Evaluation of the Sysmex UF-100 automated urinalysis analyzer

Department of Pathology, Virginia Commonwealth University/Medical College of Virginia, Richmond, VA.
^a Address correspondence to this author at: Department of Pathology, Medical College of Virginia, P.O. Box 980250, Richmond, VA 23298-0250. Fax 804-828-2812; e-mail JBENEZRA{at}HSC.VCU.EDU.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
Urinalysis is a high-volume procedure that currently requires significant labor to examine microscopic sediment. We evaluated the Sysmex UF-100 automated urinalysis analyzer for performing this task. Instrument accuracy was assessed by comparing continuous counts of microscopic elements from the UF-100 with ranges of cells (per low-power field or high-power field) from manual microscopy performed on centrifuged urines. Counts showed good agreement between methods (gamma statistic: 0.880–0.970) for all microscopic elements in 252 urine samples. Within-run imprecision of cell counts expressed as CV (mean cell count/µL) was for erythrocytes (RBC) 31% (5), 18% (50), 2.4% (800); for leukocytes (WBC) 14% (10), 11% (100), 8.5% (400); for squamous epithelial cells (SEC) 18% (5), 12% (30), 7.0% (100); for casts 45% (1), 17% (4); for bacteria 2–12% (entire range of 40–2500). Between-run imprecision on quality-control cell suspensions expressed as CV (mean cell count/µL) was for RBC 6.1% (50), 2.7% (256); for WBC 26.9% (54), 4.9% (228). Cells counted on dilution were 99.1% of expected for RBC, 102.0% for WBC, and 121.8% for bacteria. Carryover was <0.04% for RBC, <0.03% for WBC, <0.14% for SEC, <0.29% for bacteria. We conclude that the UF-100 can automatically perform reliable quantitative microscopic urinalysis in batches without operator interaction.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The microscopic analysis of urine sediment provides essential information to clinicians about disease states in patients. Unfortunately, it lacks precision and has wide interobserver variability (1)(2)(3)(4). Moreover, it is a very labor-intensive procedure, adding greatly to the cost of providing laboratory services (5). In the attempt to automate the process of microscopic urinalysis, image-based analysis systems have been developed (1)(3)(4)(6)(7). However, even though these image-based automated systems offer increased speed of throughput and precision, there is still the need for an operator to visually inspect images of cells and casts as the sample is processed by the instrument; this severely limits their usefulness in terms of freeing technician time to perform other tasks within the laboratory. Therefore, the need for a truly walkaway microscopic urine analyzer is great. To this end, we evaluated the UF-100 (Sysmex Corp.) automated urine analyzer, a flow cytometer-based walkaway instrument that performs automated microscopic urinalysis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
sysmex uf-100
The Sysmex UF-100 is an automated analyzer that performs a microscopic urinalysis by flow cytometry. In brief, after identifying the specimen by its bar code, the UF-100 aspirates 800 µL of urine and performs the analysis. Cells, bacteria, and casts are measured by electrical impedance for volume, forward light-scatter for size, and by fluorescent dye for nuclear and cytoplasmic characteristics. The cells and formed elements are categorized in multidimensional space on the basis of their size, shape, volume, and staining characteristics. The results are displayed in scattergrams on a screen, and a hard copy of the results may be obtained. Problematic specimens are identified for manual microscopic urinalysis. Although the instrument does not have chemical analysis capabilities, it can be electronically linked to an automated strip-reader to generate an integrated report; in this study, all urine samples were analyzed chemically on a different instrument (data not shown).

materials and methods
We studied 252 freshly collected urine samples submitted for diagnostic urinalysis to our laboratory in a tertiary-care university hospital. After the routine diagnostic microscopic urinalysis was performed, the samples were analyzed the same day on the UF-100 analyzer; the use of such discarded material has been approved by our institution's IRB panel. To reduce interobserver variability, the same technologist performed all the microscopic urinalyses with the same microscope, using the KOVA^® system (Hycor Biomedical) (8). The areas of a low-power field (LPF) and high-power field (HPF) were determined with a measured scale to be able to correlate the cells or particles seen in a microscopic field (usually measured as cells or particles/LPF or HPF) with the quantitative results of the UF-100, which are provided in units of cells (or particles)/µL.¹

Between-run quality-control samples consisted of suspensions of particles, provided by the manufacturer, in the sizes of bacteria, erythrocytes (RBCs), leukocytes (WBCs), and squamous epithelial cells (SECs). Within-run imprecision was determined by analyzing specimens with various concentrations of RBCs, WBCs, SECs, and bacteria 12–20 times (usually 20) each on the UF-100 analyzer. Linearity was determined by analyzing in triplicate specimens and their dilutions of 1:2, 1:4, 1:5, 1:10, 1:20, and 1:40; the slope, intercept, and proportion of measured value to expected value were determined with EP-Evaluator (Rhoads). Carryover analysis was performed by analyzing the specimen in triplicate, followed by three blank specimens; the percent carryover was determined by the formula:

Carryover = (blank 1 - blank 3)/(specimen 3 - blank 3)

Statistical analysis, including the gamma statistic (9) for measuring correlation and the McNemar test for measuring change in the distribution of two dichotomous variables by the test, was performed by using SPSS^® for Windows. Gamma is a measure of association between two variables measured on an ordinal level, and can be thought of as the probability that a random pair of observations is concordant minus the probability that the pair is discordant, assuming the absence of ties. Gamma is symmetric and ranges between 0 and 1.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
comparison with manual results
For RBC data (Table 1 ), the gamma statistic was 0.963. When we analyzed the data as clinical positive vs negative results (5 RBC/HPF = 26 RBC/µL), the two methods differed significantly (McNemar test, P <0.00001). As can be seen from the Table , the UF-100 identified more RBCs in the samples than did microscopic urinalysis; this is most likely due to incomplete pelleting of the sediment for manual urinalysis, as well as the nonstandardization of cell resuspension after centrifugation.

For the WBC data (Table 2 ), the gamma was 0.970. Analyzing the data by clinical positive vs negative results (5 WBC/HPF = 26 WBC/µL), we again found significant differences between methods (P <0.00001).

For the SEC data (Table 3 ), the gamma was 0.880. With results categorized by clinical positive vs negative (5 SECs/LPF = 1.95 SECs/µL), the methods differed (McNemar test, P <0.00001). Samples in which the UF-100 "overcalled" epithelial cells contained casts and Trichomonas organisms. The protein in these samples most likely caused WBCs to aggregate, resulting in a possibly falsely high SEC count.

The UF-100 detected many types of casts, including hyaline, granular, and cellular casts. For these data (Table 4 ), the gamma was 0.880. Analyzing the data by clinical positive vs negative results (1 cast/LPF = 0.39 casts/µL), we found significant differences between the two methods (McNemar test, P <0.00001).

precision
Within-run CVs for the RBC analysis are shown in Table 5 . The CV ranged from 33% for low numbers of RBCs (5 RBCs/µL) to <3% at RBC concentrations >660/µL. Between-run precision was 6% at a measured mean of 50 RBC/µL and 2.7% at a measured mean RBC of 256/µL.

Within-run precision for the WBC analysis is shown in Table 5 . The CV ranged from 24% for low numbers of WBCs (4 WBCs/µL) to 6% at a WBC count of 853/µL. Between-run precision was 27% at 54 WBC/µL and 4.9% at a WBC of 228/µL.

Within-run precision ranged from 24% at 3.42 SECs/µL to 2.9% at 180 SECs/LPF. Between-run precision was 9.1% at 22 SECs/LPF and 7.5% at 88 SECs/LPF.

Within-run precision for casts varied from 75% for samples with <0.42 casts/µL to 17% for samples with 4.6 casts/µL.

The within-run CVs for bacterial counts (Table 5 ) was 2–12% over the concentration of bacteria tested. Between-run CV was 29% at 54 bacteria/µL, 37% at 60 bacteria/µL, and 26% at 146 bacteria/µL.

linearity
Dilutions of urine samples were made as described in Materials and Methods. For casts, dilutions of 1:2 and 1:4 were made from a sample with 18 casts/µL, and the specimens were analyzed in triplicate.

For samples with RBC concentrations of 3864 to 15 956 cells/µL, the measured values were 87–105% of expected. For three samples with WBC concentrations of 5698 to 5735 cells/µL, values were 92–112% of expected. For bacteria (two samples), results were 111% (at 1809 cells/µL) and 133% (4433 cells/µL) of expected, and for casts (one sample) 106%.

carryover
Carryover studies were performed on four urine samples of various RBC, WBC, SEC, cast, and bacterial counts. Carryover ranged from -0.13% to 0.04% for RBC of 13 to 2404 cells/µL, -0.62% to 0.03% for WBC counts of 12 to 855 cells/µL, -1.6% to 0.14% for SEC of 3.0 to 72 cells/µL, 0.00% to 0.44% for 0.26 to 28 casts/µL, and -0.23% to 0.29% for bacterial counts of 277 to 2744 cells/µL. These results show that no substantial carryover was detected in any of the samples, demonstrating that the carryover was not systematic.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
Manual analysis of urine sediment, although clinically useful, is fraught with methodological problems. At least 11 different factors may contribute to imprecision in manual urinalysis (1), ranging from centrifugation to the different interpretation of a cell or cast in a urine sediment by different technologists (2). In addition, the process is time consuming, requiring approximately 6 min of technologist time per specimen (5). To this end, there have been attempts to automate the process, thereby improving accuracy, precision, and throughput.

The Yellow IRIS^® urinalysis instrument has automated the microscopic evaluation of urine sediment. Several studies have demonstrated that it has increased precision over routine manual urinalysis, and that it can detect more abnormalities than conventional microscopy (1)(3)(4)(6)(7). However, a technologist must be physically present at the instrument to characterize events, and some technologist interpretation of images is required, thus leading to imprecision and inaccuracy. The UF-100 instrument analyzes urine cells and particles on the basis of flow cytometric principles, analogous to the manner by which modern hematology analyzers perform complete blood and differential cell counts.

The UF-100 analyzer in this study detected more RBCs, WBCs, and SECs than did manual microscopy; similar findings have also been reported for the Yellow IRIS (1). This probably represents true detection of cells not identified by manual microscopy, since routine urinalysis has centrifugation, decantation, and resuspension steps that are either incomplete or lead to cellular loss and lysis. Nonetheless, agreement between the manual and UF-100 methods for these cellular elements was excellent (gamma = 0.88–0.97). Samples in which the UF-100 "overcalled" epithelial cells contained casts and Trichomonas organisms. The protein in these samples possibly caused WBCs to clump, resulting in a possibly falsely high SEC count.

Detection of casts with the UF-100 was less robust than was the detection of cellular elements; this has also been seen with the Yellow IRIS (6), and may be due to differences in the volume of urine sample analyzed. We chose not to evaluate accuracy of bacteria detection, since the "gold standard," quantifying bacterial counts in culture, measures something qualitatively different, live bacteria, from what the UF-100 urine analyzer measures, live and dead bacterial particles. This is a limitation of all automated urinalysis analyzers (10). Nonetheless, the UF-100 flags the presence of casts, crystals, and bacteria, alerting the laboratory to their presence and the need to characterize them further under the microscope.

Automated urinalysis performed by the UF-100 displayed good precision for analyzing cellular elements, bacteria, and casts; the CVs were much less than those reported historically for manual urinalysis (1)(3)(4), and are comparable with those seen with other automated instruments. Moreover, the assays were linear over clinically useful ranges, and no carryover was found with the UF-100 analyzer for the analytes measured.

Comparison of the UF-100 with manual microscopy, the gold standard, is somewhat difficult, since this standard is somewhat tarnished by several methodological steps that lead to imprecision and inaccuracy (1)(2). Our studies show that the UF-100 urine analyzer's results are comparable with those of manual microscopy and published reports of the performance of the Yellow IRIS urine analyzer. In addition, counts of cellular elements with the UF-100 are comparable with those obtained on unspun urines analyzed under the microscope with a hemocytometer (manuscript in preparation). We conclude that the UF-100 can perform accurate and precise quantification of microscopic elements in urine with little or no operator interaction.

	Acknowledgments

 
We acknowledge the assistance of the Sysmex Corporation, which provided technical assistance, control material, and a stipend for evaluation of the UF-100 urine analyzer.

	Footnotes

 
¹ Nonstandard abbreviations: LPF, low-power field; HPF, high-power field; RBC, erythrocyte; WBC, leukocyte; and SEC, squamous epithelial cell.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Carlson DE, Statland BE. Automated urinalysis. Clin Lab Med 1988;8:449-461. [ISI][Medline] [Order article via Infotrieve]
2. Winkel P, Statland BE, Jorgenson J. Urine microscopy: an ill-defined method examined by a multifactorial technique. Clin Chem 1974;20:436-439. [Abstract]
3. Roe CE, Carlson DA, Daigneault RW, Statland BE. Evaluation of the Yellow IRIS. An automated method for urinalysis. Am J Clin Pathol 1986;86:661-665. [ISI][Medline] [Order article via Infotrieve]
4. Deindorfer FH, Gangwer JR, Laird CW, Ringold RR. "The Yellow IRIS" urinalysis workstation—the first commercial application of "automated intelligent microscopy". Clin Chem 1985;31:1491-1499. [Abstract/Free Full Text]
5. Manual for laboratory workload recording method. Skokie, IL: College of American Pathologists, 1984:140..
6. Elin RJ, Hosseini JM, Kestner J, Rawe M, Ruddel M, Nishi HH. Comparison of automated and manual methods for urinalysis. Am J Clin Pathol 1986;86:731-737. [ISI][Medline] [Order article via Infotrieve]
7. Wargotz ES, Hyde JE, Karcher DS, Hitlan J-P, Wilkinson DS. Urine sediment analysis by the Yellow IRIS automated urinalysis workstation. Am J Clin Pathol 1987;88:746-748. [ISI][Medline] [Order article via Infotrieve]
8. McGinley M, Wong LL, McBride JH, Rodgerson DO. Comparison of various methods for the enumeration of blood cells in urine. J Clin Lab Anal 1992;6:359-361. [ISI][Medline] [Order article via Infotrieve]
9. Goodman LA, Kruskal WH. Measures of association for cross-classification. J Am Stat Assoc 1954;49:732-764.
10. Poropatich CO, Mendoza SM, Hitlan JJ, Wilkinson DS. Inconsistent detection of bacteriuria with the Yellow IRIS automated urinalysis workstation. Lab Med 1988;19:499-501.