Metformin (Figure 1) is chemically 1,1-dimethylbiguanide hydrochloride, a biguanide antidiabetic appearing as white or almost white crystals with melting point of 222 – 226 ºC and content range 98.5 – 101.0 % dried substance, freely soluble in aqueous medium, slightly soluble in alcohol, and insoluble in acetone and in methylene chloride solvents. 1 Physicochemical features of metformin showed that the drug has pKa values of 2.8 and 11.5, appeared very largely as water loving cationic species at physiological pH values. The metformin pKa values renders metformin a stronger basic in nature than most other basic drugs with less than 0.01% of the drug as unionized form in blood. The lipid solubility of the unionized species of the drug is very little as shown by its low LogP value (Log10 of the partition coefficient of the unionized form between octanol and water) of -1.43. These parameters indicate low lipophilicity and, consequently, rapid passive movement of metformin through cell membranes is much unlikely. The LogP of metformin is less than that of the other biguanide phenformin (-0.84) because the di-methyl substituents on metformin makes it less lipophilic than the bulky phenyl ethyl side chain in phenformin. More derivatives of metformin are presently being investigated with the aim of producing prodrugs that are more lipophilic with better oral absorption. 2, 3

Some developed methods for determination of metformin in biological fluids have been reported in the literature. An HPLC method for determination of metformin in brain sections and plasma of rats treated with polysaccharides was reported to be developed and validated. They used ranitidine HCl as internal standard and pentafluorophenylpropyl column (HS F5, 150 × 4.6 mm id), a mobile phase consisting of ammonium acetate-acetonitrile in ratio (47.8:52.2), rate of flow 1.3 mL/min, injection volume 20 µL and detection λ_max 233 nm. The method successfully resolved metformin from IS with retention time (9.56 and 4.91 mins) and (12 and 6 mins) for metformin and internal standard in plasma and brain samples respectively. 4 Metformin was also reported to be quantified in HPL-MS/MS method for comparative pharmacokinetic study among 3 metformin formulations in healthy Mexican volunteers in a single-dose, randomized, open-label, 3-period crossover study. The method utilized C18 (4.6 × 100 mm id) column, a mobile phase consisting of acetonitrile-formic acid (40:60), column temperature of 21°C, flow rate 0.8 mL/min and retention time of metformin 1.8 min. 5 The concentrations of metformin in plasma were also determined using HPLC method in a pharmacokinetic interaction study between metformin and teneligliptin in healthy adults. The column used was (µBondapak Phenyl, 300 ×3.9 mm, 10 mm; Waters) isocratic elution. The employed mobile phase is ternary consisted of acetonitrile-6.7 mM potassium dihydrogen phosphate-triethylamine (500:1500:1) adjusted with phosphoric acid to pH 7.6. The analyte was detected using (SPD-6A; Shimadzu Corporation, Kyoto, Japan) spectrophotometer at λ_max of 234 nm.6 A RP-HPLC method for estimation of metformin in plasma was reported that utilized propranolol as IS, C18 (150 × 4.6 mm id) column at 40°C, flow rate 1 mL/min, binary mobile phase consisted of phosphate buffer pH 7.0-acetonitrile in ratio (50:50) and retention time of 7.5 and 9.5 minutes for metformin and IS respectively. The method was applied only for bioequivalence studies using 2 metformin products and suggests that, it should be used for pharmacokinetic study of metformin. 7 A HPLC method for concurrent determination of metformin and three ACE inhibitors (lisinopril, captopril and enalapril) was also reported. The method utilized caffeine as IS and C18 (250 × 4.6 mm id) column at temperature 25°C, pH adjusted to 3.0 with phosphoric acid and mobile phase composition of acetonitrile-water in ratio (50:50). The flow rate was 1 mL/min and detection λ 218 nm. The method LODs were 3.26 and 0.98 ng/mL and LOQs 9.26 and 2.22 ng mL^-1 for lisinopril and metformin respectively. The retention time of the analysed drugs was not mentioned and the displayed chromatogram makes it difficult to ascertain that because the chromatogram is not well labelled. The method was reported to be applied in the determination of the drugs in their finished product and suggested its application for pharmacokinetic studies of the drugs. 8 To the best of our literature search, there is no report on analytical method for the analysis of metformin in human saliva. This research is aimed at developing and validating an RP-HPLC method for the analysis of metformin in human saliva which may be utilized for pharmacokinetic and bioequivalence studies involving the drug in saliva. The non-invasive procedure for sample collection in this method presents a more convenient way to patients and ease of compliance to study as compared to blood collection using invasive techniques that are painful and may be harmful to the subject unless strict safety measures are followed.

The optimized chromatographic conditions of the developed RP-HPLC method are shown in Table 1 while the chromatograms obtained are presented in Figure 2, Figure 3 and Figure 4. Calibration parameters for RP-HPLC method are shown in table 2. Calibration curve of metformin in saliva is shown in Figure 5. The results of the validation parameters for the method are shown in Table 3.

The peak height ratios of metformin and internal standard against their corresponding concentrations were determined. The linear relationship between the peak height signals (A) and their corresponding concentrations (C in ng/mL) is given by the regression equation of the type A = Cy + x (Table 2). The coefficient of correlation (r)is approaching unity(0.9987). This shows a direct proportionality relationship between peak height signals and concentrations which give the good correlation. The solvent methanol-water (80:20) appears to be the best as it serves as solvent for both metformin and caffeine as well as utilized solvent.

For the optimization of chromatographic conditions, various shackles were systematically countered to get to optimum conditions. Mobile phase being the polar part of reverse phase HPLC system always has a profound effect on the separation of drug molecules which are mostly polar in nature. Due to this, the selection of mobile phase was critically considered. Different ratios of methanol-water were used to observe the separation between metformin and IS (caffeine). We achieved the best separation with the mobile phase composition methanol and water in the ratio 80:20 (v/v) without buffers. It was observed that detection wavelength for both metformin and IS responded well at a λmax of 232 nm when scanned in various mobile phases, thus it was selected as optimum wavelength. The peak parameters such as height, asymmetry and tailing were considered while maintaining flow rate, baseline drift etc. This analytical procedure involved the use of internal standard as there was extraction in of the drug after addition of the internal standard in human saliva to account for loses if any. The uniformity of the system operation throughout the analysis was checked by initially equilibrating the column with mobile phase prior to injection of the sample to be analysed into the chromatographic system. Theoretical plates, tailing factor, resolution and repeatability were checked prior to starting analytical work every time. All the factors were found satisfactory and according to the set guidelines. 9

The low % RSD (≤ 0.70) computed shows the precision of the method. With good technique and reliable methodology, the precision should be < 15 % CV.10 There was a reported % CV of 9.8 and 8.11 in HPLC method for determination of metformin in brain sections and plasma of rats 4. This shows that the precision of each of the developed methods is satisfactory. The accuracy (≤ 1.00) of this method expressed as the measure of percentage relative error are within the range (1 – 5 %) for moderately accurately procedure. 10 The average percentage recovery for the method was found to be 99.98 % showing that the method has good recovery especially when compared with the 98.20 % reported by Valentina P, et al.7 in an HPLC method for the analysis of metformin in human plasma. The LOD (≤ 0.22 ng/mL) and LOQ (≤ 0.67 ng/mL) for the developed method are satisfactory. This shows that the method is sensitive for the analysis of the drug inhuman saliva.

From the results obtained, it can be concluded that RP-HPLC method for the analysis of metformin in human saliva was developed and validated. All the validation parameters fall within the acceptable limit for analytical method; hence the proposed method can successfully be applied for quantitative estimation of metformin in pharmacokinetic, interaction and bioequivalence studies involving metformin.

The authors declare no conflicts of interest

The authors will also like to thank Malam Saidu Bala Riti the Technologist in charge of HPLC laboratory, Department of Pharmaceutical and Medicinal Chemistry, Ahmadu Bello University, Zaria, Nigeria for running all the samples.