Multiplex Liquid Chromatography-Tandem Mass Spectrometry Assay for Simultaneous Therapeutic Drug Monitoring of Ribavirin, Boceprevir, and Telaprevir

Abstract

New directly acting antivirals (DAAs) that inhibit hepatitis C virus (HCV) replication are increasingly used for the treatment of chronic hepatitis C. A marked pharmacokinetic variability and a high potential for drug-drug interactions between DAAs and numerous drug classes have been identified. In addition, ribavirin (RBV), commonly associated with hemolytic anemia, often requires dose adjustment, advocating for therapeutic drug monitoring (TDM) in patients under combined antiviral therapy. However, an assay for the simultaneous analysis of RBV and DAAs constitutes an analytical challenge because of the large differences in polarity among these drugs, ranging from hydrophilic (RBV) to highly lipophilic (telaprevir [TVR]). Moreover, TVR is characterized by erratic behavior on standard octadecyl-based reversed-phase column chromatography and must be separated from VRT-127394, its inactive C-21 epimer metabolite. We have developed a convenient assay employing simple plasma protein precipitation, followed by high-performance liquid chromatography coupled to tandem mass spectrometry (HPLC-MS/MS) for the simultaneous determination of levels of RBV, boceprevir, and TVR, as well as its metabolite VRT-127394, in plasma. This new, simple, rapid, and robust HPLC-MS/MS assay offers an efficient method of real-time TDM aimed at maximizing efficacy while minimizing the toxicity of antiviral therapy.

INTRODUCTION

Hepatitis C virus (HCV) infection is a leading cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma. It is estimated that up to 210 million individuals worldwide are chronically infected with HCV and that 50% of these infections are due to genotype 1 HCV (1). The current standard of care for genotype 1 HCV infection with pegylated alpha interferon plus ribavirin (PEG-IFN-α/RBV) eradicates the infection in only about 40% of cases and is associated with substantial side effects (2, 3). HCV therapy entered a new era with the FDA and European Medicines Evaluation Agency approval of the directly acting antivirals (DAAs) telaprevir (TVR) and boceprevir (BOC) (4, 5). Both of these drugs are small lipophilic inhibitors of the HCV NS3-4A protease. They increase sustained virological response (SVR) rates to approximately 70% in treatment-naive genotype 1 HCV-infected patients (6, 7). They also significantly increase SVRs in those who previously failed therapy (8, 9).

In addition to the first DAAs BOC and TVR, other newer potent oral anti-HCV drugs (i.e., faldaprevir, simeprevir, sofosbuvir, daclatasvir, and asunaprevir) (10–12) are in the final stage of clinical development or are to be approved soon. Some of these new drugs even open the way to IFN-free regimens, which will undoubtedly revolutionize the treatment of HCV infection.

While these new agents offer unique opportunities to improve the management of HCV-infected patients, there are several important issues to consider. The currently approved DAAs have short half-lives and therefore need to be administered every 8 h together with a meal, which increases their absorption and exposure (by 2- to 4-fold for TVR), albeit with substantial interindividual variability. In addition, these drugs are extensively metabolized by cytochrome P450 and are substrates and inhibitors of P glycoprotein, and thus there is a risk of multiple interactions with various substances, notably, antiretroviral drugs, immunosuppressive agents, and antidepressant medications. Furthermore, the emergence of drug resistance mutations has been observed in patients who fail therapy (13).

Therapeutic drug monitoring (TDM) is now current practice for the optimization of treatment with a number of drugs (i.e., immunosuppressants, antibiotics, anti-HIV drugs, etc.) (14). Generally, the maintenance of circulating drug concentrations over a given threshold is crucial to ensure optimal antiviral action, since suboptimal drug concentrations allow low-level viral replication, which substantially increases the risk of viral resistance and virological failure. On the other hand, avoidance of unnecessarily high concentrations may limit the development of dose-dependent adverse effects.

RBV, a very polar drug, is an essential component of the anti-HCV therapy regimen but is commonly associated with hemolytic anemia, which requires close blood cell count monitoring and possibly RBV dose adjustment and the use of erythropoiesis-stimulating agents. Nonrandomized studies have shown that the TDM of plasma RBV levels improves the management of the therapeutic response and hematologic toxicity (15, 16), despite the publication of controversial reports (17). Correlations between plasma concentrations of RBV and anemia, as well as SVR rates, have been reported, with a suggested minimum trough level above 2 μg/ml at week 4 to secure SVR in patients infected with HCV genotype 1, and a maximum below 3.5 μg/ml to limit the risk of anemia and treatment discontinuation (18–20). In that context, several analytical methods of high-performance liquid chromatography coupled to tandem mass spectrometry (HPLC-MS/MS) or ultraperformance liquid chromatography (UPLC)-UV have been proposed for the quantification of RBV in plasma and erythrocytes (21–25).

The importance of TDM for the protease inhibitors (PIs) TVR and BOC has not yet been studied. However, the monitoring of TVR and BOC concentrations in patients undergoing combined HCV therapy seems to be warranted in specific situations such a drug-drug interactions (26), virological failure, the occurrence of dose-dependent adverse drug reactions, and selected clinical conditions (hepatic and renal failure, special patient populations). It is still uncertain whether a single cycle of TDM after treatment introduction might be beneficial to all patients.

To that end, an LC-MS/MS assay for the simultaneous measurement of TVR and BOC has been described (27) that uses the unrelated compound dimethyl celecoxib as an internal standard (IS) and a gradient program that does not allow the separation of TVR from VRT-127394, the virologically inactive C-21 epimer metabolite of TVR. A report published in abstract form by Chakilam et al. (28) briefly described the separation of TVR and its isomer VRT-127394 by a liquid-liquid extraction method with toluene, followed by evaporation under nitrogen of the organic solvent and reconstitution in a heptane-acetone mixture prior to HPLC-MS/MS analysis. Recently, a UPLC-MS/MS method has been reported that allows the reverse-phase separation of BOC, TVR, and their isomers with a solvent gradient composed of ammonia, methanol, and acetonitrile and uses quinoxaline as an IS for the quantification of DAAs (29).

In the perspective of developing a TDM program to improve the efficient management of anti-HCV drugs, quantification of the levels of all of these anti-HCV drugs in plasma must be made available to clinicians within 24 to 48 h. Quantification of different anti-HCV drugs in a single analytical run may best fulfill these requirements. An assay for the simultaneous analysis of HCV inhibitors now available presents an analytical challenge, however, because of the large differences in polarity of anti-HCV drugs, ranging from hydrophilic (RBV) to intermediate and highly lipophilic drugs (BOC and TVR, respectively) and because TVR is characterized by erratic behavior on standard octadecyl-based reversed-phase column chromatography and must also be differentiated from its major inactive metabolite, VRT-127394.

We therefore aimed to develop a simple, rapid, robust, and efficient HPLC-MS/MS assay for the simultaneous determination of RBV, BOC, and TVR in patient plasma. The extraction procedure by protein precipitation is straightforward and convenient, stable-isotope-labeled ISs are used for selective MS quantification, and the chromatographic conditions ensure that TVR and its C-21 epimer are separated. The development of this method was designed to meet the laboratory and clinical requirements of routine implementation.

MATERIALS AND METHODS

Chemicals, reagents, and plasma.

TVR, BOC, RBV, and the IS [¹³C-5]RBV were purchased from Toronto Research Chemicals, Inc. (North York, ON, Canada), and BOC-d9 was purchased from Alsachim (Strasbourg, France). TVR-d11 and the epimer VRT-127394 were kindly provided by Janssen Pharmaceuticals (Beerse, Belgium). The chemical structures of the respective molecules are shown in Fig. 1. Chromatography was performed with Lichrosolv HPLC grade methanol (MeOH) purchased from Merck (Darmstadt, Germany). Ultrapure water was obtained from a Milli-Q UF-Plus apparatus (Millipore Corp., Burlington, MA). Formic acid (FA; 98% pure) and isopropanol (IPA) for chromatography were purchased from Merck (Darmstadt, Germany). All chemicals were of analytical grade.

Fig 1.

Chemical structures of RBV, BOC, TVR, and VRT-127394. Shown are the chemical structures of the anti-HCV drugs RBV {1-[(2R,3R,4S,5R)-3,4-dihydroxy-5-(hydroxymethyl)oxolan-2-yl]-1,2,4-triazole-3-carboxamide)} (A), BOC {(1R,2S,5S)-N-(4-amino-1-cyclobutyl-3,4-dioxobutan-2-yl)-3-[(2S)-2(tertbutylcarbamoylamino)-3,3-dimethylbutanoyl]-6,6-dimethyl-3-azabicyclo[3.1.0]hexane-2-carboxamide} (B), TVR {(3S,3aS,6aR)-2-[(2S)-2-[[(2S)-2-cyclohexyl-2-(pyrazine-2-carbonylamino)acetyl]amino]-3,3-dimethylbutanoyl]-N-[(3S)-1-(cyclopropylamino)-1, 2-dioxohexan-3-yl]-3,3a,4,5,6,6a-hexahydro-1H-cyclopenta[c]pyrrole-3-carboxamide} (C), and VRT-127394 (R diastereoisomer of TVR) (D).

Blank plasma samples used for matrix effect (ME) assessment and for the preparation of calibration and control samples were obtained from citrated blood (1,850 × g, 10 min, +4°C, Beckman J6B centrifuge) collected from Vaquez disease patients on the occasion of their regular phlebotomy.

The blank plasma used for the preparation of the calibration and quality control (QC) samples was acidified with 10% FA (50 μl of 10% FA added to 950 μl of plasma). The acidification of plasma aims at preventing the conversion of TVR to its epimer VRT-127394 that occurs in vivo and in vitro. (Tibotec-Janssen, personal communication).

Equipment.

The LC system used consisted of Rheos Allegro quaternary pumps equipped with an online degasser and an HTS PAL autosampler (CTC Analytics AG, Zwingen, Switzerland) controlled by Janeiro-CNS 1.1 software (Flux Instruments AG, Thermo Fischer Scientific Inc., Waltham, MA). Separations were done on a Hypercarb 3-μm column (2.1 mm ID by 100 mm; Thermo Fischer Scientific) placed in a column oven thermostat regulated at +80°C (HotDog 5090; ProLab GmbH, Reinach, Switzerland). The chromatographic system was coupled to a triple-stage quadrupole quantum mass spectrometer (Thermo Fischer Scientific) equipped with an electrospray ionization (ESI) Ion Max interface and operated with the Xcalibur software package (version 2.0; Thermo Fischer Scientific).

IS, calibration standard, and QC solutions.

An IS solution was obtained by diluting and mixing TVR-d11, BOC-d9, and [¹³C-5]RBV stock solutions at 1 mg/ml in MeOH to obtain a single IS solution of 1,000 ng/ml for TVR-d11 and BOC-d9 and 4,000 ng/ml for RBV-c5.

Stock solutions of TVR and BOC at 1 mg/ml in MeOH and of RBV at 2 mg/ml in H₂O were prepared. Stock solutions were diluted and combined to obtain a single working solution in MeOH at final concentrations of 250 μg/ml for the HCV PIs and 1,000 μg/ml for RBV. The working solution was diluted and pooled with citrated acidified blank plasma (50 μl of 10% FA added to 950 μl of plasma) (see below) to obtain seven calibration samples (25, 50, 150, 250, 1,000, 1,250, and 2,000 ng/ml for the HCV PIs and 100, 200, 600, 1,000, 4,000, 5,000, and 8,000 ng/ml for RBV). QC sample solutions (see Table S1 in the supplemental material) were prepared in the same way as the calibration samples to obtain three plasma QC samples (low [LQC], medium [MQC], and high [HQC] QC samples, i.e., 75, 500, and 1,500 ng/ml for both HCV PIs, and 300, 2,000, and 6,000 ng/ml for RBV). The LQC sample has a concentration corresponding to three times the lowest calibration level, in accordance with FDA recommendations (30). All solutions were prepared according to the recommendations for bioanalytical method validation, which state that the total added volume must be <10% of the biological sample volume. The calibration standard and control plasma samples were stored as 100-μl aliquots at −80°C.

LC-MS/MS conditions.

Mobile phase A was composed of ultrapure H₂O plus 0.1% FA, mobile phase B was a 50:50 mixture of H₂O-MeOH plus 0.1% FA, and mobile phase C was IPA plus 0.1% FA.

The mobile phase was delivered by stepwise gradient elution according to the sequence reported in Table S2 in the supplemental material. The thermostat-regulated column oven was set at +80°C, and the autosampler was maintained at +10°C. The injection volume was 10 μl. The optimal potential settings and MS/MS transitions were chosen by the direct infusion into the MS/MS detector of a solution of analytes at a concentration of 1 μg/ml in MeOH. The first (Q1) and third (Q3) quadrupoles were set at a 1-amu mass resolution (full-width half maximum = 0.7). The scan time and width were 0.04 s and 1.0 m/z, respectively, and each chromatographic peak was the result of ca. 30 scans. The ionization conditions were as follows. The capillary temperature was set at 350°C. The ESI spray voltage was set at 4 kV, and the source-induced dissociation was set at 10 V. The sheath and auxiliary gas (nitrogen) flow rate were set at 35 and 10 (arbitrary units), respectively. The tube lens voltages ranged from 58 to 97 V, and the Q2 collision gas (argon) pressure was 1 mtorr (1 torr = 1.333 × 10² Pa). Chromatographic data acquisition, peak integration, and quantification were performed with the Xcalibur LC-Quan software of the Excalibur package. The mass spectrometer was operated with the ESI source Ion Max in the positive mode. Samples were analyzed via the selected reaction monitoring detection mode, by employing the transition of the [M+H]⁺ precursor ions to product ions. The optimal settings are listed in Table 1.

Table 1.

Instrument method of LC-MS/MS analysis of antivirals and their respective stable-isotope-labeled ISs

Compound	Product (m/z)	Fragment (m/z)	CE (eV)a	Tube lens voltage	Mean RT (min)b
RBV	245	113	17	58	2.5
[13C-5]RBV	250	113	17	58	2.5
BOC	520.4	308.1	25	82	6.1
BOC-d9	529.5	308.1	25	82	6.1
TVR	680.5	322.1	25	97	8.2
TVR-d11	691.5	322.1	25	82	8.2

^aCE, collision energy.

^bRT, retention time.

A seven-point calibration standard curve was calculated and fitted by quadratic log-log regression of the peak area ratios (drug peak area/IS peak area) versus concentrations. The concentration calibration range was selected to cover the clinically relevant range of concentrations expected in patients.

Blood sample collection.

Blood samples were collected from patients as part of their usual medical follow-up and/or as part of multicenter research protocol Swiss HIV Cohort Study (SHCS) 688 (Prevalence, Relevance and Determinant of HCV Drug Resistance in the SHCS) within the framework of the SHCS, which was approved by the Institutional Ethics Committees.

Blood samples (3 ml) from patients receiving antiviral therapy (RBV, TVR-RBV, or BOC-RBV) were collected in Monovettes (Sarstedt, Nümbrecht, Germany) containing citrate as an anticoagulant. Samples were centrifuged without delay at 1,850 × g (3,000 rpm) for 10 min at +4°C (J6B centrifuge; Beckman Coulter, Nyon, Switzerland). The plasma was separated, and 2.0-ml aliquots were transferred into 5-ml polypropylene test tubes to which 100 μl of 10% FA was added before storage at −80°C. At the initial step of validation with TVR-RBV patient samples, this acidification procedure was done with a separate plasma aliquot kept solely for TVR assay, for which isomerization is an issue. After having formally ascertained that acidification does not affect the concomitant RBV measurements (see below), the whole plasma from TVR-RBV patient samples was acidified prior to storage at −80°C.

Plasma extraction procedure.

A 100-μl aliquot of acidified plasma was mixed with 50 μl of IS working solution (containing TVR-d11, BOC-d9, and RBV-c5) and 450 μl of MeOH and carefully vortexed. The mixture was centrifuged at +4°C for 10 min at 20,000 × g (14,000 rpm) with a benchtop centrifuge (Benchtop Universal 16R; Hettich, Bäch, Switzerland). A 300-μl aliquot of the clear supernatant was transferred into glass HPLC vials that were tightly closed with crimp seals. A volume of 10 μl was injected into the HPLC-MS/MS apparatus for analysis.

Analytical method validation.

The method validation procedure used was based on the recommendations posted online by the Food and Drug Administration (FDA) (30), as well as on the recommendations in references 31 and 32. Other, more recent recommendations for method validation by Matuszewski et al. (33, 34) were also considered, as were those reported in reference 35.

LLOQ.

The lowest limit of quantification (LLOQ) of each drug was determined as the minimal concentration in plasma that could be quantified with a ±20% deviation between the measured and nominal concentrations, in accordance with FDA recommendations (30).

Accuracy and precision.

Three concentrations for QC samples were used to cover the wide ranges of the calibration curves corresponding to drug concentrations reported to occur in clinical samples, i.e., LQC, MQC, and HQC samples. The concentrations selected for the LQC samples were three times the respective LLOQs (i.e., the lowest calibration concentrations), in accordance with the FDA recommendations (30). Replicate analysis (n = 6) of the QC samples was used for assessment of intra-assay precision and accuracy. Interassay accuracy and precision were determined by duplicate analyses of the three QC samples repeated on 6 different days. Precision was calculated as the coefficient of variation (%CV) of the measured concentrations within a single run (intra-assay) and among different assays (interassay), and accuracy was calculated as the bias or percentage of deviation between experimental and nominal concentrations. The analytical run was validated only if the percentages of deviation (bias) of the back-calculated (experimental) concentrations from the nominal concentrations of each calibration standard and QC sample were less than ±15%.

Quantification of antiviral concentrations in plasma and whole blood over time under different storage conditions.

Quantification studies under different storage conditions included the following. The stability of TVR in acidified plasma (950 μl of plasma acidified with 50 μl of 10% FA) and nonacidified plasma at room temperature (RT) for up to 8 h was investigated to assess the interconversion of TVR to VRT-127394. QC sample concentrations in plasma over time at RT and +4°C were measured for up to 48 h. Variations of antiviral concentrations were expressed as percentages of the initial concentration measured immediately after preparation. Analyses were performed in duplicate at each subsequent time point. Antiviral concentrations in whole-blood samples at RT and +4°C were measured. A 10-ml volume of citrated whole blood was spiked with antivirals at the LQC, MQC, and HQC concentrations, distributed as 550-μl blood aliquots, and stored for up to 48 h at RT and +4°C. Immediately after preparation, three blood aliquots were centrifuged at 1,850 × g (3,000 rpm) for 10 min at +4°C and plasma was frozen at −80°C for the determination of time zero (T₀) values. Subsequent blood aliquots were collected 1, 2, 3, 4, 8, 24, and 48 h after preparation. Plasma was separated as described above. All plasma aliquots were analyzed in the same analytical run, and the quantification of antiviral concentrations over time was expressed as the percentage of deviation from the initial values at T₀. The relative percentages of TVR and its epimer VRT-127394 in patient plasma were measured by centrifugation immediately after the collection of blood either subjected or not to the above-described acidification procedure with FA. For the quantification of antiviral concentrations in plasma samples after multiple freeze-thaw cycles, three series of QC samples at LQC, MQC, and HQC concentrations underwent three freeze-thaw cycles. Frozen plasma samples were allowed to thaw at RT and were subsequently refrozen for approximately 2 h. Antiviral concentrations were measured in aliquots obtained from each of the three consecutive freeze-thaw cycles. For the quantification of antiviral concentrations in plasma samples frozen at −80°C, QC samples at the LQC, MQC, and HQC concentrations were stored at −80°C for 3 months and measured with reference to freshly prepared plasma calibration samples.

MEs, extraction yield, and overall recovery.

In the initial step of method validation, the MEs were examined qualitatively by the simultaneous postcolumn infusion of the antivirals and corresponding IS into the MS/MS detector during the chromatographic analysis of six different blank plasma extracts. The standard solution of all of the analytes at 1 μg/ml was infused at a flow rate of 10 μl/min during the chromatographic analysis of blank plasma extracts from six different sources. The chromatographic signals of each selected MS/MS transition were examined to check for any signal perturbation (drift or shift) of the MS/MS signal at the analytes' retention time.

Subsequently, the MEs were also quantitatively assessed. Three series of QC samples at LQC, MQC, and HQC concentrations were processed in duplicate as follows: (i) pure standard stock solutions dissolved in the MeOH solvent and directly injected onto a column, (ii) plasma extract samples from six different sources spiked after extraction with drugs and ISs from pure standard sample solutions, and (iii) plasma samples from six different sources (same as in series ii) spiked with drug standard solutions and ISs before extraction.

The recovery and ion suppression/enhancement of the MS/MS signal of drugs in the presence of plasma matrix (i.e., ME) were assessed by comparing the absolute peak areas of the analytes either spiked into MeOH or spiked before or after plasma extraction with six different batches of plasma, based on the recommendations of Matuszewski et al. (34).

The extraction yields of antiviral drugs and ISs were calculated as the absolute peak-area response in processed plasma samples spiked with drugs before extraction (C) expressed as the percentage of the response of the same amount of drugs spiked into blank plasma after the extraction procedure (B) (C/B ratio in percent). MEs were assessed as the ratio of the peak areas of the analytes spiked into blank plasma after the extraction procedure (B) to the peak areas of the analytes solubilized in MeOH (B/A ratio in percent). This ratio was also calculated by using B2 and A2 values (i.e., normalized with IS area). The overall recoveries of antivirals and ISs were calculated as the ratio of absolute peak-area responses of the analytes spiked into processed plasma samples before extraction (C)—such as calibration and control samples—to the peak areas of the analytes solubilized in MeOH (C/A ratio). Recovery studies were performed with plasma from six different sources spiked with analytes and respective ISs. The results normalized to the IS signal (i.e., B2 and C2) were used as an index of the effective injection volume and for matrix effect correction.

Selectivity.

Assay selectivity was assessed by analyzing plasma extracts from 10 batches of blank plasma from different sources.

RESULTS

Chromatograms.

Reverse-phase chromatographic separation of RBV, BOC, and TVR (structures are shown in Fig. 1) was done with a Hypercarb 3-μm graphite UPLC column (2.1 mm ID by 100 mm) in a thermostat-controlled oven at +80°C by using the stepwise gradient elution program described in Table S2 in the supplemental material, starting with pure H₂O plus 0.1% FA with a progressive concomitant increase in the percentage of MeOH/H₂O at 1:1 plus 0.1% FA and IPA plus 0.1% FA delivered at 0.3 ml/min. The retention times of RBV, BOC, TVR, and its isomer VRT-127394 were 2.5, 6.1, 8.2, and 8.7 min, respectively. Figure 2A shows a typical chromatographic profile of a calibration sample containing RBV, BOC, and TVR at 4,000, 1,000, and 1,000 ng/ml, respectively. A minor signal arising from the endogenous plasma compound uridine (21, 23) was observed at ca. 3 min at the MS/MS transition selected for RBV. This signal was present to variable extents in both calibration and patient samples (Fig. 2A and 3A and B, respectively), but because uridine was well separated chromatographically from RBV (eluted at 2.5 min), its presence did not interfere with the assay of RBV. Importantly, with the proposed gradient program, the signal of the inactive metabolite VRT-127394, the C-21 epimer of TVR, was satisfactorily resolved from the parent drug, which allows the selective quantification of virologically active TVR only (Fig. 2B and 3A). Typical chromatographic profiles of HCV patients receiving RBV-TVR and RBV-BOC are shown in Fig. 3A and B, respectively. The concentrations of RBV and TVR (Fig. 3A) and RBV and BOC (Fig. 3B) were 4,702, 1,958, 2,444, and 226 ng/ml, respectively.

Fig 2.

(A) Chromatograms of standard calibration samples (RBV at 5,000 ng/ml, BOC at 1,250 ng/ml, and TVR at 1,250 ng/ml). (B) Chromatogram of a plasma calibration sample containing TVR at 1,000 ng/ml to which the epimer VRT-127394 at 1,000 ng/ml was added.

Fig 3.

(A) Typical LC-MS/MS profile of plasma from a patient with chronic hepatitis C treated with PEG-IFN-α, RBV, and TVR. The TVR epimer VRT-127394 is observed at 8.7 min, after the TVR peak at 8.2 min. (B) Typical LC-MS/MS profile of plasma from a patient with chronic hepatitis C treated with PEG-IFN-α, RBV, and BOC. For details, see the text.

ISs and calibration curves.

Deuterated analogues of all antiviral drugs were used as ISs. The use of stable-isotope-labeled ISs is the first-choice approach to minimizing the influence of MEs on the accuracy and precision of a quantitative method, which is of particular importance when using ESI-MS (33, 34). The levels of control samples were selected to reflect the low, medium, and high ranges of the calibration curves. The use of stable-isotope-labeled RBV, BOC, and TVR standards has been integrated since the start of the analytical method validation procedure, and they are used throughout the routine TDM assay process.

Calibration curves over the entire range of concentrations were satisfactorily described by 1/x weighted quadratic regression of the peak-area ratio of antiviral drugs to their ISs versus the concentrations of the respective antiviral drugs in each standard sample. Over the concentration range considered, the regression coefficient (R²) of the calibration curves was always greater than 0.999 with back-calculated calibration samples within ±15% (±20% at LLOQ).

Precision, accuracy, LLOQ, and limit of detection (LOD).

The precision and accuracy values determined with the LQC, MQC, and HQC samples are shown in Table 2. The mean intra-assay precision was less than 5.9%. Overall, the mean interday precision was good, with CVs within 2 and 5%. The intra-assay and interassay deviations (bias) from the nominal concentrations of QC samples of the three antiviral drugs are summarized in Table 2. The deviations from the nominal levels were between −2.2 and 4.2%, −2 and 4.3%, and −2.9 and 0.9% for RBV, BOC, and TVR, respectively.

Table 2.

Precision and accuracy of LHQ, MHQ, and HQC samples determined by repeated analysis performed on 6 different days (interassay) and on the same day (intra-assay)^a

Compound and nominal concn (ng/ml)	Intra-assay (n = 6)			Interassay (n = 6)
	Measured concn (ng/ml)	Precision (%CV)	Accuracy (% bias)	Measured concn (ng/ml)	Precision (%CV)	Accuracy (% bias)
RBV
300	299	4.2	−2.2	301	5	2.5
2,000	2,036	1.6	4.2	2,038	2	6.1
6,000	6,022	2.7	0.4	6,056	2.4	0.9
BOC
75	73.5	4	−2	76.1	3.5	3
500	503	1	0.7	502.6	2.3	1.8
1,500	1,528	5.9	4.3	1,520	3	0.9
TVR
75	72.7	5.1	−2.9	72.2	4.7	1.5
500	499	2.2	−0.1	498	2	0
1,500	1,513.7	3.1	0.9	1,517	4	3

^aAll analysis were performed in duplicate.

The lowest achievable LODs of RBV, BOC, and TVR were 5, 5, and 25 ng/ml, respectively. The LLOQs of RBV, BOC, and TVR were 25, 25, and 100 ng/ml, respectively, and accordingly correspond to the lowest levels of calibration.

MEs and recovery.

The MEs were first examined qualitatively as illustrated in Fig. 4. The transition signal selected shows no significant shifts or drift for TVR and BOC, whereas a substantial increase in the signal intensity of the MS transition selected for RBV was observed at the retention time of RBV, presumably because of the elution of endogenous polar components from plasma that are eluted early on a graphite reversed-phase-type column. The assessments of the MEs, extraction recovery, and overall method recovery of all of the drugs are reported in Table 3. For all of the analytes, the extraction recovery was within 79.7 and 96.3%. The mean MEs were 249, 113, and 110% for RBV, BOC, and TVR, respectively. Most importantly, the B2/A2 ratio for RBV was 103.2%, indicating that the use of [¹³C-5]RBV successfully corrects for the important ME observed for the polar drug RBV. The mean overall process efficiencies (PEs) were 210% for RBV and around 100% for the PIs. For all of the analytes, the variability of the PE was always very low (<7%). (Table 3).

Fig 4.

Evolution of the m/z transition signals of the antiviral drugs and respective IS during the chromatographic analysis of six different blank plasma extracts with simultaneous postcolumn infusion of each antiviral drug at 1,000 ng/ml. For clarity, the chromatographic profile of a QC sample is superimposed.

Table 3.

ME, extraction recovery, analysis, and process efficiency of RBV, BOC, and TVR in six different plasma matrices^a

Drug and nominal concn (ng/ml)	Mean ME (%)				Mean ext RE (%)		Mean analysis RE (%)		Mean PE (%)
	B/A	%CV	B2/A2	%CV	C/B	%CV	C2/B2	%CV	C/A	%CV
RBV
300	270.1	8.8	107.3	3.6	84.1	5.4	84.2	4.1	227.2	5.4
2,000	224.3	4.0	101.8		90.6	4.0	89.1		203.2	4.9
6,000	253.6	8.3	100.4		79.7	3.0	82.5		202.2	5.9
BOC
75	113.6	8.2	105.5	2.4	91.9	4.2	88.0	4.2	104.5	5.3
500	105.9	6.2	101.9		95.4	4.0	89.2		101.1	4.6
1,500	122.1	9.1	100.8		81.7	3.7	82.5		99.7	6.6
TVR
75	108.8	8.8	104.5	3.0	92.2	4.8	88.5	4.4	100.4	5.7
500	103.7	5.9	101.8		96.2	4.0	88.7		99.8	4.8
1,500	117.7	8.4	98.4		80.5	4.2	82.0		94.8	5.6
[13C-5]RBV, 1,000	251.5	2.9			99.9	2.2			251.5	2.3
	220.3	2.1			101.6	2.9			224	3.1
	252.6	1.9			96.6	1.4			244.2	1.8
BOC-d9, 1,000	107.6	2.3			104.5	1.6			112.5	1.9
	103.9	1.9			107	2.9			111.2	1.3
	121.1	2.5			99	1.5			120	2.6
TVR-d11, 1,000	104	2.4			104.1	1.7			108.4	2.3
	101.8	1.7			108.6	2.9			110.6	1.8
	119.5	2.8			98.2	1.5			117.4	1.9

^aAnalysis were performed in duplicate. A, peak area of standard solutions without matrix and without extraction. B, peak area of analytes spiked after extraction. C, peak area of analytes spiked before extraction. B2, ratio of the peak area of the analyte and the IS spiked after extraction. C2, ratio of the peak area of the analyte and the IS spiked before extraction. The ME is expressed as the ratio of the mean peak area of the analytes spiked after the extraction (B) to the mean peak area of the same standard solution without the matrix (A) multiplied by 100. ext RE, extraction procedure recovery calculated as the ratio of the mean peak area of the analyte spiked before extraction (C) to the mean peak area of the analytes spiked after extraction (B) multiplied by 100. Analysis RE, analysis recovery calculated as the ratio of the mean peak area ratio of the analyte/IS spiked before extraction (C2) to the mean peak area ratio of the analyte/IS spiked after extraction (B2) multiplied by 100. PE, process efficiency expressed as the ratio of the mean peak area of the analyte spiked before extraction (C) to the mean area of the same analyte standard (A) multiplied by 100.

Quantification of antiviral drugs over time under different storage conditions.

A comparison of TVR stability in plasma either acidified (with 50 μl of 10% FA/950 μl of plasma) or nonacidified is illustrated in Fig. 5A and B. These results underscore the importance of adding FA to plasma to prevent the epimerization of TVR to VRT-127394 (Fig. 5B).

Fig 5.

Stability experiments assessing the evolution of TVR (A) and VRT-127394 (B) formation in plasma either acidified (with 50 μl of 10% FA/950 μl of plasma) or not acidified.

Stability data for RBV, BOC, and TVR in whole blood and in acidified plasma are reported in Table 4. The variations in the level of each drug over time indicate that RBV can be considered stable in plasma for up to 48 h at RT and +4°C. On the other hand, BOC and TVR are stable in acidified plasma only for up to 8 h at RT and for up to 48 h at +4°C.

Table 4.

Stability of antiviral drugs in acidified plasma at RT and 4°C^a

Time (h)	Acidified plasma at RT									Acidified plasma at 4°C
	RBV			BOC			TVR			RBV			BOC			TVR
	LQC	MQC	HQC	LQC	MQC	HQC	LQC	MQC	HQC	LQC	MQC	HQC	LQC	MQC	HQC	LQC	MQC	HQC
0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	−7.3	0.0	1.4	−9.7	−3.1	0.6	−4.2	0.8	−1.3	−0.2	5.5	−3.3	−0.7	2.0	−4.6	3.3	5.2	−3.2
2	2.9	3.2	−1.6	−8.1	−1.4	−2.3	−3.7	−2.6	−3.3	1.5	4.9	−2.2	0.0	2.3	−4.1	0.8	3.8	−1.2
4	−1.3	3.0	0.9	−9.6	−4.6	−4.1	−8.6	−5.3	−4.1	2.2	6.5	−2.6	1.1	1.7	−4.2	2.6	4.7	−2.5
8	1.2	3.2	0.6	−12.8	−6.7	−7.2	−12.0	−8.6	−10.6	3.2	8.2	−4.2	−0.1	3.5	−4.2	2.8	5.2	−0.7
24	0.1	6.5	0.8	−27.5	−19.9	−24.9	−24.2	−19.9	−23.3	3.7	3.4	−2.2	−1.6	0.5	−4.6	−1.5	0.1	−2.8
48	−4.7	2.7	1.5	−45.2	−36.8	−43.7	−43.9	−39.9	−44.2	3.9	2.2	−3.7	−2.7	−2.7	−4.8	0.8	−0.7	−5.8

^aPercent deviation from the starting level at T₀ is shown. Analyses were performed in duplicate.

In contrast, the analysis of whole blood (see Table S3 in the supplemental material) indicates that BOC is stable in whole blood for up to 24 h at RT and for up to 48 h at +4°C. However, these in vitro studies carried out with spiked whole blood revealed significant decreases in RBV of −24.6% at +4°C and −41.8% at RT after 1 h and a mean decline in TVR of −20.5% after 2 h at RT, with a concomitant increase in the epimer VRT-127394.

We have ascertained whether the acidification step with FA, recommended by the manufacturer for the stabilization of TVR (Tibotec-Janssen, personal communication), could be omitted to simplify the overall sample handling by nursing staff and lab technicians in the routine clinical setting. The mean relative percentages of TVR and its epimer VRT-127394 in patient plasma obtained after centrifugation performed immediately after blood collection were 66 and 67% in plasma, with and without acidification with FA, respectively. This indicates that under carefully controlled conditions (i.e., immediate plasma centrifugation after blood collection prior to storage at −80°C), the acidification procedure would not be necessary for TVR.

Overall, these stability results indicate that centrifugation of whole blood must be performed immediately (definitely within 30 min) after patient sample collection to avoid spurious levels of RBV and TVR in the plasma. In addition, overestimation of plasma RBV concentrations can happen if blood hemolysis has occurred, even to a small extent.

Table S4 in the supplemental material shows the variations of antiviral drug concentrations in plasma samples stored at −80°C after one, two, and three freeze-thaw cycles. The variations were always less than 15%, indicating no significant loss of drug after up to three freeze-thaw cycles. For long-term stability testing, LQC, MQC, and HQC samples prepared with acidified plasma were stored in a freezer (−80°C) for 3 months and then compared to freshly prepared QC samples. The variation was within −5.2 and −8.4%, suggesting that no analyte decomposition occurred in acidified plasma.

Selectivity.

In all of the blank plasma samples and patients analyzed so far, a single peak from an endogenous compound, reportedly uridine (21, 23, 25), was detected at 3 min at the mass transition selected for RBV, however, without perturbing the quantification of RBV eluted at 2.5 min.

Application of TDM of RBV, TVR, and BOC.

The performance of the newly developed multiplex assay for TDM of RBV, TVR, and BOC was evaluated in the setting of a routine clinical chemistry laboratory and was able to provide real-time plasma drug level measurements for patients with chronic hepatitis C on PEG-IFN-α and RBV combination therapy or triple therapy comprising BOC or TVR (Fig. 3A and B). Figure 6A, B, and C show the plasma RBV, BOC, and TVR levels, respectively, plotted against time after drug intake, that were determined at unselected times (TVR and RBV) and during the trough period (i.e., ca. 8 h after BOC intake) in patients with chronic hepatitis C during a medical visit. Application of this assay in the routine setting has already provided some insights into the marked interindividual variability in drug exposure (Fig. 6A, B, and C) with, in some instances, a correlation with adverse effects. For example, a high plasma RBV level of 5,731 ng/ml was measured 8 h after drug intake (arrow at the upper right of Fig. 6A) in a female patient receiving 400 mg of RBV twice daily who developed pronounced anemia (hemoglobin level, 83 g/liter). In the samples from patients on TVR analyzed so far (Fig. 6), which were taken at unselected times after the last drug intake, the signal intensity of the metabolite VRT-127394 corresponds to about 36% (range, 30 to 39%) of the peak area of the parent drug, TVR. This ratio is stable over the dosing interval (see Fig. S1 in the supplemental material) and is in accordance with previously published data (36).

Fig 6.

Drug levels in plasma plotted against time after the intake of the anti-HCV drugs RBV (A), BOC (B), and TVR (C).

DISCUSSION

We have developed and validated a multiplex assay method that requires simple plasma protein precipitation with MeOH, followed by direct injection of the supernatant onto a Hypercarb graphite UPLC column prior to detection by MS/MS analysis to perform the simultaneous quantification of RBV and the two HCV NS3-4A PIs TVR and BOC. The development of a multiplex assay represented an analytical challenge since RBV, BOC, and TVR are three molecules with marked differences in polarity, precluding a priori a single generic sample treatment for their simultaneous bioanalysis. Thus, the major achievement of our preanalytical treatment was to optimize the biological extraction procedure for all three drugs together.

HCV PIs, BOC, and especially TVR are characterized by erratic behavior on standard octadecyl-based reversed-phase column chromatography, as evidenced by the limited number of reports on the assay of these new drugs that have been published so far (27, 28, 37). In our study, several chromatographic substrates (C₁₈, propylcyano, bridged ethyl siloxane/silica hybrid [BEH], pentafluorophenyl, hydrophilic interaction liquid chromatography [HILIC], etc.) were found not to be suitable for providing a satisfactory peak shape for all three anti-HCV drugs with suitable separation of TVR and its inactive epimer metabolite. Of note, it appears that a C₁₈-based column such as Atlantis BEH and T3 (Waters) provides a chromatographic profile consisting of as many as eight peaks for pure TVR. This suggests the existence of putative TVR conformers or atropoisomers occurring in solution that, for some reason, are presumably stable enough to be separated in these recent C₁₈ packing materials. Such separation performance paradoxically represents more of a limitation than an advantage in the case of TVR, especially given its known propensity to epimerize at C-21 to give VRT-127394. Finally, a Hypercarb graphite column heated at +80°C provides a satisfactory peak shape for TVR and BOC even if we chose it initially because it had been reported as suitable for highly polar molecules (21, 23). Importantly, the Hypercarb graphite column must be kept at +80°C in a thermostat-regulated oven for this assay, not only to reduce the important chromatographic back pressure but also to improve the peak shapes of BOC, TVR, and its epimer. This column was found to have excellent stability under such conditions (more than 800 injections without technical problems), but 0.1% FA was used instead of the trifluoroacetic acid (0.1 to 10%) recommended by the Hypercarb column manufacturer to improve peak shapes and reduce potential carryover issues. It is advisable to use a thermostated oven that can automatically switch off in case of incidental LC pump stoppage, to avoid excessive temperature elevation of the solvent remaining immobilized into the column. Optimization of the late elution gradient program with IPA to increase the lipophilicity of the mobile phase to be adapted to the apolar HCV PIs results in the excellent separation of BOC and TVR, with the latter also being well separated from its major metabolite, VRT-127394. Indeed, the separation of TVR from its epimer was a priority in this assay's development. As previously stated, TVR (the S stereoisomer) can epimerize at position C-21 both in vitro and in vivo to yield a mixture of S and R C-21 epimers in human plasma at a reported 60:40 ratio at equilibrium (36). Since VRT-127394 is 30-fold less active than TVR (36), it must be chromatographically separated from TVR to avoid spurious overestimation of the level of the active antiviral species in plasma.

Of note, BOC also occurs as two isomers in vivo (37) but since the drug BOC, unlike TVR, is administered to patients as a diastereoisomeric mixture, there is limited interest in differentiating between the two isomers, especially since all of the pharmacokinetic parameters established so far are for BOC as a diastereoisomeric mixture.

The important ME observed for RBV was associated with limited interplasma variability over the concentration range (Table 3). Since all calibration samples were prepared in plasma, the largest part of this important ME was accounted for and circumvented. Finally, the use of [¹³C-5]RBV as an IS further decreases the impact of this ME, as shown by the mean B2/A2 ratio of about 100% (Table 3).

Preanalytical and stability issues had to be thoroughly investigated before considering the implementation of this assay in a routine setting. The stability studies indicate that BOC is stable in whole blood at both 4°C and RT, whereas a marked decrease in plasma RBV and TVR levels is observed when spiked whole blood is left at either RT or +4°C. In these in vitro experiments, performed after the addition of the drugs to blank whole blood, the observed decreases in RBV and TVR levels are likely to occur through different mechanisms. RBV is thought to decrease in plasma possibly because of progressive uptake into red blood cells, a phenomenon that is reduced but not prevented at 4°C. A similar phenomenon has been previously observed for ganciclovir, another nucleotide antiviral (38, 39). In patient samples, however, the equilibrium between RBV levels in plasma and red blood cells is already attained; therefore, red blood cell hemolysis must be avoided to prevent a spurious increase in RBV levels in plasma because of its release from lysed red blood cells. On the other hand, the decrease in plasma TVR levels is probably a consequence of its rapid epimerization into VRT-127394 (which can be progressively detected chromatographically) in the absence of FA in whole-blood samples.

Thus, quantification of BOC in whole blood stored at RT and +4°C is accurate for up to 24 and 48 h, respectively. In contrast, blood samples containing RBV and TVR need to be processed immediately in order to avoid a significant alteration of plasma drug levels, notably, for RBV in the presence of hemolysis. Blood samples from patients on TVR may be stored at +4° but definitely for less than 120 min to avoid unacceptable interconversion into its metabolite prior to centrifugation and plasma acidification.

Of importance, we have finally found that if patient blood samples are centrifuged immediately after collection, the relative percentages of TVR and VRT-127394 remain stable in the collected plasma samples stored at −80°C, even if the plasma acidification step with FA is omitted. The frozen plasma samples can thus be conveniently shipped to the laboratory, where the acidification step will be carried out with thawed plasma just before its analysis. This simplifies the overall handling of samples from patients on TVR, which is most suitable in the case of multicenter studies.

Although calibrators and QC samples are prepared in plasma acidified with FA for the stabilization of TVR, experiments have shown that the presence or absence of FA added to clinical plasma samples does not affect the simultaneous quantification of RBV and BOC.

Accuracy of quantification is not significantly influenced when plasma samples are subjected to three freeze-thaw cycles, which allows repetition of the analysis with stored frozen samples if confirmatory results are required.

Thus, the above-described experiments provide important information about the handling of blood samples by nursing personnel for shipment and for preanalytical processing and storage in the laboratory. Obviously, the poor stability of RBV and TVR in whole-blood samples will dictate the practical and logistical constraints to the implementation of a suitable TDM service for anti-HCV drugs. Immediately after whole blood is drawn from patients (mostly in an outpatient setting), samples must be sent to the laboratory without delay for centrifugation (RBV and TVR) and immediate acidification with FA (TVR). If the acidification step (mandatory for TVR) is not possible, the plasma must be stored immediately after centrifugation at −80°C and shipped frozen.

Overall, across the concentration ranges chosen for the establishment of the calibration curves, the precision and accuracy of the lowest calibration samples were, for each antiviral drug, within the ±20% limit recommended by the FDA. Of note, the calibration ranges were selected primarily to cover the clinically relevant ranges of drug concentrations reported in the literature (16, 36, 40, 41–46) because samples will presumably be obtained at random times after the last drug intake during a medical visit (i.e., not necessary during the trough period).

As HCV PIs have just entered into clinical use, there is limited information about whether TDM may help to maximize the clinical SVR rate and to reduce the risk of HCV resistance. Their marked interindividual pharmacokinetic variability observed so far, the narrow therapeutic range reported for RBV, and the high frequency of side effects noticed with triple therapy advocate for a formal evaluation of the benefit of TDM for both RBV and the new class of HCV PIs, which could appear of key importance in optimizing the treatment response. TDM might be particularly important in a setting of concomitant antiretroviral and immunosuppressant therapies because of the complex drug-drug interactions expected in such situations. It might also improve our understanding of the relationships between dose concentrations and responses in terms of efficacy or toxicity during triple therapy of HCV infection. This should allow us to weigh the respective risks and benefits associated with various concentrations and to standardize the concentration exposure ensuring an optimal SVR on the one hand and a minimal risk of side effects, notably, severe anemia, on the other hand.

To the best of our knowledge, this is the first multiplex assay proposed for new, currently used anti-HCV drugs. Its convenience and simplicity make it ideal for routine implementation in clinical laboratory settings. The analysis of drugs of the same therapeutic class in a single chromatographic run has the advantage of a unique sample extraction procedure and a shorter analytical run, thus reducing the overall turnaround time. Moreover, it saves time by the establishment of simultaneous calibration curves, it is applicable to blood samples from patients receiving both single-drug RBV, and newer RBV-BOC and RBV-TVR combination regimens, and since analytical results are obtained on a daily basis, TDM interpretation can be delivered rapidly. In fact, this multiplex method for RBV and the first DAAs TVR and BOC provides the foundation for the establishment and use of further multiplex real-time assays for the next-generation anti-HCV drugs under development.

The initial application of this assay in the routine clinical setting will open the way to a formal randomized clinical trial evaluating the clinical usefulness of TDM for new anti-HCV drugs combinations.