Development and validation of HPLC/UV-procedures for quantification of metronidazole in blood and urine

Metronidazole belongs to the group of antiprotozoal medicines and widely used for the treatment of infectious diseases; the medicine has a number of side effects manifested by usual symptoms of acute intoxication, especially when interacting with other drugs and alcohol. Aim. To apply the system of MiLiChrome® A-02 HPLC-analyzer widely used in the Ukrainian laboratories of forensic toxicology for the metronidazole quantitative determination in biological fluids and carry out validation of the procedures developed. Materials and methods. The sample preparation of blood and urine was carried out by extraction with acetonitrile and 2-propanol followed by separation of the organic layer under the conditions of the aqueous phase saturation with ammonium sulfate. Previously blood and urine were treated with acids. Isolation was carried out in the strong acid, neutral and weak alkaline medium. Results and discussion. To find the optimal conditions of the sample preparation such validation parameters as specificity/selectivity and recovery were determined. The results of the blank samples analysis were acceptable for all variants of the sample preparation procedures. Recovery values were reproducible for all procedures of analysis studied, but efficacy of metronidazole isolation was variable – from 85 % to 97 %. The results of verification of metronidazole stability showed the necessity to carry out all measurements within 12 hours after obtaining the solutions to be analyzed. The results of determination of linearity, accuracy and precision were the evidence of acceptable systematic and random errors of the HPLC/UV-procedures studied in the variant of the method of calibration curve, method of standard and method of additions. Conclusions. The set of HPLC/UV-procedures for the metronidazole quantitative determination in blood and urine has been developed. Validation of the procedures developed has been carried out.

At the Analytical Chemistry Department of the National University of Pharmacy (Kharkiv, Ukraine) the research in the field of creating the unified standardized validation procedures for methods of the analyte quantification in different biological matrices for application in forensic toxicology is carried out [1][2][3][4][5][6][7][8][9]. The main purposes of such unified standardized validation procedures are to provide development of the method with the optimal pre-specified parameters and acceptance of the method as legitimate.
As a part of the research mentioned the experiment for development of the method for metronidazole quantification in blood and urine using HPLC was performed.
Metronidazole belongs to the group of antiprotozoal medicines and is widely used for the treatment of infectious diseases caused by Trichomonas, Lamblia, Leishmania, etc., as well as in the treatment of peptic ulcer associated with Helicobacter pylori [10][11][12]. The medicine has a number of side effects manifested by usual symptoms of acute intoxication (giddiness, nausea, vomiting), especially when interacting with other drugs and alcohol [13][14]. And the case of interaction with alcohol may be lethal for a patient even when the therapeutic dose is taken [15]. According to unpublished data metronidazole is identified sometimes in human biological matrices in toxicological examinations in Ukraine.
Chemically, metronidazole is 2-methyl-5-nitroimidazole-1-ethanol and has the structural formula as shown in Fig. HPLC is used to analyze metronidazole in pharmaceuticals and biological fluids rather widely [16][17][18]. The main disadvantage of the present procedures is their application exclusively for metronidazole quantification; both chromatographic conditions and sample preparations are specially chosen to analyze metronidazole. It is usual situation for pharmacokinetic studies, but in forensic toxicology it is impossible to use individual procedures for each analyte, it is necessary to use unified methods of sample preparation and unified screening chromatographic conditions, so-called HPLC-analyzer system.
Thus, this research was conducted to develop some HPLC-procedures of metronidazole quantification in blood and urine using different traditional methods of sample preparation [19][20] and approved in Ukraine for application in forensic toxicology. The HPLC-analysis was carried out using the system of MiLiChrome® A-02 HPLC-analyzer [21] implemented in practice of forensic medical laboratories in Russia and Ukraine. The step-by-step validation of the procedures developed was performed according to the approaches offered by us [1][2][3][4][5][6][7][8][9] to choose the optimal variants of sample preparation provided high accuracy and precision, enough sensitivity and specificity, etc. Another aim of our experiment was to accumulate the experience of application of the standardized validation procedures offered for the method development.

Calibration and model samples
Blank-samples were prepared by spiking 5 samples (10.00 mL) of the corresponding matrix (blood or urine) obtained from different sources with 1.00 mL of distilled water.
The method of calibration curve (MCC) and the method of standard (MS) -Scheme 1 The stock solutions 1 and 2 (2500 μg/mL) were prepared by dissolving 250.0 mg of metronidazole in distilled water, and the solutions were diluted to 100.0 mL with the same solvent. The stock solutions 1 and 2 were diluted with distilled water to prepare: • the process solutions 1-7 having the concentrations of 50; 100; 150; 200; 250; 300; 350 μg/mL, respectively; • the process solutions 8-11 having the concentrations of 50; 100; 200 and 350 μg/mL, respectively. Three batches (in 11 samples each) of the corresponding matrix (blood or urine) obtained from three different sources were used to prepare the cali-bration samples 1-7 and the model samples 8-11 by spiking 10.00 mL of the matrix with 1 mL of the process solutions 1-11, respectively.
After spiking all samples were vortexed for 1 hour and stored for 24 hours at ambient temperature before the sample processing.
The method of additions (MA) -Scheme 2 The stock solution 3 (2500 μg/mL) was prepared by dissolving 250.0 mg of metronidazole in distilled water, and the solution was diluted to 100.0 mL with the same solvent. The of addition solution 1 (150 μg/mL) was prepared by dissolving 75.0 mg of metronidazole in distilled water, and the solution was diluted to 500.0 mL with the same solvent. The stock solution 3 was diluted with distilled water to prepare the process solutions 12-17 having the concentrations of 125; 125; 250; 375; 500; 500 μg/mL, respectively.
Three batches (in 6 samples each) of the corresponding matrix (blood or urine) obtained from three different sources were used to prepare the model samples 12.1-17.1 by spiking 25.00 mL of the matrix with 1.00 ml of the process solutions 12-17, respectively. After spiking all samples were vortexed for 1 hour and stored for 24 hours at ambient temperature before the sample processing.
To prepare the model samples 12.2-17.2 10.00 mL of the model samples 12.1-17.1 were spiked with 1.00 mL of the solution of addition 1 directly before the sample processing.

Reference and model solutions
The stock solutions 4 and 5 (100 μg/mL) were prepared by dissolving 50.0 mg of metronidazole in 0.01 M hydrochloric acid solution, and the solutions were diluted to 500.0 mL with the same solvent. The reference solution (8 μg/mL) was prepared by diluting 8.00 mL of the stock solution 4 to 100.0 mL with 0.01 M hydrochloric acid solution. The stock solution 5 was diluted with 0.01 M hydrochloric acid solution to prepare the model solutions 1-4 having the concentrations of 2; 4; 8 and 14 μg/mL, respectively.
Blood and urine sample preparation for the metronidazole determination -Scheme 3 Blood: 10.00 mL of blood was diluted with 20.00 mL of distilled water and processed with 10.00 mL of 10 % trichloroacetic acid aqueous solution. The mixture was vortexed for 1 hour, then centrifuged for 5 minutes at 5000 rpm.
Urine: 10.00 mL of urine was processed with 10.00 mL of 0.1 M hydrochloric acid solution and then acidified with 6 M hydrochloric acid solution to pH ≤ 2.
The next stages were the same for blood and urine. Procedure 1-1: The supernatant was processed twice with 10.00 mL of acetonitrile and vortexed for 1 hour each time. After adding 2 g of ammonium sulfate the mixture was filtered through the paper filter (wetted with acetonitrile) into a separating funnel and salted-out by adding ammonium sulfate till its dissolution stops. The top organic layer was separated, filtered through the paper filter with 1 g of anhydrous sodium sulfate into a 25.0 mL measuring flask, and diluted to the volume with acetonitrile.
Procedure 1-2: The supernatant was neutralized with 25 % ammonium hydroxide solution to pH = 7. The next stages were as for Procedure 1-1.
TLC-purification: 10.00 mL of the organic extract obtained were evaporated at 80 °С to complete removal of the organic layer; a dry residue was dissolved in ~0.5 mL of chloroform and applied quantitatively on the start line of the chromatographic plate in the form of a band of 2 cm in width. 10 μL of metronidazole standard ethanol solution (1 mg/mL) were applied in the point ("testifier") near the band. The plate was eluted in chloroform twice and then dried out using the mixture of chloroform and methanol (90 : 10) as a mobile phase; the "testifier" band was developed in UV-light, and the spot of brown color in the area of R f = 0.35-0.55 was observed. The sorbent was carefully removed from the plate part with the area of 3 × 1 сm at the level of the "testifier" into a glass bottle with 10.00 mL of 0.01 M hydrochloric acid solution, the bottle content was vortexed for 5 min and filtered through the paper filter wetted with 0.01 M hydrochloric acid solution (eluate). Method valida-tion -Schemes 4 and 5 The complete validation of the method developed was carried out using matrix (calibration and model) samples [1][2][3][4][5][6][7][8][9].
Stability In process stability of metronidazole was verified in the way of chromatographing the eluate obtained for the model sample 10 -immediately and in 1, 12, 24, 36 and 48 hours after its preparation, and the systematic error δ stability was calculated [9] and assessed.
Specificity/selectivity Blank-samples prepared using 5 different sources of blood or urine were analyzed; the summarized peaks areas within the range of t R ± 0.5 min from the corresponding chromatograms were compared with the mean peak area of metronidazole from the chromatograms of the model samples 8 and 9. The value of δ blank was calculated [9] to confirm the specificity/selectivity of the procedures of the analysis developed.
Recovery Recovery R was determined at the levels of low, medium and high concentrations in the way of ana- lyzing the model samples 8-11 and comparing their peaks areas with the peaks areas for the model solutions 1-4, respectively. Reproducibility and significance of the recovery values were assessed [9]. The experiment described was carried out within at least 3 runs/days following the requirements to repeatability of peaks areas for replicate experiments [9]. Linearity/calibration model To assess linearity/calibration model the calibration samples 1-7 were analyzed within 3 runs/days following the requirements to repeatability of peaks areas for replicate experiments [9]. The correlation coefficient, rest standard deviation and absolute term for linear dependences were calculated [3,4,9] as within-run (R k c , RSD k 0 , a k ) and between-run (R c , RSD 0 , a) parameters, and then compared with the corresponding acceptability criteria.
Accuracy and precision MCC. The concentrations of calibration samples 1-7 and model samples 8-11 were recalculated using the corresponding within-run linear dependences or between-run linear dependence, and the values "found/ given" RR i were used to determine the confidence intervals (within-run precision), the total confidence intervals , (between-run precision), the systematic errors δ k (within-run accuracy), and the total systematic errors δ intra , δ sample (between-run accuracy) [5,6,9]. MS. The ratios Z i for the calibration samples 1-7 were calculated and used to determine the confidence intervals Δ k Z (within-run precision), the total confidence interval (between-run precision), the systematic errors δ k (within-run accuracy) and the total systematic error δ intra (between-run accuracy) [7,9]. MA. The model samples 12.1-17.1 and 12.2-17.2 were analyzed within 3 runs/days. The concentrations of the model samples 12.1-17.1 were recalculated, and the values "found/given" were used to determine the confidence intervals (withinrun precision), the total confidence interval (between-run precision), the systematic errors δ MA,k (within-run accuracy), and the total systematic error δ intra MA (between-run accuracy) [8,9].
The values of confidence intervals and systematic errors were compared with the corresponding acceptability criteria.

Limit of quantification (LOQ)
The lowest point on the calibration curve was accepted as LOQ [9].

Results and discussion
The HPLC/UV-method for metronidazole determination was proposed by authors before [22] and its specificity in relation to other 5-nitroimidazoles was shown. The suitability of the abovementioned analytical procedure for further work with biological fluids was assessed using the validation procedure by model solutions [22].
It was suggested to carry out metronidazole isolation from blood and urine by amphiphylic solvent extraction with the subsequent separation of the organic layer under the conditions of the aqueous phase saturation with an electrolyte; ammonium sulfate was used as an electrolyte.
Previously, blood and urine were processed with the corresponding acids (10 % trichloroacetic acid solution for blood and 0.1 M hydrochloric acid solution for urine). This way for processing biological fluids is accepted in the Ukrainian forensic and toxicological laboratories for the general analysis. Our modification of these sample preparation procedures is dilution of blood with water in 3 times before processing with 10 % trichloroacetic acid solution -to reduce the analyte co-precipitation due to decrease of the contact area between the analyte and blood cells.
To choose the optimal isolation conditions such amphiphylic solvents as 2-propanol and acetonitrile were used in the experiment. Owing to metronidazole amphoteric properties and proceeding from our results [23] isolation was carried out in the strong acid (pH = 2), neutral (рН = 7) and weak alkaline (pH = 9) medium; carrying out isolation of analytes from biolo-gical objects in the weak acid, neutral or weak alkaline medium (instead of the strong acid or alkaline medium) resulted in decreasing of co-extraction processes of the biological matrix components in a number of cases [24]. It is necessary to note that the shift of the рН real value in alkaline side was observed for the mixtures of electrolytes saturated solutions with amphiphylic solvents [25].
Thus, 6 sample preparation procedures were studied. To find the optimal conditions of sample preparation we determined such validation parameters as specificity/selectivity and recovery according to Scheme 4.

Method validation
The validation provides application of the normalized coordinates [26]: i. e. transition from the equation A i (S i ) = b 1 · C i + a 1 to the equation Y i = b 2 · X i + a 2 ; it allows to calculate the validation characteristics, which do not depend on the analyte and specific character of the method of analysis.
The analytical range D of the methods application is 25-175 % [9]; as 100 % the mean therapeutic metronidazole concentration in blood [19] is taken; the number of concentration levels g equals 7 in constant increments of 25 % [9].
Acceptability criteria for validation parameters were formed on the basis of systematic application of the "insignificance concept" [26] -the confidence interval Δ 2 was insignificant compared with the confidence interval Δ 1 at the conventional level p = 95 %, if the following inequality was correct: Δ 2 ≤ 0.32 · Δ 1 , and proceeding from the value of extreme uncertainty Δ As for the method in analytical toxicology, which equaled 25 % and 20 % [19,20] -for the lowest point of the analytical range of the method application and for the rest of range.
In the MCC the acceptability criteria for the linear dependence and precision were found proceeding from the equality of uncertainty of plotting the calibration curve Δ cal and uncertainty of the analysis of the sample to be analyzed Δ sample [26], whence it was as follows: (2) (3) The method of validation by matrix samples consists of two phases [9]: • the preliminary phase -determination and estimation of in process stability of the analyte in the solution to be analyzed, specificity/ selectivity and recovery for the procedure; • the main phase -determination and estimation of linearity, accuracy, precision and determination of LOQ for the procedure. This method also contains the total uncertainty assessment.
For normalization of the experimental data obtained the same reference solution with the analyte concentration of was used, but its peak area was corrected taking into account the value of recovery R (its significance and value were shown at the preliminary stage of validation).

Stability
The results of verification of in process stability of metronidazole in the solution to be analyzed showed the necessity to carry out all measurements within 12 hours after obtaining the solutions to be analyzed; in 12 hours the systematic error was high enough, but within the acceptability criteria; in 24 hours the systematic error increased significantly.
These data were taken into account when determining all validation parameters and should be used in the sequel.

Specificity/selectivity and recovery
The results of analysis of blank-samples and the assessment of systematic error caused by matrix influence were acceptable for all variants of the sample preparation procedures -we fixed the absence of peaks with the retention time, which was coincident with (or near to) the metronidazole retention time, on the chromatograms of blank-samples.
Subjectively, the procedures 1-2 and 2-2 (proceeding with amphiphylic solvents at pH = 7) were characterized by the lowest level of co-extractive substances by the picture of TLC-purification, but the procedures 1-1 and 2-1 (proceeding with amphiphylic solvents in the strong acid medium) had the worst results.
Recovery values determined in the preliminary phase of validation were reproducible for all procedures of analysis studied. But efficacy of metronidazole isolation was variable -from 85 % (procedure 2-1) to 97 % (proceeding with acetonitrile at pH = 2 and 9 for urine).
The total results of recovery determination for all procedures studied are given in Tab. 1.
Thus, based on the complex results of the specificity/selectivity and recovery assessment the procedures of the sample preparation, which include acetonitrile or 2-propanol extraction in the weak alkaline (рН = 9) medium followed by separation of the organic layer under the conditions of the aqueous phase saturation with ammonium sulfate are optimal and recommended for application by us.

Linearity/calibration model, accuracy and precision
The results of determination of R c , RSD 0 , a were positive for all variants of the sample preparation procedures and the solvent used, as well as for MCC, MS and MA (Tab. 2).
The values of accuracy and precision in the variant of MCC, MS and MA for all procedures studied are presented in Tab. 3-5. The results of determination of accuracy and precision are the evidence of acceptable systematic and random errors of the HPLC/UV-procedures studied. Concerning MCC, MS and MA application -we observe the following tendency: MS -the best precision and the worst accuracy, MA -the best accuracy and the worst precision, but the difference is less distinct and clear; MCC is the optimal variant.

Limit of quantification (LOQ)
LOQ of the procedures developed is 5 µg/mL of metronidazole in biological sample.

Table 1
The results of recovery assessment for metronidazole determination procedures in blood and urine by the method of HPLC/UV  Table 2 The results of linearity verification for metronidazole determination procedures in blood and urine by the method of HPLC/UV  The results of accuracy and precision verification for metronidazole determination procedures in blood and urine by the method of HPLC/UV (extraction at pH = 2)  The results of accuracy and precision verification for metronidazole determination procedures in blood and urine by the method of HPLC/UV (extraction at pH = 7)  The results of accuracy and precision verification for metronidazole determination procedures in blood and urine by the method of HPLC/UV (extraction at pH = 9)

Total uncertainty
The results of the total uncertainty assessment show the acceptability of the procedures developed. The least values of the total uncertainty are fixed for MCC; for MS and MA they are at the same level.

Conclusions
1. The HPLC/UV-procedures of metronidazole quantitative determination in blood and urine using the standard sample preparation with application of amphiphylic solvent (acetonitrile and 2-propanol) for the analyte isolation at рН = 2, 7 and 9 with further separation of the organic layer under the conditions of the aqueous phase saturation by ammonium sulfate have been developed.
2. Validation of the procedures developed has been carried out using calibration and model samples by such parameters as stability, specificity calibration model, accuracy and precision using different analytical and standardization methods -MCC, MS and MA; and application of the validation scheme possibility offered by us before has been confirmed.
3. The HPLC/UV-procedures of metronidazole quantitative determination developed satisfy the acceptability criteria for all validation parameters. Carrying out the preliminary phase of validation allowed us to eliminate the insufficient sample preparation and avoid fulfillment of the main validation phase for these procedures.
Conflict of interests: authors have no conflict of interests to declare.