Characterization and Separation of Platinum-Based Antineoplastic Drugs by Zwitterionic Hydrophilic Interaction Liquid Chromatography (HILIC)–Tandem Mass Spectrometry, and Its Application in Surface Wipe Sampling

: Platinum-based antineoplastic drugs (PtADs) are among the most important and used families of chemotherapy drugs, which, even showing severe side effects and being hindered by drug resistance, are not likely to be replaced clinically any time soon. The growing interest in the occupational health community in antineoplastic drug (AD) surface contamination requires the development of increasingly fast and easy high-throughput monitoring methods, even considering the lack of harmonized legally binding regulation criteria. Thus, a wipe sampling method together with zwitterionic hydrophilic interaction liquid chromatography (HILIC-Z)–tandem mass spectrometry (MS/MS) analysis was developed for the simultaneous evaluation of oxaliplatin, cisplatin, and carboplatin surface contaminations. A design of experiments approach was used to optimize the chromatographic conditions. Limits of quantiﬁcation ranging from 2 to 5 ng/mL were obtained from interday and intraday repetitions for oxaliplatin and carboplatin, and between 170 and 240 ng/mL for cisplatin. The wipe desorption procedure is equivalent to other AD sampling methods, enabling a fast sample preparation, with an LC-MS/MS analysis time of less than 7 min.


Introduction
During the last 50 years, oncology has gone through remarkable changes, resulting in transforming malignant germ-cell testicular tumors from highly fatal to nearly uniformly cured neoplasms [1,2]. Cisplatin (cis-[PtCl 2 (NH 3 ) 2 ]), first synthesized in 1844 by Italian chemist Michele Peyrone [3,4], was introduced in clinical trials in the late 1970s [5]; this clinical landmark was rightfully attributed to the identification of its anticancer potential by Barnett Rosenberg in his experiments dating to 1965 [6][7][8]. Cisplatin was accepted for pharmacological use in 1978, and since then, 25 platin drugs have entered clinical trials. Both the "second generation" and "third generation" of platinum-based antineoplastic drugs (PtADs), respectively, carboplatin, approved by the United States (US) Food and Drug Administration (FDA) as Paraplatin in 1989 [9], and oxaliplatin, authorized for clinical use in the European Union in 1999 and the US in 2002 [10], gained broadly worldwide range of 0.1 to 1 ng/mL [36]. However, analyte speciation is achievable only by coupling with LC techniques [37].
Due to PtADs' low absorbance and high instability, the development of the LC-UV method is problematic. In fact, it suffers from poor sensitivity (a few µg/mL) [38][39][40]. The addition of a derivatization step improved sensitivity by up to two orders of magnitude [41,42]. Molecular MS has emerged in the last decade as a major tool to characterize metallodrug speciation [43]. More sensitive LOQs, between 2 and 25 ng/mL, were reached with LC-MS but only for single PtAD determination and often through derivatization [41,44,45].
Simultaneous analysis of cis-, carbo-, and oxaliplatin by liquid chromatographytandem mass spectrometry (LC-MS/MS) by triple quadrupole has not yet been proposed.
In recent decades, an alternative strategy for the LC analysis of polar compounds has been found in hydrophilic interaction chromatography (HILIC) [46]. Andrew Alpert suggested that during the separation process, the hydrophilic analytes are partitioned between a water-enriched layer adsorbed on the surface of the stationary phase and a more hydrophobic mobile phase [47]. Depending on the triad compound functional groups, mobile phase composition, and stationary phase chemistry, different retention mechanisms, including strong electrostatic, dipole-dipole interactions, adsorption, and formation of hydrogen bonds [48], are thought to contribute together to the overall retention [49,50]. In addition, the dominating mechanism is influenced by many interacting factors such as the ionic strength and pH of the mobile phase that can be modulated by the addition of salts, acids and bases, buffers, type of organic solvent, and temperature [51,52]. These factors make the development of HILIC methods the ideal basis for applying chemometric tools such as design of experiments (DoE) and multivariate modeling. However, over the years, only a few researchers adopted these techniques for studying HILIC retention mechanisms and method development [53].
In the present study, we report on a shell particle-zwitterionic-HILIC LC-MS/MS method developed and validated to determine oxaliplatin, cisplatin, and carboplatin. Our method, coupled with the known technique of wipe sampling, provides a robust and rapid procedure for assessing PtAD surface contamination and personnel exposure in the hospital environment.
The desorption solution (DS), consisting of a mixture of methanol:water 50:50 (v/v), was used to elute the wipe samples [32].

Instruments
The LC system consists of a Shimadzu Nexera X2 equipped with a DGU-20A5R degasser unit, two LC-30AD pumps, SIL-30AC autosampler, a CBM-20A system controller, SPD-M20A diode array detector, and CTO-20AC column oven. The used tandem mass spectrometry system was a Shimadzu LCMS 8050 triple quadrupole equipped with an electrospray source (ESI) [54]. Instrument control and data acquisition were carried out using the software LabSolution ® ver. 5.97 (Shimadzu Corp., Kyoto, Japan).
Mobile phase pH was measured through a Hanna ® Instruments portable pH meter mod. HI8424, equipped with an HI1230B pH electrode and an HI7662temperature probe.
Climatic Cabinet Sartorius SCC400L was used to weigh the IS, chemical standards, and ammonium formate salt.

Standard Solutions and Calibration Levels
Stock solutions of carboplatin, oxaliplatin, and daunorubicin (IS) were prepared at 1 mg/mL using the DS mixture, while cisplatin was prepared at the same concentration using water. All the stock solutions were stored at −20 • C. Diluted solutions of carboplatin, cisplatin, oxaliplatin, and IS were prepared from the stock solutions up to a concentration of 2 µg/mL in DS and used to optimize MS parameters.
The calibration mixture of analytes (MixPt solution) was prepared by adding aliquots of the pharmaceutical preparations by Teva (reported in Section 2.1) to obtain a 1 µg/mL concentration for carboplatin and oxaliplatin and 40 µg/mL for cisplatin. The IS working solution was made by diluting IS stock solution with a DS mixture up to 1 µg/mL.
A seven-level calibration curve for each analyte was prepared by adding 10 µL of IS working solution and proper volumes of MixPt, and diluted with DS mixture to reach a final volume of 1 mL. By following this procedure, the analyte concentrations of the calibration solutions were: 0, 5, 10, 20, 30, 40, 50 ng/mL for carboplatin and oxaliplatin, and 0, 200, 400, 800, 1200, 1600, 2000 ng/mL for cisplatin. Three levels of internal quality control (CQI) solutions were prepared by diluting MixPt with the DS mixture, to obtain 4 ng/mL (low level), 25 ng/mL (medium level), and 45 ng/mL (high level) for carboplatin and oxaliplatin, and 160 ng/mL (low level), 1000 ng/mL (medium level), and 1800 ng/mL (high level) for cisplatin. Each CQI solution contained IS in a concentration of 10 ng/mL.
The experimental design study for optimizing the chromatographic method was performed using the freshly prepared highest calibration.

Sample Preparation
Three sets of six replicates of samples were prepared to evaluate the matrix effect (ME) and the recovery (RE) for each analyte [55]. Set1 was obtained by adding 100 µL of the MixPt solution directly to a wipe cartridge [32], letting it dry for 10 min at room temperature, and desorbing it with 1.99 mL DS mixture. Set2 was made by desorbing the blank wipe cartridge with 2 mL of DS mixture, transferring 1.89 mL of this solution to a vial, and adding 100 µL of MixPt. Finally, Set3 was prepared by diluting 100 µL of MixPt solution with 1.89 mL of DS mixture. To each vial, 10 µL of IS working solution was added.
A solution of 5 mg/mL toluene in DS mixture was prepared to evaluate the void volume and the volume of the water layer adsorbed on the stationary phase. Each injected solution was filtered through a 0.2 µm GHP Acrodisc ® syringe filter before analysis (Pall Corporation).
The method was tested in an actual sampling procedure inside an antineoplastic drug preparation unit after consistent use of PtADs. Wipe samples were prepared and collected, according to Dugheri et al. [32], at the beginning and the end of the work shift. Each collection consisted of 29 sampling points, chosen among the most likely contaminated surfaces.

Experimental Design
All computations were performed using the Chemometric Agile Tool (CAT, open source, free, developed from R, The R Foundation for Statistical Computing, available at https://www.r-project.org/, accessed on 22 March 2021) [56].

Chromatography and Instrument Parameters
Concerning the chromatography, an Agilent Poroshell 120 ® HILIC-Z 2.1 · 100 mm 2.7 µm particle size (Agilent Technologies, Santa Clara, CA, USA) (S.N. USHXF02517) column was used. The mobile phase consisted of a mixture of acetonitrile and water, with organic modifier ranging from 85% to 90%, at various ammonium formate concentrations and different pH values. The pH of the mobile phase was adjusted with formic acid or with ammonium hydroxide. Finally, the mobile phases were sonicated for 15 min to clear dissolved gasses from the solution and avoid gas bubbles in the LC pumps. The eluent was delivered isocratically at a flow rate of 0.6 mL/min, in a total run time of 5.5 min.
The void time of analysis and thus the retention factor (k) were assessed by estimating the water layer adsorbed on the surface of HILIC-Z stationary phase. For this purpose, a 5 mg/mL toluene solution was analyzed under two different elution conditions: firstly, utilizing an isocratic flow of pure acetonitrile, and secondly applying the optimized multiple reaction monitoring (MRM) analysis mobile phase as the solvent, which was acetonitrile 90%, H 2 O 10% with 20 mM ammonium formate. Ten injections for each condition were repeated to establish the difference between the retention times of the analyte in acetonitrile (t R(ACN) ) and the mobile phase (t R(Phase) ). Toluene was detected at a wavelength of 254 nm by a photodiode array (PDA) detector. The water layer volume (V w ) and the volume ratio (β) were then calculated according to Guo et al. [57] through the following equations: Sample injection was carried out using a co-injection pretreatment program. The pretreatment optimization was fulfilled by repeating 3 MRM runs of the high calibration standard level for each injection condition and evaluating its response in peak shape and column overload. Co-injection was tested by loading the injection loop with 3 mixed aliquots made with increasing volumes of the sample (between 1 and 5 µL) and 8 µL of acetonitrile each, with a total injection volume of 27 to 39 µL. The MS/MS experiment co-injection pretreatment program consisted in three drawings of 2 µL from the sample vial alternated with three drawings of 8 µL from another vial containing acetonitrile.
The column thermostat was set at a temperature of 30 • C. The settings of the ESI source, operating in positive ion mode, were the following: interface voltage 4 kV, nebulizing gas flow 3 L/min, heating gas flow 10 L/min, interface temperature 300 • C, desolvation temperature 526 • C, desolvation line temperature 250 • C, heat block temperature 400 • C, and drying gas flow 10 L/min.

MS/MS Experiments
Positive Q3 scan spectra of the 2 µg/mL solutions of analytes were acquired in a range from 300 to 500 m/z with a scan time of 0.100 s.
The product ion scan (PIS) spectra of the 2 µg/mL solutions of analytes were acquired to study the fragmentation of each molecule in an m/z range from 60 to 400 with a 0.100 s scan time and using 270 kPa of argon as collision gas. Each solution was analyzed via flow injection analysis by increasing collision energies (CEs): from 5 to 55 V for oxaliplatin and carboplatin and from 5 to 35 V for cisplatin. Each run was performed with 5 µL of sample injection and 0.5 µL/min of flow, with an acquisition time of 0.6 min. The collision breakdown curves, graphics that show the relative abundance of each fragment ion produced by the analytes as the function of increasing CE, were built using the relative intensity values of each signal present in the MS/MS spectra.
The LC-MS/MS analyses were performed in positive mode MRM using a dwell time of 109 msec. The transition ions and energies are reported in Table 1.

Performance Evaluation of LC-MS/MS Methods
The method precision was assessed as intraday and interday repeatability (intermediate precision) by the relative standard deviation percent (RSD%) of replicate analysis of low, medium, and high CQI levels. The intraday repeatability was evaluated from six sets of calibration and CQI solutions processed in a single day. To assess the interday repeatability of the method, three groups of calibration and CQI solutions were freshly prepared and analyzed every day for six days. Calibration curves were obtained by plotting the peak area ratios (PARs) between the analyte and IS quantitation ions versus the nominal concentration of the calibration solution. A linear regression analysis was applied to obtain the best fitting function between the calibration points. Limits of detection and quantitation (LOD and LOQ) were calculated according to the International Council for Harmonisation (ICH) guidelines using the approach based on the standard error of the y-intercept and slope of the regression [58].
The accuracy was determined by the ratio between the determined and added amounts of analytes to CQI solutions and expressed as mean RE percent. ME and RE were calculated for each analyte by comparing the mean results of Set1, Set2, and Set3 described according to the following formula: IS was added to Set1, Set2, and Set3 to check chromatographic and injection conditions but was not used to evaluate ME and RE.

Experimental Design
The chromatographic method was initially studied by the conventional trial-and-error approach. After some days of experiments, however, the need for a systematic approach to find suitable conditions for the elution of the three target compounds became apparent. Therefore, a D-optimal design was adopted to exploit all the possible experiments already performed and add only the experiments necessary to compute a model for studying three responses, namely the retention time, the peak width, and the sensitivity (measured by the peak areas) of the detection of the three analytes [59,60].
Four factors were considered, as reported in Table 2. The factors selected were the content of organic in the mobile phase (ACN from 85 to 90%), the amount of ammonium formate (FA, from 5 to 20 mM), the pH of the aqueous mobile phase (3.0 to 7.0), and the column compartment temperature (from 10 to 30 • C). The model selected was the canonical full factorial 2 4 with two-term interactions. Therefore, eight experiments carried out by the heuristic approach could be used (experiments numbered from 1 to 8) to complete a balanced dataset, adding only seven experiments (experiments numbered from 9 to 15) as shown in Table 2. After computing the model, six additional independent experiments were carried out for model validation. The results obtained (reported in Table S1) showed that only the models describing retention and peak width were applicable, whereas the four factors did not significantly affect the detection sensitivity. Regarding the retention times of the three analytes, the models computed (Supplementary Material, Table S1) show that the percent of ACN, of course, had a significant impact on retention (i.e., the higher the %ACN, the stronger the retention of the analytes and the longer the retention times). More surprisingly, pH also had a significant effect on the retention of the three analytes. Carboplatin was the more strongly retained when pH was raised from 3 to 7, whereas cisplatin retention was the least susceptible to the mobile phase pH variation. Although the effect of temperature on retention times was statistically non-significant and the corresponding coefficients in the model were numerically very small, the interaction between temperature and pH instead resulted in significant and important factors determining the analytes' retention times. In particular, the interaction contributed more to increasing retention when both temperature and pH were at their high levels (pH 7 and 30 • C) or their low levels (pH 3.0 and 10 • C).
Out of the three models studied for peak width, the only valid model computed was that of cisplatin. A possible explanation for this outcome is that cisplatin is the less retained analyte between the three, and thus the peak width is less affected by measurement uncertainty, resulting in smaller variance associated with the model coefficients. The peak width model for cisplatin showed that ACN% of the mobile phase, temperature, and their interaction were the more relevant effects, and also statistically significant. Narrower peak widths were obtained when the lower % of ACN (85%) and the lower temperature (10 • C) were applied. However, the ACN%-temperature interaction effect counteracted the former two (see Supplementary Table S1, note the signs of the model's coefficients).
After evaluating the results, the best compromise to obtain the maximum retention and narrower peaks for the three analytes was using a mobile phase containing 90% acetonitrile, 20 mM ammonium formate at pH 6.4, and keeping the column temperature at 30 • C.

Chromatographic Conditions
The best co-injection program obtained consisted in three drawings of 2 µL from the sample vial alternated by three drawings of 8 µL from another vial containing acetonitrile, with a total injection volume of 30 µL. The pretreatment co-injection program allowed an increase in the total sample injection volume from 1.5 µL to 6 µL without altering peak shape or causing column overload.
The chromatographic profiles obtained display a good resolution between the platin molecules and a partial overlapping between cisplatin and the IS, as shown in Figure 1. Table 3 reports the retention time values, peak width, theoretical plate number, and retention factor. Table 3. The analytes' retention time (R T ), peak width (Width 1/2 ), tailing factor 5% (T f ), asymmetry factor 10% (A f ), and their relative standard deviation (RSD), the calculated plate number (N), and the retention factor (k).  Since the water layer adsorbed on the column particles is part of the stationary phase, the toluene retention time observed in the analysis conditions (tR(Phase)) can be considered the void time of analysis and is thus used to calculate the retention factor. Table S2 in the Supplementary Materials section reports the observed values for tR(Phase), tR(ACN), Vw, and β.

Mass Spectrometry
The  Figure 2. The analysis of these data allowed for selecting the most suitable product ions and their optimal CE to set up the respective MRM methods.  The data in Table 3 (see also the chromatogram in Figure 1) evidence that the separation between the three platinum-containing analytes is complete. Although cisplatin co-elutes with the IS, this is not an issue since the MS detection perfectly discriminates the two analytes. The analysis takes only about five minutes, and the precision of all chromatographic parameters (retention times, peak width, tailing factor, asymmetry factor) is satisfactory. The figure for the number of theoretical plates (N) is reported to document the column efficiency of the method at the time of its initial development. The data can be used (and were used) to build control charts to monitor the column efficiency in routine applications. The retention factors within the interval of 1 < k < 10 were judged to be suitable for the goals of the separation based on the laboratory's internal expertise.
Since the water layer adsorbed on the column particles is part of the stationary phase, the toluene retention time observed in the analysis conditions (t R(Phase) ) can be considered the void time of analysis and is thus used to calculate the retention factor. Table S2 in the Supplementary Materials section reports the observed values for t R(Phase) , t R(ACN) , V w , and β.

Mass Spectrometry
The  Figure 2. The analysis of these data allowed for selecting the most suitable product ions and their optimal CE to set up the respective MRM methods.

Mass Spectrometry
The most abundant signal of the unmodified molecule cluster was chosen as the parent ion for the MS/MS analysis, [M + H] + = 398.05 m/z for oxaliplatin, [M + H] + = 372.1 m/z for carboplatin, and [M + NH4] + = 317.9 m/z for cisplatin. Collision breakdown curves were created for each molecule with the data obtained from PIS analysis and are reported in Figure 2. The analysis of these data allowed for selecting the most suitable product ions and their optimal CE to set up the respective MRM methods.

Calibration Curves
Pharmaceutical preparations have been used instead of pure standards to create the calibration curves to obtain a more accurate view of the compound's response, i.e., one that takes account of excipients which may cause a significant variability in the analyte signal. Table S3 reports the obtained linear regression data LOD and LOQ values for each analyte for interday and intraday repetitions. Table S4 shows the results for ME and RE experiments. The values for ME ≥ 81% and RE ≥ 74% are satisfactory for the present study's aims. Table S5 reports accuracy and precision figures of merit achieved for the three CQI levels during validation. The obtained precision (RSD between 3 and 9%) and accuracy (ratio between 91 and 111%) only show margins of error exceeding 10% for concentrations close to the LOQ. In the frame of the present study, these results were thus considered acceptable.

In-Field Method Application
From a panel of 58 samples, 4 end-shift wipes showed contamination inside the laminar flow hood (LFH). A scheme of the positive results is shown in Table S6 in the Supplementary Materials. Cisplatin contamination was detectable but lower than the LOQ limit.

Discussion
The analytical methods for biological monitoring of ADs that are currently available are not sensitive or specific enough for platinum-based compounds [61]. Consequently, the best choice to evaluate occupational exposure is surface monitoring. A wipe test is a widely used method in industrial hygiene and, currently, the only one able to detect ADs at levels as low as pg/cm 2 . Healthcare workers' exposition risk depends on AD toxicity and how these drugs enter the body; this should guide their handling protocols.
The necessity to improve the sensitivity and quantification of polar compounds has led to the increased utilization of HILIC combined with ESI-MS detection. The acetonitrilerich mobile phases employed in HILIC provide favorable conditions for efficient droplet formation and desolvation within the MS source. The zwitterionic (sulfoalkylbetaine) stationary phases contain both negatively charged sulfonate groups and positively charged quaternary ammonium groups in an equimolar ratio, separated by a short alkyl spacer. The sulfobetaine phase retains water strongly, allowing for creating an enriched water layer on its surface even at a high acetonitrile content in the mobile phase, which plays a critical role in retaining polar compounds [62][63][64]. Unfortunately, at the same time, the reconstitution of the thick water layer requires a long time after a gradient run. This last issue led to the exclusion of the gradient elution, which would have made the analysis time extremely long, favoring the isocratic one. This choice, even if causing broader peaks and low retention for less well-retained molecules, allowed both good separation and a symmetric peak shape, along with short times for analysis. No carry-over effect or pressure rise caused by column contamination was observed during the analysis.
From earlier studies [32], to obtain a correct wipe desorption, a 50:50 water:methanol solution is needed to extract both hydrophilic and hydrophobic ADs. With the future perspective of unifying the studied molecules with other AD classes, this desorbing solution has been chosen for this study even if the hydrophilicity of PtADs would have suggested a higher water percentage. However, both ways would have led to chromatographic issues caused by the high water quantity in the injection volume. This would result in a poor peak shape, along with the possible loss of retention of the less well-retained substances. This issue could be resolved with injection volumes greater than 1.5 µL. The problem was solved thanks to the co-injection program (described in Section 3.2) possible with the SIL-30AC autosampler to modify the pretreatment program and mix the sample, right inside the injection loop, with minor amounts of acetonitrile present in another vial. This way, the autosampler works as a sample preparation unit, balancing the percentage of organic solvent in the injected solution and increasing the injection sample volume from 1.5 to 6 µL without affecting the chromatography. The loop capacity, in this case, 50 µL, must be considered when creating a co-injection method, since larger volumes would be excessive and thus partially discharged.
Though the simple suction of 6 µL of sample and 24 µL of acetonitrile at once would have required a faster pretreatment, the "sandwich-like" modality (2 + 8 µL · three times) is preferable because it allows better mixing of sample and solvent before the injection. In both ways, the pretreatment program increases the total time of analysis by 40 s per sample at most. It is essential to ensure that the decreasing percentage of water in the mixture does not lead to precipitation of the compounds inside the sample loop and cause obstructions in the column.
For what concerns the MS development, platinum-based compounds are undoubtedly challenging molecules. PtADs, especially cisplatin, are easily subjected to modifications and addition of molecules of water and acetonitrile to the coordination complex [63], which leads to the presence of many signals in the positive scan analysis of each pure standard. Furthermore, the natural occurrence of five stable isotopes of platinum, 192 Pt,194 Pt,195 Pt,196 Pt,and 198 Pt, leads to peculiar isotopic clusters, also complicating the setup of MS/MS methods.
After fragmentation studies and selection of the optimal transitions to set MRM analysis, the obtained abundances for carboplatin and oxaliplatin allowed us to use a range of concentrations between 5 and 50 ng/mL due to their steady coordination structure. At the same time, cisplatin's low signal intensity required higher levels of the drug to create an acceptable calibration curve (200-2000 ng/mL).
The LODs and LOQs obtained are comparable to those found in the literature range of ng/mL-even if it is difficult to compare them since sample matrixes are different [44,45]. ME and RE results showed that wipe sampling is a suitable option for PtAD surface monitoring. These results show that a small percentage of the drug is retained from the wipe matrix, but the recoveries are still encouraging for a high-throughput monitoring plan.
As reported in Table S5 in the Supplementary Materials, the method has good repeatability, precision, and accuracy in the interday and intraday repetitions, with a larger variability-still under 15%-only in proximity to the LOQ concentrations.
The analysis of the wipe samples from the AD preparation unit showed a modest contamination inside the laminar flow hood which, in agreement with previous results [33], is one of the surfaces with a higher risk of contamination.

Conclusions
The inclusion of cisplatin in the group list of probably cancerogenic to human agents (2A) by the International Agency for Research on Cancer and the risk of toxicity due to dermal absorption concerning all PtADs require the development of an analytical method able to identify and quantify the target analytes to assess surface contamination. Herein is reported the development and optimization through a chemometric model of the LC-MS/MS method for the simultaneous separation and quantification of carboplatin, cisplatin, and oxaliplatin at the ng/mL level. The separation of the analytes was performed using a solid-core zwitterionic-HILIC column. Reliable qualitative and quantitative analysis can be performed with a fast and simple sample preparation and run time shorter than 7 min. The pretreatment co-injection program allowed us to increase the injection volumes without altering the chromatographic profile of the substances and improving method sensibility. The DoE approach permitted a substantial reduction of the number of tests needed for the method's development. Despite the molecular instability of platinum-based complexes, the obtained results showed adequate precision and accuracy. The present LC-MS/MS method can thus be proposed as a high-throughput procedure for surface contamination monitoring of PtADs.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/separations8050069/s1, Table S1. Models coefficients computed and validation studied on six independent replicates of experiments in the point with coded coordinates (1, 1, 0.7, 0.6) corresponding to the experimental conditions selected for the analyses, i.e., mobile phase containing ACN 90% (v/v), ammonium formate 20 mM, pH 6.4, and column temperature set at 30 • C. Table S2: Shows the observed values from which was calculated the column volume ratio in the chromatography operating conditions. Table S3. Linear regressions data, R 2 , LOD, and LOQ of each analyte. Table S4. Data results for matrix effect (ME) and recovery (RE). Table S5. Data results of precision and accuracy for the three CQI levels. Table S6

Conflicts of Interest:
The authors declare no conflict of interest.