Reagents and materials

All organic solvents and NaCl salt were of analytical grade and were purchased from Merck KGaA (Germany) and used as received. Milli-Q^® water (18.3 M'Ω cm^–1) was used throughout the experiment after filtering through 0.22 mm Nylon membrane. 1,1,3,3-tetraethoxypropane (TEP) (99%) as precursor of malondialdehyde preparation was also purchased from Merck KGaA (Germany). 1,1,3,3-tetramethoxypropane (TMP) was obtained from Sigma-Aldrich (St. Louis, MO, USA) and used as internal standard.

TEP was used to prepare a MDA stock solution. A volume of 10 μL of TEP was accurately diluted to 10 mL with 0.1N HCl in a screw-capped test tube and incubated in boiling water bath for 15 min and then rapidly cooled with tap water. A working solution of MDA was prepared by pipetting 1 mL of this solution into a 100 mL volumetric flask and diluted to volume with ultrapure water. This solution contains 1 mg of MDA per litter which is stable for few days at 4 °C.

Apparatus

A Varian 450-GC gas chromatograph (Varian Inc., USA) equipped with a flame ionization detector was used for all analyses. The GC was fitted with a Varian capillary column (30 m × 0.32 mm, film thickness 0.25 μm). The gas chromatography conditions were as follows: (1) the injector port was operated in split mode with a split ratio of 50:1 and it was kept at 200 °C; (2) the FID temperature was 250 °C; (3) the initial oven temperature was 60 °C for 1 min, and increased to 140 °C at 20 °C min^-1 then raised to 180 °C at 40 °C min^–1, and remained for 1 min at this final temperature; (4) usage of high-purity nitrogen as a carrier gas (1.8 mL min^–1). Hydrogen and air were used as detector gases at 30 and 300 mL min^–1, respectively. A vortex mixer (50 Hz) from Labnet International, INC (USA) was used. Also for this work a Universal 32 R, Hettich Zentrifugen (Germany) centrifuge was employed . A Denver (Germany) model UB-10 pH meter was used for pH measurements.

VADLLME procedure

6 mL of the aqueous standard solution (1 mg L^-1) or sample solution is transferred into a 10-mL glass extraction vessel. A mixture of 450 μL acetic acid as dispersive solvent with 150 μL benzyl alcohol as extracting solvent is transferred into the vessel. The mixture was then vigorously shaken on vortex agitator for 1 min at max speed rate, a cloudy solution is obtained. Then, the mixture is centrifuged for 4 min at 2000 rpm. As a result of centrifugation, benzyl alcohol droplets were precipitated at the bottom of the centrifuge tube. Deposited phase was transferred to a micro-tube by a conventional sampler, from which, 0.9 μL was mixed with 0.1 μL of TMP (as internal standard) inside a GC microsyringe and injected into the GC. The microsyringe was cleaned with methanol three times before next injection to avoid formation of air bubbles and the carryover of compounds between extractions. In all cases, the analytical signal was recorded as the area ratio of the analyte peak to the internal standard peak. Calibration was performed using aqueous calibration solutions subjected to the same VADLLME procedure described above.

Optimization of VADLLME

With the aim of achieving the best efficiency of the proposed method, different factors affecting extraction efficiency were investigated, including the type and volume of the extraction and dispersive solvents, vortex and centrifugation times, and volume, ionic strength and pH of the sample solution. A uni-variate approach was employed to optimize influential factors in order to simplify the optimization procedure. A series of experiments were designed for this purpose as discussed below.

Type of the extraction solvent

The selection of an appropriate extraction is a critical point for all microextraction processes. While the extraction solvent needs to be polar enough to have a large equilibrium distribution constant for MDA, it should have low solubility in aqueous, good extraction capability of the compound, low volatility and good chromatographic behavior. Based on these facts, six solvents including methyl isobutyl ketone, o- xylene, n- pentanol, n-octanol, n- heptanol and benzyl alcohol were selected. The results are shown in Figure 1. The best extraction efficiency was achieved when benzyl alcohol was utilized as an extraction solvent. This solvent is suitable because benzyl alcohol is a polar, protic solvent with relatively high dielectric constant which can easily elute highly polar MDA. Thus, in the present study this solvent was used for all extractions.

Figure 1 Effect of dispersive solvent on the extraction efficiency. Conditions: extraction solvent volume, 150 μL; dispersive solvent, acetic acid; dispersive solvent volume, 450 μL; vortex time, 60 s; centrifuge time, 4 min; sample volume, 4 mL; pH, 3. 

Type of the dispersive solvent

Methanol, ethanol, acetone, acetonitrile, and acetic acid were selected and examined as dispersive solvent. Also, extraction without the dispersive solvent was evaluated. From the five dispersive solvents tested, the acetic acid showed the best chromatographic behavior; also the highest efficiency was achieved with this solvent. Therefore, subsequent experiments were accomplished by using it. In vortex assisted DLLME, micro volumes of a low-density organic solvent are dispersed into an aqueous sample using vortex mixing, a mild emulsification procedure. The well micro droplets formed ensure fast partitioning rates, i.e. short equilibration times, due to the shorter diffusion distance and larger specific surface area.

Response surface methodology for optimization of the extraction of MDA

Response surface methodology (RSM) is an affordable and reliable technique to optimizing certain processes. This method leads to a reduction of designed experiments to study the effect of operation factors. To study the parameters affecting the pre-concentration of MDA, the seven factors in five-levels were used for design of experiments (DOE) (Table 1). The input variables were benzyl alcohol volume (A) (50- 200 μL), acetic acid volume (B) (150-450), vortex time (C) (20- 100 s), centrifugation time (D) (2- 10 min), volume of sample solution (E) (3- 7 mL), amounts of NaCl (F) (10- 40 g L^-1) and pH (G) (2-6). The factor levels were coded as −2 (-α, low), −1, 0 (central point), +1, +2 (+α, high).

Table 2 indicates the analyses of the experimental results of the second- order polynomial model for the MDA yield. The behavior of the system is described using the following quadratic equation.

Where Y is the natural logarithm of predicted response (Y=Ln(Response)), β₀ is the constant, X₁, X₂, …, X_k are the coded independent variables, β_i is the linear effect, β_ii is the quadratic effect, β_ij demonstrates the coefficient of the interaction factor, e is the random error or allows for description or uncertainties between predicted and determined value [²⁸].

The most commonly selected technique in the RSM method is central composite design (CCD). Design Expert software was applied for RSM regression analysis and optimization of input parameters by observed response (relative peak area). The statistical testing of the model, which inclusive linear, quadratic and interaction coefficient, was carried out with ANOVA analysis by F-test to achieve the empirical correlation between input and output parameters. To evaluate the goodness of fit of the model, each term of model was tested statistically which confirmed the significance of F- values by p ≤ 0.05. The values of R², adjusted R² and predicted R², lack of fit and adequate precision of models were obtained to check the quality of the proposed polynomial. To visualize the input-output relationships, the response surface plot and contour plot were drawn.

The quadratic model in terms of actual value variables is indicated in equation (2).

The obtained values for the critical points are as follows: benzyl alcohol volume = 100 μL, acetic acid volume = 251 μL, vortex time = 79.85 s, centrifugation time 4.02 min, volume of sample solution= 5.98 mL, amounts of NaCl= 10 g L^-1 and pH= 3.03.

Using the sum of squares (SS) values of the corresponding term (Eq. 3), the percent contribution (PC %) of each of the individual term in final model were computed (Table 3) [²⁹].

The results of the response surface model fitting in the form of analysis of variance (ANOVA) are showed in Table 3. The ANOVA results determine which the model was significant, as evident from the Fisher's F _test (F_model = 4282.818) by a very low probability value (p_model= < 0.0001).

Design- Expert software generated two dimensional response surface plots indicated in Figure 2. With the correlation coefficient (R²) the goodness of the model was checked. The adjusted R² value of 0.999728 indicated which only 0.0272 % of the total variation was not described using the model. Hence, the graph in Figure 3 shows a good correlation the value of between the experimental and predicted data of the response (R² =0.999). The lack-of-fit determine the failure of the model to represent value in the experimental domain at point that is not included in the regression [²⁹]. The not significant data of lack-of-fit (>0.05) revealed which the quadratic model is statistically significant for the response.

Figure 2 Response surface-2D/ contours showing the effect of independent variable on the relative peak area of MDA. 

Figure 3 Correlation plots between predicted and experimental values. 

Analytical performance for determination of MDA by VADLLME

Analysis of human blood plasma

To evaluate the applicability of the VADLLME-GC method, it was applied to the analysis of blood plasma sample of humans without any pretreatment. In the first step, 250 μL of blood plasma was transferred into 100 μL of 6 M NaOH, and then heated in a hot water bath at 60 °C for 30 minutes. After cooling to 2 °C, 500 μL of 20% w/v trichloroacetic acid was added. The mixture was vortexed for 1 min and the sample was left for 10 min at ambient temperature for completion of the reaction. The mixture was then centrifuged at 10,000 rpm for 10 min at 4 °C. The resulting deproteinized supernatant solution was transferred into a 6 mL vial and diluted to volume with ultrapure water and then analyzed [³¹], in which MDA was detected at the concentration of 284.7 μg L^-1 (RSD=6.0%, n=3). The GC chromatogram obtained from VADLLME for MDA extraction without sample spiking is depicted in Figure 4. For validation of this analysis, the most common method of MDA analysis, which is spectrophotometry [³¹, ³²] was used. This gave the MDA concentration 297.3 μg L^-1. As can be seen, the measured value is in good agreement with the standard method of MDA analysis at 95% confidence interval.

In order to investigate the validity of the proposed method for different concentrations, blood plasma samples were also spiked at three levels of 5, 25 and 50 μg L^-1 with MDA. All the steps mentioned above were performed to remove proteins. Results are presented in Table 5. As can be seen, good recoveries between 95.8% to 106.0% were achieved which indicate that different levels of MDA can be measured successfully by using VADLLME-GC method. Relative standard deviation (n=3) better than 7.15% was obtained.

Figure 4 GC chromatogram obtained from VADLLME extraction. Extraction conditions: extraction solvent, benzyl alcohol; extraction solvent volume, 150 μL; dispersive solvent, acetic acid; dispersive solvent volume, 450 μL; vortex time, 60 s; centrifuge time, 4 min; sample volume, 6 mL; pH, 3; MDA = malondialdehyde; TMP = 1,1,3,3-tetramethoxypropane (internal standard). 

Table 5 Analysis of MDA in a spiked plasma sample by the VADLLME extraction. 

MDA added (µg L-1)	MDA found (µg L-1)	Recovery (%)	RSD (%), n=3
- 5	284.7 290.0	- 106.0	6.05 7.15
25	309.1	97.6	5.87
50	332.6	95.8	6.70