Development and validation of a multiplex UHPLC‐MS/MS assay with stable isotopic internal standards for the monitoring of the plasma concentrations of the antiretroviral drugs bictegravir, cabotegravir, doravirine, and rilpivirine in people living with HIV

Abstract The widespread use of highly active antiretroviral treatments has dramatically changed the prognosis of people living with HIV (PLWH). However, such treatments have to be taken lifelong raising issues regarding the maintenance of both therapeutic effectiveness and long‐term tolerability. Recently approved or investigational antiretroviral drugs present considerable advantages, allowing once daily oral dosage along with activity against resistant variants (eg, bictegravir and doravirine) and also parenteral intramuscular administration that facilitates treatment adherence (eg, long‐acting injectable formulations such as cabotegravir and rilpivirine). Still, there remains a risk of insufficient or exaggerated circulating exposure due to absorption issues, abnormal elimination, drug‐drug interactions, and others. In this context, a multiplex ultra‐high performance liquid chromatography coupled to tandem mass spectrometry (UHPLC‐MS/MS) bioassay has been developed for the monitoring of plasma levels of bictegravir, cabotegravir, doravirine, and rilpivirine in PLWH. A simple and convenient protein precipitation was performed followed by direct injection of the supernatant into the UHPLC‐MS/MS system. The four analytes were eluted in less than 3 minutes using a reversed‐phase chromatography method coupled with triple quadrupole mass spectrometry detection. This bioassay was fully validated following international guidelines and achieved good performances in terms of trueness (94.7%‐107.5%), repeatability (2.6%‐11%), and intermediate precision (3.0%‐11.2%) over the clinically relevant concentration ranges (from 30 to 9000 ng/mL for bictegravir, cabotegravir, and doravirine and from 10 to 1800 ng/mL for rilpivirine). This sensitive, accurate, and rapid UHPLC‐MS/MS assay is currently applied in our laboratory for routine therapeutic drug monitoring of the oral drugs bictegravir and doravirine and is also intended to be applied for the monitoring of cabotegravir/rilpivirine levels in plasma from PLWH receiving once monthly or every 2‐month intramuscular injection of these long‐acting antiretroviral drugs.


Funding information
Swiss National Science Foundation, Grant/ Award Number:  cabotegravir/rilpivirine levels in plasma from PLWH receiving once monthly or every 2-month intramuscular injection of these long-acting antiretroviral drugs.

K E Y W O R D S
antiretroviral therapy, long-acting injectables, pharmacokinetics, therapeutic drug monitoring,

| INTRODUCTION
Optimal efficacy and good tolerability are key points during the development of antiretroviral (ARV) drugs. 1 Yet, besides therapeutic effectiveness and drug safety profile, long-term adherence is required to achieve viral suppression. 2,3 The development of long-acting injectable (LAI) formulations can overcome the adherence issue 4 by maintaining effective plasma concentrations over months. Thus, LAI has the potential to improve adherence thereby preventing drug resistance. In addition, LAI can improve patients' privacy and reduce social stigmas associated with daily intake of ARV drugs. It has been stated that about as much as 50% to 70% of people living with HIV (PLWH) would be interested in LAI formulations when available. 5 Cabotegravir and rilpivirine are the first two drugs of LAI formulation, currently in final phase of clinical development. 6,7 Cabotegravir is a potent HIV integrase strand transfer inhibitor (INSTI), 8 while rilpivirine is non-nucleoside HIV reverse transcriptase inhibitor (NNRTI). Long plasma half-life of both substances made them good candidates for the development of LAI formulations administered monthly 9,10 or every 2 months. 11 In addition to HIV treatment, LAI-ARV drugs are also investigated separately in the indication of pre-exposure prophylaxis (PreP).
Whether used for treatment or prevention, important pharmacokinetic variability was shown following intramuscular injection of cabotegravir and rilpivirine in clinical trials. 9,10,[12][13][14] These clinical studies have generally included carefully selected PLWH, who may not reflect the complex situation in a real-life clinical setting. In particular, drug-drug interactions (DDIs) are likely to occur, 15 also with LAI-ARV drugs, and we have at present very limited information on their actual clinical importance, prompting the monitoring of ARV plasma levels when new comedications at risk of DDIs are introduced in patients on LAI-ARV drugs. Besides, intersubject variability may be more pronounced particularly in special population (ie, underweight or obese patients, hepatic or renal impairment, aging, or pregnancy).
In addition to these novel injectable formulations, ARV developments are also focused on improving the safety and tolerability profile. The last-generation ARV drugs bictegravir (a potent unboosted INSTI 16 ) and doravirine (a next-generation NNRTI 17 ) represent attractive oral therapeutic options because of their improved tolerability profiles. Both bictegravir and doravirine are substrates of CYP3A4 and can consequently be victims of DDIs. However, there is currently a lack of data concerning the magnitude of DDIs with these novel ARV drugs. Yet, in the next few years, most PLWH in middle-and high-income countries will switch to one of these last-generation ARV therapies, either oral or LAI formulations.
The availability of liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) methodologies for the determination of ARV concentrations in human plasma is a key aspect for drug pharmacokinetic studies and therapeutic drug monitoring (TDM) in patients.
Several assays have been previously developed for the measurement of rilpivirine as oral formulation. [18][19][20] To the best of our knowledge, only two LC-MS/MS assays have been published for the quantification of bictegravir in human plasma. 21,22 In addition, although cabotegravir and doravirine plasma concentrations have been determined in several studies, 23,24 no publication has been dedicated to the development and the validation of such LC-MS/MS methodologies.
In this article, we aimed at developing and validating a simple and fast multiplex assay by ultrahigh-performance liquid chromatography coupled to tandem mass spectrometry (UHPLC-MS/MS) for the simultaneous determination of the latest generation ARV drugs bictegravir, cabotegravir, doravirine, and rilpivirine in human plasma.

| Stock solutions preparation
Each analyte was weighed and dissolved in the required volume of solvent. Stock solutions of bictegravir (1 or 5 mg/mL), cabotegravir (1 mg/mL), and doravirine (2 mg/mL) were prepared in DMSO.
Rilpivirine powder was dissolved in a mixture of DMSO:MeOH 1:1 to obtain the final concentration of 0.5 mg/mL. These stock solutions were stored at −20 C for bictegravir, cabotegravir, and doravirine.
The stock solution of rilpivirine was stored at +4 C, as currently done for the routine monitoring of rilpivirine plasma concentrations in the framework of our TDM service. 18 One working solution (WS) at 100 μg/mL for bictegravir, cabotegravir, and doravirine and 20 μg/mL for rilpivirine was prepared All solutions were stored at −20 C.

| Plasma pre-treatment procedure
Protein precipitation was operated by adding a 300-μL volume of the IS-WS to 100 μL of calibration, validation, or patient plasma samples.
The mixture was vortexed and centrifuged at 18 620g (14 000 rpm) at +4 C for 10 minutes with a benchtop centrifuge (Benchtop Mikro 220R centrifuge, Hettich, Bäch, Switzerland). Three hundred microliters of the supernatant were directly transferred into an HPLC vial with insert.

| Analytic conditions
The mobile phases (ie, H 2 O + 0.1% FA (A) and ACN + 0.1% FA (B)) were delivered at a flow rate of 300 μL/min, following this multistep gradient: first, linear gradient from 40% to 60% B in 3 minutes, up to 95% B in 0.2 minutes, followed by an isocratic stage at 95% B for 0.8 minutes. Then, solvent B was reduced to 40% (initial conditions) in 0.1 minute, followed by a re-equilibration step up to 5 minutes (total F I G U R E 1 Chemical structures of the analyzed antiretroviral drugs analysis time). Samples were stored at +5 C in the autosampler, and the injection volume was 7 μL.
Polarity switching capability enabled ESI positive (spray voltage 3900 V) and negative (spray voltage 3400 V) analysis in the same sample injection. ESI source parameters were optimized as follows: the ion transfer tube and vaporizer temperatures at 300 C and 150 C, respectively; sheath, auxiliary, and sweep gas flow rates at 45, 25, and 0 (arbitrary units), respectively. The first (Q1) and third (Q3) quadrupoles operated with a mass resolution of 1.  An LC-MS/MS method is considered devoid of significant ME if RSD value is <4%.

| Trueness, precision, accuracy profiles, limits of quantification, and linearity
Trueness and precision of the method were assessed over three different days. Several regression models were fitted to adequately describe the response concentration profile. The selection of the best calibration model was based on the estimations of trueness and precision, the narrowest β-expectation tolerance interval, and the lowest LLOQ. 33 Concentrations of the validation standards were back-calculated with the daily calibration curve. The trueness (systematic error) was defined as the percentage of deviation between the calculated concentrations of validation standards and the nominal value. The precision (random error) was estimated by two components: the repeatability (intraday variances) and intermediate precision (intraday and interday variances). [34][35][36] Precision parameters were reported as RSD at each concentration level. 33 The total error encompassed both systematic and random errors and was evaluated using accuracy profiles. βexpectation tolerance intervals represent the concentration range where β% of future results is expected to lie. [37][38][39] Using data obtained during the validation phase, this approach allows to confidently predict the future results that will be obtained during the routine use of the method. Based on the absolute accuracy profiles, LLOQ was graphically interpolated as the lowest concentration for which the β-expectation tolerance interval crosses the acceptance limits (±30%). 25,26,40 Finally, the capacity of the method to give quantitative results proportional to nominal concentrations was evaluated by ordinary least square regression on the plot representing back-calculated concentrations vs nominal concentrations. This defines the linearity of trueness and was assessed each day of validation.

| Measurement uncertainty
An analytical result should also be reported with respect to its mea-

| Stability studies
Stability studies included bench-and long-term stabilities. The stability of plasma at room temperature (RT) and in the fridge (+4 C) up to 48 hours was evaluated. In addition, stability after three freeze/thaw cycles was assessed by thawing frozen samples at RT for 1 hour and refreezing them during 1 hour, three times in a row. Furthermore, plasma samples were submitted to thermal viro-inactivation process (60 min at +60 C in a water bath) since this procedure has been shown to efficiently inactivate HIV particles present in the samples. 43,44 Finally, medium stability was evaluated with plasma samples frozen at −20 C and −80 C during 6 weeks. Analyses were performed in triplicate. The mean of the concentrations obtained after each stability study were compared with the mean concentration of samples prepared at time 0.

| Analytical method development
The optimization of the LC-MS/MS assay aimed at improving sensitivity while minimizing runtime. First, standard solutions of each analyte at 5 μg/mL in MeOH were directly infused into the MS detector in order to select optimal MS/MS, as reported in Table 1. LC-MS/MS transitions for bictegravir and rilpivirine differed from reported values.  3 25 In the only published bioassays for bictegravir, mass spectrometer operated in ESI positive mode, while in our study, bictegravir sensitivity was higher in the negative mode. 21   ACN. Different injection volumes of pretreated samples ranging from 3 to 10 μL were tested, and a volume of 7 μL was finally chosen as the best compromise between suitable sensitivity and satisfactory peak shape. As shown in Figure 2, an adequate separation of the four analytes was achieved in less than 3 minutes, with satisfactory peak shapes.

Compound ESI polarity (+/−) Precursor Ion (m/z) Product Ion (m/z) Collision Energy (V) Typical Retention Time, min
The MS part of the analytical assay was optimized (as reported in Section 2.6) by choosing the appropriate ESI source parameters to improve sensitivity while minimizing background noise.
Finally, IS concentrations were selected to obtain satisfying ISnormalized response functions, by avoiding variability due to low IS concentrations and by circumventing a significant contribution of IS signal to analyte signal in case of excessive IS concentrations.

| Selectivity and carryover
The good selectivity of the chromatographic method was demonstrated with the absence of interference at the retention times of the four analytes when analyzing human blank plasma from 10 different sources.
The injections of a blank plasma processed with IS-WS or the highest calibration standard processed with MeOH:ACN (1:1) did not reveal any significant signal on the analytes or IS transitions, respectively, demonstrating the absence of cross talks.

| ME, ER, and PE
As shown in Figure 3, no major interferences (ie, ion suppression or enhancement) were observed at analytes' retention times. This result supports the suitability of the chromatographic method, preventing an impact of endogenous plasma components on the ionization process of the four analytes and ISs.
Quantitative results of the assessment of IS-nME, IS-nER, and is-nPE are summarized in Table 2. The IS-nME of the analytes was considered satisfactory and varied from −6% to 12%, while RSD values were lower than 6%. Regarding IS-nER and IS-nPE, acceptable results were observed with values ranging from −15% to 4% and −16% to 5%, respectively, with RSD lower than 10%. Overall, matrix-matched calibration along with the use of isotopically labelled IS was found to adequately limit MEs issues.

| Trueness, precision, and accuracy profile
A β value of 80% was chosen for the establishment of βexpectation tolerance intervals, representing the fraction of future results that would be expected to fall within the obtained tolerance intervals during routine application of the method. 46 As demonstrated in Figure 4, accuracy profiles obtained for each compound lie within the acceptance limits of ±30% for biological samples. 26 Since the β-expectation tolerance interval of bictegravir, cabotegravir, and doravirine does not cross the acceptance limits of ±30%, the LLOQ was defined as the concentration of the lowest validation sample (ie, 30 ng/mL). Considering an accuracy of ±30% (total error), the lowest concentration measurable in human plasma (LLOQ) was 10 ng/mL for rilpivirine.
Linearity was considered satisfactory since slopes and intercepts ranged from 0.96 to 1.03 and −75.4 to 40.2, respectively. In addition, determination coefficient (R 2 ) were all higher than .99.

| Measurement uncertainty
The absolute uncertainty vs concentration profiles were best described by polynomial (bictegravir, cabotegravir, and doravirine) and power regression models (rilpivirine). The relative uncertainty of each compound at each validation levels is shown in Table 3. With a confidence level of 95%, the unknown true value located at maximum ±27.2%, ±41.4%, ±14.8%, and ±21.4% around the measured result for bictegravir, cabotegravir, doravirine, and rilpivirine, respectively.

| Stability studies
Stability of analytes in plasma is reported in Table 4. Results demonstrated that analytes did not significantly degrade after storage of plasma samples at room temperature and +4 C for up to 48 hours. In addition, no significant alteration of plasma concentrations was observed after three consecutive freeze-thaw cycles (variation < ±15%). The thermal viro-inactivation process had no significant influence on analytes concentrations, with variations comprised between −6.5% and 10.6%. Finally, medium-term stability studies showed no significant influence of degradation after 6 weeks of freezing at −20 C and −80 C.

| Clinical applications
The proposed LC-MS/MS assay has been applied to patient's samples obtained for clinical purposes, within the framework of our TDM service. A typical chromatographic profile of a plasma from an HIVinfected patient receiving bictegravir 50 mg once daily is shown in Figure 5A.  F I G U R E 6 Steady-state pharmacokinetic profile of bictegravir when administered at 50 mg once daily. Continuous green line represents the population median prediction. Dark green and light green shaded area represent the 50% and 90% prediction interval, respectively. Bictegravir plasma concentrations obtained from our therapeutic drug monitoring (TDM) service in patients included in the Swiss HIV Cohort Study has been superimposed (black points) [Colour figure can be viewed at wileyonlinelibrary.com]