| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Institute for Laboratory Medicine and Pathobiochemistry, and
2
Department of Paediatric Nephrology, University Clinic Charité of the Humboldt University Berlin, Schumannstrasse 21, 10117 Berlin, Germany;
3
Dedicated to Prof. Dr. E. Köttgen on the occasion of his 60th birthday.
a author for correspondence: fax 49-30-2802-8422, e-mail pranav.sinha{at}charite.de
None of the methods currently available for evaluating hematuria and leukocyturia is perfect. Urinalysis by test strip combined with automated particle counting is an attractive approach that may identify false-negative results of both techniques. We wished to evaluate the possibility that microscopic evaluation of urine samples could be substantially reduced by this approach.
The fully automated urine flow cytometer UF-100 classifies urinary particles on the basis of their light scattering, fluorescence, and impedance properties. The instrument counts erythrocytes, leukocytes, bacteria, epithelial cells, and casts and flags the presence of pathological casts, small round cells (SRCs), yeast-like cells, crystals, and spermatozoa. The instrument is intended to replace, to an extent, routine urine microscopy. The operating principles of the UF-100 (1); its precision, accuracy, and analytical sensitivity (2); and its potential for differentiation between renal and postrenal hematuria (3) have been published previously (4).
The precision of particle counting in microscope chambers is poorer than counting by flow cytometry (5). The analysis of urine samples by UF-100 and the test strip analyzer Clinitek Atlas showed discordant results in a small but not negligible number of cases. This result has also been described by Ben-Ezra et al. (4). We therefore rechecked the false results in erythrocyte and leukocyte counts by examining discrepant results from both instruments by microscopic evaluation and evaluated a workflow for the routine urine analysis using the combination of UF-100 and Clinitek Atlas.
We analyzed 288 mainly pathological urine specimens by both test strip analyzer (Clinitek Atlas; Bayer Diagnostics) and UF-100 (Sysmex). The samples were examined within 3 h of arriving at the laboratory; no preservatives were used. We additionally reviewed 261 by microscopy.
The ranking system was used to compare the results from UF-100 and
Clinitek Atlas (Table
1A)
. This ranking system was devised because cells were
counted very precisely by UF-100 and, therefore, a differentiated
classification was required to compare the test strips results.
|
The workflow strategy was to use both methods together and to detect implausible results by rank value differences between the Clinitek Atlas and UF-100 for erythrocytes and leukocytes separately. Differences in rank values of two or more were regarded as medically relevant. Specimens exhibiting this difference were reviewed by microscopy.
Discordant cases with lower rank values for strip results and higher rank values for flow cytometry were reviewed by microscopy. Microscopy was also performed in cases with higher ranks for strip results if large numbers of cells were seen. If cells were not seen in the UF-100, fresh samples were requested and examined. In some cases, the presence of lysed cells was assumed or excluded in concordance with the disease. Microscopic review was also performed when warning messages indicated that manufacturer specifications were being exceeded. The microscopy counts were performed after centrifugation (500g for 5 min; Madaus System). The cells were stained with Sternheimer supravital solution (Alcian Blue and Pyronin B; Oy Reagena) (6)(7).
Our workflow strategy and rank classifications were then tested prospectively on 635 unselected routine specimens obtained over a 5-day period.
The dysmorphism and isomorphism criteria were examined on the UF-100 in 120 patients with confirmed renal or postrenal hematuria. The diagnostic sensitivity of these criteria for evaluating the localization of the hematuria was determined against phase contrast microscopy and established clinical diagnoses (8)(9)(10).
Statistical calculations were carried out using SPSS for Windows (11) and the frequencies of binomial distribution (12).
Red blood cell (RBC) and white blood cell counts were compared using both methods in 288 urine samples with a high frequency of pathologic results. In most cases (263 of 288 samples), there was agreement between the two methods based on our ranking system. The valid results were within the limits of 1.5 rank difference (see Table 1B). The 25 discordant cases are presented in Table 1B. The traditional workflow would have required microscopic review in 67% of the samples. Using the new combined approach, we reduced the microscopic review rate significantly, to 9% (P <0.05).
This approach was further validated using 635 random samples. In this case, the microscopic review rate was 36% (n = 230). The use of both instruments reduced the review rate to 6% (n = 39; P <0.05). Microscopy also clarified the situation in 25 specimens with discordant results (Table 1B).
Of the 263 concordant cases between the test strip and the UF-100, 248 were checked microscopically. Differing results for 6 samples for erythrocytes and 6 samples for leukocytes were obtained from 11 urine samples. In all cases, microscopic review revealed more cells than did the instruments.
The review algorithm of the UF-100 suggested microscopy in 142 samples (133 manufacturer-defined flags and 103 user-defined flags). In most cases, user-defined flags were caused by increased numbers of SRCs and "pathologic casts" (Table 1B). In some of the specimens showing a review message for pathological casts, we found urate crystal aggregates. When compared with microscopic reviews, the UF-100 always flagged the presence of pathological casts, SRCs, or yeast cells. Most of the manufacturer-defined review flags (n = 66) were attributable to particle counts exceeding the upper limit of the UF-100 measurement range (>40 000/µL). These samples required reanalysis following dilution.
Other warning messages, e.g., "nonlysed RBC 20%" appeared in 48 of 635 samples. This message indicates strong lysis of RBCs, and the RBC value is extrapolated from exact fluorometric measurement of nonlysed RBCs (20%) and RBC fragments (ghosts, 80%). In some other cases, however, bacteria may be misclassified as RBC fragments, and the number of RBCs can be spuriously extrapolated by the UF-100.
The warning message "Abnormal DC Sensitivity" occurred almost exclusively in urine samples from newborns (7 of 14 samples) when the conductivity of the sample was abnormally low or high. Because our working algorithm assumed examination by both test strip and the UF-100, this warning flag could be ignored when test strip and UF-100 results were in concordance. This reduced the necessary microscopic checks of flagged samples from 142 to 11 samples.
In this study, the localization of hematuria was based on the clinical diagnosis and morphological observations. These classifications were compared to the morphology flags obtained in the UF-100. The UF-100 has fixed algorithms for morphology flagging. In our study, the UF-100 correctly classified 37 of 53 postrenal (70%) and 66 of 86 renal hematurias (77%). In 20 cases, the UF-100 could not correctly classify hematuria because of low RBC counts.
These findings indicate that the combination of test strips with UF-100 flow cytometry can reduce and practically eliminate false-negative or -positive results that are obtained (~2%); these samples can be reviewed microscopically. The major reason for discrepant results was misclassification of particles (yeast or bacteria for lysed RBC). Scrutiny of the scattergrams, however, is crucial for correct classification of these particles. In addition to improved accuracy, our strategy yields quantitative results for particular elements in urine. This may be an improvement to the microscopic counts currently used for monitoring of renal diseases.
Because microscopic review of urine specimens requires 6–8 min per specimen, our approach reduces manual labor, and the majority of specimens can be analyzed using automated techniques. Furthermore, the reliability is improved by automation. The UF-100 displays several flags for microscopic review. In most cases, these reviews can be circumvented by using test strips in parallel and diluting samples with high total cell counts. However, the manufacturer-defined review flags for SRCs and pathological casts can be confirmed accurately only by microscopy. In our opinion, these findings are of less diagnostic importance because quantitative urinary protein determination offers higher diagnostic reliability in renal disease.
Renal and postrenal hematuria can be distinguished relatively well based on the isomorphism or dysmorphism flags of the UF-100. However, delay in analysis because of transportation problems combined with low RBC counts may also be a critical factor.
In summary, the combined sequential analysis of urine sample with a test strip analyzer and the UF-100 flow cytometer appears to be better than the standard procedures used to date. The main advantages are that the majority of specimens can be analyzed automatically, thus reducing manual labor and turnaround times. In specific cases, however, special microscopic techniques such as sediment analysis with or without supravital staining can still be used as auxiliary techniques.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Gai, G. B. Piccoli, G. P. Segoloni, and G. Lanfranco Microscopic Urinalysis and Automated Flow Cytometry in a Nephrology Laboratory Clin. Chem., September 1, 2003; 49(9): 1559 - 1560. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
