HPLC with Post-Column Derivatization with Alizarin for Determination of OATD-02, an Anticancer Arginase Inhibitor in Clinical Development

Abstract

The aim of this study was to develop an analytical method for selective determination of OATD-02 by high-performance liquid chromatography (HPLC) with post-column derivatization and fluorescence detection (FLD). OATD-02, a new boronic acid derivative, is a highly potent anticancer arginase inhibitor in clinical development. Chromatographic analysis of OATD-02 poses problems because this molecule has weak ultraviolet absorption. The derivatization reaction was based on the reaction between boronic acid from OATD-02 and alizarin solution. The optimized mobile phase consisted of a mixture of sodium bicarbonate in water and acetonitrile at a flow rate of 0.50 mL/min. Alizarin solution in methanol was delivered at a flow rate of 0.50 mL/min. The fluorescent complexes were detected by a fluorescence detector (excitation and emission wavelengths at 470 and 580 nm, respectively). The present method demonstrated proper values for selectivity, linearity, recovery (>99%), precision (RSD: 0.6%), sensitivity (LOD: 20 µg/mL and LOQ: 50 µg/mL), stability of solutions, and robustness.

1. Introduction

In the drug development life cycle, analytical methods development is an essential and indispensable element. A number of guidelines regulate qualitative and quantitative aspects of analytical methods development [1]. Drug approval by regulatory authorities requires the drug candidate to prove control of the entire process of drug development, starting from development of the active pharmaceutical ingredient (API) synthetic process to final-quality release of the investigational medicinal product (IMP), by using validated analytical methods [2]. Developing a specific analytical method to control the quality of a new drug requires key steps, starting with determining the type of method and characterizing the compound. Any analytical method has to be suitable for its intended use and should accurately quantify the active pharmaceutical ingredient without interference from process impurities or other potential impurities. OATD-02, a new boronic acid derivative, is the highly potent and selective small-molecule arginase inhibitor currently in clinical trials as anticancer agent, involved in both tumor immunity and metabolism. Because of direct antitumor efficacy in animal models, it was selected as a clinical candidate for cancer immunotherapy [3,4]. OATD-02 (Figure 1) is difficult to analyze because of its high polarity, zwitterionic nature and poor solubility in most organic solvents. Loosing of the chemical groups being UV active chromophores during the synthesis process also caused detection problems. As for analytical challenges in the drug development process, it is necessary to develop a sensitive, accurate and precise analytical method capable of monitoring related substances in API and drug products [5]. OATD-02 is a new compound; therefore, it was necessary to develop a new analytical method. The direct UV detection method at short wavelengths (near 200 nm) was difficult to used due to low sensitivity and mobile phase cut-off. The second approach used UV detection method of amino acids occurred in the OATD-02 molecule with creation of copper (II) complexes. It was not a specific and selective method because of the impurities without amino acid groups (OAT-9012, OAT-9013, OAT-9016, OAT-9017). This UV method used copper (II) salts as an additive to the mobile phase with isocratic elution, so the total analysis time was longer than 30 min and the peaks are broadened.

High-performance liquid chromatography (HPLC) methods with UV-Vis detection are the most simple and the most widespread analytical procedures in pharmaceutical applications. Although it suffers from some limitations, particularly for molecules that do not possess UV chromophores. Those compounds having weak chromophores can be successfully detected by UV at short wavelengths (near 200 nm). However, mobile-phase constituents that have high UV cut-offs should be avoided because they might blind the detection of compounds with weak chromophores [6]. International pharmacopoeias mostly employed HPLC with UV detection for this purpose but there is a problem when the drugs are lacking a chromophore [7]. Moreover, many impurities, which may be genotoxic, either lack a UV chromophore or offer an insufficient UV response at a low concentration. It would be a reluctant conclusion that the applicability of the conventional HPLC–UV is limited for quantification of related substances in API or drug products [8]. The control of impurities in a drug substance is an essential part of pharmaceutical drug development. The potential impurities in an API could have significant impact on toxicology studies and drug safety. For a better understanding of the chemical reactions and the synthetic process, it is necessary to obtain data from impurity analyses. Basically, the impurities are structurally related to the drug substance and require very sensitive and efficient chromatographic systems, they especially exist at a very low concentration in ratio to API [9]. Many papers have reported on the liquid-chromatographic determination of drugs and drug candidates [10,11,12,13,14]. Moreover, there are many publications describing the use of high-performance liquid chromatography with different detectors for analysis of different compounds lacking strong UV chromophores [15,16,17,18,19]. Some of these methods [10,11,12] have utilized a very expensive mass spectrometric detector, which may be not suitable for routine quality control of medicinal preparations. Other papers [15,16,20] have been published on the liquid chromatographic determination of drugs using a ELSD detector, which is widely used for the analysis of non-UV-absorbing compounds. However, ELSD has significant limitations in precision, sensitivity, dynamic range, and the nature of calibration curves [21]. The charged aerosol detector [17,22,23] and the refracting index detector [13,24,25] are the other choices if the UV detector is insensitive to drugs with weak chromophores. Generally, CAD response to analyte concentration is not linear over a large concentration range [15]. RI, while commonly used, has considerable limitations in sensitivity and is not compatible with gradient elution [23]. There are several papers [14,18,19] which have reported on the liquid chromatographic determination of new compounds using fluorescence detection. The problem, however, is to search through the many papers that describe the determination by derivatization of an active pharmaceutical ingredient having weak chromophores. After the discovery of bortezomib, other drugs with a boronic acid group have been developed. Boron-containing compounds have been found to possess broad biological activities such as anticancer, antibacterial, or antiviral and the low toxicity of the boronic acid moiety itself has been postulated [26]. For detection of boronic acid groups in molecules, thin-layer chromatography [27] and high-performance liquid chromatography with a post-column reaction with alizarin were used [28]. The aim of the work was to develop and validate the HPLC method for determination of OATD-02 and its impurities.

2. Materials and Methods

2.1. Reagents

The HPLC-grade acetonitrile, methanol and water were purchased from Merck (Darmstadt, Germany). Sodium bicarbonate (NaHCO₃) were provided by Chempur (Piekary Slaskie, Poland). Alizarin was purchased from Sigma Aldrich (Sant Louise, MI, USA). OATD-02 was synthetized at Molecure SA (formerly OncoArendi Therapeutics SA, Warsaw, Poland). OATD-02 standard was synthetized and defined as a purified compound that is well characterized and quality reference standards (purity, residual solvents, water content, identification, structural confirmation). The crude mixtures from synthesis contained all the process impurities—OAT-9012, OAT-9013, OAT-9014, OAT-9015, and OAT-9016—in different proportions. The mixtures, which were used for method optimization, contained OAT-9014, OAT-9015, OAT-9016 and OAT-9017.

2.2. Instruments

Analyses were performed on a Shimadzu HPLC system with a fluorescence detector. Data acquisition was performed using LabSolution software version 5.106.

2.3. HPLC and Derivatization Conditions

The chromatographic column used was XSelect CSH C18 (2.5 µm, 100 mm × 3 mm, Waters). The separation was achieved on a gradient method. Mobile phase A contained 10 mM sodium bicarbonate in water and mobile phase B contained a mixture of 10 mM sodium bicarbonate in water: acetonitrile in the ratio of 20:80 v/v. The flow rate of the mobile phase was 0.5 mL/min. The column temperature was maintained at 30 °C and the samples temperature was ambient. The applied elution conditions were: 0 min, 0% B; 0–2 min, linear gradient from 0% to 6% B; 2–4 min, 6% B isocratic; 5 min, 10% B; 9 min, 10% B; 18 min, 100% B; 18.1–23 min, 0% B. The detection was monitored by a fluorescence detector (excitation and emission wavelengths were set at 470 and 580 nm, respectively) and the injection volume was 5 µL. The effluent of HPLC column was connected to a T-junction, where alizarin solution was delivered from another pump. In the post-column derivatization the effluent from the analytical column was mixed with reagents then directed through a reaction coil where the product was formed and subsequently it was detected by a fluorescent detector. The alizarin solution was prepared in concentration of 350 µM (alizarin in methanol) and it was delivered at a flow rate of 0.50 mL/min. The reaction coil was made of PEEK tubing (0.018 mm ID, length 65 cm).

2.4. Method Development

The main objective of this work was to develop a selective HPLC-FLD method for the separation and accurate quantification of OATD-02 and its related substances in an active pharmaceutical ingredient. Method development was quite challenging mainly due to the presence of many impurities and strong polarity of the compounds The reversed-phase chromatographic mode with an appropriate analytical column was chosen for separation of OATD-02 and its impurities. Various stationary phases in different dimensions with various particle sizes, with expected differences in their selectivity, were tested. Among all columns, the HPLC Waters XSelect CSH C18 column (2.5 µm, 100 mm × 3 mm) was chosen as the optimal column for the retention and separation of the OATD-02. Due to its Charged Surface Hybrid (CSH) technology and hydrophobic interactions with the C18 ligand, it provides ionic interactions with the analytes [29], which was found to be useful for the retention of OATD-02. The next step in the method development process was the mobile phase selection, which was initially an aqueous solution of ammonium salts with methanol but was later replaced with 10 mM sodium bicarbonate with acetonitrile, mostly due to better retention, peak shapes and resolution between the polar compounds. The flow injection analysis was performed in order to determine emission and absorption maxima of the fluorescent complex. The optimum wavelength of excitation and emission for the fluorescent complex of alizarin-OATD-02 were, respectively, 470 nm and 580 nm. Alizarin is a fluorescent compound, which dissolves well in organic solvents and forms a fluorescent complex with boronic acid moiety present in OATD-02.

2.5. Preparation of Sample Solution

The sample solutions were freshly prepared on a daily basis with water as diluent. An accurately weighted 1 mg of OATD-02 sample was dissolved in 1 mL of diluent. For validation preparation, accurate appropriate amounts of OATD-02 samples were weighed and transferred into a volumetric flasks and diluted to an appropriate volume of diluent. The solutions were mixed and filtered through a 0.45 μm disposable syringe filter unit before injection into the chromatographic system.

2.6. Method Validation

The method was validated for parameters such as specificity, linearity, accuracy, precision, the LOD/LOQ, stability of solutions and robustness. Linearity was performed by injecting prepared sample solutions of OATD-02 of different concentrations in the range of 0.8–1.2 mg/mL to the HPLC system. The peak area was plotted against the concentration and the generated regression line equation along with the R² was reported. To evaluate the linearity and range of the method, five different concentrations were prepared: 0.8, 0.9, 1.0, 1.1 and 1.2 mg/mL. Three separate injections of each concentration were analyzed under the same conditions and the average result was reported. Specificity of the method was evaluated by injecting the blank solution and sample solution to ensure the peak of OATD-02 can be well separated from each other and free of interference from the blank solution. Accuracy and precision (repeatability) analyses were performed by assaying triplicate preparations for accuracy solutions (80%, 100%, 120%) against the assay solution of the well-characterized standard material and calculation of the recovery results. Instrumental precision was established by five replicate injections of the sample solution (1 mg/mL). The RSD was then calculated for the generated peak areas. The same analytical conditions were used by different analysts and on different columns. The intermediate precision of the method was established by testing six test solutions against the assay solution of the well-characterized standard material and calculation of the recovery results. The sensitivity of the method was established by measuring the detection limit (LOD) and the quantification limit (LOQ). Serial dilutions of OATD-02 solution were made and then each dilution was injected to the HPLC system. The LOD is expressed as a concentration that gives a signal-to-noise ratio over 3:1, while the LOQ of the sample can be determined with a signal-to-noise ratio over 10:1. The stability of the sample solution was evaluated after 3 days by assaying the sample solution against the fresh prepared sample solution and calculated the % assay of OATD-02. Robustness of the developed analytical method was assessed by small changes in the experimental conditions. The chromatographic changes include mobile phase composition, flow rates, temperature of the column, emission and excitation wavelengths.

3. Results

3.1. Optimal Conditions for HPLC

3.1.1. Effect of Mobile Phase pH on Retention of OATD-02

The influences of mobile phase on separations were investigated and summarized. OATD-02 has two pKas at approximately 1.6 and 7.1. The peak shapes were narrow and symmetrical when the mobile phase was comprised of acetonitrile and 10 mM sodium bicarbonate solution. Accordingly, different mobile phases were tested: ammonium formate, sodium acetate with acetonitrile or methanol. The pH values of the aqueous buffers were measuring before mixing them with the organic modifier. The pH of the mobile phase had impact on resolution, selectivity and peak shapes of analyzed compounds. At a lower pH of approximately 6.3 in ammonium formate solution, basic analytes such as OATD-02 were protonated and eluted more quickly without sufficient resolution from others compounds. At a higher pH in sodium acetate of approximately 7.3, OATD-02 were more retained, because there were a mixture of neutral and protonated amine groups. Peak tailing was observed because the pKa of OATD-02 is similar to the pH of the buffer solution and the analyte eluted as both a charged and uncharged species. When the acetonitrile and sodium bicarbonate (10 mM) were used as mobile phases, a significant improvement in peak shape of OATD-02 was observed compared with the combination of acetonitrile and ammonium formate or sodium acetate as shown Figure 2 (A), (B), (C). The addition of sodium bicarbonate to the mobile phase was necessary to obtain the best separation and run time as shown in Figure 2 (C). At the highest pH in sodium bicarbonate, approximately 8.3, in OATD-02 there were mostly uncharged amine groups and all basic compounds were more retained without tailing peaks. Finally, acetonitrile and sodium bicarbonate (10 mM) were chosen as the optimal chromatographic mobile phase. Chromatographic analysis of the samples was performed within 23 min. As seen in Figure 2 (C), the impurities and OATD-02 in crude mixture were separated well with a resolution higher than 2.0 among each other.

The most important chromatographic parameters (area, high, efficiency expressed as number of theoretical plates, resolution, peak width and peak symmetry) obtained in different mobile phase compositions are summarized in

Table 1. Comparison of the analytical parameters for OATD-02 and related substances in different mobile phase compositions.

10 mM HCOONH4
#Peak	Name	Retention Time, Min	Area, µV × Min	Height, µV	Tailing Factor	Efficiency	Rs	Width (1/2)), Min	S/N
1	OAT-9015	1.86	351,598	29,603	-	321	-	-	6
2	OAT-9014 epi-OATD-02	2.01	355,836	20,317	-	12	0.1	-	4
3	OATD-02	3.99	2,824,963	61,963	-	196	1.1	0.58	13
4	OAT-9016	5.67	84,399	10,901	1.47	10,689	2.5	0.12	2
10 mM CH3COONa
#Peak	Name	Retention Time, Min	Area, µV × Min	Height, µV	Tailing Factor	Efficiency	Rs	Width (1/2)), Min	S/N
1	OAT-9015	4.57	111,014	17,364	1.62	11,161	-	0.09	6
2	OAT-9014 epi-OATD-02	5.67	904,701	73,438	1.20	5206	4.5	0.18	26
3	OATD-02	8.21	5,217,590	208,729	0.73	1965	4.8	0.41	73
4	OAT-9016	9.28	186,428	15,203	1.40	13,737	2.0	0.19	5
10 mM NaHCO3
#Peak	Name	Retention Time, Min	Area, µV × Min	Height, µV	Tailing Factor	Efficiency	Rs	Width (1/2)), Min	S/N
1	OAT-9015	4.75	96,289	18,010	1.38	15,272	-	0.08	9
2	OAT-9014 epi-OATD-02	6.12	735,667	101,383	1.45	15,615	7.8	0.11	49
3	OATD-02	8.57	4,894,797	662,409	1.49	30,321	12.5	0.11	318
4	OAT-9016	12.96	167,195	21,002	1.54	55,442	21.0	0.12	10

. Regarding the peak resolution, number of theoretical plates and peak width, the best results for all peaks were obtained with the sodium bicarbonate in mobile phase.

3.1.2. Optimization of Detection and Derivatization

The optimum wavelength of excitation and emission for the fluorescent complex of alizarin-OATD-02 were, respectively, 470 nm and 580 nm. The effect of the concentration of alizarin on the intensity of the peak for OATD-02 and the baseline on the chromatogram was examined. The different concentrations of alizarin from 200 µM to 500 µM were evaluated and it was noted that the alizarin concentration of 350 µM enabled obtaining the most intensive peak of the OATD-02 with the least increase in the background baseline. This was very important in the end of run time analysis, where the background baseline could suppress the intensity of impurities at lower concentration levels.

3.1.3. Influence of the Stationary Phase

Various stationary phases were tested: Luna C18, 3 µm, 100 mm × 2 mm, 100 Å; Kinetex Phenyl-Hexyl, 5 µm, 250 mm × 4.6 mm, 100 Å; Thermo Scientific Syncronis C18, 3 µm, 100 mm × 2.1 mm, 100 Å; Kinetex XB-C18 2.6 µm, 100 mm × 3 mm, 100 Å; Waters XSelect HSS T3, 2.5 µm, 100 mm × 2.1 mm, 100 Å and Zorbax Extend-C18, 3.5 µm, 50 mm × 2.1 mm, 100 Å and the flow rates were adjusted to column diameters. The comparison of the main chromatographic parameters for OATD-02 and related substances on different columns was presented in Figure 3A–C.

The narrowest peaks, especially for OAT-9016, was achieved for XSelect CSH-C18. Considering the results displayed in Figure 3B the highest efficiencies of the main compound and the latest impurity were obtained on the XSelect CSH-C18 column. Good peak shape can be defined by tailing factor, which was presented in Figure 3C. The tailing factors below or approximately 1.5 for the main compound and impurities were achieved on the columns: Syncronis C18, XSelect HSS-T3 and XSelect CSH-C18. The application of the Luna C18 column allowed separation of all compounds, but the intensity of the last peak of OAT-9016 was insufficient, which can cause analysis problems for low concentrations. The only two compounds were retained on phenyl-hexyl stationary phase. It happened because the stationary phase with a phenyl group is intended for aromatic compounds. Phenyl-type stationary phases show better selectivity for aromatic compounds than C18 stationary phases because π–π interactions occur between the aromatic analyte and the phenyl group of these stationary phases [30]. Therefore, the analysis of the OATD-02 molecule, which is not an aromatic compound, on a phenyl-hexyl column did not yield high-intensity chromatographic signals. In the results obtained for the mixture of OATD-02 and impurities using the Kinetex XB-C18 column, only two of the three impurities can be seen. In addition, the chromatographic signals obtained were broadened. Despite endcapping, i.e., binding of residual silanol groups, there may still be free silanol groups, and this may lead to undesirable interactions with the analytes and, consequently, to peak asymmetry. For the Synchronis C18 column, all the compounds analyzed in the mixture were successfully retained and separated with not broadened peaks. Due to double endcapping of Zorbax Extend-C18 stationary phase with a porosity of 80 Å, symmetrical peaks were obtained for all analyzed compounds with good intensity. The use of the HSS T3 column for the analysis of the OATD-02 mixture and impurities allowed for the separation of all compounds, and the obtained signals were symmetrical. The use of the XSelect CSH-C18 column to analyze the OATD-02 mixture and three impurities enabled obtaining the best separation results from all comparable columns. The Waters XSelect CSH-C18 column was designed using BEH, a stationary phase with an ethylene-bridged hybrid inorganic-organic packing, which allows for better separation and longer retention of polar compounds [31]. The CSH (Charged Surface Hybrid) technology consists in additional modification of the column surface with charge, which makes possibility to using this column for analysis of compounds with different polarity and in a wide pH range of the mobile phase (pH 1–11).

Table 2. Table summarizing chromatographic parameters of determination of OATD-02 in crude samples on different columns.

OATD-02
Type of Column	kR	Area, μV × Min	Height, μV	Tf	N/m	Wh1/2, Min	S/N
Luna C18	10.0	5,114,031	666,204	1.52	223,820	0.11	278
Kinetex Phe-hex	3.0	1,138,000	87,869	4.01	89,284	0.15	34
Syncronis C18	10.4	3,997,187	548,064	1.29	260,230	0.10	344
Kinetex XB-C18	6.9	4,661,412	670,962	1.73	321,730	0.10	328
XSelect HSS T3	10.1	4,941,138	779,688	1.57	303,920	0.09	440
Zorbax Extend-C18	10.4	4,555,981	644,835	1.53	175,340	0.10	239
XSelect CSH-C18	6.6	4,894,797	662,409	1.49	303,210	0.11	318

summarizes the obtained most important chromatographic parameters for the determination of OATD-02 in crude samples on different columns. The peak areas of OATD-02 were the largest values on the columns: Luna C18, XSelect HSS T3 and XSelect CSH-C18, but only on the XSelect CSH-C18 column was the tailing factor of OATD-02 peak below 1.5. Moreover, the XSelect CSH-C18 can be used in the widest range of mobile phase pH values (under pH 11), which is very important when using a basic mobile phase with sodium bicarbonate.

The comparisons of the selectivity differences of stationary phases were presented in Figure 4A,B as the selectivity charts for reversed phase packings. In Figure 4, the logarithm of selectivity between nearest compounds vs. the retention factor of each compound was plotted. Different stationary phases can be discriminated by the alpha value of two compounds. Moreover, it is useful for finding stationary phases with similar selectivity properties. The retention of basic analytes can be used as a measure for the silanophilic activity of stationary phases [32]. The retention factor of OATD-02 or OAT-9016 on the x-axis is a measure of the hydrophobicity of stationary phases. The relative retention alfa OATD-02 /OAT-9014 or OAT-9016/OATD-02 shown on the y-axis is a selectivity comparison. On selectivity chart Figure 4A for retention of OATD-02 the most hydrophilic stationary phase is phenyl-hexyl phase, because of occurrence of aromatic rings, but the tailing factor of OATD-02 on this stationary phase is over 4. Furthermore for OAT-9016 compound there is no retention on this stationary phase, probably there is small broadened peak, which cannot be integrated. On selectivity chart Figure 4B for retention of OAT-9016 the most suitable stationary phase is CSH-C18 phase. At basic pH of sodium bicarbonate the effect of free silanol groups, which could cause unwanted additional interaction is weaker, because the basic analytes became uncharged and contribute lower to electrostatic interactions [33].

Retention in reversed-phase liquid chromatography (RP-LC) can be described by the analyte–stationary phase interactions: hydrophobicity, steric resistance, hydrogen bonding of basic or acidic analytes to, respectively, acidic or basic sorbent sites, and ion-interaction. The mechanism of retention in the reversed phase is primarily hydrophobic interactions. If hydrophobicity could be the only stationary phase property which affected selectivity, a plot of log k for one column vs. another column would give a straight line, with all the points on the line [34]. For further comparison of the retentivity of two C18 columns, the data of log k for the Waters XSelect CSH-C18 column with ion-exchange properties plotted vs. log k for the Zorbax Extend-C18 column were illustrated in Figure 5. The slope of this plot depends on the hydrophobicity of the CSH-C18 column compared to the Zorbax Extend-C18 column. There are deviations from this plot, which reflect differences in column selectivity for the four described compounds. These dependencies support that retention is influenced not only by hydrophobicity, but especially for charged analytes.

The Waters XSelect CSH-C18 column (2.5 µm, 100 mm × 3 mm) was chosen as the optimal column for the best peak shapes (tailing factor < 1.5), retentions and column efficiency for the OATD-02 and its impurities. The OATD-02 sample solution was injected at different mobile phases and different HPLC conditions using different stationary phases, flow rates, and mobile phases. The best obtained HPLC chromatographic condition was summarized in

Table 3. The best chromatographic conditions for OATD-02 separation.

Parameters	Results
Column	Waters XSelect CSH C18 2.5 µm, 100 mm × 3 mm
Mobile phase A	10 mM NaHCO3 (840 mg/L) in H2O
Mobile phase B	10 mM NaHCO3 (840 mg/L) in H2O:ACN (20:80)
Flow rate	0.5 mL/min
Mobile phase C	350 µM alizarin in MeOH
Post-column reagent flow rate C	0.5 mL/min
Column temperature	30 °C
Detection	Fluorescence, λex= 470 nm, λem= 580 nm
Injection volume	5 µL
Sample solvent	water
Sample concentration	1 mg/mL
Run time	23 min
Gradient	Time	Conc. %B
	0 min	0 %B
	2–4 min	6 %B
	5–9 min	10 %B
	18 min	100 %B
	18.1–23 min	0 %B

.

3.2. Method Validation

The developed method was validated according to the International Conference of Harmonization (ICH) guideline [2]. Validation parameters were tested and the results were within the acceptable criteria, which proves that the analytical method is a validated method. To evaluate the linearity range of the method, five different concentrations were prepared: 0.8, 0.9, 1.0, 1.1 and 1.2 mg/mL. The peak area under the curve was plotted against concentration. The developed method showed to be a linear relation (R² = 0.999); the regression line equation was found to be y = 47,791x + 2569. The specified range was established by confirming that the analytical method provides an acceptable linearity from 80 to 120 percent of the test concentration [2], which was set to 1 mg/mL. This depended on the intended application of the analytical procedure for the assay of a drug substance. The linearity in the lower concentration range was also checked. The method was also evaluated for accuracy, the LOD, the LOQ, precision, recovery validation parameters. The mean percentage recovery (preparing three concentrations of the sample: 0.8, 1.0, 1.2 mg/mL, which were injected in the HPLC in triplicate) was 99.2%. The RSD was calculated for every sample concentration (0.8, 1.0, and 1.2 mg/mL) and was found to be 0.66%, 0.36%, and 0.12%, respectively (

Table 4. The accuracy results for OATD-02.

Sample #	Recovery
	Accuracy III (80%), %	Accuracy II (100%), %	Accuracy I (120%), %
1	99.6	99.4	99.0
2	98.5	99.5	99.2
3	99.6	98.8	99.2
RSD, %	0.66	0.36	0.12
Mean Accuracy, %	99.2
SD	0.38
RSD, %	0.39

). The relative standard deviation for repeatability were calculated 0.4%.

Instrumental precision was assessed by injecting one concentration five times, and the relative standard deviation (RSD) of the area was calculated and found to be 0.6%, which was within the acceptable criteria. The results, mean, standard deviation and relative standard deviation were presented in

Table 5. The injection precision results for OATD-02.

Injection #	Peak Area, μV × Min
1	4,718,036
2	4,734,661
3	4,767,180
4	4,781,615
5	4,775,214
Mean	4,755,341
SD	27,590
RSD, %	0.58

.

For intermediate precision, the two analysts’ results for the mean percent recovery and RSD were 99.5% and 2.0%, respectively. According to the results, the sample solution was stable under normal laboratory condition for 96 h (the % assay of OATD-02 in the test solution was 99.5%).

The method was also found to be robust under the varied parameters; the injected concentration of OATD-02 was 1 mg/mL. Robustness was checked by systematic variations in the organic proportion (+5%), column temperature (+5 °C), another lot number of column, the flow rate of mobile phase and alizarin solution (±0.2 mL/min), detector wavelength λ_ex and λ_em (±2 nm). The efficiency and tailing factor were then compared to the original method, which was presented in

Table 6. The robustness results for OATD-02.

Temperature	Retention Time, Min	Efficiency	Tailing Factor
30°C	8.61	29,238	1.47
35°C	8.38	30,358	1.50
Flow, mL/min	Retention time, min	Efficiency	Tailing factor
0.4	9.80	28,664	1.50
0.5	8.61	29,238	1.47
0.6	7.83	30,013	1.46
Eluent composition	Retention time, min	Efficiency	Tailing factor
0% B	8.61	29,238	1.47
5% B	8.18	27,328	1.49
Column S/N#	Retention time, min	Efficiency	Tailing factor
1573018118510	8.61	29,238	1.47
1733216418402	8.55	35,545	1.50
Flow rate of alizarin, mL/min	Retention time, min	Efficiency	Tailing factor
0.4	8.60	29,183	1.48
0.5	8.61	29,238	1.47
0.6	8.62	29,706	1.48
Wavelength λex/λem	% change in peak area	Efficiency	Tailing factor
468/580	1.6	29,305	1.48
470/580	-	29,238	1.47
472/580	1.7	29,299	1.48
470/578	1.9	29,353	1.47
470/582	1.9	29,412	1.48

. The method can be regarded as robust to the above small changes and the acceptance criteria for plate number ≥20,000 and tailing factor ≤1.5 were met.

The sensitivity of the method was assessed by the LOD and the LOQ. The minimum quantity of OATD-02 which the method can detect is expressed as the LOD and was determined by injection of diluted samples. The signal-to-noise ratio (S/N) of 3:1, which represents the LOD, was found to be 0.02 mg/mL. The signal-to-noise ratio 10:1 was determined as the LOQ and was found to be 0.05 mg/mL. The developed method was found to be selective for the generated peak of the derivatized OATD-02 from other reaction reagents (related substances). The peak was well separated from the other eluting peaks. Moreover, the peak of OATD-02 was symmetrical with an acceptable theoretical plate (N > 3000) and tailing factor (<1.5).

4. Discussion

Derivatization Mechanism

Alizarin shows weak fluorescence by itself, but when it interacts with a boronic acid, a fluorescent boronic ester is formed. The related substances and side products in the OATD-02 synthesis process contained a boronic acid function and they could form fluorescence complexes with alizarin dye (Figure 6). Post-column derivatization with alizarine and fluorescence detection was used, because of fast and efficient generation of fluorescent esters. The derivatization reaction is selective so that derivate products can be easily seen against a complex background. Therefore, the post-column derivatization technique is used with fluorescence detection to increase sensitivity and selectivity in HPLC analysis.

The analytical method package for active pharmaceutical ingredient must meet the criteria, which are set out in the guidelines such as the ICH guidelines [1]. Use of boronic acid compounds is increasing recently because of their anticancer, antibacterial, and antiviral activity, as well as their application as sensors and delivery systems. Previously, compounds with boron were thought to be toxic, but now they are widespread as biologically active compounds and pharmaceutical agents [35,36]. OATD-02, an arginase inhibitor, was selected as a clinical candidate for cancer immunotherapy, because of direct antitumor efficacy in animal models [3]. Its structure contains chromophore groups with a low molar absorption coefficient in the UV and visible light range. The selected compound is polar and non-soluble in most organic solvents (unsuitable for NP-HPLC). However, OATD-02 contains boronic acid, and with forms a fluorescent complex the alizarine. This approach has enabled using HPLC with a fluorescence detector (FLD) and with post-column derivatization. It is important to find good consistency of hydrophobic and silanophilic properties of stationary phases to reach good chromatographic repeatability of results for different compounds. To control the ionization state of the analyzed compounds, the effect of the mobile phase pH on retention was checked. At a lower mobile phase pH, below the pKa of the basic compounds, they were mostly ionized and likely to participate in hydrogen bonding, so they spent less time in hydrophobic interactions with stationary phase and in result have less retention. At a higher pH in sodium acetate, there were both ionized and neutral forms of the compounds, the charged form had less hydrophobic interaction while uncharged form had hydrophobic interaction and in result the broad and asymmetrical peaks were observed. Moreover, the few free silanol groups present in the stationary phase participated in ionic interactions with ionized compounds. At the highest pH in sodium bicarbonate in mobile phase, OATD-02 and other basic impurities were mostly deprotonated and were more retained without tailing peaks. The pH of the mobile phase based on sodium bicarbonate is above 8 and OATD-02 has deprotonated carboxylic group and uncharged amine groups, whereas OAT-9016, which has an amide group instead of a carboxylic group is retained more strongly on the non-polar stationary phase. OAT-9015 contains a hydroxy group instead of tertiary amine which causes the lower retention time due to ionization of the hydroxyl group. OAT-9014, which is an epimer of OATD-02, but due to the steric resistance of three functional groups has short retention time than OATD-02. Compounds with hydroxyl and amines groups could display the ability to interact with stationary phases through hydrogen bonding. These compounds can additionally interact with the silanol groups found on the surface of the silica. Comparison of column selectivity with combined hydrophobic and hydrogen bond capacity, can help improve resolution. Focusing on manipulation of hydrophobic retention and hydrogen bond capacity is useful for analysis of polar compounds. The chromatogram obtained in best chromatographic conditions (Table 3) was presented in Figure 7. All the impurities that were detected and characterized in the synthetic process were well separated.

5. Conclusions

To determine the assay and impurities profile of OATD-02, a HPLC method with fluorescence detection was developed. The results indicated that the developed method displays good sensitivity, precision and accuracy to be used to detect OATD-02 in drug substances. An easy-to-use HPLC method with post-column derivatization and fluorescence detection was developed and validated for the OATD-02 drug candidate. The method allows the specific detection of the OATD-02 molecule in drug substances, even in crude mixtures from organic synthesis. We successfully derivatized OATD-02 with a post-column derivatization system using alizarin solution. There is no complicated sample preparation procedure. The derivatized OATD-02 compound is quantified at a very low concentration and the proposed method was successfully validated. Our developed method showed high precision, linearity and selectivity.