Skip to main content
NCBI home page
RSC Advances logoLink to RSC Advances
. 2026 Aug 4;16(39):44555–44566. doi: 10.1039/d6ra03853k

Plug-and-play coaxial nested-tube supported liquid membrane extraction for the determination of nonsteroidal anti-inflammatory drugs in human plasma

Huifen Huang a, Mengyuan Lv b, QianQian Wang c, Qiwen Zhang d, Yongjie Yang d, Di Chen b,e,✉, Zifei Qin d,✉
PMCID: PMC13434744  PMID: 42553468

Abstract

A robust coaxial nested-tube supported liquid membrane extraction (CNT-SLME) method was developed for determining nonsteroidal anti-inflammatory drugs (NSAIDs) in human plasma. The device was adapted from commercial nucleic acid purification columns using industrial-grade polypropylene fabric as a stable, hydrophobic support for a dihexyl ether liquid membrane. This “plug-and-play” three-phase system (acidic donor/organic membrane/alkaline acceptor) facilitates efficient analyte enrichment and reduced matrix interference. The setup allows simultaneous processing of multiple samples using a standard multi-well shaker, overcoming the low-throughput bottlenecks of traditional liquid-phase microextraction. Under optimized conditions, the method yielded satisfactory recoveries (46.89–66.63%), negligible matrix effects (86.50–103.51%), excellent linearity (R2 > 0.998), and detection limits of 3.2–55.3 ng mL−1. Validation in spiked plasma demonstrated high accuracy (recoveries 90.05–104.69%) and precision (RSDs ≤ 8.26%). Given its simplicity, cost-effectiveness, and batch-processing capability, CNT-SLME provides a practical platform for routine therapeutic drug monitoring and clinical pharmacokinetic studies.

A plug-and-play coaxial nested-tube supported liquid membrane extraction platform enables simple, low-solvent, and batch-compatible determination of NSAIDs in human plasma.graphic file with name d6ra03853k-ga.webp

1. Introduction

Nonsteroidal anti-inflammatory drugs (NSAIDs) are a class of therapeutic agents widely utilized in clinical practice for their potent antipyretic, analgesic, and anti-inflammatory properties.1,2 While highly effective at conventional doses, their prolonged or high-dose administration is frequently associated with serious adverse effects, including gastrointestinal injury, renal impairment, and cardiovascular events.3–5 Consequently, the accurate quantification of NSAIDs in human plasma is essential for therapeutic drug monitoring (TDM), ensuring optimal clinical outcomes while minimizing the risks of systemic toxicity.6

Currently, high-performance liquid chromatography (HPLC) coupled with ultraviolet (UV) or mass spectrometric (MS) detection is the gold standard for NSAID analysis.7,8 These techniques provide excellent sensitivity and selectivity for the simultaneous determination of multiple analytes within complex biological samples.9 However, the direct analysis of NSAIDs in human plasma remains a significant challenge due to the inherent complexity of the matrix.10 High concentrations of proteins and endogenous lipids can cause severe chromatographic interference and reduce detector sensitivity.11 Therefore, sample preparation is a critical prerequisite in the analytical workflow, serving the dual purpose of matrix cleanup and analyte enrichment. Although traditional techniques such as liquid–liquid extraction (LLE) and solid-phase extraction (SPE) are widely established, they are often criticized for being time-consuming, labor-intensive, and dependent on large volumes of hazardous organic solvents.12 To address these limitations, various miniaturized strategies, including solid-phase microextraction (SPME), liquid-phase microextraction (LPME), and their modified derivatives, have been developed.13 These techniques offer distinct advantages such as reduced solvent consumption, high enrichment factors, and significantly simplified workflows.

Among the various LPME modalities, supported liquid membrane (SLM) extraction has emerged as a particularly powerful three-phase alternative.14 SLM offers superior selectivity by integrating extraction and back-extraction into a single, continuous step. Within this domain, hollow fiber liquid-phase microextraction (HF-LPME) is a representative technique; however, its practical utility is often hampered by the tedious manual assembly of fragile fibers and poor mechanical stability, which limits its user-friendliness and increases the risk of operation failure in large-scale clinical applications.15 While several automated and parallelized systems have been developed recently to improve throughput, many existing SLM configurations still rely on specialized, expensive membrane modules or complex custom-made hardware. Such systems are often difficult to replicate and operate for non-specialist laboratory personnel. Consequently, there remains a pressing need for a “ready-to-use” extraction platform that combines the high selectivity of SLM with the robustness and throughput required for routine clinical batch analysis.

In this study, a robust and cost-effective coaxial nested-tube supported liquid membrane extraction (CNT-SLME) protocol was established for the determination of NSAIDs in human plasma. The core innovation lies in the adaptation of commercial nucleic acid purification columns into a “plug-and-play” extraction device. This setup utilizes industrial-grade polypropylene fabric (PPF) as a stable, hydrophobic SLM support. This unique configuration overcomes the long-standing trade-off between operational simplicity and sample throughput by enabling the simultaneous, batch-scale processing of multiple samples on a standard laboratory multi-well shaker. The surface morphology and chemical properties of the PPF support were comprehensively characterized, and the extraction parameters were systematically optimized via response surface methodology (RSM). Finally, the performance of the CNT-SLME method was rigorously validated and successfully applied to the monitoring of NSAIDs in clinical plasma samples.

2. Materials and methods

2.1. Chemicals and materials

Standards of nonsteroidal anti-inflammatory drugs (NSAIDs), including loxoprofen (≥98%), naproxen (≥99%), flurbiprofen (≥98%), and diclofenac sodium (≥98%), were obtained from Adamas Reagent Co., Ltd (Shanghai, China). Sulindac (≥98%) and ibuprofen (≥98%) were purchased from Aladdin Biochemical Technology Co., Ltd (Shanghai, China). The molecular structures of these compounds, along with their octanol–water partition coefficients (log P) and acid dissociation constants (pKa), are summarized in Table S1.

Organic solvents and reagents, including dihexyl ether, 1-octanol, n-dodecane, and toluene, were provided by Adamas Reagent Co., Ltd. Formic acid and phosphoric acid (H3PO4) were purchased from Bide Pharmatech Co., Ltd (Shanghai, China), and sodium hydroxide (NaOH) was supplied by Macklin Biochemical Technology Co., Ltd (Shanghai, China). HPLC-grade acetonitrile was purchased from Tedia Company, Inc (Fairfield, OH, USA). Ultrapure water (H2O) was obtained from a Milli-Q water purification system (Millipore, Billerica, MA, USA).

Polypropylene fabric (PPF), commercially marketed as polypropylene nonwoven filters, was obtained from Tiantai Nanhe Filtration Material Co., Ltd (Zhejiang, China). This industrial-grade material, primarily designed for oil filtration, has a pore size of 0.2 µm and serves as the supporting membrane in the construction of the subsequent microextraction platform.

2.2. Sample preparation

Blank plasma samples were obtained from healthy, drug-free volunteers. Specifically, venous blood was collected into an EDTA-coated anticoagulant tube and immediately centrifuged at 4000×g for 10 min at 4 °C. The obtained plasma was stored at −80 °C until further analysis.

Before use, plasma samples were allowed to equilibrate to room temperature (∼20 °C). For spiked plasma samples, appropriate amounts of the NSAID standards were added to the blank plasma and vortex-mixed for 30 seconds to achieve the desired concentrations. To reduce viscosity and control pH, plasma samples were diluted 1 : 10 with phosphate buffer (20 mM, pH 2.5) before the subsequent extraction procedure.

2.3. Coaxial nested-tube supported liquid membrane extraction procedure

The CNT-SLME procedure was designed to enable facile and batch-scale sample processing through a three-phase extraction system. This setup integrates a donor phase, an acceptor phase, and a SLM within a coaxial nested-tube configuration. The detailed construction and operational workflow are illustrated in Fig. 1, and a more detailed assembly procedure is provided in Fig. S1.

Fig. 1. Schematic illustration of (A) the CNT-SLME device construction and (B) the analytical workflow for the extraction of NSAIDs from plasma.

Fig. 1

Open in a new tab

The extraction device was adapted from a commercial nucleic acid purification column, which consisted of an inner insert tube and an outer collection tube. Specifically, the original membrane in the inner insert tube was replaced with a PPF disk serving as the SLM support, which was secured at the bottom of the inner tube using a plastic retaining ring to ensure mechanical stability during shaking. The conical tip of the inner tube was excised to fully expose the PPF disk, thereby maximizing contact between the membrane and the surrounding solutions. The device was then completed by nesting the modified inner tube into a standard outer collection tube, forming the coaxial nested-tube configuration.

To perform the extraction, 60 µL of dihexyl ether was carefully impregnated onto the PPF disk to establish a stable organic liquid membrane. Subsequently, 1.0 mL of the donor phase adjusted to pH 2.5 was added to the outer collection tube, followed by the insertion of the assembled inner tube. Then, 300 µL of the acceptor phase at pH 13.5 was introduced into the inner compartment, completing the three-phase configuration of acidic donor phase/hydrophobic organic membrane/alkaline acceptor phase. The assembled device was placed on a multi-well shaker at an agitation speed of 1000 rpm for 106 min, facilitating the diffusion of NSAIDs from the donor phase into the SLM and their subsequent partitioning into the acceptor phase. Following extraction, a 150 µL aliquot of the acceptor solution was collected and acidified with 2 µL of formic acid, yielding a final pH of approximately 3 before LC-UV analysis.

2.4. LC-UV analytical conditions

Chromatographic analysis of the NSAIDs was performed using an Agilent 1260 Infinity HPLC system (Agilent Technologies, Santa Clara, CA, USA), which was equipped with a quaternary pump, an autosampler, a thermostatted column compartment, and a variable wavelength detector (VWD). Separation was achieved on an Agilent ZORBAX SB-C18 column (250 × 4.6 mm, 5 µm) maintained at a constant temperature of 35 °C. The isocratic elution was carried out using a mobile phase consisting of acetonitrile and 0.1% (v/v) aqueous formic acid in a ratio of 55 : 45 (v/v). The flow rate was set at 1.0 mL min−1 with a total chromatographic runtime of 15 min. The detection wavelength was monitored at 230 nm, and the sample injection volume was 10 µL.

3. Results and discussion

3.1. Characterization of PPF

The surface morphology of the PPF was first characterized to evaluate its structural suitability as a support for the hydrophobic organic liquid membrane. As shown in the SEM micrograph (Fig. 2A), the PPF consists of a disordered and overlapping network of microfibers with diameters ranging from approximately 5 to 20 µm. This three-dimensional fibrous structure is beneficial for the stable immobilization of the hydrophobic organic extractant, such as dihexyl ether, through capillary forces and hydrophobic interactions.

Fig. 2. Surface morphology and functional group analysis of the PPF: (A) SEM image, and (B) FTIR spectrum.

Fig. 2

Open in a new tab

The chemical identity of the support was further confirmed by FTIR spectroscopy (Fig. 2B). The spectrum displays characteristic vibrational bands of polypropylene,16 including intense peaks in the range of 2838–2949 cm−1 corresponding to the symmetric and asymmetric stretching vibrations of aliphatic –CH2– and –CH3 groups. The bands at approximately 1373 cm−1 and 1454 cm−1 are attributed to CH2 bending and CH3 deformation modes, further confirming the methyl-substituted backbone of the PPF. The absorption near 1165 cm−1 is assigned to C–C skeletal stretching, while additional peaks at 995, 972, 841, and 810 cm−1 correspond to CH bending and rocking modes. These FTIR results confirmed the polypropylene nature of the PPF support, while the absence of polar functional group signals suggested its intrinsic hydrophobicity.

To further demonstrate the suitability of PPF as a hydrophobic SLM support, an additional visual wettability test was performed. As shown in Fig. S2, the dyed water droplet remained nearly spherical on the PPF surface, whereas the dyed dihexyl ether droplet rapidly wetted and penetrated the fabric. This result directly confirmed the hydrophobic and oleophilic characteristics of the PPF, which are critical for immobilizing the organic liquid membrane while preventing penetration of the aqueous donor and acceptor phases. Therefore, the PPF support was suitable for maintaining a stable three-phase CNT-SLME system.

3.2. Preliminary evaluation of the effectiveness of CNT-SLME

The extraction capability and chromatographic cleanliness of the CNT-SLME device were evaluated using aqueous solutions spiked with six NSAIDs at a concentration of 4 µg mL−1 (curve (b) of Fig. 3). For reference, a standard solution at the same concentration was analyzed to establish the characteristic retention times of each analyte (curve (a) of Fig. 3). Additionally, a blank aqueous sample was processed using the full CNT-SLME procedure to assess potential background interference from the device materials or the organic extractant (curve (c) of Fig. 3). The preliminary experiments were conducted under the following initial conditions: a 60 µL organic liquid membrane of dihexyl ether, a shaking frequency of 800 rpm for 120 min, a donor phase of 1 mL (pH 2), and an acceptor phase of 400 µL (pH 13).

Fig. 3. LC-UV chromatograms of curve (a), a 4 µg per mL NSAID standard solution; curve (b), the same NSAID standard solution (4 µg mL−1) after treatment with the CNT-SLME procedure; and curve (c), a blank solution subjected to the same procedure.

Fig. 3

Open in a new tab

As shown in curve (b) of Fig. 3, all six target NSAIDs (sulindac, loxoprofen, naproxen, flurbiprofen, diclofenac sodium, and ibuprofen) were successfully transferred from the donor phase to the acceptor phase through the dihexyl ether-impregnated PPF disk and ultimately partitioned into the acceptor phase. The retention times of the extracted analytes were in excellent agreement with those of the corresponding standards, indicating that the three-phase coaxial extraction process was highly effective. The sharp and symmetrical peak shapes obtained after extraction (curve (b) of Fig. 3) indicated the feasibility of mass transfer of the analytes from the acidic donor phase, through the hydrophobic liquid membrane, and into the alkaline acceptor phase. In contrast, the blank extract (curve (c) of Fig. 3) exhibited a remarkably clean chromatographic baseline with no detectable interference peaks within the retention windows of the target compounds. This result confirms that dihexyl ether is chemically compatible with the PPF support; it neither dissolves the fibers nor leaches into the aqueous phases in a manner that would interfere with NSAID detection. Collectively, these findings underscore the stability of the liquid membrane system and demonstrate its feasibility for further method optimization and application in human plasma analysis.

3.3. Optimization of extraction conditions via single-factor experiments

To determine the optimal extraction conditions, several key factors influencing extraction efficiency were systematically evaluated, including the type and volume of the organic solvent, the pH of the donor and acceptor phases, the shaking frequency, the extraction time, and the phase volumes. These parameters were optimized through single-factor experiments conducted in triplicate (n = 3). The mean extraction recoveries and their corresponding standard deviations were compared to identify the optimal settings. Initial parameters included a 60 µL dihexyl ether liquid membrane, a 1.0 mL donor phase (pH 2), and a 400 µL acceptor phase (pH 13), with shaking performed at 800 rpm for 120 min.

3.3.1. Selection of the membrane solvent and its volume

The selection of an appropriate membrane solvent is a critical step in the development of the CNT-SLME procedure because it constitutes the liquid barrier responsible for the selective transport of analytes. In this study, four potential membrane solvents with varying polarities and viscosities, including dihexyl ether, 1-octanol, n-dodecane, and toluene, were evaluated to determine their influence on the extraction efficiency of the six target NSAIDs. The results indicated that dihexyl ether yielded the highest extraction recoveries across all analytes (Fig. 4A). This superior performance is likely due to its optimal balance of hydrophobicity and polarity, which facilitates high partition coefficients for the NSAIDs while ensuring the long-term stability of the liquid phase within the polypropylene fabric pores. In contrast, the significantly lower recoveries observed with n-octanol and toluene may be attributed to their higher water solubility or insufficient affinity for the target compounds within this specific three-phase system.

Fig. 4. Optimization of CNT-SLME parameters for six NSAIDs: (A) extractant type, (B) dihexyl ether volume, (C) donor phase pH, (D) acceptor phase pH, (E) shaking frequency, (F) extraction time, (G) donor phase volume, and (H) acceptor phase volume.

Fig. 4

Open in a new tab

The volume of the membrane solvent was subsequently optimized within the range of 30 to 60 µL to ensure complete coverage of the PPF support (Fig. 4B). As the volume of dihexyl ether increased from 30 µL to 50 µL, a marked increase in extraction recovery was observed. This indicates that a minimum volume is necessary to form a continuous and stable liquid membrane across the fiber network. Beyond 50 µL, the recoveries reached a plateau, suggesting that the support was fully saturated. To ensure maximum stability and reproducibility during batch-scale processing, a volume of 60 µL was selected as the optimal membrane solvent volume for all subsequent experiments.

3.3.2. Optimization of donor and acceptor phase pH values

The pH values of the donor and acceptor phases are among the most influential parameters in the CNT-SLME procedure, as they determine the ionization state and the corresponding partition coefficients of the target analytes.17

For acidic compounds such as NSAIDs, which typically possess pKa values in the range of 4.00 to 4.85, the donor phase must be sufficiently acidic to ensure that the analytes exist in their neutral form, thereby facilitating their extraction into the hydrophobic membrane solvent. As shown in Fig. 4C, the extraction recoveries of the six NSAIDs reached their maximum at a donor phase pH of 2.0. When the pH was increased beyond 3.0, a significant decrease in extraction recovery was observed because a larger fraction of the analytes became ionized, which reduced their solubility in the dihexyl ether membrane. The slight recovery increase observed for diclofenac sodium at pH 5 compared with pH 4 may be associated with its specific molecular structure, including the presence of a nitrogen atom that could influence its pH-dependent distribution behavior. Nevertheless, the overall trend supported the selection of pH 2.0 as the optimal donor-phase pH.

Simultaneously, the acceptor phase pH must be maintained at a highly alkaline level to deprotonate the NSAIDs upon their arrival at the membrane–acceptor interface. This deprotonation effectively traps the analytes in their ionic form within the acceptor phase and prevents back-extraction into the organic phase. According to Fig. 4D, the extraction recoveries increased sharply as the acceptor phase pH rose from 10.0 to 13.0, reaching a stable plateau between pH 13.0 and 14.0. This trend indicates that a pH of 13.0 is sufficient to ensure complete ionization and efficient trapping. Consequently, a donor phase pH of 2.0 and an acceptor phase pH of 13.0 were selected for further studies.

3.3.3. Optimization of shaking frequency and extraction time

The mass transfer kinetics in the CNT-SLME system are primarily governed by the shaking frequency and the extraction time. Shaking frequency is a critical parameter as it provides the mechanical agitation necessary to reduce the thickness of the stagnant aqueous boundary layers at the membrane interfaces, thereby accelerating the diffusion of analytes. As shown in Fig. 4E, the extraction recoveries for the six NSAIDs improved significantly as the shaking frequency was increased from 400 to 1000 rpm. This enhancement is attributed to the increased convective mass transfer and faster partitioning of the target compounds from the donor phase into the organic membrane. A frequency of 1000 rpm was selected for further studies as it provided optimal recovery while maintaining the mechanical stability of the liquid membrane within the PPF support. This fixed agitation condition was used to provide a reproducible hydrodynamic window for the subsequent extraction-time evaluation and RSM optimization.

The effect of extraction time was evaluated over a range of 30 to 150 min to determine the duration required to reach thermodynamic equilibrium. As illustrated in Fig. 4F, the recoveries for all analytes increased steadily and reached a stable plateau at 90 min. This stabilization suggests that the partitioning of the NSAIDs between the acidic donor phase and the alkaline acceptor phase had reached near-equilibrium conditions. To achieve the best balance between high enrichment factors and practical analysis speed, an extraction time of 90 min was adopted for subsequent batch processing.

3.3.4. Volumes of the donor and acceptor phases

The volumes of the donor and acceptor phases directly influence the extraction recovery and the resulting enrichment factor, thereby affecting overall analytical sensitivity. Due to the physical constraints of the coaxial nested-tube design, the donor phase cannot properly interact with the supported liquid membrane when the volume is below 0.8 mL. Conversely, volumes exceeding this threshold may lead to spillage from the tube during high-frequency shaking. Therefore, donor volumes in the range of 0.8 to 1.1 mL were systematically evaluated. For the donor phase, increasing the solution volume typically enhances the enrichment factor; however, in this specific coaxial setup, larger volumes result in reduced agitation amplitude at the interface. This leads to insufficient contact between the donor phase and the supported liquid membrane, limiting internal convection and reducing mass transfer efficiency. As shown in Fig. 4G, the peak areas initially increased with volume, reached a maximum at 1.0 mL, and then slightly decreased. Consequently, a donor phase volume of 1.0 mL was selected as the optimal value to provide an appropriate balance between enrichment capacity and extraction efficiency.

For the acceptor phase, decreasing the volume enhances the enrichment factor by concentrating the analytes into a smaller space, resulting in higher peak areas. Acceptor volumes in the range of 200–500 µL were evaluated, as shown in Fig. 4H. The results indicated that the peak areas for all six NSAIDs peaked at 200 µL and decreased significantly as the volume increased to 500 µL. This trend is attributed to the dilution effect, where a larger volume of the acceptor phase reduces the final concentration of the target analytes. Although 200 µL provided the highest sensitivity, 300 µL was selected as the optimal volume for subsequent experiments. This choice was made to ensure ease of operation and provide a sufficient volume for reliable sample retrieval, while still maintaining a high enrichment factor and an excellent analytical response.

3.4. Multivariate optimization using response surface methodology

Response surface methodology (RSM) was employed to systematically evaluate the combined effects of key variables on the extraction performance of the CNT-SLME system and to elucidate potential interactions among them.18 Compared with traditional one-factor-at-a-time approaches, RSM provides a more comprehensive assessment by fitting a quadratic predictive model and visualizing factor interactions through three-dimensional (3D) response surfaces, thereby facilitating the identification of true optimal conditions within the experimental domain.

Based on preliminary single-factor screening, donor-phase pH, acceptor-phase pH, and extraction time were identified as the most influential factors affecting extraction recovery. Shaking frequency was fixed at 1000 rpm because it provided the highest recoveries without compromising membrane stability; consequently, possible frequency–time interactions were not included in the Box–Behnken design (BBD) model. It should be noted that the present BBD was developed under a fixed shaking frequency, with the main extraction parameters as the focus. The potential interaction between shaking frequency and extraction time was therefore not specifically considered and could be further explored in future studies using a more comprehensive response-surface design. A BBD was applied to construct the quadratic regression model, with the corresponding experimental levels summarized in Table S2, the experimental conditions and results listed in Table S3, and the analysis of variance (ANOVA) presented in Table S4. The model demonstrated high statistical significance (p < 0.0001), while the lack-of-fit term was not significant (p = 0.2410), indicating excellent model fitness and reliable predictive capability for optimizing extraction conditions.

Analysis of interaction effects revealed that the interaction between donor-phase pH and acceptor-phase pH was statistically significant (p = 0.0008). This observation aligns with the response surface shown in Fig. 5A, where the curved, non-parallel contour lines indicate a strong synergistic effect between a moderately acidic donor environment and a highly alkaline acceptor environment in enhancing transmembrane transport and recovery. In contrast, interactions involving extraction time were not significant (p > 0.05), consistent with the relatively smooth surfaces in Fig. 5B and C. These plots show that extraction time mainly exerts an individual quadratic effect, improving recovery up to a point, followed by a decline once equilibrium or membrane saturation is approached.

Fig. 5. 3D response surface plots showing the interaction effects of key variables on the extraction recoveries of NSAIDs: (A) interaction between donor-phase pH and acceptor-phase pH, (B) interaction between donor-phase pH and extraction time, and (C) interaction between acceptor-phase pH and extraction time.

Fig. 5

Open in a new tab

Based on numerical optimization, contour analysis, and visual inspection of the response surfaces, the optimal extraction conditions were determined to be a donor-phase pH of 2.5, an acceptor-phase pH of 13.5, and an extraction time of 106 min. Under these optimized conditions, the absolute extraction recoveries ranged from 46.89% to 66.63% (Fig. 6A), with an average recovery of 55.29% (Table S5). The model-predicted value was highly consistent with the experimental mean, with deviations falling within acceptable analytical variability. These results further confirm the robustness of the fitted model and the reliability of the RSM-based optimization strategy.

Fig. 6. Evaluation of extraction performance and matrix effects: (A) extraction recoveries of the six target NSAIDs under the optimized conditions, and (B) matrix effects of the six NSAIDs in human plasma.

Fig. 6

Open in a new tab

The observed absolute recoveries can be attributed to the inherent mass-transfer limitations of the three-phase SLM system. In this process, NSAIDs must sequentially partition from the acidic donor phase into the hydrophobic organic membrane and then into the alkaline acceptor phase, making the overall extraction efficiency dependent on partition equilibrium, membrane diffusion, and interfacial mass transfer. Therefore, the proposed CNT-SLME method was not intended to achieve exhaustive extraction, but rather to provide a practical balance among extraction efficiency, matrix compatibility under diluted plasma conditions, operational simplicity, and batch-processing capability.

The stability of the target NSAIDs under the optimized alkaline acceptor-phase conditions was further evaluated using standard solutions. As shown in Fig. S3, the peak area ratios between the freshly prepared 4 µg mL−1 mixed NSAID standard solution and the same solution kept at pH 13.5 for 106 min remained within 93.82–98.78%, confirming good stability of the NSAIDs under the optimized alkaline conditions.

3.5. Method validation

To comprehensively assess the reliability and applicability of the developed CNT-SLME method, this study evaluated key analytical performance parameters, including matrix effects, linearity, limit of detection (LOD), limit of quantification (LOQ), accuracy, and precision.

Initially, the influence of the plasma matrix on the analytical signal was evaluated to assess potential interference. The matrix effect was determined by comparing the peak areas of the analytes in spiked plasma samples with those in spiked pure water at equivalent concentration levels. As shown in Fig. 6B and summarized in Table S6, the calculated matrix effects for the six NSAIDs ranged from 86.50% to 103.51%. These values fall well within the acceptable range (typically 85–115%),19 indicating that co-existing substances in the human plasma matrix did not significantly influence the CNT-SLME process or the subsequent chromatographic performance. The good matrix compatibility observed here may be attributed to the selective transport characteristics of CNT-SLME under the present sample preparation conditions, which helped limit the carryover of residual plasma components into the acceptor phase. During plasma dilution with phosphate buffer at pH 2.5, no obvious visible protein precipitation, particulate matter, or membrane blockage was observed. Moreover, even if minor protein precipitation occurred under the acidic condition, the precipitated proteins would remain in the donor phase and could not pass through the hydrophobic organic liquid membrane into the acceptor phase. Therefore, they would not directly enter the final analytical solution. These results confirmed that the acidified plasma matrix was compatible with the CNT-SLME procedure and did not cause significant interference with quantitative determination. Consequently, aqueous standard solutions were employed as a simplified calibration strategy for method development in this proof-of-concept study, whereas matrix-matched calibration should be used for future routine clinical applications.

The linearity of the method was evaluated by constructing calibration curves for the six NSAIDs using spiked aqueous solutions. As summarized in Table 1, the method exhibited excellent linearity within the investigated concentration ranges, with coefficients of determination (R2) ≥ 0.998. The LODs and LOQs were calculated based on signal-to-noise ratios (S/N) of 3 and 10, respectively. The LODs ranged from 3.2 to 55.3 ng mL−1, while the LOQs ranged from 10.7 to 184.2 ng mL−1. These values demonstrate the high sensitivity of the CNT-SLME procedure, particularly for an LC-UV-based system. Given that therapeutic concentrations of NSAIDs in plasma typically fall within the low to mid µg mL−1 range, the achieved sensitivity is more than sufficient for clinical requirements.20,21

Table 1. Linear ranges, regression data, LODs and LOQs for the determination of NSAIDs.

Analyte Linear range (µg mL−1) Linear equation R 2 LOD (ng mL−1) LOQ (ng mL−1)
Sulindac 0.1–20 y = 58.515x + 0.4832 0.999 10.3 34.5
Loxoprofen 0.1–20 y = 31.59x − 0.388 0.998 14.1 46.9
Naproxen 0.05–20 y = 231.6x − 0.4087 0.999 3.2 10.7
Flurbiprofen 0.1–20 y = 41.932x + 0.209 0.999 18.5 61.5
Diclofenac sodium 0.1–20 y = 48.403x − 0.554 0.999 21.6 72.0
Ibuprofen 0.2–20 y = 15.493x − 0.1542 0.998 55.3 184.2
Open in a new tab

The precision and accuracy of the CNT-SLME method were rigorously evaluated by analyzing spiked human plasma samples at three representative concentration levels (0.2, 1 and 10 µg mL−1), as summarized in Table S7 and illustrated in Fig. 7. The intra-day relative recoveries for the six target NSAIDs ranged from 90.05% to 104.69% with RSD values below 6.90%, while the inter-day evaluations yielded relative recoveries between 91.96% and 104.39% with RSDs ranging from 0.73% to 8.26%. All precision and accuracy values fell well within the 15% limit, with recovery distributions for all analytes at the three spiking levels tightly clustered around the 100% target. These results confirm that the CNT-SLME procedure is a robust, stable, and reproducible tool for the therapeutic drug monitoring of NSAIDs in complex human plasma matrices. Although aqueous calibration was adopted as a simplified strategy in this proof-of-concept method-development study, matrix-matched calibration, incorporation of a suitable internal standard where feasible, and LC-MS/MS detection should be considered for future routine clinical applications to better account for inter-individual plasma variations and minor variations in extraction or injection efficiency, while further improving selectivity and sensitivity.

Fig. 7. Accuracy and precision profiles for the target NSAIDs: (A) distribution of relative recoveries obtained from intra-day analyses at three concentration levels (0.2, 1, and 10 µg mL−1), (B) relative standard deviations (RSD, %) for intra-day analyses at the same concentration levels, (C) distribution of relative recoveries obtained from inter-day analyses at three concentration levels (0.2, 1, and 10 µg mL−1), (D) relative standard deviations (RSD, %) for inter-day analyses at the same concentration levels.

Fig. 7

Open in a new tab

3.6. Plasma sample analysis

The practical applicability of the proposed method was evaluated by analyzing human plasma samples. To assess method selectivity and accuracy, unspiked plasma samples, as well as plasma samples spiked at two specific concentration levels (0.2 and 1.0 µg mL−1), were processed and analyzed. To minimize the impact of individual variability and ensure broad applicability, the experiments were replicated using plasma samples obtained from five different individuals.

Representative chromatograms are illustrated in Fig. 8. As shown in the chromatogram of the unspiked plasma (curve (a) of Fig. 8A), no target NSAIDs were detected, and no interfering peaks from endogenous substances were observed at the retention times of the target analytes, indicating good method selectivity. In contrast, distinct peaks corresponding to the target NSAIDs were clearly observed in the plasma samples spiked at 0.2 µg mL−1 (curve (b) of Fig. 8A) and 1.0 µg mL−1 (curve (c) of Fig. 8A). The peak intensities in traces b and c showed a good correlation with the spiked concentrations, further confirming the sensitivity and reliability of the method.

Fig. 8. Analysis of NSAIDs in human plasma samples: (A) chromatograms obtained from curve (a), blank plasma; curve (b), plasma spiked at 0.2 µg mL; and curve (c), plasma spiked at 1.0 µg mL; (B) violin plot illustrating measured concentrations from five different individual plasma lots spiked at 0.2 µg mL−1 and 1.0 µg mL−1 levels.

Fig. 8

Open in a new tab

Furthermore, the concentrations of the target NSAIDs in the spiked samples, determined using the established calibration curves, were found to be in good agreement with the nominal spiking levels (Fig. 8B and Table S8). The measured concentrations were in good agreement with the theoretical values, with all relative bias values within ±10.0%. These results indicate that, under the current conditions, the method was not substantially affected by matrix interference and was reliable for the quantitative analysis of NSAIDs in plasma samples.

3.7. Comparison with previously reported liquid-phase extraction methods

As the CNT-SLME procedure developed in this study belongs to the field of liquid-phase microextraction, it was compared with several representative liquid-phase extraction-based strategies previously reported for the determination of NSAIDs to highlight its technical and operational advancements (Table 2). The comparison primarily focuses on key analytical performance and practical features.

Table 2. Comparison of the proposed method with representative previously reported methods for the determination of NSAIDs in biological fluidsa.

Sample matrix Sample preparation method Solvent consumption Detection technique LOD (ng mL−1) Ref.
Urine LLME 4 mL ACN HPLC-DAD 40–180 22
Plasma DLLME 220 µL dichloromethane HPLC-DVD 160–500 23
Urine DLLME 250 µL dichloromethane GC-MS 0.1–0.2 24
Plasma LLME 300 µL dichloromethane LC-MS/MS 1.0–25.0 25
Urine HF-LPME 1-Octanol LC-DVD 60 26
Urine 3P-HF-LPME Dihexyl ether LC-UV 5–15 27
Plasma CNT-SLME 60 µL dihexyl ether LC-UV 3.2–55.3 This study
Open in a new tab
a

LLME, liquid–liquid microextraction; DLLME, dispersive liquid–liquid microextraction; HF-LPME, hollow fiber liquid-phase microextraction; 3P-HF-LPME, three-phase hollow fiber liquid-phase microextraction; CNT-SLME, coaxial nested-tube supported liquid membrane extraction; HPLC, high-performance liquid chromatography; DAD, diode array detector; UV, ultraviolet detection; MS, mass spectrometry; ACN, acetonitrile.

Despite their widespread use, existing liquid-phase extraction methods often face significant operational bottlenecks. Traditional LLE is notorious for being time-consuming and labor-intensive, often requiring large volumes of toxic organic solvents. While DLLME has improved extraction speed, the process remains highly dependent on manual labor and requires repetitive centrifugation for each individual sample, which inherently limits its sample throughput in clinical batch analysis. Furthermore, conventional HF-LPME is frequently hampered by the tedious assembly of fragile fibers and poor mechanical stability, leading to low user-friendliness and a higher risk of operation failure during large-scale applications. Even within the realm of SLM techniques, many existing configurations rely on specialized, expensive membrane modules or complex, custom-made devices that are difficult to replicate and operate for non-specialist laboratory personnel.

In contrast, the proposed CNT-SLME method overcomes these limitations through its superior simplicity, cost-effectiveness, and batch-scale processing capability. The core of this approach is a ready-to-use coaxial nested-tube design adapted from mass-produced commercial purification columns. This “plug-and-play” configuration eliminates the need for any complex manual assembly or specialized modules, making the device exceptionally easy to use. Crucially, the nested-tube setup allows for the simultaneous processing of multiple samples using a standard multi-well shaker, drastically improving laboratory efficiency and enabling true batch-scale analysis. By employing industrial-grade PPF as a robust and economical support for the liquid membrane, the method maintains good matrix compatibility and sensitivity comparable to more complex techniques. Collectively, these features establish CNT-SLME as a highly practical and scalable platform for batch-oriented plasma analysis, while further matrix-matched calibration, internal-standard-assisted validation, and coupling with LC-MS/MS would be desirable for routine therapeutic drug monitoring in clinical settings.

It should be noted that the extraction time of 106 min represents a relatively long duration compared with some rapid liquid-phase extraction methods. This is mainly associated with the mass-transfer characteristics of the three-phase SLM system, in which NSAIDs must sequentially partition from the donor phase into the organic membrane and then into the acceptor phase. However, the CNT-SLME procedure allows multiple samples to be processed simultaneously on a standard multi-well shaker, and the extraction step requires minimal manual intervention once the devices are assembled. Therefore, although the single-run extraction time is relatively long, the method remains practical for batch sample preparation due to its parallel-processing capability, low solvent consumption, and simplified operation.

4. Conclusion

A robust and cost-effective CNT-SLME protocol was established for the determination of NSAIDs in human plasma. By repurposing commercial purification columns and utilizing industrial-grade polypropylene fabric as a stable SLM support, a “plug-and-play” device was realized that eliminates the need for tedious manual assembly or specialized membrane modules. The defining feature of the CNT-SLME system is its batch-scale processing capability, which enables the simultaneous extraction of multiple clinical samples on a standard laboratory shaker without requiring individual centrifugation or complex handling. Systematic optimization and comprehensive validation confirmed that the method offers good sensitivity, accuracy, and precision. Overall, the CNT-SLME approach overcomes the long-standing trade-off between operational simplicity and sample throughput in liquid-phase microextraction. Its inherent stability, combined with the use of economical materials, makes it a promising platform for high-efficiency therapeutic drug monitoring and clinical research in complex biological matrices. For future routine clinical applications, the incorporation of matrix-matched calibration, a suitable internal standard, and LC-MS/MS detection would further strengthen quantitative robustness, selectivity, and sensitivity.

Ethical statement

All experiments involving human blood samples were conducted in accordance with the Declaration of Helsinki, applicable national regulations, and institutional guidelines. The study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University. Informed consent was obtained from all human participants involved in this study.

Conflicts of interest

The authors declare that they have no competing interests.

Supplementary Material

RA-016-D6RA03853K-s001

Acknowledgments

This work was supported by the Henan Provincial Science and Technology Research Project (262102311173), and Joint Construction Project of Henan Medical Science and Technology Research Project of China (LHGJ20220426, LHGJ20240280).

Data availability

All data underlying the results are available as part of the article and no additional source data is required.

Supplementary information (SI): chemical information and structures of the six NSAIDs (Table S1); factors, experimental runs, recovery results, and analysis of variance for the Box–Behnken design (Tables S2–S4); extraction recoveries, matrix effects, intra- and inter-day recoveries and precision, and analytical results for human plasma samples (Tables S5–S8); and supporting figures showing the assembly and operational workflow of the CNT-SLME device (Fig. S1), wettability evaluation of the PPF support (Fig. S2), and stability of the NSAIDs under alkaline conditions (Fig. S3). See DOI: https://doi.org/10.1039/d6ra03853k.

References

  1. Tena-Garitaonaindia M. Rubio J. M. Martínez-Plata E. Martínez-Augustin O. de Medina F. S. Biomed. Pharmacother. 2025;187:118069. doi: 10.1016/j.biopha.2025.118069. [DOI] [PubMed] [Google Scholar]
  2. Dong J.-Y. Wang R.-J. Ran Q. Yang X.-B. Han W.-C. Dong F.-Q. Chen D. Sustainable Chem. Pharm. 2026;49:102311. doi: 10.1016/j.scp.2026.102311. [DOI] [Google Scholar]
  3. Klomjit N. Ungprasert P. Eur. J. Intern. Med. 2022;101:21. doi: 10.1016/j.ejim.2022.05.003. [DOI] [PubMed] [Google Scholar]
  4. Bindu S. Mazumder S. Bandyopadhyay U. Biochem. Pharmacol. 2020;180:114147. doi: 10.1016/j.bcp.2020.114147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Ji T. Li Z. Liu Z. Chen Z. J. Chromatogr. A. 2023;1705:464226. doi: 10.1016/j.chroma.2023.464226. [DOI] [PubMed] [Google Scholar]
  6. Amiri A. Ghaemi F. J. Chromatogr. A. 2021;1648:462168. doi: 10.1016/j.chroma.2021.462168. [DOI] [PubMed] [Google Scholar]
  7. Parikh N. H. Solanki J. Parikh P. K. Ranch K. Gajjar A. Dave B. Maheshwari K. Renukuntla B. Boddu S. H. Arabian J. Chem. 2024;17:105446. doi: 10.1016/j.arabjc.2023.105446. [DOI] [Google Scholar]
  8. Zhang K. Li B. Chen Y. Mai H. Biomed. Anal. 2026;3:10. doi: 10.1016/j.bioana.2026.05.001. [DOI] [Google Scholar]
  9. Siqueira Sandrin V. S. Oliveira G. M. Weckwerth G. M. Polanco N. L. D. H. Faria F. A. C. Santos C. F. Calvo A. M. Metabolites. 2022;12:751. doi: 10.3390/metabo12080751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Wang D. Li M. Zhang X. Chai S. Chen D. Anal. Bioanal. Chem. 2026;418:2095–2105. doi: 10.1007/s00216-026-06335-2. [DOI] [PubMed] [Google Scholar]
  11. Wang Y. Lv M. Liu Y. Xia X. Zhang N. Chen D. Biomed. Anal. 2025;2:120. doi: 10.1016/j.bioana.2025.09.003. [DOI] [Google Scholar]
  12. Wang D. Li M. Dai T. Zhang X. Chen D. Microchem. J. 2025:114222. doi: 10.1016/j.microc.2025.114222. [DOI] [Google Scholar]
  13. Lv M. Wang Y. Qi C. Ma Y. Wang H. Chen D. Anal. Bioanal. Chem. 2025;418:91–109. doi: 10.1007/s00216-025-06179-2. [DOI] [PubMed] [Google Scholar]
  14. Lorenzo-Parodi N. Kaziur-Cegla W. Gjelstad A. Schmidt T. C. Anal. Bioanal. Chem. 2023;415:1765. doi: 10.1007/s00216-023-04579-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Bombana H. S. dos Santos M. F. Muñoz D. R. Leyton V. J. Chromatogr. B. 2020;1139:121973. doi: 10.1016/j.jchromb.2020.121973. [DOI] [PubMed] [Google Scholar]
  16. Lv M. Wang Y. Liu Y. Ma L. Xu S. Ma C. Chen D. J. Food Compos. Anal. 2025;148:108493. doi: 10.1016/j.jfca.2025.108493. [DOI] [Google Scholar]
  17. Ramos-Payan M. Maspoch S. Llobera A. Anal. Chim. Acta. 2016;946:56. doi: 10.1016/j.aca.2016.09.040. [DOI] [PubMed] [Google Scholar]
  18. Robles-Apodaca S. M. González-Vega R. I. Ruíz-Cruz S. Estrada-Alvarado M. I. Cira-Chávez L. A. Márquez-Ríos E. Del-Toro-Sánchez C. L. Ornelas-Paz J. d. J. Suárez-Jiménez G. M. Ocaño-Higuera V. M. Processes. 2024;12:2729. doi: 10.3390/pr12122729. [DOI] [Google Scholar]
  19. Wang B. Xu S. Li W. Liu Y. Li Z. Ma L. Xu X. Chen D. Talanta. 2024;271:125706. doi: 10.1016/j.talanta.2024.125706. [DOI] [PubMed] [Google Scholar]
  20. Helmy S. A. El-Bedaiwy H. M. Ther. Drug Monit. 2014;36(1):100–107. doi: 10.1097/FTD.0b013e31829dcbed. https://dx.doi.org/10.1097/ftd.0b013e31829dcbed [DOI] [PubMed] [Google Scholar]
  21. Davies N. M. Skjodt N. M. Clin. Pharmacokinet. 2000;38:377. doi: 10.2165/00003088-200038050-00001. [DOI] [PubMed] [Google Scholar]
  22. Hassan M. Alshana U. J. Pharm. Biomed. Anal. 2019;174:509. doi: 10.1016/j.jpba.2019.06.023. [DOI] [PubMed] [Google Scholar]
  23. Han S. Y. Jeong K. M. Yoo D. E. Jin Y. Kim E. M. Kim S. H. Lee S. Y. Kim S. H. Zhao J. Lee J. J. Sep. Sci. 2017;40:4050. doi: 10.1002/jssc.201700451. [DOI] [PubMed] [Google Scholar]
  24. Avino P. Notardonato I. Passarella S. Russo M. V. Appl. Sci. 2020;10:5441. doi: 10.3390/app10165441. [DOI] [Google Scholar]
  25. e Souza M. A. C. Reis N. F. A. de Oliveira Pacheco I. C. P. Martins M. A. P. Gloria M. B. A. Pianetti G. A. Fernandes C. J. Pharm. Biomed. Anal. 2022;217:114845. doi: 10.1016/j.jpba.2022.114845. [DOI] [PubMed] [Google Scholar]
  26. Horta R. P. do Amaral B. Peralta-Zamora P. G. Silva B. J. G. J. Chromatogr. Sci. 2018;56:564. doi: 10.1093/chromsci/bmy023. [DOI] [PubMed] [Google Scholar]
  27. Cha Y. B. Myung S.-W. Bull. Korean Chem. Soc. 2013;34:3444. doi: 10.5012/bkcs.2013.34.11.3444. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

RA-016-D6RA03853K-s001
RA-016-D6RA03853K-s001.pdf (148.7KB, pdf)

Data Availability Statement

All data underlying the results are available as part of the article and no additional source data is required.

Supplementary information (SI): chemical information and structures of the six NSAIDs (Table S1); factors, experimental runs, recovery results, and analysis of variance for the Box–Behnken design (Tables S2–S4); extraction recoveries, matrix effects, intra- and inter-day recoveries and precision, and analytical results for human plasma samples (Tables S5–S8); and supporting figures showing the assembly and operational workflow of the CNT-SLME device (Fig. S1), wettability evaluation of the PPF support (Fig. S2), and stability of the NSAIDs under alkaline conditions (Fig. S3). See DOI: https://doi.org/10.1039/d6ra03853k.

Articles from RSC Advances are provided here courtesy of Royal Society of Chemistry

RESOURCES