Evaluation of the Clinitek 200 urinary test-strip reader in the analysis of dog and rat urines in pre-clinical toxicology studies

Transcription

1 240 Laboratory Animals (1993) 27, Evaluation of the Clinitek 200 urinary test-strip reader in the analysis of dog and rat urines in pre-clinical toxicology studies A. PAQUIGNON, G. TRAN & J. P. PROVOST Department of Biochemistry, Clinical Chemistry Laboratory, Pfizer, and Centre de Recherches, BP 159, Amboise, France Summary The performance of a clinical urinary test-strip reader Clinitek 200 was evaluated for dog and rat urines, in the context of pre-clinical toxicology studies. No major discrepancies were found between data generated by visual estimation or automatic measurement. Analysis of spiked samples showed good agreement between actual concentrations and Clinitek 200 responses for ketone bodies and glucose although a lack of sensitivity was found for the latter. Results for proteins showed over- or underestimation in dog and rat urines respectively at low concentrations, and overestimation at high concentrations in both species. Reproducibility of responses was excellent for ketone bodies, glucose and proteins but was weaker for haemoglobin and bilirubin. High bilirubin concentrations were found to interfere with the haemoglobin reaction. The ph measurements were found to be accurate only around ph 7. Specific gravity measurements were unreliable. Overall, the Clinitek 200 as a screening tool proved sufficiently reliable in the measurement of all parameters tested, with the exception of specific gravity. Keywords: Laboratory animals; Urinalysis; Test strip; Automated analysis The clinical chemistry analyses performed in the course of pre-clinical toxicology studies, with Correspondence to: lp Provost, Department of Biochemistry, Clinical Chemistry Laboratory, Pfizer, Centre de Recherches, BP 159, Amboise, France. Received 27 April 1992; accepted 25 September 1992 novel drug entities, generally include the analysis of urine constil uents. These studies generate large numbers of samples; for example in the course of a carcinogenicity study using rats, over 500 samples of urine need to be analysed. To meet these demands, qualitative or semiquantitative examinations of urinary constituents with reagent strips are performed. When changes in colour are estimated visually, each reactive pad on such a test strip must be compared individually with a standard set of colours, and at various times after wetting the pad in order to optimize the conditions for each parameter. These analyses are, therefore, tedious and relatively timeconsuming, and because of the subjective nature of the readings they are of doubtful accuracy (Peele et al., 1977). The introduction of automatic reagent strip readers, which llse the reflectance principle, may overcome some of these drawbacks. Automated analysers increa;e sample throughput (300 reagent strips can be processed per hour), and, within the context of regulatory toxicology studies, the printout of data produced by these machines improves compliance with the requirements of Good Laboratory Practice. Finally, as these analysers are generally equipped with an RS 232 interface, data can be automatically transferred to a main frame computer for data processing and statistical analysis. However, these automatic strip readers and the accompanying reagent strips are intended for human use. Whilst a number of studies have been perform~:d to assess the reliability of these systems in a human clinical context (Peele et al.,,,1977; Grinstead et al., 1987), their suitability for use with dog and rat urine has to our knowledge yet to be investigated.

2 Evaluation of the Clinitek 200 urinary test-strip reader 241 In this study, we examined the use of the Clinitek 200* reader for the analysis of dog and rat urines. We have assessed the following points: recovery and within-run reproducibility of responses using spiked samples, comparison between visual and automatic reading for ketone bodies, glucose, proteins, haemoglobin and bilirubin, and we compared results with those obtained by direct measurement of ph and specific gravity. Materials and methods Animals and sampling conditions Urine samples were obtained from untreated male and female beagle dogs (Marshall Farms, New York, USA) and rats (Crl: COBS-VAF-CD (SO) BR, Charles River, France). No distinction was made between sexes in these experiments. Body weights ranged from 6 to 9 kg for dogs and from 200 to 600 g for rats. During the sampling period, the animals were placed in individual stainless steel metabolism cages; dogs were deprived of food and water, rats were deprived of food but had free access to tap water. Dogs were maintained at a room temperature of 19± 3 C and rats at 20 ± 2 C with a relative humidity of 60 ± The air was changed in the animal rooms 14 times per hour. Urine was collected overnight (for approximately 16h) in plastic containers without addition of a preservative, and analysed within 1h following collection. Measurements using the Clinitek 200 and test-strips Reagent strip analysis, using Multistix 10-50* reagent strips was performed according to the manufacturer's instructions. Visual estimations were performed by a single trained observer. Automatic readings were performed using a Clinitek 200 urine analyser; a reference strip, supplied by the manufacturer, with pads of known reflectance was used to calibrate the instrument. Printed non-numeric messages from Clinitek and Multistix are registered trademarks of Ames Co., Division of Miles Laboratories, Inc., Elkhart, USA. Clinitek 200 and their significance were: Neg (Negative), T (Traces), Sm (Small), M (Moderate), Lg (Large). Additionally, specific gravity was measured using a UR-l hand-refractometer (Atago, Japan), and the ph with a digital ph meter (Orion, USA). For ph measurements (using a ph meter) and for densitometry, urine samples were centrifuged (10 min at approximately 2000 rpm). Preparation of 'spiked' urines Spiked samples of urine were prepared by addition of the test analyte to urine samples which were negative for that parameter by reagent strip analysis. Ranges of final concentrations were as follows: ketone bodies, mg/i; glucose, O 5-20 g/i; proteins, 0'075-3 g/i; haemoglobin, O' mg/i; bilirubin, 4-64 mg/1. A zero concentration grade was assumed when no analyte was added. Materials used to 'spike' these urines were as follows: acetoacetic acid, a-d ( + ) glucose and dog and rat haemoglobin (Sigma Chemical Company, St Louis, USA); standard solutions of bovine serum albumin (Preciset) and human bilirubin (precilib) (Boehringer Mannheim France, Meylan, France). Results Comparison of Clinitek 200 and visual reading of reagent strips The data, summarized in Table 1, was generated using (unspiked) urine samples from control animals; it should be noted that we were not able to include glucose-positive urine. In both species, there was agreement between the automated (Clinitek 200) and visual estimates of analyte; however, the automated reading overlapped the visual reading by up to 3 sections of the reagentstrip scale. Analysis of spiked urine samples from dogs and rats using the Clinitek 200 We made 10replicate measurements with different urine samples to test the reproducibility of the reflectance readings. Data for the analysis of urine samples which had been spiked with

3 242 Paquignon, Tran & Provost Table 1. Comparison of Clinitek 200 and visual readings of reagent strips Ketone bodies (mg/l) Proteins (gil) Haemoglobin Dog Rat Visual reading Clinitek Visual reading C/initek N Neg T N Neg T / Neg I Neg ]8 T 5 4 I T II N Neg T 0,3 N Neg T 0 3 >3 Neg Neg T T ' I I 12 I 4 7 > ]8 N Neg T Sm M N Neg T Sm M Neg Neg T 4 I 1 2 T Sm 1 1 Sm II 11 M 4 I 3 M 2 2 Lg 1 1 Bilirubin N Neg 8m N Neg Sm Neg Neg Sm Sm 10 9 I N: Number of readings made. different concentrations of one of 5 analytes (ketone bodies, glucose, protein, haemoglobin, and bilirubin) are shown in Table 2. Ketone bodies. In both species, there was good agreement between the actual concentrations of this analyte and those determined with the teststrip reader. Glucose. Over the glucose concentration range used, spiked urine from dogs and rats returned similar results. When urine was spiked with O 5 gil of glucose, 9/10 samples were reported as not containing glucose by the Clinitek. Samples containing 1gil glucose gave results as 'trace', indicating that this concentration is probably the lowest level detected by the Clinitek reader. Proteins. In the analysis of dog urine, O' 075 gil protein gave a 'trace' response whereas that spiked with O' 15 gil protein gave a value of '0, 3 gil' protein; with rat urine spiked with 0,075 or 0 15 gil protein the Clinitek readings obtained were for a 'trace' amount. With samples from both species spiked with O' 3 gil protein level, there was good agreement between the added protein value and Clinitek readings. However, at higher concentrations, the Clinitek overestimated the protein concentration; all the samples spiked with O' 5 gil gave values of 'I gil' and those samples spiked with I gil gave values of 'greater than 3 gil'. Haemoglobin. All dog urine samples spiked with O' 3 mg/l of haemoglobin gave 'trace' values; in simil"rly spiked rat urine, 6/10 samples gave 'trace' values and 4/10 samples returned 'negative' values. At all other concentrations, there appeared to be some agreement between the added haemoglobin level and the Clinitek readings. However, for a given concentration, the readings from the Clinitek system were split between 2 succl~ssivescale readings; for example,

4 Evaluation of the Clinitek 200 urinary test-strip reader 243 Table 2. Automated analysis of spiked urine samples from dogs and rats Added value Dog Rat Ketone bodies (mg/i) Neg Tr Neg Tr I I 9 80 Glucose (g/i) Neg Tr >10 Neg Tr 2'5 5 >10 0,5 9 I 9 I 1 2'5 I , Proteins (g/i) Neg Tr 0,3 >3 Neg Tr 0'3 >3 0' ' , ' I I Haemoglobin (mg/i) Neg Tr Sm M Lg Neg Tr Sm M Lg 0, , I 9 Bilirubin (mg/i) Neg Sm M Lg Neg Sm M Lg * * * 9* 64 10* 10* Values for which an interference with the reactive pad for haemoglobin was found. with urine spiked with 60 mgll haemoglobin, bilirubin, 5110 and 1/10 samples gave negative 7/10 and 6/10 Clinitek readings indicated a values; however, in similarly spiked rat urine, the 'medium' haemoglobin value and the remaining Clinitek readings were for 'small' amounts samples were reported as 'large' for dog and rat present. At the other concentrations of bilirubin urines respectively. added, the agreement between the spiked bilirubin concentration and the Clinitek readings was Bilirubin. In dog urine spiked with 4 or 8 mg/l satisfactory for both species. A surprising

5 244 Paquignon, Tran & Provost ph-meier 9 Refractometry A ' 1.010~ 6 7 B Value from Clinilek B L -' ' ' ~' , : :025 1_030 Value from Clinitek Refractometry B ~--"----~- ' Value from Clinitek Fig. 1. Comparison between Clinitek 200 and ph-meter for dog (a) and rat urines (b) ::U~_~, Value fromclinilek Fig. 2. Comparison between Clinitek 200 and refractometry for dog (a) and rat urines (b). finding was that at bilirubin concentrations of 64 mg!l in dog urine and of 8 mg!l or greater in rat urine, the reagent strip was positive for haemoglobin (flagged figures) although no haemoglobin had been added to the urine. Comparison between the Clinitek 200 and direct methods of measuring ph and density Comparisons of ph values for canine and rat urine measured using the Clinitek 200 and a ph meter, are shown in Figs la and lb. With rat urine, measurements of ph with the Clinitek 200 gave values around ph 7 which were in agreement. Below ph 6, 5, the Clinitek system underestimated the ph, whilst for alkaline samples (ph 7, 5 to 9) it overestimated the ph. These errors in ph estimation were marked, e.g. at ph 7, 5, values from the Clinitek system could differ by a ph value of 1 5 from the value obtained with a ph meter. Similar trends to overestimate and underestimate the ph were observed for d.ogs. Data for specific gravity measurements using the Clinitek 200 versus refractometry are shown in Fig. 2. The refractometer measured specific gravity up to a value of in increments of O'005 whereas the reagent strip estimated density up to 1'030 in increments of 0,005. There was poor agreement between the 2 sets of measurements for rat and dog urines. In addition, data from the Clinitek system was heterogeneous, particularly for dog urine. Discussion Comparison of Clinitek 200 and visual readings using unspiked urine samples For urines collected during toxicology studies, there were no major discrepancies when data generated by visual examination of test-strips were compared with those generated by the

6 Evaluation of the Clinitek 200 urinary test-strip reader 245 Clinitek 200. However, it should be stressed that visual discrimination between two successive steps of colouring may be difficult, especially for 'border-line' responses. Also the Clinitek 200 may allow a greater discrimination by automatic measurements (and hence improve accuracy) than data generated by visual estimation of reagent strip reactions. Analysis of spiked urine samples from dogs and rats There was good agreement between added concentrations and Clinitek 200 readings for ketone bodies. For glucose, the correlation was acceptable over the range 1 to 20 gil; below 1 gil, the Clinitek 200 system is unlikely to detect glycosuria in either species. There were differences in protein measurements with dog and rat urine; with O' 15 g proteinll urine, the protein content was overestimated in dog urine but underestimated in rat urine. Conversely, at higher protein levels (~0'5 gil), protein content was over-estimated in urine from both species. Protein estimation using test-strips relates primarily to albumin (Honnadel & Koller, 1987), and by reaction with a ph indicator, tetrabromophenol blue. Since the protein composition of laboratory animal urine varies with age and sex (Alt et al., 1980), a potential source of error using these test-strips is a gross underestimation of (.\(2u-globulin, which is a major urinary protein in adult male rats (Borghoff et al., 1991). Moreover, highly alkaline urine can interfere with the buffering system of the protein reactive pad in test-strips and hence affect colour development. These pitfalls have previously been described for the analysis of rat urine using test-strips (Evans & Parsons, 1986). The data from this study show that the agreement between actual and measured values for bovine albumin in urine from either dogs or rats is poor at the concentrations used. Quantitative methods for analysis of urinary protein should be used whenever accurate monitoring of protein excretion is required. The reproducibility of 10 successive readings using spiked urine samples was excellent in most cases, with the exception of haemoglobin (dog and rat urine, all levels of concentration) and for bilirubin in dogs at the lowest tested (4 mg/l) level. However, as these parameters are estimated on a qualitative scale, it is possible that this arbitrary range corresponds to border-line levels of response for the Clinitek 200 system. At the same time (a small) electronic drift or differences in colour development could also account for these phenomena. It is important to point out that at a high concentration of bilirubin, positive interferences were seen on the haemoglobin reaction pads. This is a surprising finding since haemoglobin is detected by its peroxidase-like activity; an explanation for this interference may be the presence of small quantities of interfering substances in the bilirubin standards used in this study. The reasons for the species differences are unclear. ph and density Our results are in agreement with and extend the findings of Grinstead et 01. (1987) for ph measurement in human urine. That ph values in the range 6 to 7 were optimal with the Clinitek 200 system was expected from data generated with freshly obtained human urine (Schumann, 1984). However, since the ph of urine from dogs and rats often falls outside this range (for example, the 5 and 95% percentile values for rat urine in our laboratory are ph 6' 3 and 8, 3; n == 260), then the use of the Clinitek 200 system to monitor urinary ph accurately in these 2 species is questionable. Since the Clinitek 200 system reads ph values by increments of o 5 ph units within its working range, the accuracy does not compare with that of a ph meter which measures to the nearest 0'1 ph unit or less. However, we consider that when only an approximate estimation of ph is required (as in screening tests for example), then the Clinitek 200 system is sufficiently reliable. The total lack of agreement between the Clinitek 200 system and refractometry for measurement of specific gravity can be explained by the fact that the density test pads on reagent strips react with cationic substances, and are

7 246 Paquignon, Tran & Provost not sensitive, or only weakly so, to other urinary components such as proteins or organic acids. Our findings are in contrast with results obtained with human urine where there was good correlation between the two methods (Audoin et al., 1985). This difference is probably due to differing urinary composition in various species. Moreover, the narrow working range of I to for Clinitek 200 is unsuitable for high density urines obtained with dogs which were deprived of water during the sampling period. In conclusion, the Clinitek 200 is a sufficiently reliable instrument for the qualitative or semiquantitative estimation of ketone bodies, glucose, haemoglobin, bilirubin and ph in dog and rat urines. In contrast, the measurement of specific gravity :.s unreliable. References Alt JM, Hackbarth H, Deerberg F & Stolte H (1980) Proteinuria in rats in relation to age-dependent renal changes. Laboratory Animals 14, Audouin W, Couture G, Girard lc & Forest lc (1985) Evaluation de la densite urinaire par la bandelette reactive de la Cie Ames et comparaison avec la densite don nee par Ie refractometre et avec l'osmolalite urinaire. Clinical Biochemistry 18, Borghoff SJ, Miller AB, Bowen lp & Swenberg la (1991) Characteristics of chemical binding to Ci2u -globulin in vitro; evaluating structure-activity relationships. Toxicology and Applied Pharmacology 107, Evans GO & Parsons CE (1986) Potential errors in the measurement of total protein in male rat urine using test strips. Laboratory Animals 20, Grinstead GF, Scot: RE, Stevens BS, Ward VL & Wilson DM (1987) The Ames Clinitek 200/Multistix 9 urinalysis method compared with manual and microscopic methods. Clinical ChemistlY 33, Honnadel DC & K,)l1er A (1987) Urine protein, total. In: Methods in Clinical Chemistry (eds LA Kaplan, AJ Pesce), pp St Louis: CV Mosby Peele ld, Gadsden IlH & Crews R (1977) Evaluation of Ames "Clini-Tek". Clmical Chemistry 23, Peele JD, Gadsden RH & Crews R (1977) Semi-automated vs. visual reading of urinalysis dipsticks. Clinical Chemistry 23, Schumann GB (1984) Examination of urine. In: Clinical Chemistry, ThEOry, Analysis and Correlation (cds LA Kaplan, Al Pesce), pp St Louis: CV Mosby