Voltammetric methods for electrochemical characterization and quantification of artemether-based antimalarials

Every year substandard and falsified (SF) artemisinin derivative-based antimalarials are responsible for the loss of 450 000 deaths and billions of GBP. The lack of infrastructure and funds to support pharmaceutical quality control in many low-and-middle-income countries contributes to this problem. This work assesses fitness for purpose of voltammetric methods for identification and quantification of artemether in the presence of excipients. Electrochemical characterization of artemether using cyclic voltammetry shows that the reduction of artemether is chemically irreversible within the potential range of −0.4 V to −1.4 V. A chronocoulometric quantification algorithm for artemether is created and tested with pure artemether, as well as filtered and unfiltered Riamet® tablets. Filtration of Riamet® tablets provides no additional benefit for the quantification of artemether in Riamet®. In addition, artemether's response to pH indicates possible protonation and coupled homogeneous chemistry. Finally, sodium sulfite is an effective means of removing dissolved oxygen and improving artemether signal resolution in air-equilibrated PBS. This concludes that electrochemical analysis is a promising method for artemether identification and quantification.


Introduction
The expected range in concentration of dissolved oxygen is necessary to determine as signal interference from dissolved oxygen is expected as the reduction potential of oxygen on glassy carbon electrodes (-0.6 V vs. Ag/AgCl 8 ) occurs near the reduction potential of artemether (-1.2 V vs. Ag/AgCl) in phosphate buffer with a pH of 7.55.Nevertheless, there is a large overvoltage for the reduction of oxygen, therefore potentials significantly more negative than E0 for oxygen are required for measurable analyte current 9 .
The concentration of dissolved oxygen can be predicted with the Clausius-Clapeyron equation, which describes the relationship between temperature and vapor pressure of gasses 10 : Where ∆ '() is the enthalpy of vaporization of the gas (  *% ).
Henry's Law constant ( # in M/atm) relates the partial pressure of a species in the gas phase (pg in atm) with the concentration of that species in the aqueous phase (ca in M) 11 : 2)  # ≝  ( / + Combining Henry's Law constant with the Clausius-Clayeron equation yields the van't Hoff equation 11 : Where  # ⊖ represents Henry's Law constant under standard conditions ( ⊖ = 298.15K).
The temperature dependance of dissolved oxygen can be explained by Henry's Law as a function of temperature 11 : 4) The partial pressure of oxygen can be calculated with the following relationships 12 : The partial pressure of oxygen will vary based on temperature, altitude, and relative humidity.Assuming high altitude (3,000 m), high temperature (50°C) and 100% humidity,   2  is 12,300 Pa and   is 70,108 Pa 13 .Using the above equations, the concentration of dissolved oxygen is calculated as 92 μM.On the other hand, at sea level with a low temperature (5°C),   2  is defined as 860 Pa 14 , so the concentration of dissolved oxygen is calculated as 375 μM.Thus, the concentration of dissolved oxygen will vary from 92 μM to 375 μM.

a) Supporting electrolyte selection
Phosphate-buffered saline (PBS) with a neutral pH was chosen for the supporting electrolyte solution.PBS was selected because of its long-term stability in environments with a high ambient temperature (above 40°C) and humidity 15 .

b) Solvent Selection
Dimethyl sulfoxide (DMSO) was selected as the solvent as it maximizes the signal from the API in the presence of excipients.
Literature states the first step of the API recovery process is to dissolve the drug in a solvent which is a good solvent for the API and a poor solvent for as many excipients as possible 16 .This will allow the API to dissolve in the solvent while the excipients are isolated as solids.This solid-liquid extraction helps to purify the API from the excipients when the drug is filtered.
Artemisinin and its derivatives are reported to be sparingly soluble in aqueous buffers, and literature recommends that the artemisinin derivative is first dissolved in a solvent before dilution in the buffer 17,18 .For the solvent to be accessible for low-resource settings, the solvent must be non-toxic, not require refrigeration and be readily available in target markets.
Although dimethylformamide and methanol are good solvents for artemether 19 , they are not ideal for field use due to their known toxicity.Ethanol and dimethyl sulfoxide (DMSO) both meet the defined criteria and are potential solvents for artemether 19,20 .Zhang et al. reports successful dissolution of the artemisinin with ethanol 21 .Ethanol is a slightly better solvent for solid-liquid extraction of artemether from excipients than DMSO, as it is a poor solvent for 5/7 and 4/7 excipients respectively 22 .
Artemether was found to be more stable in ethanol (degradation after 60 minutes) than in DMSO (degradation after 30 minutes).While ethanol showed promising results with pure artemether, DMSO was ultimately selected as the signal from Riamet® dissolved with ethanol was poor.This could be due to ethanol's inability to break down particles of Riamet® tablets and free the API into solution.
The concentration of the API/DMSO stock solution was 16.757 mM.This concentration was chosen because lower concentrations (8.379 mM) prevented the larger Riamet® particles from breaking down, while higher concentrations (33.515 mM) caused artemether to decay quickly, both of which resulted in poor signal from artemether.

Comparison of pure artemether, filtered Riamet® tablets and unfiltered Riamet® tablets with cyclic voltammetry and chronocoulometry
An AM-LUM formulation was chosen because it is the most widely prescribed ACT in Sub-Saharan Africa 23 , as well as the availability of Riamet® in the UK 24 .

a) Filter selection
A 0.22 µM Millipore filter was selected based on the particle size of the excipients and the signal generated from the filtered drug solution.The ideal filter selected for API extraction is a filter with a pore size smaller than the excipients but larger than the analyte 16 .The particle size for the target analyte, artemether, is 1.14 µM.Artemether particles are smaller than six out of the seven excipients found in Riamet® tablets.Therefore, ideal pore size should fall within the range of 2 µM to 60 µM.S3 Comparison of variability for intercept (adsorption) and slope (sensitivity) for artemether, filtered Riamet® and unfiltered Riamet® for quantification with total charge and Anson slope.3.5 Sodium sulfite in air-equilibrated PBS effect on artemether with cyclic voltammetry

Dissolved oxygen effect on artemether with cyclic voltammetry
Fig. S2 Reduction peak current density and reduction peak potential extracted from Fig. S1 with scan rates from 50 mV/s to 500 mV/s.(left) Peak current density as a function of square root scan rate.(right) peak potential as a function of log scan rate.

Fig. S1
Fig. S1 Raw cyclic voltammetry scans from 0.20 mM artemether in nitrogen-equilibrated PBS (pH 7.54) with scan rates varying from 50 mV/s to 1000 mV/s

Fig. S3
Fig. S3 Blank cyclic voltammetry scans for air-, nitrogen-and oxygen-equilibrated PBS with pH of 7.55

Table S4
Comparison of means for intercept (adsorption) and slope (sensitivity) for artemether, filtered Riamet® and unfiltered Riamet® for quantification with total charge and Anson slope.

Table S5 &
S6 Pairwise comparisons with each sample type with normalized linear regression models for total charge and Anson slope.