Optimized miniaturized air-assisted liquid–liquid microextraction for determination of non-steroidal anti-inflammatory drugs in bio-fluid samples

Abstract

A miniaturized air-assisted liquid–liquid microextraction (M-AALLM) method was developed for the simultaneous extraction of amlodipine (Am), atorvastatin (At), and ibuprofen (Ib) in some human bio-fluid samples prior to their determination using high performance liquid chromatography (HPLC). Since all steps are performed in a capillary tube, not only is there no need to use a disperser solvent, and the volumes of the extraction solvent and the sample solution are less than those in the other liquid-phase microextraction methods, but also a much better dispersion occurs. Several significant parameters including the type and volume of the extraction solvent, pH and ionic strength of the sample solution, and number of extraction cycles were investigated and optimized using a central composite design (CCD) strategy. The optimal extraction efficiencies were achieved using 2.6 μL of n-octanol, 50 μL of the sample solution including 3.25 g mL⁻¹ of NaCl, and 4 extraction cycles (∼20 s). The limits of detection, linear dynamic ranges (with r² > 0.99), recoveries, and repeatabilities were obtained to be 3 ng mL⁻¹, 0.01–235 μg mL⁻¹, 92–95%, <4.7% (intra-day precision), and <5.9% (inter-day precision), respectively. The results obtained indicate that the proposed M-AALLME method has desirable merits to apply as a simple and fast method for simultaneous determination of the tested drugs in the human plasma and urine samples.

---

1. Introduction

Non-steroidal anti-inflammatory drugs (NSAIDs) form a group of analgesic, anti-pyretic, and anti-inflammatory agents are used with great frequency in both humans and animals since they do not induce sedation, respiratory depression or addiction. This type of drugs alters renal function if given in high enough doses, reducing the renal blood flow and glomerular filtration rate thus causing sodium retention. In salt-sensitive subjects, this retention will cause the blood pressure to rise. For this reason, the effect of NSAIDs on blood pressure is greater in hypertensive individuals and in those treated with anti-hypertensive medication.¹ Thus, in people who consume these drugs at the same time, the increase in blood pressure is affected more by NSAIDs. To investigate the interaction between these drugs as well as their effects on each other, they must simultaneously be isolated and separated.

Amlodipine (Am), atorvastatin (At), and ibuprofen (Ib), as commonly used NSAIDs, have been widely used in the treatment of pain and inflammation in the rheumatic disease to treat adult rheumatoid arthritis and other musculoskeletal disorders (Fig. 1).

Fig. 1 Chemical structures of (a) amlodipine, (b) atorvastatin, and (c) ibuprofen.

Several methods have been reported for the direct analysis of NSAIDs in different pharmaceutical formulations and biological fluids. These works have generally considered the analysis of merely one compound or at most a few, and have been carried out using capillary electrophoresis,² spectrofluorimetry,³ flow-injection fluorimetry assay,⁴ or chromatographic separation (GC or HPLC) coupled with mass spectrometry.^5,6

Most GC methods are time-consuming, and have recently been replaced by HPLC with UV-visible detection,⁷ electro-chemical detection,⁸ or mass spectrometry.⁹

However, analysis of the drugs in biological fluids has always been a challenge to analytical chemists due to the complexity of the sample matrix and low levels of the drugs. For example, urine samples contain a wide variety of components including salts, aromatic acids, catecholamine metabolism, etc., and plasma samples consist of large molecular weight proteins, lipids, fats, etc. Many components in these matrices are often not compatible with the mobile phases used in HPLC. Hence, a direct injection of a urine or serum sample onto an HPLC system without a prior sample preparation is harmful.¹⁰ In this way, a simple and efficient sample preparation method is usually necessary to extract, clean-up, and concentrate the analytes of interest from biological matrices.

Some conventional extraction methods such as liquid–liquid extraction (LLE) or solid-phase extraction (SPE) require complex, laborious, and time-consuming procedures. Additionally, LLE requires large volumes of organic solvents, rendering it a potential danger to the environment and also the human health.¹¹ In order to overcome these disadvantages, the research works have focused on the miniaturized extraction methods, which lead to solvent and sample savings and a less time-consuming analysis.^12,13 The new approaches have fostered the evolution of a different family of techniques that strain the ability of the term “microextraction” to describe adequately the volume of the extracting phase, which is very small in relation to the volume of the sample solution.¹⁴

Drop-to-drop microextraction (DDME) has been suggested for the analysis of very small amounts of liquid samples, especially the biological ones. The general procedure involves using a drop of 0.5 μL of the solvent to extract the analytes from a sample in a conical or V-shaped micro-vial.¹⁵ However, since only very small amounts of the sample is frequently used, the concentrations of the analytes must be higher than those for the other extraction methods. On the other hand, the limited contact area between the extraction solvent and the sample solution is the main disadvantage of the DDME method.

Air-assisted liquid–liquid microextraction (AALLME) was introduced, for the first time, by Farajzadeh et al. in 2012 for the extraction of some triazole pesticides in real samples.¹⁶ This method is similar to dispersive liquid–liquid microextraction (DLLME), in which an appropriate mixture of an extraction solvent and a dispersive water-miscible solvent is rapidly injected into the aqueous sample solution by a syringe, and a cloudy state is subsequently formed. The main priority of AALLME over DLLME is the elimination of the dispersive solvent, which is toxic and increases the solubility of the analytes in the sample solution. In this method, a few microliters of the organic solvent (denser or lighter than water) is transferred into the aqueous sample solution in a conical centrifuge tube, and the mixture is then repeatedly withdrawn into a glass syringe and pushed out into the tube. By this action, fine organic droplets are formed, and the extraction solvent is entirely dispersed in the sample solution. After centrifugation of the cloudy solution formed, the extractant is settled down at the bottom of the centrifuge tube or gathered on the surface of sample and used for further analysis.^17,18 AALLME has a great potential to be considered as a miniaturized microextraction method if the volumes of the extraction solvent and the sample solution used can be reduced simultaneously.

In the current work, considering the prominent advantages of the DDME and AALLME methods, using a capillary tube as the extraction vessel and a micro-syringe for distribution of a small volume of the solvent in a low volume of the sample solution, a novel miniaturized AALLME (M-AALLME) method was developed and performed in a short period of time and with minimal laboratory devices. To optimize the parameters affecting the method performance, a central composite design (CCD) was applied, which allowed testing the effective factors including the sample solution pH value, solvent volume, salting effect, and extraction cycle numbers with the least number of observations.¹⁹ Finally, the optimized M-AALLME method was successfully examined for simultaneous extraction of amlodipine, atorvastatin and ibuprofen in the human plasma and urine samples.

2. Experimental

2.1. Reagents and solutions

Standards of At, Am and Ib were supplied from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade water, acetonitrile, and NaCl were obtained from Merck (Darmstadt, Germany). Undecanol, 1,2-dichlorobenzene, toluene, chloroform, and n-octanol (as the extraction solvents) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ultra-pure water was supplied from Merck (Darmstadt, Germany). Syringe filters (0.2 μm) were purchased from Sigma-Aldrich (St. Louis, MO, USA).

2.2. Preparation of calibration standards and quality controls

A stock solution of each one of the drugs (1000 μg mL⁻¹) was prepared in methanol and stored at 4 °C. The working solutions of the analytes (0.05–220 μg mL⁻¹) were prepared by serial dilution in ultra-pure water. The blank plasma and urine samples were spiked to yield a seven-point calibration curve (0.05, 1, 10, 50, 100, 150, and 220 μg mL⁻¹) for the analytes.

Two levels of quality control (QC) materials were prepared in-house by spiking the plasma and urine samples with the working solutions of the analytes. The bio-fluid samples were spiked with the working solutions to yield the final concentrations of 0.05 and 220 μg mL⁻¹ (low QC and high QC, respectively). The QC materials were stored at −20 °C in 5 mL aliquots.

2.3. Apparatus

A Knauer HPLC system (Berlin, Germany) equipped with a K-1001 HPLC pump, a D-14163 degasser, and a K-2600 UV detector was used. The chromgate software (version 3.1) for the HPLC system was used to acquire and process the chromatographic data. The chromatographic analysis was performed on an ODS III column (250 mm × ID 4.6 mm, 5 μm). The pH values for the solutions were measured on a model PHS-3BW pH-meter (Bell, Italy). A model EBA20 centrifuge 212 (Hettich, Germany) was used to accelerate the phase separation.

2.4. Chromatographic conditions

An optimum separation of the analytes was achieved by a gradient elution with a binary mobile phase (A, acetonitrile; B, 0.1 mol L⁻¹ of a phosphate buffer solution of pH 3.0) and a flow rate at 1 mL min⁻¹. The gradient elution program was 0–3 min 45% A; 3–3.5 min 75% A; 3.5–12 min 75% A; 12–15 min 60% A; HPLC loop and injection volumes of 5 μL; and a detection wavelength of 211 nm.

2.5. Sample preparation

2.5.1 Plasma samples. The human plasma samples were prepared from five healthy volunteers in our laboratory including (i) blank plasma (without analytes) and (ii) spiked plasma. Frozen human plasma samples were taken out of a −20 °C freezer, kept at room temperature for 30 min, and allowed to thaw. A total of 500 μL of plasma was transferred into a 1.5 mL polypropylene tube, to which 20 μL of a standard working solution (50 ng mL⁻¹) and 600 μL of acetonitrile were added. The resulting solution was centrifuged at 6000 rpm for 5 min. A total of 500 μL of the supernatant solution was then transferred into another tube, to which 500 μL of water with 0.1% formic acid was added, and the tube was then centrifuged at 6000 rpm for 10 min. Finally, 50 μL of the filtered supernatant solution was transferred into a capillary tube (0.8–1.1 mm ID, 90 mm length, and 90 μL volume) for extraction and clean-up.

2.6. Urine samples

The human urine samples were also prepared from five healthy volunteers, similar to the plasma samples. The urine samples were collected and kept at −20 °C. Before analysis, the sample was thawed, centrifuged for 10 min at 5000 rpm, and filtrated using a syringe filter. Subsequently, a certain volume of the standard drug of a known concentration was added to the urine sample. Finally, 50 μL of the supernatant solution was transferred into a capillary tube for extraction and clean-up.

2.7. Extraction process

The proposed method is quite simple and consists of three well defined steps, namely:

2.7.1 Sample loading. 50 μL of the sample solution and 2.6 μL of n-octanol (as the extraction solvent) were transferred into a capillary tube using a micro-syringe.

2.7.2 Extraction. The mixture of the sample solution and the extraction solvent was rapidly sucked into a micro-syringe, and then injected into the capillary tube via the syringe needle (for four times). As a consequence, the mixture turned cloudy.

2.7.3 Phase separation. In order to break down the resulting emulsified solution, the mixture was centrifuged for 4 min at 6000 rpm. The fine solvent droplets were gathered as the upper phase of the aqueous phase, which was collected using a micro-syringe.

In order to homogenize the extraction solvent and increase the reproducibility, the collected volumes were diluted to 5 μL by methanol before HPLC analysis (Fig. 2).

Fig. 2 Schematic set-up of the miniaturized air-assisted liquid–liquid microextraction method.

2.8. Experimental design and data analysis

A central composite design (CCD) was employed to optimize the extraction method. When the number of effective parameters are lower than/or equal to four, CCD is preferred.²⁰ Data analysis was performed using Design-Expert trial version 7.0.0. (Stat-Ease Inc., Minneapolis).

3. Results and discussion

Selection of an appropriate extraction solvent is of great importance in the M-AALLME method. The physico-chemical properties of the extraction solvent control the emulsion conditions, and, consequently, the extraction efficiency. The primary requirements of an adequate extraction solvent for the proposed method are low water solubility, being less dense than water, and having high extraction capability for the analytes of interest. Moreover, a low level of toxicity and a good chromatographic behavior are the other desirable properties. Based on these criteria, 1-octanol, toluene, n-heptane, cyclohexane, 2-dodecanol, 1-undecanol, and n-hexane were tested. The results obtained showed that 1-octanol gave the best extraction efficiency (Fig. 3).

Fig. 3 Effect of the type of extraction solvent on the extraction efficiency (in terms of peak area).

To determine the effective parameters and achieve the best method performance, CCD was used to obtain the optimal experimental conditions with a minimum number of experiments. Peak area was used as a criterion for the extraction efficiency. In this way, four independent parameters namely pH of the sample solution (A), volume of the solvent (B), number of extraction cycles (C), and amount of salt (D) were investigated and optimized (Table 1).

Table 1 Variables and values used for central composite design

Variables	Symbols	Code levels
		−1	0	+1
pH	A	2.5	3.5	4.5
Volume of extraction solvent (μL)	B	1.75	2.5	3.25
Number of extraction cycles	C	2.25	3.5	4.75
Amount of salt (g mL^−1)	D	3.25	5.5	7.75

---

For the four independent variables, the total number of required tests was calculated as:

N = 2ⁿ + 2n + n_c = 2⁴ + 2 × 4 + 6 = 30 (1)

in which 2⁴ factorial design augmented with 2 × 4 star points and 6 central points.

In order to validate the model and obtain the interaction between the variables and responses, the analysis of variance (ANOVA) was applied (Table 2), which revealed that the effect of D was not significant (P > F < 0.0001), while factors A, B, and C were significant for the analysis.²¹ The p-value probability was also relatively low (p < 0.05), indicating a high confidence level (95%).²² The lack-of-fit test was designed to determine whether the selected model was adequate in describing the experimental data or if a more complicated model should be used. The test was performed by comparing the variability of the current model residuals with the variability between the observations (area counts) at replicate values of the independent variables.²³

Table 2 Reduced response models and statistical parameters obtained

Source	Sum of squares	df	Mean square	F-Value	p-Value
					Prob > F
Amlodipine
Model	64 231.26	8	8028.908	99.42904	<0.0001	Significant
Lack-of-fit	1471.54	16	91.97	2.050981	0.2193	Not significant
Residual	1695.75	21	80.75
Pure error	224.2128	5	44.84257
CV% = 9.35	R^2	0.97	RAdj^2	0.96	RPred^2	0.92
Equation	R = +220.82 + 29.92A + 7.92B + 14.67C − 2.78D − 19.09A^2 − 15.81B^2 − 24.75C^2 + 11.32D^2
Atorvastatin
Model	423 789.1	8	52 973.64	98.02416	<0.0001	Significant
Lack-of-fit	9827.11	16	614.19	2.02	0.2249	Not significant
Residual	11 348.7	21	540.41
Pure error	1521.58	5	304.32
CV% = 10.39	R^2	0.97	RAdj^2	0.96	RPred^2	0.92
Equation	R = +562.39 + 77.34A + 20.49B + 37.86C − 7.59D − 49.07A^2 − 40.78B^2 − 63.76C^2 + 27.78D^2
Ibuprofen
Model	113 553.6	8	14 194.201	99.752015	<0.0001	Significant
Lack-of-fit	2573.932	16	160.87	1.97	0.2388	Not significant
Residual	2988.19	21	142.29
Pure error	414.261	5	82.852192
CV% = 9.09	R^2	0.97	RAdj^2	0.96	RPred^2	0.92
Equation	R = +287.03 + 40.05A + 10.61B + 19.56C − 3.82D − 25.40A^2 − 21.21B^2 − 33.03C^2 + 14.30D^2

---

The qualities of the fitted polynomial models were examined on the basis of the determination coefficients (R²). Since R² always decreases when a regression variable is eliminated from a regression model, in statistical modeling, the adjusted R² (R_Adj²), which takes the number of regression variables into account, is usually selected.²⁴ In the present work, the amounts of R², R_Adj², and predicted R² (R_Pred²) for all the response models were well within the acceptable limits. R_Adj² is an adjustment for the number of terms in the respective model, and its higher values indicate a better accordance with the experimental data and the fitted model.²⁵ The results obtained showed that there were no significant differences between R_Adj² and R_perd², revealing that the experimental data showed a good fitness with the second-order polynomial equations (Table 2).

The % CV is a measurement that expresses standard deviation as a percentage of the mean. As a general rule, a model can be considered reasonably reproducible if its % CV is less than 10%.²⁶ In general, a % CV higher than 10 indicates that variation in the mean value is high and does not satisfactorily develop an adequate response model.²⁷

A close inspection of Fig. 4a reveals that the residuals are generally close to a straight line, which indicates the normal distribution of the error, and supports the fact that the model adequately fits the data. These plots are very important, and it is required to check the normality assumption in the fitted model. This ensures that the model provides an adequate approximation to the optimization process. It is clear that no obvious pattern is followed in the residual vs. the predicted response (Fig. 4b).^25,28 The graph represents the normal distribution of errors in a specified range (between +3 and −3 standard deviation), which is indicative of the lack of a systematic error.

Fig. 4 (a) Plot of predicted values vs. actual values for atorvastatin (based on peak area) (b) plot of residuals vs. predicted response for atorvastatin.

3D surface plots were constructed, as shown in Fig. 5a–c. These plots showed visually the effects and interaction of two independent variables on the responding variable as the third independent variable was fixed at the central experimental level of zero.²⁹ The variables giving quadratic and interaction terms with the largest absolute coefficients in the fitted models were chosen for the axes of the response surface plots to account for the curvature of the surfaces. The effect of pH and solvent volume on the peak area of At (as a representative analyte) is given in Fig. 5a. With increase in the pH and volume of solvent extraction up to certain values, the analyte response increases and then is constant (pH < 4.3, lower than the pK_a values for the analytes). This is due to the presence of acidic functional groups in the analyte structures, which kept them at their molecular forms in the sample solution in acidic forms.

Fig. 5 Typical 3D response surface plots for atorvastatin as a representative analyte. (a) pH vs. volume of extraction solvent, (b) pH vs. extraction cycle, (c) volume of extraction solvent vs. extraction cycle.

The effects of volume of the extraction solvent and the number of extraction cycles on the peak areas of At are given in Fig. 5c. With increase in the volume of the extraction solvent and the extraction cycles up to certain values, the peak area increases firstly and then decreases. However, when a constant volume of the extraction solvent is used, the analyte peak area slightly decreases after reaching the equilibrium status. This might be due to the increase in the solubility of the extractant in the aqueous sample. As expected, with increase in the number of the extraction solvent, the extraction efficiency increases as well. After reaching a maximum peak area, the efficiency remains constant or slightly decreases.

A desirability of 0.92 (D = 0.92) was obtained after the modeling and optimization steps. Based on the desirability obtained, the best responses were reached when the extraction conditions were 2.6 μL of the extraction solvent and 50 μL of the sample solution with pH 4.2, 3.25 g mL⁻¹ of NaCl, and 4 extraction cycles (∼20 s).

3.1. Method validation

3.1.1 Accuracy. In order to investigate the method validation, the spiking recovery (SR) values were calculated according the following equation:³⁰

SR (%) = (C_found − C_real)/C_added (2)

where C_found, C_real, and C_added are the concentration of the analyte after addition of a known amount of the standard to the real sample, concentration of the analyte in the real sample, and concentration of a known amount of the standard spiked to the real sample, respectively. The standard addition method was used to determine the analytes concentrations in the real sample. The summarized quantitative results are listed in Table 3 (n = 3). The high spiking recovery values revealed that no significant matrix effect was observed in the real samples. Consequently, the proposed method showed a satisfactory accuracy and a good capability in the analysis of the analytes in the understudied samples.

Table 3 Recovery of the analytes from samples taken from healthy volunteers (n = 3)^a

Sample	Added (μg mL^−1)	Amlodipine		Atorvastatin		Ibuprofen
		Found ± S.D^b (μg mL^−1)	Recovery (%)	Found ± S.D (μg mL^−1)	Recovery (%)	Found ± S.D (μg mL^−1)	Recovery (%)
a Extraction conditions: extraction solvent: 2.6 μL of n-octanol; sample pH = 4.2, sample volume: 50 μL; number of extraction cycles = 4, salt addition = 3.2 g mL^−1.b Standard deviation.
Plasma 1	10	9.7 ± 0.45	97 ± 3.8	10.1 ± 0.46	101 ± 4.0	9.9 ± 0.46	99 ± 4.4
	50	49.5 ± 2.4	99 ± 4.1	49.0 ± 2.5	98 ± 3.9	50.0 ± 2.7	100 ± 4.1
	150	151.5 ± 7.3	100 ± 3.9	148.5 ± 7.6	99 ± 3.8	147.0 ± 7.4	98 ± 3.9
Plasma 2	10	9.8 ± 0.46	98 ± 3.9	10.0 ± 0.44	100 ± 3.9	9.9 ± 0.42	99 ± 4.3
	50	48.5 ± 2.5	97 ± 4.3	48.0 ± 2.4	96 ± 4.1	49.5 ± 2.6	99 ± 4.6
	150	148.5 ± 7.0	99 ± 4.2	148.5 ± 7.2	99 ± 4.5	153.0 ± 7.9	102 ± 3.9
Plasma 3	10	10.1 ± 0.54	101 ± 4.6	9.9 ± 0.52	99 ± 3.8	10.2 ± 0.54	102 ± 4.1
	50	49.5 ± 2.7	99 ± 3.8	50.0 ± 2.8	100 ± 4.0	48.5 ± 2.5	97 ± 4.2
	150	145.5 ± 8.0	97 ± 4.0	147.0 ± 7.5	98 ± 4.1	148.5 ± 8.1	99 ± 4.3
Plasma 4	10	9.9 ± 0.46	99 ± 4.2	9.8 ± 0.45	98 ± 4.0	9.9 ± 0.47	99 ± 4.3
	50	51.0 ± 2.8	102 ± 4.1	49.5 ± 2.7	99 ± 4.5	48.5 ± 2.8	97 ± 4.0
	150	144.0 ± 7.3	96 ± 3.8	150.0 ± 8.1	100 ± 3.9	151.5 ± 8.2	101 ± 4.2
Plasma 5	10	9.8 ± 0.52	98 ± 4.1	9.9 ± 0.55	99 ± 3.9	10.1 ± 0.54	101 ± 4.2
	50	50.0 ± 2.8	100 ± 4.5	48.5 ± 2.6	97 ± 4.0	48.5 ± 2.7	97 ± 4.0
	150	151.5 ± 7.7	101 ± 4.0	150.0 ± 7.9	100 ± 4.4	147.0 ± 8.2	98 ± 4.1
Urine 1	10	10.2 ± 0.55	102 ± 4.2	9.8 ± 0.52	98 ± 3.8	10.1 ± 0.49	101 ± 4.0
	50	49.5 ± 2.6	99 ± 3.9	50.0 ± 2.7	100 ± 4.2	49.5 ± 2.6	99 ± 3.9
	150	151.5 ± 8.1	101 ± 4.0	148.5 ± 8.0	99 ± 4.0	144.0 ± 7.8	96 ± 4.4
Urine 2	10	9.9 ± 0.53	99 ± 4.0	9.7 ± 0.53	97 ± 3.7	9.9 ± 0.51	99 ± 4.2
	50	50.5 ± 2.7	101 ± 4.3	50.0 ± 2.7	100 ± 4.0	48.0 ± 2.5	96 ± 3.9
	150	147.0 ± 8.0	98 ± 3.9	148.5 ± 8.1	99 ± 4.0	153.0 ± 8.1	102 ± 4.1
Urine 3	10	10.1 ± 0.54	101 ± 3.8	9.8 ± 0.51	98 ± 4.0	9.9 ± 0.53	99 ± 4.1
	50	49.5 ± 2.5	99 ± 4.2	51.0 ± 2.7	102 ± 4.1	50.5 ± 2.8	101 ± 4.0
	150	151.5 ± 8.1	101 ± 4.3	148.5 ± 8.0	99 ± 3.9	147.0 ± 7.9	98 ± 4.3
Urine 4	10	9.8 ± 0.48	98 ± 4.0	9.9 ± 0.51	99 ± 4.1	9.8 ± 0.52	98 ± 4.2
	50	50.5 ± 2.7	101 ± 4.1	51.0 ± 2.7	102 ± 4.0	50.0 ± 2.8	100 ± 4.3
	150	151.5 ± 7.7	101 ± 4.3	147.0 ± 7.9	98 ± 4.4	148.5 ± 7.8	99 ± 3.8
Urine 5	10	9.9 ± 0.50	99 ± 4.2	10.0 ± 0.54	100 ± 4.2	9.9 ± 0.52	99 ± 4.1
	50	48.5 ± 2.6	97 ± 4.3	49.0 ± 2.5	98 ± 3.9	50.5 ± 2.7	101 ± 4.5
	150	151.5 ± 8.0	101 ± 4.0	153.0 ± 7.9	102 ± 4.4	145.5 ± 7.8	97 ± 3.8

---

3.1.2 Linearity, limits of detection, and limits of quantification. Standard calibration curves were prepared over the concentration range of 0.01–216 μg mL⁻¹ for At, 0.01–205 μg mL⁻¹ for Am, and 0.01–235 μg mL⁻¹ for Ib (obtained using seven concentrations in each range). The peak areas were treated by the linear least square regression analysis.

The limit of detection (LOD) and limit of quantification (LOQ) were calculated as 3 × (σ/S) and 10 × (σ/S), respectively, where σ is the standard deviation of the blank and S is the slope of the calibration curve (Table 4).

Table 4 LOD, LOQ, linear dynamic range (LDR), intra-day and inter-day precisions, and squared correlation coefficient (r²) for proposed method under optimum conditions

Analyte	LOD (μg mL^−1)	LOQ (μg mL^−1)	LDR (μg mL^−1)	Intra-day precision^a (% RSD)	Inter-day precision^a (% RSD)	Correlation coefficient (r^2)
a n = 5.
Amlodipine	0.003	0.01	0.01–205	4.6	5.5	0.998
Atorvastatin	0.003	0.01	0.01–216	4.4	5.8	0.998
Ibuprofen	0.003	0.01	0.01–235	4.7	5.9	0.997

---

3.1.3 Precision. The precision of the proposed method, as determined by the coefficient of method variations, was expressed by studying the repeatability (intra-day and inter-day precisions) (Table 4). The intra-day precision was evaluated by analyzing five replicates of the QC samples at three concentrations on the same day, and the inter-day precision was established by analyzing five replicates of the QC samples at three concentrations in three consecutive days (Table 4).

3.2. Evaluation of method performance

Under the optimized experimental conditions, the proposed method was successfully used to extract Am, At, and Ib in the human plasma and urine samples, simultaneously. The urine and plasma samples were spiked with 0.05 and 10 μg mL⁻¹ for both Am and At and 100 μg mL⁻¹ of Ib, and the extraction process was repeated for three times for each sample. A statistical comparison was done on the assay results obtained from the proposed method using the student's t-test. The values for t_cal were less than the t-table values, which confirmed the accuracy of the data. Typical chromatograms showed that there was no significant interference in the peaks for the drug standards after and before spiking the analytes in blank urine and plasma samples after extraction (Fig. 6).

Fig. 6 Chromatograms of standards (a), blank urine sample after extraction (b), spiked urine sample after extraction (c), blank plasma sample after extraction (d), spiked plasma sample after extraction (e). Extraction conditions: extraction solvent: 2.6 μL of n-octanol; sample pH = 4.2, sample volume: 50 μL; number of extraction cycles = 4, salt addition = 3.2 g mL⁻¹.

4. Conclusion

The purposed method, which is a statistically-based experimental design method, proved to be a valuable approach for the extraction and clean-up of trace amounts of drugs in biological liquid samples such as human plasma and urine. Since the method requires just a small volume of an organic solvent, it could be considered as an environmentally friendly method and good applicability for the determination of selected compounds in human plasma and urine samples. Comparison of the proposed method with other reported AALLME methods shows that it has some interesting advantages including the following ones: (i) low amounts of the extraction solvent and sample solution are consumed, (ii) it is simple, and can be performed in a short period of time, (iii) the analytical merits are comparable to other extraction methods, (iv) no toxic disperser solvents such as ethanol, acetone, acetonitrile, and methanol – used in the other LPME-based methods – are used in the proposed method (Table 5). Application of an experimental design also provides a large amount of information concerning the factor-response behavior of the method with a minimum number of experiments. These characteristics are of key interest for laboratories doing routine analysis of this type of analytes in the bio-fluid samples.

Table 5 Comparison of proposed method with other published procedures

Method/detection technique	Samples	Analytes	Extraction solvent	Volume of extraction solvent	Number of extraction cycles	Extraction time	LOD	References
Automated air-assisted liquid-phase microextraction/stopped flow spectrophotometry	Water	Chromium(VI)	Toluene	250 μL	4	∼6 min	4.5 μg L^−1	31
Low-toxic air-agitated liquid–liquid microextraction using a solidifiable organic solvent/gas chromatography	Human plasma and wastewater	Amitriptyline and imipramine	1-Dodecanol	14 μL	13	2 min	5.0–7.0 ng mL^−1	32
Tandem air-agitated liquid–liquid microextraction/high-performance liquid chromatography	Human plasma and wastewater	Diclofenac, ibuprofen, and mefenamic acid	1,2-Dichloroethane	37 μL	17	7 min	0.1–0.3 ng mL^−1	33
Air-assisted liquid phase microextraction based on switchable hydrophilicity solvent/atomic absorption spectrometer	Road dust, tap water, waste water, sea water and river water	Palladium	Triethylamine	750 μL	5	3 min	0.07 μg L^−1	34
Air-assisted liquid–liquid microextraction coupled/flame atomic absorption spectrometry	Water	Lead	Carbon tetrachloride	210 μL	8	<1 min	1.36 ng mL^−1	35
Air-assisted liquid–liquid microextraction/gas chromatography	Tap water, river water, petrochemical wastewater, refinery wastewater, and municipality wastewater	Phenolic compounds	1,1,1-Trichlorethane	40 μL	20	<2 min	0.1–0.4 μg L^−1	36
Air-assisted, low-density solvent-based liquid–liquid microextraction and solidified floating organic droplets/spectrophotometry	Fruit juices	Carotenoids	1-Dodecanol	40 μL	7	<1 min	0.04 μg mL^−1	37
Ionic-liquid-mingled air-assisted liquid–liquid microextraction based on solidification of floating organic droplets	Environmental water and honey	Benzoylurea insecticides	1-Dodecanol and [P14,6,6,6]PF6	30 μL of 1-dodecanol and 10 μL of [P14,6,6,6]PF6	10	6 min	0.01–0.1 μg L^−1	38
Optimized miniaturized air-assisted liquid–liquid microextraction	Human plasma and urine	Amlodipine, atorvastatin and ibuprofen	n-Octanol	2.6 μL	4	20 s	3 ng mL^−1	The present work

---

Compliance with ethical standards

The experimentations in this study absolutely served for maintaining, sampling, and analysis in accordance with the ethical guidelines and recommendations for biomedical research and human laboratory of Declaration of Helsinki. Also the research board of research & technology deputy of Semnan University has approved all results, and the consent of all participants was obtained for the research work involving human subjects.

Acknowledgements

The authors would like to thank Semnan University Research Council for the financial support of this work.

References

1. E. J. Kuipers, L. Lundell, E. C. Klinkenberg-Knol, N. Havu, H. P. Festen, B. Liedman, C. B. Lamers, J. B. Jansen, J. Dalenbäck and P. Snel, N. Engl. J. Med., 1996, 334, 1018–1022.
2. V. Pucci, R. Mandrioli, M. A. Raggi and S. Fanali, Electrophoresis, 2004, 25, 615–621.
3. G. M. Escandar, A. J. Bystol and A. D. Campiglia, Anal. Chim. Acta, 2002, 466, 275–283.
4. M. I. Albero, C. Sauchez-Pedreño and M. S. Garcia, J. Pharm. Biomed. Anal., 1995, 13, 1113–1117.
5. H. H. Maurer, F. X. Tauvel and T. Kraemer, J. Anal. Toxicol., 2001, 25, 237–244.
6. M. Locatelli, V. Ferrone, R. Cifelli, R. Carmine Barbacane and G. Carlucci, J. Chromatogr. A, 2014, 1367, 1–8.
7. B. Barfi, A. Asghari, M. Rajabi and N. Mirkhani, RSC Adv., 2015, 5, 106574–106588.
8. A. Loudiki, W. Boumya, H. Hammani, H. Nasrellah, Y. El Bouabi, M. Zeroual, A. Farahi, S. Lahrich, K. Hnini, M. Achak, M. Bakasse and M. A. El Mhammedi, Mater. Sci. Eng., C, 2016, 69, 616–624.
9. A. Rúbies, L. Guo, F. Centrich and M. Granados, Anal. Bioanal. Chem., 2016, 408, 5769–5778.
10. S. L. Prabu and T. N. K. Suriyaprakash, Extraction of Drug from the Biological Matrix: A Review. In: Applied Biological Engineering – Principles and Practice, 2012, p. 479.
11. X. You, Z. Xing, F. Liu and N. Jiang, J. Chromatogr. A, 2013, 1311, 41–47.
12. B. Barfi, A. Asghari, M. Rajabi, A. Barfi and I. Saeidi, J. Chromatogr. A, 2013, 1311, 30–40.
13. A. Barfi, H. Nazem, I. Saeidi, M. Peyrovi, M. Afsharzadeh, B. Barfi and H. Salavati, J. Pharm. Biomed. Anal., 2016, 121, 123–134.
14. H. Lord and J. Pawliszyn, J. Chromatogr. A, 2000, 902, 17–63.
15. H. F. Wu, J. F. Yen and C. C. Chin, Anal. Chem., 2006, 78, 1707–1712.
16. M. A. Farajzadeh and M. R. A. Mogaddam, Anal. Chim. Acta, 2012, 728, 31–38.
17. B. Barfi, A. Asghari, M. Rajabi and S. Sabzalian, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2015, 998, 15–25.
18. B. Barfi, A. Asghari, M. Rajabi, A. Goochani Moghadam, N. Mirkhani and F. Ahmadi, J. Pharm. Biomed. Anal., 2015, 111, 297–305.
19. M. Rajabi, M. Kamalabadi, M. Jamali, J. Zolgharnein and N. Asanjarani, Hum. Exp. Toxicol., 2013, 32, 620–631.
20. T. Rakić, I. Kasagić-vujanović, M. Jovanović, B. Jančić-stojanović and D. Ivanović, Anal. Lett., 2014, 47, 1334–1347.
21. M. A. Farajzadeh, M. Bahram and M. R. Vardast, Clean: Soil, Air, Water, 2010, 38, 466–477.
22. P. Fernández, S. Seoane, C. Vázquez, A. M. Bermejo, A. M. Carro and R. A. Lorenzo, Anal. Bioanal. Chem., 2011, 401, 2177–2186.
23. J. Parajo, J. Alonso, M. Lage and D. Vazquez, Bioprocess Eng., 1992, 8, 129–136.
24. J. Zhang, Z. Liang, S. Li, Y. Li, B. Peng, W. Zhou and H. Gao, Talanta, 2012, 98, 145–151.
25. Q. K. Beg, V. Sahai and R. Gupta, Process Biochem., 2003, 39, 203–209.
26. R. H. Myers, D. C. Montgomery and C. M. Anderson-Cook, Response Surface Methodology: Process and Product Optimization Using Designed Experiments, A Wiley-Interscience Publication, 4th edn, 2016.
27. S. Dutta, A. Bhattacharyya, A. Ganguly, S. Gupta and S. Basu, Desalination, 2011, 275, 26–36.
28. Y.-K. Liu, X.-Y. Jia, X. Liu and Z.-Q. Zhang, Talanta, 2010, 82, 1212–1217.
29. H. Sereshti, V. Khojeh and S. Samadi, Talanta, 2011, 83, 885–890.
30. D. T. Burns, K. Danzer and A. Townshend, Pure Appl. Chem., 2002, 74, 2201–2205.
31. M. Alexovič, V. Andruch, I. S. Balogh and J. Šandrejová, Anal. Methods, 2013, 5, 2497–2502.
32. A. Asghari, Z. Saffarzadeh, M. Bazregar, M. Rajabi and L. Boutorabi, Microchem. J., 2017, 130, 122–128.
33. M. Bazregar, M. Rajabi, Y. Yamini, A. Asghari and M. Hemmati, Anal. Chim. Acta, 2016, 917, 44–52.
34. M. Ezoddin, K. Abdi and N. Lamei, Talanta, 2016, 153, 247–252.
35. G. Fakhriyan, H. Zavvar Mousavi and S. M. Sajjadi, Anal. Methods, 2016, 8, 995–1002.
36. H. Ghorbanpour, A. Yadeghari, L. Khoshmaram and M. A. Farajzadeh, Anal. Methods, 2014, 6, 7733–7743.
37. P. Sricharoen, N. Limchoowong, S. Techawongstien and S. Chanthai, Food Chem., 2016, 203, 386–393.
38. M. Yang, X. Xi, X. Yang, L. Bai, R. Lu, W. Zhou, S. Zhang and H. Gao, RSC Adv., 2015, 5, 25572–25580.

---