ANALYST, FEBRUARY 1986, VOL. 111 167 Determination of Vitamin C in Urine by Flow Injection Analysis* Fernando Lazaro, Angel Rios, M. D. Luque de Castro and Miguel Valcarcel Department of Analytical Chemistry, Faculty of Sciences, University of Cordoba, Cordoba, Spain A spectrophotometric method with flow injection analysis is described for the determination of vitamin C in urine, which presents substantial improvements over the existing conventional manual and automatic methods. Determinations can be carried out at the pg ml-1 level; the method has an average recovery error of +2.5% and a sampling frequency of 90 samples h-1 can be attained. Keywords: Vitamin C determination; flow injection analysis; urine The determination of vitamin C in its three forms (ascorbic acid, AA; dehydroascorbic acid, DHAA; and 2,3-diketogulonic acid, DKGA) has been carried out in numerous ways.l The most frequent method is based on the titration of ascorbic acid with 2,6-dichlorophenolindophenol, 2,6-DCPIP2J; nevertheless, this method is only useful for samples with small amounts of foreign species as it is subject to numerous interferences from reducing species such as sulphur dioxide, tannins, cysteine , sulphydryl compounds, certain metal ions, plant pigments, reductants and similar compounds.For this reason, depending on the sample matrix, this method and others that eliminate possible interferents have been used so far. For the determination of vitamin C in urine and blood the method most commonly used is that of Roe and K ~ e t h e r , ~ which is based on the formation of coloured osazones in a concentrated sulphuric acid medium, via the reaction between 2,4-dinitrophenylhydrazine and 2,3-diketogulonic acid, an oxidised form of ascorbic acid.Nevertheless, in spite of overcoming most interferences this procedure does have several disadvantages: it is complex and time consuming (over 180 min per analysis are required); and it is subject to other interferents such as hexoses, pentoses, glucuronic acid, reductants, hystidine and several other amino acids, although most of them do not interfere at the normal level in which they are normally found in urine and blood. To improve this method several modifications have been suggested: the use of different oxidising agents such as b r ~ m i n e , ~ 2,6-DCPIP6 and benzoquinone7; different reaction conditionss; the use of different acids6; or the use of different separative techniques.5 Automatic methods using reagents such as 2,6-DCPIP,9 o-phenylenediamine (DPD)10 and 2,4-dinitrophenol (2,4-DNP)llJ2 have also been developed with the aid of Technicon AutoAnalyzers.These methods are the most similar to that proposed in this paper, in which the chloramine T - KI - starch reaction and flow injection analysis (FIA) are used jointly. This latter development is an easily automatable technique,l3 which, in addition, is inexpensive. These facets, together with the absence of interferents in the method, make this a good alternative to routine vitamin C analysis. Experimental All reagents and apparatus used in this work were the same as described previously,l4 except for a Metrohm Dosimat E535 automatic burette.The FIA manifold employed has also been described previously.14 Procedure Urine samples were obtained from individuals who had been given a pharmaceutical compound containing 0.5 g of vitamin * To whom correspondence should be addressed. C. Samples were collected in a polyethylene flask to which a final concentration of 500 mg 1-1 of oxalic acid had been added. The urine samples were subsequently diluted 1 + 5 with 0.9 M H2SO4 and pumped to their confluence with a KI - starch stream until they filled the loop of the injection valve, which was inserted into the chloramine T stream. The concentrations of vitamin C in urine shown in Tables 1 and 3 were obtained with diluted samples.Results and Discussion Firstly the two methods proposed in reference 14 were applied to the determination of vitamin C in urine; however, the FIA titration method (in which KBr - methyl red solution is used as the indicator) is subject to major interferences owing to its low selectivity. Urine samples with a concentration of vitamin C of 60 pg ml-1, as determined by the normal FIA method, provided concentrations over 6000 pg ml-1 with the FIA titration technique. This result was predictable as sulphydryl compounds interfere at the same analyte level because of the strong oxidising character of Br2, a product yielded in the indicator reaction (chloramine T - KBr) which acts as an oxidant for vitamin C. Conversely, the normal FIA method, which uses KI - starch solution as the indicator, provided good results and was therefore chosen for the application of the method to real samples.Stability of the Samples The study was performed on urine samples diluted 1 + 5 with 0.9 M &SO4. Oxalic acid, a reducing agent, was used as a preservative.15 Concentrations between 0 and 2000 pg ml-l were used in this study. For concentrations of oxalic acid equal to or higher than 500 pg ml-1, the analytical signal yielded by the sample remained constant for at least 24 h, whilst in the absence of a preservative a decrease in the signal of 10% was observed during the same time interval. A concentration of 500 pg ml-1 was therefore chosen for subsequent experiments. The reproducibility of the method was studied at two analyte concentrations (30.30 and 65.70 pg ml-1) with nine different samples in each instance and at two different times (1.0 and 3.5 h). The values obtained for the relative standard deviation (r.s.d.) were as follows. After 1 h: 30.30 pg ml-l (r.s.d.= +0.93%); 65.70 pg ml-1 (r.s.d. = +0.32%). After 3.5 h: 30.30 vg ml-1 (r.s.d. = +0.900/,); 65.70 pg ml-1 (r.s.d. = +0.30%). The precision was good in both instances but was improved by increasing the analyte concentration. For the study of the recovery of vitamin C in urine, six samples were taken at different times from an individual who had taken 0.5 g of vitamin C . Concentrations of 20,40 and 80 pg ml-l of vitamin C were added to each sample. The results obtained are listed in Table 1; a good recovery was observed (the average recovery = 99.6% with an average error of 2.24%). These data indicate the absence of interferents in this168 ANALYST, FEBRUARY 1986, VOL.111 Precision 1.3 , i Table 1. Results for the recovery of vitamin C in urine 10 Vitamin C addedlpg ml-1 Vitamin C found in urine 20.00 40.00 80.00 Urine diluted 1 + 5/ sample pg ml-1 Found/pg ml-1 Recovery, Yo Found/pg ml-1 Recovery, YO Found/pg ml-1 Recovery, YO 1 20.97 40.06 95.5 60.00 97.6 102.00 102.0 2 29.03 49.68 103.2 69.68 101.6 109.02 100.0 3 36.13 56.13 100.0 74.84 96.8 115 .OO 98.6 4 39.35 59.68 101.6 78.71 98.4 119.50 100.2 5 41.61 61.95 101.6 80.65 97.6 119.50 97.4 6 59.35 78.71 96.8 102.00 106.6 137.50 97.7 Table 2. Comparison between the FIA and conventional titration methods for the determination in vitamin C in synthetic samples FIA method Titration method Error of FIA Amount of method relative vitamin C Amount Relative Amount Relative to titration addedlpg ml-1 found/pg ml-1 error, YO found/yg ml-1 error, '/O method, '/O 20.00 40.00 50.00 60.00 70.00 80.00 100.00 120.00 140.00 Mean error, YO 19.04 43.79 49.56 56.35 68.64 79.78 106.34 120.53 135.98 -4.8 9.5 -0.9 -6.1 -1.9 -0.3 6.3 0.4 -2.9 3.7 21.54 41.57 50.73 62.70 73.97 85.25 106.38 126.11 145.13 5.7 3.9 1.5 4.5 5.7 6.5 6.4 5.1 3.6 4.8 -11.6 5.3 -2.3 - 10.1 -7.2 -6.4 0.0 -4.4 -6.3 4.6 Table 3.Comparison between the FIA and conventional titration methods for the determination of vitamin C in urine samples (diluted 1 + 5 ) and containing 9.26 pg ml-1 of vitamin C Amount of vitamin C addedlyg ml - 1 10.00 30.00 40.00 50.00 60.00 70.00 90.00 110.00 130.00 Mean error, Yo FIA method Titration method Amount found/pg ml- 21.07 39.26 47.85 59.88 71.01 78.60 102.69 119.43 136.29 Relative error, Yo 9.4 0.0 -2.9 1.0 2.5 -0.8 3.5 0.1 -2.2 2.5 Amount found/pg ml-1 21.84 41.57 51.43 61.29 69.57 79.61 95.81 114.34 134.56 Relative error, Yo 13.4 5.9 4.4 3.4 0.4 0.4 -3.5 -3.7 -3.4 4.3 Error of FIA method relative to titration method, YO -3.5 -5.5 -6.7 -2.3 2.1 -1.3 7.2 4.0 1.3 3.8 method, which is attributable to the weak oxidising character of the I2 formed in the reaction.Comparison of Methods The FIA method used in this work was compared with the conventional titration technique16 (Fig. 1) for synthetic samples and urine samples. In addition, both methods were compared with the determination of vitamin C by direct weighing.Analysis of synthetic samples On analysing nine different samples by the two methods average errors of 3.7 and 4.8% were obtained by the FIA and conventional titration methods, respectively, both rela- tive to direct weighing (Table 2). In the conventional titration procedure the errors were always positive, possibly owing to the difficulty in determining the end-point, as it is not based on a colour change, but in the persistence of the colour. Moreover, the kinetics of the reaction are very slow in the vicinity of the equivalence point, so that it is necessary to wait for a certain time after each addition of titrant before making the measurements. This shortcoming is not present in the FIA I Sensitivity - 600 Decreased sample volume Rapidity Decreased IT1 consumption reagent 1 1 Fig.1. Improvements of the suggested procedure over the conven- tional titration procedure method, which affords a sampling frequency of 90 samples h-1 compared with 6 samples h-1 in the conventional titration. In addition, the titration technique requires the use of anANALYST, FEBRUARY 1986, VOL. 111 169 automatic burette (reading to 0.01 ml) owing to the small reagent volume needed. Analysis of urine samples The same study performed on urine samples with final concentrations of 9.26 pg ml-1 after a 1 + 5 dilution (Table 3) yielded average errors of 2.7 and 4.3% for the FIA and conventional titration methods, respectively, both relative to direct weighing. These errors are slightly smaller than those obtained for synthetic samples.The average error of the FIA method relative to the conventional method is 3.8% in this instance. These results reveal a smaller error with the proposed method compared with the conventional method, for both synthetic and urine samples. When the proposed method is compared with automatic methods such as that of Pelletier and Brassard12 it can be concluded that the proposed method has several advantages over the latter: although the latter technique offers higher sensitivity and similar precision and recovery, the working scheme is very complex (13 channels are necessary as opposed to 3 in the FIA method); maintenance is expensive and time consuming (it requires daily washing with 75 ml of HN03 and distilled water for 40 min); the flow cell must be washed weekly with dichromate solution and the tubing system must be changed after 120 determinations, whilst the FIA manifold only requires washing for 5 min with distilled water after a working day and allows up to 10000 determinations to be performed (90 determinations h-1, 8 h a day, for 14 d) before changing the tubing system.In addition, the maximum sampling rate achievable is 13 samples h-1 (with the use of a sampler) compared with 90 samples h-1 for the suggested method. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Cooke, J. R., and Moxon, R. E. D., in Counsell, J. N., and Hornig, D. H., Editors, “Vitamin C,” Applied Science, Barking, 1981, pp. 167-198. Harris, L. J., and Ray, S. N., Biochem. J . , 1933, 27, 303. Birch, T. W., Biochem. J . , 1933, 27,590. Roe, J. H., and Kuether, C. A., J. Biol. Chem., 1943,147,399. Zloch, Z., and Ginter, E., 2. Klin. Chem., 1970, 8, 302. Pelletier, O., J. Lab. Clin. Med., 1968, 72, 674. Saari, J. C., Anal. Biochem., 1966, 15, 537. Roe, J. H., J. Biol. Chem., 1961, 236, 1611. Egberg, D. C., J. Sci. Food Agric., 1973, 24, 789. Dunmire, D. L., J. Assoc. Off. Anal. Chem., 1979, 62, 64F. Aeschbacker, H. U., and Brown, R. G., Clin. Chem., 1972,18, 965. Pelletier, O., and Brassard, R., Adv. Autom. Anal. Technicon Znt. Congr., 1973, 9, 73. VBlcarcel, M., and Luque de Castro, M. D., “Flow Injection Analysis: Principles and Applications,” Ellis Horwood, Chichester, in the press. LBzaro, F., Rios, A., Luque de Castro, M. D., and VBlcarcel, M., Analyst, 1986, 111, 163. Haddad, P., J. Chem. Educ., 1977, 54, 192. Verma, K. K., and Gulati, A. K., Anal. Chem., 1980,55, 1045. Paper A51244 Received July 8th, 1984 Accepted August 28th, 1985