Introduction

Carbamazepine is a neutral iminostilbene compound, which is highly lipid soluble. It has been used as a first-line medication for partial seizures and has shown indications for treatment of bipolar disorder and trigeminal neuralgia. Side effects of carbamazepine include hyponatremia, cardiac conduction abnormalities, ataxia, nystagmus, and cognitive impairment. According to its narrow therapeutic range of 4–12 μg/mL, the levels exceeding 15 μg/mL cause toxic effects. Therefore, therapeutic drug monitoring should be established to optimize dosing regimens in order to reduce adverse reactions and to improve seizure control [1,2,3]. Several analytical methods have been reported for the determination of carbamazepine in different matrices (e.g., serum, urine, dried blood spot (DBS), dried saliva spot, pharmaceuticals, and waste water). These methods included, for instance, spectrophotometry, high-performance thin-layer chromatography (HPTLC), high-performance liquid chromatography (HPLC), HPLC–MS/MS, near-infrared spectroscopy (NIR), electrochemical sensor, and enzyme-linked immunosorbent assay (ELISA) [4,5,6,7,8,9,10,11,12,13]. Moreover, due to its simplicity, speed, high efficiency, and low sample/reagent consumption, CE has also been applied for the determination of carbamazepine in tablets [14], serum, and plasma [3, 14,15,16]. All studies employed micellar electrokinetic chromatography (MEKC) for analysis of carbamazepine because of the analyte’s hydrophobicity and pKa values (2.3 and 13.9) [17]. The MEKC separation is based on the different distributions of analytes between the pseudostationary phase of micelles and the bulk phase of background electrolyte (BGE), which enables the separation of neutral and charged analytes. With sodium dodecyl sulfate (SDS) micelles, the migration order is anions, neutral analytes, and cations. Anions remain mostly in the bulk solution due to electrostatic repulsion from the micelles. The neutral analytes’ separation depends on their hydrophobicity, while cations have strong electrostatic attraction with micelles. Due to the usually low blood levels of target analytes, on-line preconcentration technique, i.e., sweeping, has been utilized in order to enhance the method sensitivity. The sweeping mechanism is based on electrokinetic focusing of analytes possessing strong affinity to the micellar phase and on modifying conductivity of a sample plug by an additive, e.g., high salt concentration. At the interface between the high-conductivity sample zone and the low-conductivity separation zone, the velocity of the target analyte is changed upon interaction, resulting in analyte sweeping or focusing into a narrow band [18,19,20].

DBS collection as a blood microsampling technique illustrates many advantages over conventional blood sampling, including less invasive blood withdrawal from a heel or a finger prick, collection of small amounts of blood, and simple storage and transportation. Moreover, DBS sampling also promotes chemical stabilization of pharmaceuticals and their metabolites over a long period of time [21]. Nevertheless, typical preparation procedures for DBS samples require multiple steps and involve elution, protein precipitation, and centrifugation followed by a specific extraction to isolate target analytes from the matrix, which can be time consuming, laborious, and costly. Furthermore, the multiple-step sample preparation procedure could lead to a sample/analyte loss. To overcome the above-reported limitations, automated DBS processing has been recently developed, which includes no or only minimum operator’s intervention and performs the entire procedure from DBS elution until analytical quantification [22, 23].

Another approach might be the application of molecularly imprinted polymers (MIPs) since they offer distinct advantages over both liquid–liquid extraction and solid-phase extraction, such as enhancement of selective binding with the target analyte, high extraction efficiency, and simple/flexible synthesis of MIPs. MIPs offer tailor-made selective binding sites in synthetic polymers. Addition of template molecules facilitates the formation of recognition sites during polymerization, which are removed prior to use [24]. The combination of MIPs and interpenetrating polymer networks (IPNs) is capable of a more specific binding with the analyte and an efficient interference removal. Generally, IPNs comprise two or more polymers in the networks, which are at least partially interlaced on a molecular scale without covalent bond formation. However, the network cannot be separated unless chemical bonds in the polymers are broken [25]. Synthetic MIPs with different polymers have been used as extraction sorbents of carbamazepine in various matrices, e.g., urine [26, 27], blood [28], and environmental waters [29]. Moreover, MIPs have also been reported for the extraction of other analytes, e.g., atenolol [30], creatinine [31], cortisol [32], and myoglobin [33] from blood samples.

In this study, the combination of a paper-based MI-IPN as a DBS collection and on-spot microextraction device with sweeping-MEKC was presented for the analysis of carbamazepine in DBS samples. The MI-IPN was synthesized by radical polymerization of tetramethyl orthosilicate (TMOS) and styrene using carbamazepine as the template, and the resulting MI-IPN was coated on a filter paper support. Previously, the MI-IPN in a packed syringe was utilized as microextraction sorbent combined with HPLC-UV analysis of carbamazepine in urine samples [26]. However, the developed paper-based MI-IPN could reduce the MI-IPN preparation time by direct deposition of the reaction mixture onto the paper support, which eliminates grinding and packing of MI-IPN into the syringe and reduces solvent evaporation time. After DBS extraction from the MI-IPN device, KCl solution was added to the extract and the mixture was analyzed by MEKC. The extraction efficiency obtained from the paper-based MI-IPN was also compared with the commercial DBS collection cards. The proposed method was rapid, simple, and selective and provided disposability of the collection/extraction devices, high extraction efficiency, and interference removal. To the best of our knowledge, the combination of DBS sampling with MEKC for the determination of carbamazepine has not been reported and the presented MI-IPN-MEKC approach presents an interesting alternative for clinical applications, including therapeutic drug monitoring of carbamazepine in blood.

Materials and methods

Chemicals and reagents

Carbamazepine, 2,2-azobis(2-methylpropionitrile) (AIBN), TMOS, and styrene were purchased from Sigma (Steinheim, Germany). Boric acid, sodium tetraborate decahydrate, SDS, and sodium hydroxide were from Fluka (Buchs, Switzerland). Potassium chloride, methanol (MeOH), acetonitrile (ACN), and acetic acid (AcOH) were purchased from Lach-Ner (Neratovice, Czech Republic), and hydrochloric acid was from VWR, Avantor (Stříbrná Skalice, Czech Republic). Deionized (DI) water with resistivity higher than 18 MΩ·cm was prepared by exchange of ions in a mixed-bed ion exchanger water purification system G 7749 (Miele, Gütersloh, Germany).

Instrumentation

CE separations were performed on a 7100 CE instrument (Agilent Technologies, Waldbronn, Germany) and controlled by a PC through Agilent ChemStation software. Separations were carried out using fused-silica capillaries (50/375 μm i.d./o.d.) with a total length (Ltotal) of 48.5 cm and an effective length (Leff) of 40 cm from Polymicro Technologies (Phoenix, AZ, USA). Detection was by a diode array detector using a detection wavelength of 214 nm. All experiments were performed at positive polarity voltage. For daily use, the capillary was rinsed with 1 M NaOH, 0.1 M NaOH, DI water, and BGE for 5 min each. Between runs, the capillary was rinsed with 0.1 M NaOH, DI water, and BGE for 1, 1, and 2 min, respectively.

Standards, BGE solutions, and sample preparations

Stock solution of 1000 μg/mL carbamazepine was prepared by dissolving the pure chemical in ACN. The stock solution was protected from light, stored at 2–8 °C, and used within 7 days. Then, working standard solution was prepared by diluting the stock standard solution with a KCl solution. The stock solution of KCl (400 mM) was prepared by transferring 0.7455 g KCl into a 25-mL volumetric flask and filled to the mark with DI water. Next, the KCl stock solution was transferred to mix with the carbamazepine stock solution to obtain working standard solutions with KCl concentrations of 40, 60, 80, and 100 mM. BGE solution containing 20 mM SDS in 30 mM borate buffer (pH 8.9) was prepared by transferring 0.09 g boric acid and 0.575 g sodium tetraborate decahydrate into a 50-mL volumetric flask. The borate buffer was adjusted to pH 8.9 by adding 1 M HCl or 1 M NaOH, and the flask was filled to the mark with DI water. Subsequently, 0.144 g SDS was transferred into a 25-mL volumetric flask and filled to the mark with the borate buffer.

Blank DBS samples were prepared by pipetting 20 μL of drug-free capillary blood directly onto a Whatman® 903 protein saver card (GE Healthcare Ltd., Cardiff, UK), GenCollect™ 2.0 (Ahlstrom-Munksjö, Helsinki, Finland), and the developed MI-IPN-coated Whatman® Grade 1 paper. For spiked DBS samples, 15 μL of drug-free capillary blood was mixed with 5 μL of carbamazepine standard solution in DI water and the resulting mixture (20 μL) was spotted onto the three DBS cards. All DBSs were dried at room temperature for 2 h. Then, the DBSs were punched out using a hole puncher (12 mm) for the commercial DBS cards, while 12-mm discs were pre-punched before coating with the MI-IPN and DBS collection. Each 12-mm disc contained the entire DBS, and the discs were transferred to Swinnex® polypropylene filter holders (13 mm, Millipore, NSW, Australia). Carbamazepine was flushed out with 200 μL MeOH:AcOH (95:5 v/v) [26] or 50–80% (v/v) ACN. A mean volume collected after extraction from the Whatman® Grade 1 paper-based MI-IPN was 133.6 ± 4.1 μL (n = 10). One hundred microliters of the extract was diluted with 120 μL of appropriate KCl solution and 180 μL of DI water for MEKC analysis. The DBS samples were collected from volunteers at the Institute of Analytical Chemistry who signed a written informed consent. Carbamazepine concentrations reported in this manuscript relate to the original concentrations in the capillary blood.

Preparation of paper-based MI-IPN

In order to prepare MI-IPN, a hybrid of homogeneous polystyrene and silica gel polymer was synthesized by in situ radical polymerization. Firstly, 5–100 mg carbamazepine was dissolved in 4 mL ACN, which was thoroughly mixed with a solution containing 390 μL TMOS, 50 mg AIBN, and 390 μL 0.1 M HCl by stirring prior to addition of 220 μL styrene monomers [26]. The final concentrations of carbamazepine in the reaction mixture ranging from 4.2 to 84 mM were investigated. The non-imprinted-interpenetrating polymer network (NI-IPN) was also prepared with the absence of carbamazepine. Then, 20 μL of the polymer hybrid solution was sequentially pipetted to 12-mm paper discs (10 × 20 μL for each disc), which were pre-punched from Whatman® qualitative filter paper, Grade 1 sheet (60 cm × 60 cm; GE Healthcare Ltd., Cardiff, UK) with pore size of 11 μm and thickness of 180 μm. The complete pipetting procedure for simultaneous preparation of 60 MI-IPN discs took approximately 2 h. The MI-IPN- and NI-IPN-coated discs were placed on the syringe filter holder (Fig. 1) and were washed with 3 mL each of MeOH and DI water in order to remove carbamazepine (as the template) and unreacted monomers, and to condition the polymer prior to use.

Fig. 1
figure 1

Schematic diagrams of molecularly imprinted-interpenetrating polymer network (MI-IPN) structures

Full size image

The morphology of the cellulose paper and the MI-IPN/NI-IPN-coated papers was observed by scanning electron microscope (SEM, model MIRA3, Tescan, a. s., Brno, Czech Republic) using the In-Beam mode at 1 kx and 30 kx magnification.

CE analysis

The CE conditions for the analysis of carbamazepine were modified from Izzo et al. [15]. The initial conditions were 25 mM SDS in 30 mM borate buffer at pH 8.9, injection at 50 mbar for 5 s, separation voltage of 20 kV, capillary temperature of 25 °C, and detection wavelength of 214 nm. Further, several factors affecting analytical performance were investigated including SDS concentration in the BGE solution (15–25 mM), applied voltage (20–25 kV), and sample loading (150–750 mbar·s). The optimized CE conditions were determined from resolution (Rs), number of theoretical plates (N), and relative standard deviation (RSD) values of migration times (tm) and peak areas. The total run time (including capillary flushing, sample injection, and MEKC separation) was less than 15 min per sample.

The optimized CE conditions were validated in terms of linearity, accuracy, precision, and lower limit of quantitation (LLOQ) [34, 35] by measuring eluates of spiked DBS samples. Calibration curves of DBS samples spiked at 4–800 μg/mL carbamazepine were established by plotting peak areas against seven different concentrations. Five injections were performed for each concentration. Within-run accuracy and precision were assessed by the determination of DBSs spiked at 4 concentration levels including LLOQ and low, medium, and high quality control (QC) samples. Five samples per level were analyzed. Between-run accuracy and precision were evaluated at the same concentration level as within-run accuracy and precision by performing triplicate injections of each level on two different days. The accuracy was reported as percent of the nominal value. The mean concentration of the QC samples should be within ± 15% of the nominal value, except for the LLOQ, which should be within ± 20% of the nominal value. Method precision, calculated from the %RSD, should not exceed 15%, except for the LLOQ, which should not exceed 20%.

Results and discussion

Paper-based MI-IPN

The initial synthesis procedure of MI-IPN using in situ radical polymerization was modified from a previous study [26], which used the synthesized MI-IPN as a sorbent packed in a syringe for extraction of carbamazepine, naproxen, and dexamethasone from urine samples. In the reaction mixture, AIBN was used as the azo polymerization initiator, which decomposed into two isobutyronitrile radicals enabling the free radical polymerization. Generally, the useful temperature range for AIBN is 60–120 °C. However, the radicals could be produced at a slower rate at lower temperatures [36]. In this system, the TMOS sol–gel matrix and polystyrene were independently cross-linked, which could be defined as a non-covalent IPN (Fig. 1). The polystyrene penetrated through the sol–gel network at the nanometer to the micrometer level and the template molecule imprinted on the hybrid polymers. The template and unreacted monomers were washed out from the MI-IPN, whereas only the unreacted monomers were removed from the NI-IPN. After the template removal, the specific cavities of carbamazepine were formed. The traditional template to monomer ratio of approximately 1:4 is commonly used for the preparation of MIPs [37]. The previous study employed the template concentration of 105 mM, which resulted in the template to styrene to TMOS ratio of 1:4.2:5.7 [26]. In our experiments, by increasing the template concentration from 4.2 to 42 mM, the extraction efficiencies increased from 29.3 ± 0.9% to 92.1 ± 2.3% (n = 3). In the absence of template (i.e., paper-based NI-IPN), the extraction efficiency of 22.6 ± 0.7% (n = 3) was achieved, which is in a good agreement with the previous study. The paper-based NI-IPN and MI-IPN using the template concentration less than 42 mM provided weak and nonspecific bindings of carbamazepine. On the other hand, by increasing the template concentration to 84 mM, the amount of specific cavities of carbamazepine did not increase or the reaction mixture yielded enough specific binding sites to bind the total target analyte at the template concentration of 42 mM since the extraction efficiency (89.3 ± 3.1%, n = 3) has not improved. Therefore, the template concentration of 42 mM was selected as an optimum for synthesizing the paper-based MI-IPNs. In comparison with the previous study, the quantity of the template was reduced by using the template to styrene to TMOS ratio of 1:9:12. Moreover, the preparation time was significantly shortened by direct deposition of the reaction mixture onto the discs, which eliminated grinding and packing of the MI-IPN into the syringe and reduced solvent evaporation time. The solvent evaporation rate improved from 1 week [26] to 30 min (in this work) when the reaction mixture was sequentially pipetted onto the paper discs.

Figure 2 shows the SEM images of the plain cellulose paper and the same paper coated with the synthesized MI-IPN. The enlarged inset demonstrates coating of the cellulose fiber surface with the MI-IPN but also the fact that the inter-fiber network was filled with the MI-IPN. The morphology of the MI-IPN illustrated a porous polymer providing a large surface area with the specific recognition feature towards the template molecule, which is advantageous for the extraction sorbent. On the other hand, a rigid polymer was obtained in the paper-based NI-IPN, which presumably exhibits no or reduced porosity and thus a lower affinity for carbamazepine extraction (Fig. S1 in the Electronic Supplementary Material, ESM). Similar differences among the MI-IPN and NI-IPN structures were observed previously for different polymeric materials and are visualized in Fig. S2 in the ESM.

Fig. 2
figure 2

SEM images of the Whatman® Grade 1 filter paper uncoated and coated with the MI-IPN

Full size image

CE (MEKC) analysis

The experimental conditions for the MEKC analysis of carbamazepine were 30 mM borate buffer at pH 8.9 containing 15–25 mM SDS (as BGE solution) using positive polarity separation voltage of 20–25 kV, which was modified from Izzo et al. [15]. The concentration of carbamazepine in standard solution was 5 μg/mL during the optimization process. SDS was employed as the micelle-forming pseudostationary phase at concentrations above its critical micelle concentration of 8 mM [38]. The anionic SDS micelles migrate electrostatically towards the anode with different velocity from the bulk liquid flow in the capillary, and analytes interacting strongly with the micelles have longer migration times than analytes with no or weak interactions. Consequently, the micelles interacted with the hydrophobic neutral form of carbamazepine by partition mechanism. By increasing the SDS concentration from 15 to 25 mM, increased migration time of carbamazepine was observed, indicating that a higher amount of carbamazepine distributed in the micelles. A broad peak of carbamazepine was obtained in 25 mM SDS (tm = 9.2 min and N = 1.9 × 104), while 15 mM SDS concentration provided tm = 6.5 min and N = 2.5 × 104. The SDS concentration of 20 mM with the applied voltage of 25 kV was thus chosen since at these conditions a good carbamazepine peak shape with tm = 7.3 min (RSD ≤ 1.4%) and N = 3.8 × 104 (RSD ≤ 2.4%) was achieved (n = 3). The major metabolite of carbamazepine is carbamazepine-10,11-epoxide [15, 16], which is more polar than the parent drug (see Fig. S3 in the ESM). The actually employed MEKC method offers the separation of carbamazepine and carbamazepine-10,11-epoxide, since carbamazepine-10,11-epoxide has weaker interaction with the micelles and migrates faster than carbamazepine, and thus, no co-migration of the two compounds is expected [15]. Moreover, the developed MI-IPN provides lower selectivity to carbamazepine-10,11-epoxide because of a steric hindrance of epoxide and lower hydrophobic interactions with polystyrene in the MI-IPN. Consequently, the metabolite is partially eliminated already during DBS sampling and extraction steps.

Subsequently, sensitivity improvement for the MEKC determination of carbamazepine was examined by the increase of injection volume (150–750 mbar·s). However, the increase of the injection volume resulted in serious broadening of the carbamazepine peak with no increase in the peak height. Therefore, the sensitivity enhancement was examined by sweeping in sample matrix containing high salt concentrations. Addition of a saline solution (i.e., KCl) to the sample matrix provides a greater conductivity of the sample zone in comparison with the BGE solution. By applying separation voltage at positive polarity, anionic SDS micelles stack at the interface between the sample and the BGE solution and sweep the analyte [19, 20]. The sample matrix containing KCl at a concentration of 40–100 mM was investigated for on-line sample enrichment with the injection at 50 mbar for 15 s (Fig. 3). Increasing the KCl concentration from 40 to 60 mM led to an increase in MEKC sensitivity, while the sensitivity decreased at the higher KCl concentrations (80 and 100 mM). The highest S/N of carbamazepine (S/N = 44.5) with N = 1.0 × 105 was obtained using 60 mM KCl, which ensured about fourfold higher sensitivity in comparison to the sample dissolved in 30 mM borate buffer.

Fig. 3
figure 3

Effect of KCl concentration on sensitivity enhancement. CE conditions: BGE solution: 30 mM borate buffer, 20 mM SDS, pH 8.9; separation voltage: 25 kV; injection: 50 mbar × 15 s; UV detection at 214 nm; carbamazepine concentration: 5 μg/mL

Full size image

Applications

Three different collection cards including the Whatman® 903 protein saver card, the GenCollect™ 2.0 card, and the paper-based MI-IPN were used for DBS sample collection and storage at the ambient temperature prior to analysis. Whatman® 903 protein saver is the most commonly used card for DBS collection [39], while the GenCollect™ 2.0 card is coated to specifically allow for direct amplification of DNA and to prevent growth of microorganisms during the short-term storage at ambient temperature.

DBS discs with the spiked carbamazepine (20 μg/mL) sampled on the three types of collection cards were placed into the syringe filter holders and eluted with three different solvents. Using MeOH:AcOH (95:5 v/v) [26] as the solvent, the carbamazepine peak was broad and reduced sensitivity was observed for all types of collection cards. The broad peak resulted in lower analytical signals; moreover, this solvent provided higher background noise and the overall S/N ratios decreased. DBS samples are usually eluted with DI water, organic solvents, and mixtures thereof, and 50% (v/v) ACN and 80% (v/v) ACN were examined as the next solvents. By application of 50% (v/v) ACN, carbamazepine was recovered from all three collection cards; note, however, that only extraction from the paper-based MI-IPN resulted into analytical records free from matrix interferences. Electropherograms recorded for Whatman® 903 and GenCollect™ 2.0 DBS cards revealed co-elution of blood proteins and other macromolecular compounds, which in the case of GenCollect™ 2.0 cards completely disabled quantitative analysis of carbamazepine (Fig. 4a). By further increasing the content of the organic solvent, better removal of matrix interferences was achieved and only small peaks were detected for the two commercial cards at 80% (v/v) ACN. Using the GenCollect™ 2.0 card, the carbamazepine peak illustrated a broadened peak shoulder, which can be an unknown sample matrix component migrating with almost the same mobility as carbamazepine (Fig. 4b). The paper-based MI-IPN and Whatman® 903 protein saver card provided excellent separation of carbamazepine and minor sample matrix peaks. The extra small peak (an unknown compound from the polymerization mixture) was separated from carbamazepine with Rs > 13.0 for the paper-based MI-IPN extracts. The extraction efficiencies of carbamazepine from DBS using the Whatman® 903 protein saver card, the GenCollect™ 2.0 card, and the paper-based MI-IPN were 59.8 ± 2.8%, 47.2 ± 1.4%, and 92.2 ± 2.5% (n = 3), respectively. Therefore, the developed paper-based MI-IPN extracted with 80% (v/v) ACN offered the most efficient interference removal and the highest extraction efficiency.

Fig. 4
figure 4

MEKC analysis of DBS extracts from the MI-IPN-coated filter paper and two commercial collection cards by using a 50% (v/v) ACN and b 80% (v/v) ACN. CE conditions as for Fig. 3; DBSs spiked with carbamazepine at a concentration of 800 μg/mL

Full size image

The Whatman® 903 protein saver card as well as the GenCollect™ 2.0 card can also be coated with specific chemistries. Coating of the two commercial DBS cards with the MI-IPN was compared with that of the Whatman® Grade 1 filter paper, and the different types of papers illustrated different extraction efficiencies based on their physicochemical properties. For instance, the Whatman® Grade 1 filter paper was thinner and more porous than the Whatman® 903 protein saver card and the GenCollect™ 2.0 card resulting in the formation of a homogeneous coating material with a larger surface area for the Whatman® Grade 1 filter paper, which enabled higher extraction efficiency. The extraction efficiencies of the MI-IPN-coated Whatman® 903 protein saver card, GenCollect™ 2.0 card, and Whatman® Grade 1 filter paper were 68.4 ± 3.3%, 73.9 ± 1.9%, and 92.2 ± 2.5% (n = 3), respectively.

To examine the practical applicability of the developed approach, unspiked drug-free capillary blood (blank) and drug-free capillary blood spiked with 4 μg/mL of carbamazepine were spotted onto the Whatman® Grade 1 paper-based MI-IPN sampling discs. The spiking concentration corresponds to the minimum therapeutic concentration of carbamazepine in blood (4–12 μg/mL). Six blank DBS samples from 6 different individuals were processed and analyzed, and no interfering peaks were detected in the MEKC electropherograms. Figure 5 depicts an electropherogram of the spiked DBS sample and clearly demonstrates that the proposed approach is sensitive enough for quantitative determination of carbamazepine even at the lowest concentration of the therapeutic range. The optimized MI-IPN-MEKC method was then validated for the determination of carbamazepine in spiked DBS samples. The validation was done according to bioanalytical method validation guidelines [34, 35], and all corresponding data are summarized in Table 1. Linearity of the calibration curve was measured for drug-free capillary blood spiked with carbamazepine in the 4–800 μg/mL range. The linear model can be accepted (R = 0.9960) based on the back-calculation of carbamazepine concentrations, which were within 15% of their nominal values [35]. The within-run and between-run precisions showed RSDs of 2.8–4.6% and 3.3–5.1% (n = 5) for carbamazepine peak areas, respectively. The within-run and between-run accuracies illustrated the mean recoveries of 92.3% (RSD ≤ 4.6%) and 90.8% (RSD ≤ 5.1%, n = 5), respectively, and indicated that the sample matrix did not affect the quantitation of the investigated analyte in the samples. The LLOQ was 4 μg/mL (S/N = 5) with an RSD value of 3.5%.

Fig. 5
figure 5

Analysis of carbamazepine in DBS at the LLOQ level using MI-IPN-coated filter paper for sample collection. CE conditions as for Fig. 3

Full size image
Table 1 Validation data of the analytical method for carbamazepine determination in DBS samples collected on paper-based MI-IPN-coated Whatman® Grade 1 filter paper discs (n = 5). CE conditions as for Fig. 3
Full size table

The hematocrit content and aging of the DBS can have direct effect on the stability and the extractability of target analytes from DBSs. Blood samples with three different hematocrit values (low, medium, and high, n = 6 for each hematocrit value) were prepared by using capillary blood with addition of plasma, capillary blood with no changes, and capillary blood after removal of plasma according to ref. [21]. These blood samples were spiked with 10 μg/mL of carbamazepine and were spotted onto the Whatman® Grade 1 paper-based MI-IPN sampling discs. Extraction and MEKC analysis were performed immediately after the drying period and 2 weeks after the spots were dried out (DBSs were stored at lab temperature/humidity). Results are shown in Fig. S4 in the ESM and reveal statistically no significant difference (< 8.0%) for carbamazepine eluted from the fresh vs. the 2-week-old DBSs. Extraction of carbamazepine from DBSs with different hematocrit content resulted in slightly increased concentrations for higher hematocrits and the difference was approx. 12.0% for the low vs. the high hematocrit content, which still complies with the requirements for bioanalytical method validation [34, 35]. Higher analyte concentrations extracted from DBSs with increased hematocrit content were also observed previously [21] and might be associated with different protein/red blood cell content in the different blood samples. Thus, quantification using a standard addition method could be applicable for an improved precision/accuracy in analysis of DBSs with substantially varying hematocrit content.

Combination of the DBS sampling technique and sweeping-MEKC therefore represents a viable alternative for simplified carbamazepine monitoring. The DBS sampling using the disposable MI-IPN collection/extraction discs can be done at home by the patients and increases their comfort and willingness to participate in clinical studies. Furthermore, the developed collection/extraction discs enable rapid sample processing, which is done by a simple flushing step prior to MEKC analysis, and offer interference-free extracts suitable for direct injection into the CE instrument.

Conclusions

A simple synthetic MI-IPN was prepared by in situ radical polymerization of polystyrene and silica gel polymer prior to the deposition onto pre-cut discs of Whatman® Grade 1 filter paper. The paper-based MI-IPN facilitated the DBS collection and selective extraction of carbamazepine. Consequently, the analyte was determined by sweeping-MEKC with UV detection as a rapid and efficient analytical method. The analytical method offered good precision, accuracy, and sensitivity, which could be applied for the determination of typical therapeutic concentrations of carbamazepine in blood in order to establish appropriate dosage and to avoid adverse effects. The developed paper-based MI-IPN provided superior performance in terms of lower cost (0.12 €/sample), higher extraction efficiency, and better selectivity when compared with the commercially available DBS collection cards (i.e., Whatman® 903 protein saver card and GenCollect™ 2.0 card). Moreover, the preparation of the paper-based MI-IPN discs was characterized by a simple and a rapid synthesis since the necessary preparation steps [26] were reduced. The DBS extraction process might be further directly connected to the CE instrument by liquid handling in a sequential injection analysis mode [40], which could enable fully automated analyses of DBSs, and is the subject of our further study.