1.5	2.52 × 10−5	458	7.6
2.2	3.85 × 10−6	3,001	50.0
3	7.76 × 10−7	14,887	248.1
6	1.32 × 10−6	8,752	145.9
7	1.62 × 10−5	713	11.9
7.4	3.48 × 10−5	332	5.5
8	1.04 × 10−4	111	1.9
8.6	3.99 × 10−4	29	0.5
Plasma	8.55 × 10−5	135	2.3

FIG 2.

pH degradation rate profile of DHA. k is the rate constant per second.

In plasma, the rate constant k for degradation was higher than that of the corresponding buffer solution at pH 7.4: 8.55 × 10⁻⁵ s⁻¹ versus 3.48 × 10⁻⁵ s⁻¹, respectively. The corresponding t_1/2 values were 5.5 h (332 min) and 2.3 h (135 min).

DHA decay followed pseudo-first-order kinetics at constant temperature and ionic strength over the pH range of 1.0 to 8.6 (beyond which the reaction was rapid and the procedure unsuitable for determining the rate constant). This pseudo-first-order dependence is illustrated in Fig. 3, which shows that the log concentration versus time plots are linear and is confirmed by the absence of a significant influence of drug concentration on k_obs or t_1/2 in the range of 20 mg/liter (k = 6.22 × 10⁻⁵ ± 2.08 × 10⁻⁷ s⁻¹; t_1/2 = 11,138 ± 37 s) to 50 mg/liter (6.74 × 10⁻⁵ ± 2.08 × 10⁻⁶ s⁻¹; t_1/2 = 10,290 ± 312 s).

FIG 3.

Observed pseudo first-order plots for the hydrolysis at 37°C. Log DHA concentration over time.

Antiplasmodial DHA activity depends on temperature and pH.

DHA was incubated in plasma at 37°C or 40°C (to mimic increased body temperature during a malaria attack) for 3 and 6 h. Longer incubation was not considered since DHA activity was completely lost after 18 h in plasma (see Fig. 1A). Table 4 shows a time-dependent loss of DHA activity from 3 to 6 h incubation in plasma at 37°C and 40°C; the residual activity was <50% and decreased to 15% after 3 and 6 h of incubation, respectively (Fig. 4A), which is in agreement with DHA half-life calculated in plasma (2.3 h). The loss of activity was increased by higher temperature (40°C).

TABLE 4.

Effects of short incubation times and higher temperatures on DHA activity

Medium	IC50 (nM)a
	3 h		6 h
	37°C	40°C	37°C	40°C
Fresh DHAb	2.4 ± 0.5
PBS	3.31 ± 1.3	7 ± 1.9	4.3 ± 2.1	NDc
Plasma	6.13 ± 1.9d	12.8 ± 1.8d	10.08 ± 4.1d	38.2 ± 3.5d

^aData are the mean IC₅₀ ± SD of three different experiments in duplicate.

^bDHA (2 μM) was used as a fresh preparation or incubated for 3 or 6 h in PBS or in plasma at 37°C or 40°C. P. falciparum (W2 strain) cultures were then exposed to the drugs for 48 h, and the parasite viability was assessed by the pLDH method.

^cNo data.

^dP < 0.05 versus fresh DHA.

FIG 4.

(A) Residual activity of DHA (IC₅₀ of fresh DHA/IC₅₀ of DHA after incubation) in PBS or in plasma at different temperatures and times. (B) Correlation-regression of k_obs (degradation rate constant per second) and biological activity (IC₅₀ in nanomolar concentration) in the pH range of 7.2 to 7.6.

To better investigate small variations of pH around the physiological 7.4, DHA was incubated in phosphate buffer at pH 7.2, 7.4, and 7.6 at 37°C for 6 h, showing a progressive reduction of activity from pH 7.2 to 7.6 (Table 5). A high degree of correlation was found between IC₅₀ and k_obs in the pH 7.2 to 7.6 range, clearly indicating that the loss in DHA activity is related to increased degradation (Fig. 4B).

TABLE 5.

Influence of pH on antiplasmodial DHA activity

Fresh or incubated DHAa	IC50 (nM)b	Ratio IC50 fresh DHA/IC50 DHA after incubation
Fresh DHA	2.8 ± 0.7
pH 7.2	7.79 ± 1c	0.36
pH 7.4	10.6 ± 1.1c	0.26
pH 7.6	14.3 ± 3.5d	0.20

^aDHA (2 μM) was used as a fresh preparation or incubated for 6 h at pH values of 7.2 to 7.6. P. falciparum (W2 strain) cultures were then exposed to the drug for 48 h and the parasite viability was assessed by the pLDH method.

^bData are the mean IC₅₀ ± SD of three different experiments in duplicate.

^cP < 0.01 versus fresh DHA.

^dP < 0.05 versus fresh DHA.

Reductants promote DHA degradation.

The presence of the antioxidant ascorbic acid (or N-acetylcysteine, data not shown) in the RBC lysate strongly reduced DHA activity in the presence of normal hemoglobin (HbO) (Table 6). CO binding to Fe(II)-heme conferred partial protection against this loss in activity. Instead, adding ascorbic acid to plasma had no effects on DHA activity. Three-hour incubation times were used in these experiments, as DHA activity dropped rapidly in the presence of ascorbic acid.

TABLE 6.

Effect of ascorbic acid on DHA activity

Medium	Control		+Ascorbic acid (0.5 mM)		+Ascorbic acid (1 mM)
	IC50 (nM)a	Fresh DHA IC50/DHA IC50 after incubation	IC50 (nM)	Fresh DHA IC50/DHA IC50 after incubation	IC50 (nM)	Fresh DHA IC50/DHA IC50 after incubation
Fresh DHAb	2.25 ± 0.3
HbO	4.74 ± 0.2c	0.47	18.91 ± 6.1	0.12	>100
HbCO	5.01 ± 0.2c	0.44	8.97 ± 0.6	0.25	17.79 ± 2.3
Plasma	6.73 ± 2.3d	0.33	5.35 ± 2.3	0.42	5.57 ± 2	0.4

^aData are the mean IC₅₀ ± SD of three different experiments in duplicate.

^bDHA (2 μM) was used as a fresh preparation or incubated for 3 h in blood components in the presence of ascorbic acid. P. falciparum (W2 strain) cultures were then exposed to the drug for 48 h, and the parasite viability was assessed by the pLDH method.

^cP < 0.01 versus fresh DHA.

^dP < 0.05 versus fresh DHA.

DISCUSSION

The implications of the findings of this paper cover the use of DHA in clinics, the measurement of levels in biological fluids, and in vitro sensitivity assays. We detected a drop in DHA activity against P. falciparum in vitro after incubation in different conditions that broadly mimic physiological and pathological situations occurring during malaria infection. We also identified conditions of in vitro assays, which are likely to influence the interpretation of DHA in in vitro data. The dose of DHA used in the experiments (2 μM) is biologically relevant since it is in the range of maximal concentration of drug in serum (C_max) detected in the plasma of adults patients after oral or parenteral administration of artesunate (18).

Consistent results were obtained in this paper between experiments measuring chemical degradation and biological activity, confirming that loss in activity and decomposition of DHA are related. The chemical degradation and the corresponding reduction in biological activity are pH, time, and temperature dependent. Of all the conditions tested, plasma (and serum; data not shown) had the largest effect on drug activity.

Changes in DHA activity on P. falciparum were consistent with in vitro degradation rates, which were tested over pH values ranging from highly acidic to slightly basic as well as in human plasma. DHA appears to be more prone to decomposition than artesunate around neutral pHs and in plasma. Artesunate itself is rapidly and extensively transformed into DHA in vivo (17). In experimental conditions, the artesunate half-life (t_1/2) at pH 7.4 is 10.8 h, and in plasma it is 7.3 h (P. Olliaro and V. Navaratnam, unpublished data) compared with 5.5 and 2.3 h for DHA. This means that after 18 h incubation of artesunate, assuming a linear decay throughout, only about 31% and 18% of the drug will still be artesunate itself (the rest will be DHA and its further degradation products [6]). This translates in 89% of the activity of the parent drug remaining after incubation in PBS for 18 h for artesunate versus 38% for DHA; the respective values in plasma are 17% for artesunate and 0% for DHA.

Here, we show that DHA is stable within the pH range 2 to 6 (conditions which apply to the stomach and small intestine where the drug is absorbed), while pHs of <2 and >6 promote DHA degradation. This U-shaped pH profile is typical of reactions that are susceptible to specific acid-base catalysis. The reaction follows pseudo-first-order kinetics with a log-linear decay over time up to pH 8.6.

A possible limitation of this study is that it measured the disappearance of DHA but did not characterize the formation and disappearance of other species, notably the rearranged bioactive peroxyhemiacetal metabolite (6). The parallel between chemical decay and biological activity on P. falciparum should, therefore, allow for all the bioactive species.

To mimic conditions that may occur in vivo, we incubated DHA in phosphate buffer around the physiological pH (7.2, 7.4, or 7.6) and found a high correlation between the speed at which degradation occurs and loss in activity. DHA was relatively more stable and more active at pH 7.2, under conditions such as those encountered in malaria-related acidosis, for instance, than at pH 7.4. At pH 7.2, the DHA t_1/2 in buffer was ca. one-third longer (8.1 h instead of 5.5 h), and potency was ca. one-fourth higher (IC₅₀, 7.8 versus 10.6). This difference between the t_1/2 and IC₅₀ ratios was likely accounted for by the presence of other bioactive species, such as the above-mentioned peroxyhemiacetal metabolite. However, if acidosis might drive the equilibrium one way, intravascular hemolysis may compromise DHA activity by liberating Fe(II)-heme. Hemolysis, another condition related to severe manifestations of malaria, was reproduced in the present work by incubating the drug in erythrocyte lysate. Previous studies show that Fe(II), but not Fe(III), can decompose artemisinin (19, 20) and that peroxidic antimalarials are stable in the presence of hemoglobin but react with free heme (21). In this work, further evidence of the role of Fe(II)-heme is that flushing RBC lysate with CO, which tightly binds Fe(II)-heme inhibiting its reactivity, moderated the effects on DHA activity. These data are in agreement with a previous study, wherein it was observed that the activity of artemisinin antimalarials is increased when parasites are grown in the presence of CO and that artemisinins are unaffected by carboxy-hemoglobin or CO-heme but are decomposed by Fe(II)-hemoglobin or Fe(II)-heme (13, 22). Thus, we can expect DHA to be degraded faster and, therefore, be less active when intravascular hemolysis occurs. Also supporting the role of Fe(II)-heme is the finding that the introduction of ascorbic acid, a reducing agent, during DHA incubation in normal erythrocyte lysate (but not in CO-erythrocyte lysate or plasma in the absence of hemolysis) further reduces DHA activity. Ascorbic acid would reduce the pool of Fe(III) to Fe(II), which then in turn would degrade DHA.

The results of this study also have practical methodological implications for pharmacokinetic and in vitro drug assays. They support the idea of using methods for artemisinin determination in blood that employ oxidizing agents (i.e., potassium dichromate or hydrogen peroxide) to stabilize blood and prevent degradation (23, 24). In addition, it is important to consider factors that may favor hemolysis and exposure to heme products. These results are also important for the conduct and interpretation of in vitro assays; DHA activity dropped after incubation in the media containing fetal calf serum or AlbuMax, which are routinely used for in vitro culture studies. It is therefore important to use the drug immediately after preparation or keep the drug under conditions that prevent loss of activity (e.g., stock solution in ethanol at 4°C [25]).

Finally, DHA decomposition may bear on the rise of artemisinin resistance; chemical decomposition, with consequent decay in antimalarial activity and parasites being exposed to suboptimal DHA concentrations, may select for drug-resistant parasites in patients treated with DHA itself or artemisinin derivatives metabolized to DHA.