Determination of an Anti‐Parasitic Active Pharmaceutical Ingredient in Wastewater Effluents Using Capillary Zone Electrophoresis

ABSTRACT

Ireland has a successful pharmaceutical industry with over 100 pharmaceutical manufacturing sites across the island. Although this success has many benefits, the irreversible effects emissions from pharmaceutical manufacturing can have on the environment are a major drawback. Although known pollutants are regularly monitored with limits set out by the Environmental Protection Agency, one significant pollutant has been overlooked: pharmaceutical pollution. Detecting these pollutants and ensuring they are at a safe concentration for the environment is of utmost importance. In recent years, capillary electrophoresis is being recognised as a suitable alternative to high‐performance liquid chromatography due to its many benefits such as faster analysis, water‐based buffers and smaller sample volumes. In this paper, a capillary zone electrophoresis (CZE) method with a preconcentration step of solid‐phase extraction was developed for an anti‐parasitic active pharmaceutical ingredient (API) called ZB23. The API was successfully detected in a wastewater sample in less than 10 min using the CZE parameters of 25 mM borate buffer with a pH of 10.5, 15% MeOH, 10 kV voltage, 25 mbar for 5 s injection size, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

Abbreviations

API

active pharmaceutical ingredient

Ld

length to the detector

Lt

total length of capillary

PNEC

predicted no‐effect concentration

1. Introduction

Ireland has a successful pharmaceutical industry with over 100 pharmaceutical manufacturing sites across the island. The Irish Central Statistics Office (CSO) reported that in 2023, the food, chemical and pharmaceutical sectors accounted for 71.2% of all production in Ireland. In 2023, the pharmaceutical industry reported the highest Net Selling Value (NSV) of €58.7 billion [https://www.cso.ie/en/releasesandpublications/ep/p-iips/irishindustrialproductionbysector2023/#:~:text=ThePharmaceuticalsectorreportedthe,2021]. Although this provides Ireland with significant employment and a healthy economy, the chemical emissions it produces could have irreversible effects on the environment if not properly managed. Chemicals, such as volatile organic compounds (VOCs), total organic carbons (TOCs) and heavy metals, are routinely monitored as part of licencing requirements set out by the Environmental Protection Agency (EPA); however, one significant pollutant has been overlooked: pharmaceutical pollution.

In recent years, research has shown the detrimental effects of pharmaceuticals on the ecosystem and any organism exposed to these pollutants. Phytoplankton that were exposed to pharmaceuticals in their ecosystem had adverse effects on their growth, photosynthesis and their cellular metabolism [1], fish who were exposed to anti‐depressants had altered behavioural traits which interfered with reproduction [2, 3], and zebrafish who were exposed to antibiotics from wastewater effluents had high mortality rates and numerous deformities [4]. These are just some of the lasting effects exposure to pharmaceutical pollution has had on these organisms.

Pharmaceuticals can enter the environment through numerous pathways such as human excretion, wastewater treatment plants, hospitals and surface runoff [5, 6, 7], but, according to the World Health Organization (WHO), wastewater effluents are a leading source of pharmaceuticals in the environment [https://pubmed.ncbi.nlm.nih.gov/21987209/]. Pharmaceuticals are difficult to fully remove from wastewater effluents partially due to their complex chemical structure as well as because wastewater treatment processes are not designed to remove them. Detecting active pharmaceutical ingredients (APIs) in wastewater effluents and ensuring that they are at a safe concentration for the health and well‐being of the environment will act as a suitable alternative. At the same time, research continues on how to thoroughly and safely remove them.

Research has begun to detect these pollutants in environmental waters in recent years. High‐performance liquid chromatography (HPLC) has been the chromatographic technique of choice due to its many benefits such as its high precision and selectivity [8]. It has proven in the literature to be very successful in the detection of a wide range of pharmaceuticals, including antibiotics [9, 10, 11, 12, 13], non‐steroidal anti‐inflammatories (NSAIDs) [14, 15, 16, 17], anti‐cancer [18, 19, 20, 21], hormones [22, 23, 24, 25, 26, 27] and anti‐depressants [28, 29, 30]. Although HPLC has shown its success in detecting APIs in wastewater, it requires a large amount of organic solvent, making the analytical method not environment‐friendly [11]. Capillary electrophoresis (CE) has been gaining attention as a suitable alternative to HPLC due to its many benefits such as little to no organic solvent, smaller sample volumes, minimal sample preparation, faster analysis times and its versatility to test a wide range of analytes [31, 32, 33]. CE has shown success in detecting pharmaceuticals in environmental waters with many publications in the last few years [31, 34, 35, 36, 37].

With its many benefits over HPLC, the CE has one major disadvantage: poor sensitivity. A pre‐concentration step, such as liquid–liquid extraction (LLE), can be added to the CE method to improve sensitivity. Solid‐phase extraction (SPE) has become a successful addition to CE methods in recent years, due to its many benefits over traditional LLE such as its versatility due to its many sorbents, which allows for cleaner eluents and higher recoveries [38].

This research is a collaboration between Hovione Ltd., a pharmaceutical contract development and manufacturing company and University College Cork. Hovione routinely uses HPLC in its analysis of the API. This research aims to develop a capillary zone electrophoresis (CZE) method to detect Hovione's anti‐parasitic API ZB23 in wastewater effluents. The effects of the buffer pH and concentration on the migration time of the API, the addition of an organic modifier to address solubility issues, and the determination of the voltage and injection size were investigated to provide an efficient and environment‐friendly method.

2. Materials and Methods

2.1. Chemicals

Sodium tetraborate decahydrate, sodium acetate, 1 M sodium hydroxide (NaOH), 1 M hydrochloric acid (HCl), phosphoric acid, methanol (MeOH), acetonitrile (ACN) and acetone were purchased from Sigma‐Aldrich (Germany). Sodium phosphate dihydrate was obtained from Fluka Analytical. All chemicals were of analytical and HPLC reagent grade. The target analyte ZB23 was collected at Hovione Ltd., Ireland.

2.2. Instrumentation

Results were obtained using an Agilent CE 7100 (Waldbronn, Germany) equipped with a diode‐array detector (DAD). Fused silica capillaries (Composite Metal Services, Shipley, UK) with a 50 µm inner diameter (id) and a 75 outer diameter (od) were used for each experiment. The new capillary was preconditioned using 1 M NaOH for 10 min, deionised (DI) water for 10 min and the BGE for 10 min. The capillary was preconditioned for each run using 1 M NaOH for 3 min, DI water for 3 min and the BGE for 3 min. The samples were injected into the CE using a hydrodynamic injection for 5 s at 50 mbar. The experiments were carried out using a voltage of 10 kV. The CE software used was Agilent's Chemstation B.04.02. The pH of all samples and BGE was measured using the Metrohm 654 with the microelectrode Metrohm 6.0234.100. A Milli‐Q Advantage A10 (Millipore, Molsheim, France) water purification system with Q‐POD dispenser provided ultrapure DI water at 18.2 Ω. The solutions were sonicated using a Branson 5510. For the SPE, samples were extracted using Agilent's Vac Elut 20 Manifold (Waldbronn, Germany) and a vacuum pump purchased from Waters (Dublin, Ireland). The Bond Elut PPL 6 mL, 500 mg cartridges and Bond Elut SPE Reservoir 60 mL were purchased from Agilent (Waldbronn, Germany), and the Strata C18‐E (55 µm, 70 Å) 500 mg, 6 mL cartridges were purchased from Phenomenex (Aschaffenburg, Germany). Test tubes of 6 mL were purchased from Agilent (Waldbronn, Germany).

2.3. Sample Preparation

2.3.1. CE Solutions

The BGE solutions were prepared in 100 mL volumetric flasks with ultrapure DI water and sonicated until the solution was fully dissolved. The pH of the BGE was measured using the Metrohm pH meter and altered using either 1 M NaOH or 1 M HCl. Individual solutions of ZB23 were prepared in a 25‐mL volumetric flask. A volume of 1.25‐mL (5%) methanol was added to help dissolve the analyte, and the BGE was added up to the mark. The solution was then sonicated for 5 min to degass. All solutions were prepared daily. The solutions were filtered through a 0.2‐µm regenerated cellulose syringe filter (Agilent, Ireland) before being placed into vials.

2.3.2. SPE Solutions

For the Bond Elut PPL cartridges, 6 mL MeOH and 6 mL DI water were used to condition and equilibrate. Both solutions were eluted by gravity. A volume of 200‐mL of a spiked wastewater sample was loaded in 60 mL intervals into the reservoir and eluted under a 15 Hg vacuum using a vacuum pump purchased from Waters (Dublin, Ireland). The cartridge was washed with 10 mL of ultrapure DI water and left to dry under the vacuum for 5 min. A 6 mL solution of formic acid in 1:1 MeOH:ACN was loaded in the cartridge and left to dry for 10 min. Once dry, a 6 mL solution of ammonia hydroxide in 1:1 MeOH:ACN was added to the cartridge. The formic acid and ammonia hydroxide solutions were collected in 6 mL test tubes and poured into a round bottom flask. The round bottom flask was placed in a 35°C sand bath under nitrogen gas until the SPE eluent crystallised. Once fully crystallised, the eluent was reconstituted using 0.2 mL of MeOH and 1.8 mL of ultrapure DI water.

For the Strata C18‐E cartridges, all solutions were extracted using a 15 Hg vacuum. The cartridges were first conditioned with 6 mL MeOH and equilibrised with 6 mL DI water. A 6 mL spiked wastewater sample was loaded into the cartridge and washed using a 6 mL solution of 5% MeOH and DI water. The cartridge was then left to dry under vacuum for 5 min. The eluent was extracted using 6 mL of MeOH and collected in a 6 mL test tube for analysis.

2.4. Wastewater Sample Preparation

Wastewater samples were collected from Hovione Ltd., Cork, Ireland and stored in the refrigerator at 4°C. Before analysis, the wastewater was measured at the desired volume and filtered through a Merck MF‐Millipore MCE Membrane Filter, 0.45 µm pore size to remove any particles.

3. Results and Discussion

3.1. Effect of BGE pH

CZE separates analytes on the basis of their size‐to‐charge ratios. The pH of the BGE is an important parameter in CZE as it influences the electroosmotic flow. As the inner wall of the capillary comprises SiO⁻ groups, the buffer pH will influence the ionisation of these silanol groups. At low pH, ionisation of these anions will be more difficult as the buffer contains fewer positive ions. Using a buffer with a high pH, the ionisation of the silanol groups occurs more easily due to a higher concentration of positive ions. This allows the creation of the electroosmotic flow in less time. It was expected that the migration time of the analyte decreases as the pH increases; however, during the analysis of buffer pH with ZB23, the migration time increased as the buffer pH increased. As the pH increases, there is a higher presence of cation ions in the capillary causing a decrease in the thickness of the double layer, therefore, decreasing the zeta potential and, therefore, decreasing the electroosmotic flow [39]. The effect of the buffer pH on ZB23's migration time is presented in Figure 1.

FIGURE 1.

An electropherogram presenting the effect of buffer pH on the migration time. Peak 1 = systematic peak, Peak 2 = 1.50 × 10⁻⁴ mol/L of ZB23 in 25 mM borate buffer with a pH of 8–10.5, a voltage of 10 kV, a current of 36 µA, a power of 0.4 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

The buffers were chosen on the basis of their pH ranges to analyse a wide pH range, and, therefore, sodium acetate (pH 3–4.5), sodium phosphate (pH 5–7.5) and sodium borate (pH 8–10.5) were selected. Both acetate and phosphate buffers were deemed unsuitable for the API due to solubility issues in low pH; therefore, the best results were produced using borate buffer. To determine the optimum pH, the number of theoretical plates for pH 8–10.5 was calculated. From Figure 2, pH 9 produced a higher number of theoretical plates with 64,618; however, pH 8–9.5 was ruled out as the peaks produced had a low peak height which affected the peak shape presented in Figure 1. pHs 10 and 10.5 produced a better peak height and better peak shape. From those two pHs, pH 10.5 was determined as the optimum buffer pH due to a higher number of theoretical plates of 55,104. The number of theoretical plates is presented in Figure 2.

FIGURE 2.

A bar chart presenting the number of theoretical plates calculated for the pH 8–10.5. These results were achieved using 1.41 × 10⁻⁴ mol/L of ZB23 in 25 mM borate buffer, a voltage of 10 kV, a current of 22 µA, a power of 0.2 W, an Ld of 31.5 cm, an Lt of 40 cm and a detection wavelength of 214 nm.

3.2. Effect of BGE Concentration

Determining the optimum BGE concentration is important for the detection of ZB23 as it influences the electroosmotic flow (EOF). An increase in BGE concentration lowers the electroosmotic flow as it lowers the zeta potential due to a higher presence of cation ions in the capillary causing a decrease in the thickness of the double layer. As the BGE concentration increases, the migration time of the API increases which is presented in Figure 3.

FIGURE 3.

An electropherogram presenting the effect of the BGE concentration of the migration time of the API. Peak 1 = systematic peak, Peak 2 = 1.50 × 10⁻⁴ mol/L of ZB23 in 10, 15, 20 and 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 9, 12 and 19 µA, a power of 0.1 and 0.2 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

The BGE concentration also influences the current which is presented in Figure 4. Due to Ohm's law, increasing buffer concentration decreases the resistance of the buffer therefore increasing the current. A high current can lead to peak broadening which decreases the separation efficiency [40]. For this reason, the concentrations investigated were 10, 15, 20 and 25 mM borate buffer. All four concentrations provided successful detection of the API. Although the 10, 15 and 20 mM provided faster migration times, the 25 mM BGE provided the higher number of theoretical plates; therefore, 25 mM was chosen as the optimum BGE concentration.

FIGURE 4.

A scatter graph presenting the effect of buffer concentration on the current. These results were achieved using 1.50 × 10⁻⁴ mol/L of ZB23 in 10, 15, 20 and 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 9, 12 and 19 µA, a power of 0.1 and 0.2 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

3.3. Organic Modifier

ZB23 had solubility issues with the buffer; therefore, the addition of an organic modifier was investigated to improve the solubility of the analyte in the buffer. To keep the solutions as green as possible, the organic solvents selected were on the basis of the Pfizer solvent selection guide [41]; therefore, 5%, 10% and 15% methanol, acetonitrile and acetone were selected. Although all three organic solvents provided excellent results, 15% MeOH provided ZB23 with the highest number of theoretical plates as presented in Figure 5 and, therefore, chosen as the optimum organic modifier.

FIGURE 5.

Bar chart presenting the number of theoretical plates calculated in 5%, 10% and 15% v/v MeOH, ACN and acetone. These results were achieved using 1.50 × 10⁻⁴ mol/L of ZB23 in 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 19 µA, a power of 0.2 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm. CAN, acetonitrile; MeOH, methanol.

3.4. Optimisation of Voltage

Voltage influences the migration time of the analyte as it varies the electric field. This causes a change in the electroosmotic flow. As the voltage increases, the electroosmotic flow increases, therefore, reducing the migration time of the analyte. The effect of voltage on migration time is presented in Figure 6.

FIGURE 6.

Electropherograms of the voltage optimisation analysis overlayed by 150 mAu for the CZE method. Peak 1 = 1.40 × 10⁻⁴ mol/L of ZB23 in 15% MeOH and a 25 mM borate buffer with a pH of 10.5, a voltage of 5–15 kV, a current range of 17–55 µA, power range of 0.1–0.8 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

Voltage also influences the amount of current generated due to Ohm's law. Increasing the voltage, which increases the current, can cause an increase in heat production due to Joule heating. The effect of voltage on the current is presented in Figure 7. If the heat produced is not dispersed the temperature inside the capillary will increase which decreases the viscosity of the buffer and allows more current to flow. A voltage range of 5–20 kV was investigated. The run for 20 kV had to be stopped during the analysis as the current was very unstable which was caused by Joule heating on the capillary. The number of theoretical plates was calculated, and 10 kV was chosen as the optimum voltage due to the higher number of theoretical plates.

FIGURE 7.

A scatter graph presenting the effect of the voltage on the current. These results were achieved using 1.50 × 10⁻⁴ mol/L of ZB23 in 25 mM borate buffer with a pH of 10.5, 5% v/v MeOH, a voltage of 5, 10 and 15 kV, a current of 17, 35 and 55 µA, a power of 0.2, 0.3 and 0.4 W, an injection size of 50 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm.

3.5. Optimisation of the Injection Plug

It is recommended that the length of the injection plug should be below 1% of the total capillary length. Using the Hagen–Poiseuille equation, the injection volume and length were determined for 5, 10, 15, 20, 25, 50 and 75 mbar (Table 1):

$$\text{Hagen-Poiseuille equation:}{V}_i=\frac{\mathrm{\Delta }P{d}^4\pi tinj}{128\eta L}.$$

TABLE 1.

Table of results for the injection plug volume and length.

Injection vacuum (mbar) for 5 s	5	10	15	20	25	50	75
Volume of injection (nL)	1	2	3	4	5	10	14
Length of injection plug (mm)	0.10	0.20	0.40	0.50	0.60	1.20	1.80
Total length of capillary (mm)	400	400	400	400	400	400	400
% of total capillary	0.03	0.05	0.10	0.13	0.15	0.30	0.45

From the calculations, it was determined that all seven injection sizes were below 1% of the total capillary length. To determine the optimum injection plug, the number of theoretical plates was calculated for 5, 10, 15, 20, 25, 50 and 75 mbar. Increasing the injection size, the number of theoretical plates decreased. This is due to the hydrodynamic injection where the increase in pressure results in an increase of volume injected into the capillary. This results in an increase in the peak areas with an increase in the pressure applied. Pressure values of 5, 10, 15, 20 and 75 mbar were ruled out because of the effect on the peak shape. From 25 and 50 mbar, 25 mbar resulted in the highest number of theoretical plates and, therefore, was chosen as the optimum injection size.

3.6. CZE Method Validation

The repeatability of the method was analysed by running the method five times in a row and obtaining the relative standard deviation (RSD) of the migration time and the peak area of NB26 on the same day (inter) and a different day (intra‐repeatability). The reproducibility was evaluated by determination of the RSD. Calculating an RSD% of less than 2% indicates a reliable method was developed. The RSD of both the migration time and peak area for both the inter‐ and intra‐repeatability analyses were determined to be less than 2%. The limit of detection (LOD) and the limit of quantification (LOQ) of this CZE method were determined to be 2.25 × 10⁻⁷ and 7.44 × 10⁻⁷ mol/L, respectively. The LOD detected in this method is a lower concentration than ZB23's predicted no‐effect concentration (PNEC) of 3.83 × 10⁻⁷ mol/L, as stated on its safety data sheet (SDS).

3.7. Analysis of Spiked Wastewater Sample

Wastewater samples were collected from Hovione Ltd. and analysed using the developed CZE method. A blank of ultrapure DI water was analysed first to give a smooth baseline. A wastewater sample was analysed next to determine any peaks that could potentially interfere with the product peak. The wastewater was spiked with the CZE method LOD of 2.25 × 10⁻⁷ mol/L. The results in Figure 8 present the successful detection of the API in the wastewater sample using the developed CZE method.

FIGURE 8.

Electropherogram of the overlayed blank and spiked wastewater samples. Peak 1 = 2.25 × 10⁻⁷ mol/L of ZB23 in 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 35 µA, a power of 0.4 W, an injection size of 25 mbar for 5 s, an Lt of 40 cm, an Ld of 31.5 cm and a detection wavelength of 214 nm. DI, deionised.

3.8. SPE Analysis

From the method validation, it was determined that the method detected a concentration lower than ZB23's PNEC value of 3.83 × 10⁻⁷ mol/L. As the CZE is not as sensitive compared to other chromatographic techniques, a pre‐concentration step of SPE was added to the method to determine if the wastewater has created any matrix effects. SPE cartridges chosen for this research were Agilent's Bond Elut PPL and Phenomenex's Strata C18‐E 6 mL cartridges. Both cartridges use reversed‐phase SPE to retain and extract the analytes of interest in environmental samples, but the difference between them is their sorbents. The Bond Elut PPL uses a modified divinylbenzene polymer, whereas the Strata C18‐E uses a silica sorbent. Although both cartridges are used for water samples, the better peak shape and higher number of theoretical plates were achieved with the Bond Elut PPL cartridge. A summary of these results can be reviewed in Table 2. The differences in the peak shape are demonstrated in Figures 9 and 10.

TABLE 2.

Summary of solid‐phase extraction (SPE) results achieved from the Bond Elut PPL and the Strata C18‐E cartridges.

	Bond Elut PPL	Strata C18‐E
Number of theoretical plates	170,262	161,501
LOD	2.25 × 10−7 mol/L	5.27 × 10−6 mol/L
LOQ	7.42 × 10−7 mol/L	1.74 × 10−5 mol/L
% Recovery	98.25	42.86

Abbreviations: LOD, limit of detection; LOQ, limit of quantification.

FIGURE 9.

Electropherogram of ZB23 extracted from wastewater using the Bond Elut PPL cartridge. Peak 1 = systematic peak, Peak 2 = ZB23. Separated using a 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 35 µA, a power of 0.4 W, an injection size of 25 mbar for 5 s, an Ld of 31.5 cm, an Lt of 40 cm and a detection wavelength of 214 nm.

FIGURE 10.

Electropherogram of ZB23 extracted from wastewater using the Strata C18‐E cartridge. Peak 1 = systematic peak, Peak 2 = ZB23. Separated using a 25 mM borate buffer with a pH of 10.5, a voltage of 10 kV, a current of 35 µA, a power of 0.4 W, an Ld of 31.5 cm, an Lt of 40 cm and a detection wavelength of 214 nm.

Although the Bond Elut PPL cartridge provided a higher number of theoretical plates, the optimum SPE cartridge is determined on the basis of the LOD and the percentage recovery. The LOD for both cartridges was determined by pipetting 1 mL of the SPE eluent in a 25 mL volumetric flask. Wastewater effluent was added up to the mark. The LOD of the Bond Elut PPL cartridge was determined to be 2.25 × 10⁻⁷ mol/L which did not change from the LOD of the CZE method. A lower concentration was detected with the Strata cartridge with an LOD of 5.27 × 10⁻⁶ mol/L. To determine which cartridge provided a higher recovery, a wastewater sample was spiked with 5.64 × 10⁻⁶ mol/L of the analyte solution before running the SPE method. The eluent was analysed using the developed CZE method, and the peak area of the API was recorded. The recovery was repeated with a second spiked wastewater sample run through the SPE, and an additional 5.64 × 10⁻⁶ mol/L of the API standard was spiked into the eluent. The percentage recovery was calculated using the following equation:

$$\%\mathrm{recovery}=\frac{\mathrm{peak}\mathrm{area}\mathrm{of}\mathrm{blank}\mathrm{eluent}}{\mathrm{peak}\mathrm{area}\mathrm{of}\mathrm{spiked}\mathrm{eluent}}\times 100$$

A higher recovery was achieved using the Bond Elut PPL cartridge with a % recovery of 98.25%. The Strata cartridge had a lower recovery percentage of 42.86%. A summary of these results is included in Table 2.

4. Conclusion

The developed CZE method proved to be successful in detecting the anti‐parasitic API in less than 10 min. The developed CZE method successfully detected the API at a lower concentration than the analytes PNEC. To determine if the wastewater created any matrix effects, a preconcentration step of SPE was added to the method. Although both SPE cartridges detected the API in the wastewater effluent, a higher number of theoretical plates and a higher recovery were achieved with the Agilent PPL. Although this cartridge was determined to be the optimum cartridge between the two cartridges analysed, the LOD of the CZE method did. It can be determined that no matrix effects were created from the wastewater.

Conflicts of Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Emma O'Sullivan‐Carroll reports financial support was provided by Hovione.

Data Availability Statement

Due to the confidential nature of the commercial API used in this research, supporting data is not available.