Validated Method for the Quantification of Baclofen in Human Plasma Using Solid-Phase Extraction and Liquid Chromatography–Tandem Mass Spectrometry

Abstract

A highly sensitive and fully validated method was developed for the quantification of baclofen in human plasma. After adjusting the pH of the plasma samples using a phosphate buffer solution (pH 4), baclofen was purified using mixed mode (C8/cation exchange) solid-phase extraction (SPE) cartridges. Endogenous water-soluble compounds and lipids were removed from the cartridges before the samples were eluted and concentrated. The samples were analyzed using triple-quadrupole liquid chromatography–tandem mass spectrometry (LC–MS-MS) with triggered dynamic multiple reaction monitoring mode for simultaneous quantification and confirmation. The assay was linear from 25 to 1,000 ng/mL (r² > 0.999; n = 6). Intraday (n = 6) and interday (n = 15) imprecisions (% relative standard deviation) were <5%, and the average recovery was 30%. The limit of detection of the method was 5 ng/mL, and the limit of quantification was 25 ng/mL. Plasma samples from healthy male volunteers (n = 9, median age: 22) given two single oral doses of baclofen (10 and 60 mg) on nonconsecutive days were analyzed to demonstrate method applicability.

Introduction

Gamma-aminobutyric acid (GABA) is the primary inhibitory neurotransmitter in the nervous system and consequently is involved in many brain functions. It has two main receptor systems, the GABA_A receptor whose function is modulated by medications such as benzodiazepines and barbiturates as well as alcohol and the GABA_B receptor, which is modulated by baclofen (β-(4-chlorophenyl)-γ-aminobutyric acid). These drugs have sedative, muscle relaxant and anticonvulsant properties. Baclofen is a licensed antispasmodic, structural analog of GABA and a GABA_B receptor agonist. Baclofen is currently being investigated as a treatment for alcohol dependence and has been shown to increase abstinence and to reduce alcohol-associated craving and anxiety (1, 2). The potential mechanism for this action may be through GABA_B modulation of mesolimbic dopamine neurons, which are involved in meditating the rewarding effects of drugs and alcohol (3). Currently, there is much debate about what dose is effective in treating alcohol dependence with trials examining doses from 30 to 300 mg/day since those with alcohol dependence appear to tolerate such high doses with some believing them to be necessary for effectiveness (4). It is not clear whether the ability to be able to tolerate such high doses is due to altered metabolism of baclofen or in the sensitivity of their target, the GABA_B receptor. To inform the debate about dosing, it is therefore timely to conduct pharmacokinetic–pharmacodynamic studies for which accurate measurement of plasma concentrations of baclofen is required.

In the UK, baclofen is available in a tablet form (10 mg) and in solution for intrathecal administration (0.05–2.0 mg/mL). It is rapidly absorbed after oral ingestion. Approximately 85% of the dose is excreted unchanged in the urine with ∼15% metabolized predominantly into the deaminated form β-(p-chlorophenyl)-γ-hydroxybutyric acid, which is eliminated in both urine and feces (5). The deaminated metabolite is inactive in animals (6), therefore it is sufficient to quantitate the parent drug only. After oral ingestion, the plasma concentration of baclofen in humans reaches peak concentration in ∼2 h and has a half-life of 3–4 h (5). A study by Knutsson et al. showed that the minimum concentration in plasma which caused a therapeutic response was 90 ng/mL, and the highest concentration at which optimal therapeutic response was observed was at 650 ng/mL (7). Jones et al. observed no side-effects in patients who were gradually administered baclofen to a maximum dosage of 60 mg/day (8).

Many methods are available for the analysis of baclofen; they include ion-exchange, solid-phase extraction (SPE) or liquid–liquid extractions with analysis by gas–liquid chromatography (GLC) (7, 9–11), gas chromatography–mass spectrometry (GC–MS) (12, 13) and high-performance liquid chromatography (HPLC) with ultraviolet detection (6, 14–16). More recently, liquid chromatography–tandem mass spectrometry (LC–MS-MS) (17–19) has been employed.

Due to the amphoteric nature of baclofen, it is not possible to extract it efficiently from biological specimens. Many of the published methods require complex multiple steps clean-up and derivatization, which are time consuming and expensive (7, 9, 10, 20).

Most published methods utilized structurally similar compounds to baclofen as an internal standard (9, 10, 17, 18, 21, 22), rather than using the deuterated analog (baclofen-d₄). It is important to use the deuterated analog when analysis is performed by LC–MS-MS as it is the only way to truly compensate for potential matrix enhancement or suppression.

Many LC–MS-MS methods only monitor one product ion (17, 18, 21, 22), while others monitor two (23) or three (19) product ions for both quantification and confirmation. In cases where only one product ion is monitored, there is a potential for reporting a false-positive result. Analysis performed on an Agilent 6460 MS–MS (Agilent Technologies, UK) in the triggered dynamic multiple reaction monitoring (t-DMRM) acquisition mode enables a product ion spectrum to be generated. In the t-DMRM acquisition mode, both quantifier and qualifier product ions (primary ions) are continuously monitored for the analytes in their retention time window. When the primary product ion with the highest m/z reaches a user-defined threshold, the remaining product ions, the secondary ions (maximum of eight), of the analytes are then triggered to be monitored at three discrete points along the analyte's peak, generating a product ion spectrum specific to the analytes. This product ion scan is done in a single analytical run without compromising sensitivity.

The aim of this study was to develop a simple SPE–LC–MS-MS method using deuterated internal standard and t-DMRM mode to allow for simultaneous quantification and confirmation of baclofen in human plasma, and to analyze authentic plasma samples to demonstrate applicability of the method.

Materials and Methods

Chemicals

Baclofen (±) (1 mg/mL) was purchased from Sigma-Aldrich Company Ltd (Poole, UK). Baclofen-d₄ (10 mg) was purchased from LGC Standards (Teddington, UK). Methanol and hexane (HPLC grade), acetonitrile (LC–MS grade), 35% ammonia solution (analytical grade), 90% formic acid (analytical grade) and sodium dihydrogen orthophosphate were all purchased from VWR (Lutterworth, UK). Deionized water (15.0 MΩ cm at 25°C) was produced in-house using a PURELAB Option-R7/15 (ELGA-VEOLIA, UK) water purifier. Narc-2, 3 mL (125 mg, Bakerbond™), SPE cartridges were obtained from Trichema (Cheshire, UK).

Clinical samples

Eight healthy male volunteers completed a double-blind randomized three-way crossover study, in which they received a single oral dose of placebo (vitamin C, 100 mg), 10 mg or 60 mg baclofen with an interval of at least 1 week between each study day. One subject terminated the study after two visits (receiving placebo, 60 mg baclofen only), and these data have been included. For each volunteer, blood samples (10 mL) were taken at baseline, before any medication had been given and at 0.5, 1.0, 2.0, 3.0, 4.0 and 6.0 h after ingestion. The blood samples were then centrifuged, and the plasma was divided in two aliquots and stored at −20°C.

This study was conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from London Chelsea Research Ethics Service committee (11/LO/1973), and relevant research governance approvals were obtained.

Instrumentation and chromatographic conditions

For analysis, a 1260 Infinity Binary high-performance liquid chromatography system coupled to a 6460 Triple Quadrupole (TQ) mass spectrometer (Agilent Technologies, UK) was used. Mobile phase A was composed of deionized water–formic acid (0.1%) and mobile phase B was composed of acetonitrile–formic acid (0.1%). Chromatographic separation was performed at a column temperature of 40°C and a flow rate of 0.20 mL/min using a Poroshell 120 EC-C18 HPLC column (2.1 × 100 mm, 2.7 µm) (Agilent Technologies, UK). The injection volume was 2 µL. The sample run time was 9.10 min followed by 5 min column re-equilibration (total run time per sample: 14.10 min). The LC gradient system and TQ mass spectrometer parameters are listed in Table I.

Table I.

LC and TQ Mass Spectrometer Parameters

	Time (min)	Mobile phase %
		A	B
LC parameters
Gradient program	0.0	95.0	5.0
	0.1	95.0	5.0
	6.0	10.0	90.0
	9.0	10.0	90.0
	9.1	95.0	5.0
Post-time	5 min hold	95.0	5.0
				Max pressure limit
				600 bar
TQ mass spectrometer parameters
Ionization Mode	ESI (positive mode)
Scan type	dynamic-MRM
Gas temperature	350°C (N2 gas)
Gas flow	6 L/min
Nebulizer pressure	15 psi (103.4 kPa)
Capillary voltage	1,500 (pos)
Collision Gas	N2 (high purity: 29.0 psi/199.9 kPa)
MS1 Heater	100°C
MS2 Heater	100°C
Cycle time	500 ms
Resolution (MS1 and MS2)	Unit

Mobile phase A: deionized water:formic acid (0.1%). Mobile phase B: acetonitrile:formic acid (0.1%).

MassHunter Data Acquisition software (version B.06.00) was used to control the LC–MS-MS instrument and perform data analysis. Analysis was carried out using electrospray ionization (ESI) in the t-DMRM mode. The primary product ion transitions monitored for baclofen were as follows: quantifier m/z − 214.1 > 150.9 and qualifier m/z − 214.1 > 115.0 (retention time: 4.79 min). The primary product ion transitions monitored for baclofen-d₄ were as follows: quantifier m/z − 218.0 > 120.1 and qualifier m/z − 218.0 > 119.0 (retention time: 4.78 min). The quantifier, qualifier and triggered ion transitions monitored by MS–MS are listed in Table II.

Table II.

Ion Transitions Monitored by MS–MS for Baclofen and Baclofen-d₄ Analysis in Positive Ionization Mode

Analyte	Precursor ion (m/z)		Product ions (m/z)	Trigger threshold	Collision energy (V)	Cell accelerator voltage
Baclofen
	214.1	Quantifier ion (1°)	150.9	40	17	7
		Qualifier ion (1°)	115.0		57	7
		Triggered ions (2°)	196.6		9	4
			178.9		13	4
			172.9		5	4
			144.0		21	4
			116.0		37	4
			97.0		17	4
			89.1		69	4
			56.0		45	4
		Fragmentor (V)	79
Baclofen-d4
	218.0	Quantifier ion (1°)	120.1	50	41	7
		Qualifier ion (1°)	119.0		45	7
		Triggered ions (2°)	201.0		9	4
			154.9		21	4
			135.8		9	4
		Fragmentor (V)	84

Preparation of standard solutions

Stock solutions of baclofen (1,000 ng/mL) and the internal standard baclofen-d₄ (200 ng/mL) were prepared in deionized water. Quality control (QC) standards of baclofen were prepared in deionized water at two different concentrations: 80 ng/mL (low QC) and 800 ng/mL (high QC). All solutions were stored at −20°C.

Preparation calibrators and QC

Eight calibration standards consisting of blank, 0, 25, 50, 100, 250, 500 and 1,000 ng/mL of baclofen were prepared in extraction tubes by diluting the 1,000 ng/mL stock solution in deionized water (1 mL total). One milliliter of each QC standard was pipetted into extraction tubes undiluted.

Sample preparation

To all calibrators and QC (1 mL), 1 mL of human drug-free plasma followed by 1 mL of deionized water were added. To 1 mL of volunteers' plasma samples, 2 mL of deionized water was added. Internal standard solution (1 mL) was added to the calibrators (excluding the blank) QC and volunteer samples. To the blank, 1 mL of deionized water was added. The pH of the samples was adjusted to pH 4 by adding 1 mL of 1 M phosphate buffer (pH 4). The samples were mixed on a rotary mixer for 10 min and then centrifuged at 2,500 rpm (1,048×g) for 5 min.

SPE

The SPE cartridges were activated with 2 mL of methanol and conditioned sequentially by adding 2 mL of deionized water followed by 2 mL of 1 M phosphate buffer (pH 4). The supernatant of the plasma samples was applied to the conditioned cartridges followed by 2 mL of deionized water. Once the deionized water had passed through the cartridges, they were acidified with two × 250 µL aliquots of 0.1 M HCl. The cartridges were dried under vacuum for 5 min. Additional washing of the cartridges was performed with 0.5 mL of hexane before drying the cartridges completely. The analytes were eluted using 3 mL of methanol–ammonia (80:2) solvent mixture; the eluents were collected into extraction tubes and evaporated to dryness under nitrogen at 60°C. The samples were reconstituted in 120 µL of methanol. Aliquots of 60 µL of the reconstituted samples were taken into LC micro-vials and diluted further with 140 µL acetonitrile–deionized water (50:50) solvent mixture. The samples were injected (2 µL) onto the LC–MS-MS.

Method Validation

For validation requirements, the following criteria were determined: specificity/selectivity, stability, linearity, limit of detection (LOD) and lower limit of quantification (LOQ), intra-day and inter-day accuracy and precision, recoveries and assessment matrix effect.

Specificity/selectivity

To matrix match the calibrators, human drug-free plasma was used to prepare all calibrators and QC. The specificity of the method was confirmed by analysis of extracted drug-free plasma samples (n = 10), to demonstrate lack of chromatographic interference from endogenous compounds with the analytes of interest. Although the application relies on specific mass transitions, some drugs may co-elute and share the molecular mass and possibly some of the product ions of the analytes of interest. To establish that no interference was occurring from other drugs that maybe present in the sample, a single un-extracted drug mixture containing 62 most commonly observed drugs in routine toxicological screening was analyzed (at 8.0 ng/2 µL injection) with the method. The mixture included opiates, amphetamines, cocaine, anti-depressants, anti-psychotics and over-the-counter drugs.

Stability

Stability of baclofen in frozen plasma was determined. The first set of aliquots from all volunteers were defrosted and analyzed. For one volunteer, the second set of aliquots were defrosted and analyzed 2 months after the first set (n = 6). For two volunteers, the second set of aliquots were defrosted and analyzed 4 months after the first set (n = 18).

To determine if baclofen in plasma was stable after two freeze–thaw cycles, plasma samples from one volunteer were thawed and refrozen after primary analysis and then thawed and analyzed again after 4 months (n = 12).

Linearity, LOD and LOQ

None-weighted regression was used to establish the calibration lines, and a linear curve fit type was used. Linearity was verified by analyzing two full calibration lines on three non-consecutive days (n = 6). For the calibration line to be acceptable, the lowest quantifiable value had to be within ±20% of its theoretical value compared with its extracted values against the obtained calibration line and the rest of the calibration standards were required to be within ±15%. The LOQ was defined as the lowest concentration that could be reproducibly detected within ±20% error and a signal to noise ratio (S:N) >10. The LOD was set as the lowest concentration that could be detected with the presence of the quantifier ion, qualifier ion and also the presence of all triggered product ions at an acceptable ratio (±20%) and a S:N >3. To determine the LOD of the method, drug-free plasma fortified with 5 and 10 ng/mL of baclofen was extracted and analyzed on 10 non-consecutive days.

Accuracy and precision

For intraday analysis, the low and high QCs (n = 6) were extracted with all the calibration standards. For interday analysis, five replicates of each QC were prepared and extracted with all the calibration standards. This was repeated on three nonconsecutive days (n = 15). The percentage relative standard deviation (% RSD) of the obtained results was then calculated to determine imprecision. The percentage accuracy of the results was also calculated.

Extraction efficiency

Two QCs (80 and 800 ng/mL) were prepared in drug-free plasma (three at each concentration) and then extracted using the SPE procedure. Six drug-free plasma samples were also extracted using the SPE procedure. To all the eluents, an external standard, baclofen-d₄, was added. Baclofen was then added to the drug-free plasma eluents at 80.0 and 800 ng/mL (three at each concentration). The samples were then dried, reconstituted and analyzed. This was done on two nonconsecutive days. The percentage recovery was calculated by dividing the mean peak area ratio of the QC samples, fortified prior to SPE, by the mean peak area ratio of the samples to which baclofen was added after SPE multiplied by 100.

Assessment of matrix effect

A matrix effect experiment was carried out to determine if any potential suppression or enhancement was occurring to the ion signals of the analytes due to the sample matrix. Six different batches of drug-free human plasma sample were extracted in duplicate. To the extracted samples and to two empty extraction tubes (no matrix), 120 µL of methanol containing 1,000 ng of baclofen were added and then reconstituted as normal with the mobile phase. The average response of the baclofen in the matrix containing samples was compared with the response of the baclofen in the nonmatrix containing sample.

Results

Specificity/selectivity

No interfering signals were observed from the drug-free plasma samples (n = 10) analyzed or from the 62 most commonly encountered drugs in routine toxicological screening using the described method.

Stability

Baclofen in authentic plasma samples was stable when left frozen for 2–4 months at −20°C, and after two freeze–thaw cycles. The percentage imprecision (% RSD) of the baclofen concentration seen in the samples kept frozen (set 2) for 2 and 4 months compared with the concentrations in the original samples (set 1) was <5% (n = 24 total). The % RSD of the baclofen concentration observed before and after the freeze–thaw cycles was <15% (n = 12).

Linearity, LOD and LOQ

All lines (n = 6) were linear from 25 to 1,000 ng/mL with a correlation coefficient (r²) >0.999. Acceptable % error (±10%) was obtained throughout the calibration range. The results are summarized in Table III.

Table III.

Calibration Curve Linearity Validation Data (n = 6) for Baclofen

Theoretical concentration (ng/mL)	Accuracy (%)	Imprecision (% RSD)
25.0	106.5	4.4
50.0	103.0	2.8
100.0	99.7	3.0
250.0	99.3	1.3
500.0	99.4	1.1
1,000.0	100.2	0.3

The LOD of the method was 5 ng/mL. At this concentration, the quantifier, qualifier and all triggered product ions were present in an acceptable ratio (±20%) and with an S:N > 3. The LOQ was 25 ng/mL as this was the lowest concentration on the calibration curve.

Accuracy and imprecision

Acceptable accuracy of ≥95% and imprecision (% RSD) of ≤5% were obtained for both intraday (n = 6) and interday analysis (n = 15). The results are summarized in Table IV.

Table IV.

Intraday (n = 6) and Interday (n = 15) Validation Data for Baclofen

Baclofen QC (ng/mL)	Mean value (ng/mL)	Accuracy (%)	Imprecision (% RSD)
Intraday validation data (n = 6)
80	82	102.2	2.6
800	785	98.2	1.5
Interday validation data (n = 15)
80	80	100.6	2.4
800	786	98.2	2.6

QC, quality control standard.

Extraction efficiency

The average extraction efficiency of this method was 30% (±15.9% RSD, n = 12).

Assessment of matrix effect

The mean matrix effect was observed at −9.4% (±5.6% RSD, n = 6), indicating some ion suppression was occurring.

Method application on volunteer plasma samples

The results obtained from analyzing the samples from the volunteers after taking a single oral dose of 10 and 60 mg of baclofen are shown in Table V. All predose and placebo samples were negative for baclofen (LOD, 5 ng/mL). All results fell within the calibration range of the developed method except one which required a one in two dilutions using drug-free plasma. The range of the concentration of baclofen observed in volunteers given a single 10 mg dose was 39–204 ng/mL and 145–1,101 ng/mL for a 60 mg dose (monitored over 6 h post-administration). The average peak plasma concentration was 158 ng/mL for the 10 mg dose (n = 8), and 689 ng/mL for the 60 mg dose (n = 9). The peak plasma concentration was reached within 2 h of drug consumption for both dosages, which supports the findings from the previous studies (5, 7, 24). The results here are similar to a previous study in which a single 10 mg oral dose of baclofen, given to healthy volunteers, resulted in an average peak plasma concentration of 141 ng/mL (n = 12) within 2 h of ingestion (24).

Table V.

Data Obtained from the Analysis of Plasma Samples from Volunteers After Taking a Single Oral Dose of Baclofen (10 and 60 mg)

Time post ingestion (h)	10 mg dose (n = 8)			60 mg dose (n = 9)
	Range (ng/mL)	Mean (ng/mL)	Median (ng/mL)	Range (ng/mL)	Mean (ng/mL)	Median (ng/mL)
0.5	55–171	130	137	145–692	440	411
1.0	114–204	158	144	435–1,101	702	744
2.0	101–148	130	130	444–749	631	653
3.0	85–124	104	101	414–696	565	549
4.0	60–102	80	81	313–591	448	451
6.0	39–62	50	48	173–389	277	262

Data obtained from healthy male volunteers, age range: 21–35 years, median age: 22 years.

For all placebo and predose samples, baclofen was not detected (LOD 5 ng/mL).

Discussion

Due to the amphoteric nature of baclofen, it is difficult to extract via liquid–liquid extraction. Therefore, mixed mode SPE cartridges containing a strong cation exchanger (SO₃^–) and nonpolar solid phase (C8) were selected for extraction. Adjusting the pH of the plasma sample to pH 4 and acidifying the SPE cartridges facilitated the retention of the analytes via ionic interaction between the charged basic functional group of the analytes and the cation exchanger. Hexane wash alone was adequate for lipid removal and provided an excellent LC–MS-MS response for baclofen with an LOD of 5 ng/mL. SPE produced a clean extract which reduced potential matrix effects from occurring. Baclofen is soluble in both methanol and water; as the mobile phase contains deionized water and acetonitrile, to aid miscibility; the dried extract was first reconstituted in methanol and then in acetonitrile–deionized water (50:50) solvent mixture. Although recovery was comparatively low at 30%, this was adequate for LC–MS-MS analysis.

Most published methods using HPLC, GC, GLC and GC–MS require complex extraction which is time consuming, and many methods require some form of derivatization (6, 7, 9–11, 13, 15, 25). The described method is a quick and simple extraction process and does not require derivatization. Extraction can be performed with reagents that are readily available in most laboratories.

Although extractions for LC–MS-MS analysis in general are simpler than the methods described above, many methods only monitor one to three product ions for both quantification and confirmation (17, 18, 21, 23). The developed method not only monitors a quantifier and qualifier ion, but also up to eight other generated products ions to provide a fingerprint match for the analytes of interest and therefore minimizing false-positive results. Being able to simultaneously quantitate and confirm the analytes in a single run maximizes the time and cost efficiency of the method as the samples do not have to be re-run to obtain scan spectral data. Figure 1 shows peaks for the quantifier and qualifier ions, and the spectra for the product ions for extracted baclofen at 0 ng/mL (showing no interfering signals from the deuterated internal standard), 5 ng/mL (LOD) and 25 ng/mL (LOQ).

Figure 1.

Extracted plasma samples analyzed using LC–MS-MS with t-DMRM showing peaks for quantifier and qualifier ions of baclofen, and product ions spectral match: (a) 0 ng/mL, (b) 5 ng/mL (LOD) and (c) 25 ng/mL (LOQ).

To utilize the product ion scanning ability in t-DMRM, the scanning mode of the instrument was set as dynamic-MRM (d-MRM). This mode enabled the system to acquire only the relevant transitions in the specific retention time window rather than in a time segment of the total run time. This reduced the number of concurrent MRMs being monitored at a specific retention time and in turn increased sensitivity for the specific MRM transitions being monitored. Standard product ion scanning usually takes ∼200 ms per cycle; in the t-DMRM to monitor 10 MRM transitions it takes <50 ms. Having a fast cycle time and high dwell times means more data points are acquired across the analyte's peak increasing sensitivity for the analytes at low nanogram concentrations even if the recovery of the method is low.

Many published LC–MS-MS methods use structurally similar compounds to baclofen as an internal standard (17, 18, 21, 23). In the described method, baclofen-d₄ was used as the internal standard. Using a deuterated analog of the analyte as an internal standard is the only way to truly compensate for potential matrix enhancement or suppression. As matrix effect varies along the chromatographic run, using an alternative internal standard that elutes at a different retention time and with different ionization properties to the analyte being measured may lead to matrix effects going unnoticed and a false concentration may be reported. In the described method, matrix suppression was observed at −9.4%, indicating the importance of using baclofen-d₄. When using SPE, it is also important to use a deuterated internal standard as nonuniformity of the solid-phase can occur that may alter recovery of the analyte and the nondeuterated internal standard by different amounts.

As described, baclofen is showing promise as an effective treatment for alcohol dependence by increasing rates of abstinence as well as decreasing alcohol craving and anxiety (1, 4). However what dose is optimal is not clear. The underlying reason for many alcohol-dependent patients to be able to tolerate very high doses is not understood, and therefore, there is an urgent need to be able to measure plasma baclofen concentrations to inform this debate. The method described here is suitable as a range of baclofen concentrations can be measured. The method can also be modified to quantitate at subnanogram concentrations by altering the dilution of the final extract.

Conclusion

We report here a simple cost-effective SPE–LC–MS-MS method for simultaneous quantification and confirmation of baclofen in human plasma. The LOD is 5 ng/mL, the LOQ is 25 ng/mL and linearity is up to 1,000 ng/mL. This method provides robust spectral confirmation and can be applied to therapeutic drug monitoring, forensic cases and pharmacokinetic studies.