Chemicals
Artesunate drug substance was obtained from IPCA (Mumbai, India) with purity above 96.0% complying with the International Pharmacopoeia. HPLC gradient grade methanol was purchased from Fisher Scientific (UK). An Arium 611 purification system of Sartorius (Germany) was used to purify water at 18.2 MW.cm quality. Disodium phosphate was obtained from Merck (Germany). Formic acid (Riedelde Haën, Germany), mannitol (SigmaAldrich, USA), ammonium formate and phosphoric acid (Fluka, Switzerland) were all analytical grades.
Method development and validation
HPLC method
UPLCUV analyses were carried out on a Waters Alliance 2695 separation module equipped with a Waters 2487 Dual labsorbance detectors (all Waters, USA). The artesunate content of the products was determined at 210 nm. Data handling was performed with Empower 2 software. In a standard HPLC run, 10 mL was injected and separated on a HALO RPC18 column (50 × 4.6 mm, 1.7 µm solid core particle with a 0.5 mm porous silica layer fused to the surface) at a flow rate of 1.1 mL/min (Advanced materials technology, Wilmington, USA). An isocratic LC separation with a mobile phase consisting of a mixture of 45% (A) 10 mM m/V ammonium formate in water at pH 4.5, and 55% (B) methanol, was performed maintaining the column at 25 ± 5 °C.
Preparation of reference solutions
A stock solution was prepared by dissolving the accurately weighed 400 mg of artesunate in 50 ml 45/55% V/V 10mM ammonium formate /MeOH. The solution was further diluted with mobile phase to prepare in a range of 10120% label claim (l.c.) A 100% reference solution corresponds to 3.33 mg/mL artesunate.
Sample preparation
Samples of artesunate aqueous solution (250 µL) from each formulation containing mannitol and phosphate buffer were taken. The solutions were then centrifuged at 17000 g for 4 min at 5 °C to precipitate the phosphate in order to protect the silicabased analytical column and thus prolong its lifetime. Subsequently, the supernatant was collected and diluted with methanol and then acidified with 0.1 M formic acid to obtain a 100% label claim corresponding to 3.33mg/ml in 50/50% V/V MeOH/purified water acidified with formic acid.
Validation procedure
Prior to the validation process, the analytical HPLC method suitability test was carried out by injecting six replicates of the standard sample (3.33 mg/mL). The number of theoretical plates, resolution and tailing factors were assessed. Subsequently, the newly developed HPLC method was validated for specificity, linearity and range, accuracy, precision, limit of detection (LoD), and limit of quantitation (LoQ) in accordance with ICH guidelines (Q2A and Q2B).
Specificity
The presence of hydrolytic products, βDHA, and αDHA, close to artesunate is critical during its analysis. Thus, the specificity was determined in the presence of the matrix and related compounds, i.e a placebo, a placebo spiked sample, and a spiked placebo with DHA.
Linearity
Serial artesunate solutions were prepared and evaluated over the concentration range of 10 to 120% l.c. The calibration curve was constructed, and then the correlation coefficient, yintercept, and the slope of the regression line were obtained.
Accuracy
Accuracy was determined by spiking artesunate reference solution into the placebo samples. It was established across the specified range of the analytical procedure at 10, 50, and 100% l.c. with 3 replicated injections of each spiked placebo.
Repeatability/precision
Repeatability and intermediate precision of the method were investigated by injecting the sample solutions in triplicate at three concentration levels (3.33mg/ml, 1.66mg/ml, and 0.33mg/ml) consisting of a total of 9 and 18 chromatogram runs, respectively. The RSD of the assay results was calculated at each concentration level.
Limit of detection (LoD) and limit of quantitation (LoQ)
LoD and LoQ were calculated out of the calibration curve by using the following equations:
\(LoD=\frac{3.3*\sigma }{S} and LoQ= \frac{10* \sigma }{S}\) Eq. 1
s is the standard deviation of the yintercept of the regression line while S is the slope of the calibration curve of the analyte.
Artesunate product development for degradation kinetics study
Artesunate (40 mg/mL) was dissolved in the aqueous solution of phosphate buffer with or without mannitol according to the experimental design (Table 4). Each formulated artesunate solution was incubated at predefined temperatures in a Max Q 4000 (Thermo Scientific, San Jose, USA) at 25 and 40 °C and in a refrigerator at 5 °C. Aliquots of artesunate solutions were withdrawn at different intervals, and assayed as described earlier.
Kinetic evaluation of the hydrolysis of artesunate
The kinetics were determined by taking as an example the artesunate solution containing 0.3 M phosphate buffer at pH of 9 in presence of 0.22 mmol/mL mannitol. At the three different temperatures experimentally explored, the hydrolysis kinetics, and the observed rate constants k were determined. Arrhenius equations (using Eq. 2) were applied to determine the activation energy Ea and the frequency factor A, as well as to calculate the halflife (t1/2), and the shelflife (t0.9) of the formulation.
\(\text{l}\text{n} \text{k}=\text{l}\text{n} \text{A}\frac{\text{E}\text{a}}{\text{R}\text{T}}\) Eq. 2
Where k = specific rate constant, Ea= activation energy, A = frequency factor T = temperature (Kelvin).
In addition, the effect of buffer strength on the degradation rate of artesunate in aqueous solution was evaluated at different pH values and temperatures. A Design of Experiments (DOE) approach was hence applied using three controlled independent variables, i.e. buffer strength, pH, and mannitol concentration. The former two variables were set at 3 levels while the last variable was studied at 2 levels. The investigated levels are given in Table 1. Hydrolysis rate constants, the activation energy and frequency factor were selected as dependent factors or responses to finally model the stability of the drug. A total of 18 experimental units were thus executed in a random sequence to minimize uncontrolled influences on the estimated effects (Table S1).
Table 1. Factors and their levels investigated
Factor

Levels

1

0

+1

(A) Sodium phosphate buffer strength (mmol/mL)

0.30

0.40

0.50

(B) Ph

8

9

10

(C) Mannitol concentration (mmol/mL)

0


0.22

Statistical analysis and Optimization
Statistical analysis, modelling, and leastsquares linear regression were performed using Stata/SE Statistics for Windows, version 14, including manual Dixon and Grubbs’ tests for detection of outliers. After building the model, it was interpreted graphically by visualizing 2D contour plots and 3D response surface plots for each response using the Designexpert version 13 (USA). Furthermore, the model was interpreted statistically by the determination of the significance of the coefficients of the factors. The models were then refitted by only including the significant terms and deleting the nonsignificant terms.