urised

<-/
~t
I
,":'
I( I\~
A1
". \
__
-'--
Biotechnol. Appl. Biochem. (2009) 53.139-144
www.babonline.org
(Printed in Great Britain)
doi:IO.I~2/BA200BOI88
139
Comparison of labUMat-with-UriSed
and iQ®200 fully
automatic urine sediment analysers with manual urine analysis
Okhan Kadir Akin*l. Muhittin A. Serdarj-. Zeynep Cizmecit.
Suleyman Aydin§
Ozlem Genct and
*Kecioren Research and Training Hospital, Clinical Chemistry and Laboratory Science, Ankara 063BO, Turkey, tGulhane School
of Medicine, Department of Clinical Chemistry, Ankara, 0601B, Turkey, tKecioren Research and Training Hospital,
Microbiology and Clinical Microbiology, Ankara 063 BD, Turkey, and §Departrnent of Biochemistry and Clinical Biochemistry,
Firat University, School of Medicine, Fira!, Elazig 23 I 19, Turkey
Urine analysis is one of the most common tests for
assessing urinary-tract
and kidney diseases. In recent years there have been new developments
in the
automation
of this test. The objective of the present
study was to compare the performances
of two urine
sediment
analysers,
namely labUMat
with UriSed
(77 Elektronika
Kft, Budapest, Hungary) and iQ@200
(Iris Diagnostics,
Chatsworth,
CA, U.S.A.), with the
KOVA® method
for manual urine measurement
by
evaluating the results in terms of similar parameters
(cells or particles
per lower-power
field or high·
power field). The results obtained
using the UriSed
and iQ@200 analysers
were
more
reproducible
(7.1-30.2
and 14.9-35.4%
respectively)
than those
obtained
using the manual technique
(17.9-44.4%).
Significant correlations
were established
among the
three
techniques
in the evaluation
of leucocytes,
erythrocytes
and epithelial cells. Although the UriSed,
iQ®200 and visual.microscopic
measurements
were
in agreement,
confirmation
of the results from
automated
methods
by manual
urine analyses
is
significantly
useful, especially for pathological
cases
that were close to the limits of the techniques.
Introduction
Chemical testing of urine samples (reagent strip methods)
and identification and counting of particles have been
performed routinely to identify and monitor diseases of the
kidney and urinary tract [I]. Despite the introduction of an
automated system to count particles more than two decades
ago, manual microscopic examination of urinary sediment
is still needed. owing to the limitations of this system [2-5].
Owing to such a need. the earliest sediment analysers
utilized an image-based analysis system [5-7]. However,
even though such automated systems increase the speed of
throughput and the precision of measurement, the operator
must still inspect images of cells and casts visually as the sam-
pie is processed by instruments. This severely limits the
usefulness of such systems in terms of freeing technician
time for other laboratory tasks.
The objective of the present study was to compare the
iQ®200 microscopic analyser (Iris Diagnostics, Chatsworth,
CA, USA.), which has been in use for some years, with the
recently installed Uri5ed automation system [LabUMat (77
Elektronika Kft, Budapest, Hungary)] and with manual urine
microscopy, using similar parameters
[cells or particles
per LPF (Iower.power field) or HPF (high-power field)] for
evaluation.
Materials and methods
We studied 600 recently collected midstream urine samples
submitted to our laboratory
for diagnostic urinalysis.
The samples were collected in sterile containers and
transferred to test tubes (3 ml for the iQ®200 analyser,
5.5 ml for LabUMat with UriSed and 10 ml for routine
manual diagnostic microscopic urinalysis). No preservatives
were added. Samples were analysed within I h of arrival.
To reduce inter·observer
variability, the same technician
performed all the microscopic urinalyses with the same
microscope by using the KOVA® system (Hycor Biomedical,
Garden Grove, CA, USA.). The examined LPF and HPF
areas were determined with a measurement scale so that
the numbers of cells or particles in a field (usually measured
as cells or particles per LPF or HPF) could be correlated.
The HPF and lPF results from the iQ®200 and labUMatwith·UriSed devices were included in the present study.
In the iQ®200 analyser, urine particles in the sample
are imaged in a planar flow cell that orients and constrains
Key words: fully automatK urine sediment analYSIs.iQ(!l200 unne analyser.
KOVA ~ system, LabUMat with UriSed urine analyser, visual manual Urine
sediment analYSIs.
AbbreViations used' EC. eprthelial cell, HPF. high-power field: LPF, IO'N-power
field: NPV, negative predictive value; PP\/, positIVe predictive value. RBC.
red blood cell; WBC. white blood cell.
To whom correspondence should be sent (email [email protected]).
~ 2009 Portland Press Ltd
140
-
------------._----------
O. K. Akin and others
particles hydrodynamically within the focal plane of a
microscope objective. The iQ®2oo uses APR'" (AutoParticle Recognition), a well-trained neural network, to
classify and quantify twelve formed elements. The results
can be auto-reported
on the basis of user-defined criteria.
iQ®200 takes 500 photographs from each urine sample,
compares them with standard images, and classifies particles
on the basis of tissue, contrast, shape and size from the
image. The iQ®200 device classifies the images and presents
them to the user, allowing corrections and new definitions
to be introduced [8,9].
The LabUMat operates on the basis of microscopic
examination of a urine sample in a special disposable
cuvette. During the measurement
process, the urine
sample is transferred to the cuvette and centrifuged. Highresolution complete views of field images are then recorded
automatically by a microscope. The Uri5ed device pipettes
a 200 I.d urine specimen and creates a preparation of 0.145
mm'; the depth of the native urine in the cuvette becomes
1.1 mm. After centrifuging the preparation for 10 s at 260 g
and thereby pelleting the particles, the device analyses a
2.4 J.LIurine sample by scanning 15 field images. These
images are then evaluated by a special algorithm. The sample
is evaluated by a special neural-network-based
imageprocessing algorithm through the use of the so-called multilevel decision method. Each image is recognized in 'real time',
while the evaluation procedure is running on the image just
after recording. The evaluation takes 3-4 s per image.
This recognition software is implemented in UriSed user
software, fully developed and improved by 77 Elektronika.
The device also counts the particles by comparing the
entire image obtained from the areas examined with standard images. It shows the entire field image to the user and
also allows corrections and new definitions to be introduced.
The iQ®200 analyser conducts the sediment analysis
by utilizing flow cytometry. The UriSed analyser, conversely,
carries out this analysis by using the microscopic image of
the preparation it produces from the urine sample in full
automation. When the two devices are compared, the most
significant advantage of the iQ® 200 analyser over the
UriSed one is that it makes it possible to individually display
the sediment images by classifying them [RBCs (red blood
cells), WBCs (white blood cells) and casts]. However, the
advantage of the UriSed system is that it displays the entire
field image, just as is the case when manual microscopy
is used. The UriSed device is capable of scanning 15 field
images. Displaying the entire field image simultaneously
provides an edge in assessing the sample.
For the IQ®2oo analyser, the refractometric method
(specific-gravity measurement interval 1.000-1.050), and for
------~th~ec;-La=bOMat
wlth-OriSed. the dipstick test (LabStrip U I I
plus; Analyticon Biotechnologies AG, Lichtenfels, Germany)
(specific-gravity measurement
interval 1.000-1.030), are
~ 2009
Portland
Press Ltd
Table
I
Reference
values for WBCs.
--
RBCs and ECs
Positive
Cells
Negative
few
Moderate
WBC, (cellslHPF)
RBC, (cellslHPF)
EC, (cellslLPF)
0-5
0-2
0-2
6-10
3-5
Low
11-20
6-10
Medlum
Many
21-50
11-25
High
>51
26-50
>51
used to assess the density. Both devices conduct the urine
pH analyses through the dipstick test (pH indicator).
Liquicheck Control (urinalysis control) by Bio-Rad
Laboratories (Hercules, CA, U.S.A.) level I and level 2 was
utilized in the reproducibility study (lot numbers: level I,
6 I261; level 2, 61272). In order to analyse reproducibility
within or between the runs, the coefficients of variation
were estimated on 20 measurements
of two different
control materials [level I: 0-2 RBCs, 0-1 WBC; level 2:
20-120 RBCs, 10-50 WBCs)].
During the assessment of patient results, the reference
values given in Table I were used for WBCs, RBCs and ECs
(epithelial cells) [10].
Cases that were viewed as negative and positive by all
three techniques were accordingly recorded as negative or
positive. When the results of three techniques were not in
agreement, they were re-tested on the basis of established
cut-off values; checks were performed using strips, and
an attempt was made to establish reliable values. When
two of the three methods gave positive results, the sample
was assumed positive. Sensitivity, specificity, PPV (positive
predictive value) and NPV (negative predictive value) were
calculated on the basis of this assessment.
Specific-gravity values obtained from refractometric
analysis by the iQ®200 were utilized to study the effects
of specific-gravity and pH in cases that gave inconsistent
results. The pH values used were averages of the iQ®200
and UriSed strip pH results.
Statistical analyses were performed
using SPSS®
(version 15.0) for Windows. Lilifor's test was used for first
estimates of the population mean and variance based on the
data. Non-parametric
Gamma statistics were performed
to measure correlations and the McNemar test was used to
measure changes in distribution
of two dichotomous
variables. The independent
sample t test was used
to evaluate the differences in specific-gravity and pH
measurements in non-concordant cases. A value of P < 0.05
was considered statistically significant.
Results
The reproducibility of WBC and RBC counts was assessed
at two different levels using all three methods, and the
results are shown in Table 2.
Comparison of automatic and manual urine analysis
Table 2
Reproducibility of iQ(!)200. UriSed and manual methods
Abbreviation: 0/. coefficient of variatIon.
CellslHPF
Bet......-eenprecision
Within precision
Level I
0/(%)
Mean
1.6±0.3
1.2 ± 0.4
19.1
35.4
RBCs
WBCs
1.4±0.4
1.2 ± 0.4
RBCs
WBCs
1.6±0.7
1.5±0.7
Method
Cells
Mean
Q"'2oo
RBCs
W8Cs
UriSed
Manual
Table 3
± S.D.
Level 2
Level I
± SD.
Level 2
± SD.
± 5.D.
0/(%)
Mean
1.7 ± 0.4
1.2 ± 0.4
23.5
33.3
43.5 ± 7.8
39.6 ± 6.7
17.9
16.9
7.1
13.1
1.5 ±0.5
1.3 ±04
33.3
30.8
44.7 ± 4.5
36.9 ± 5.6
10.1
15.2
17.4
19.2
1.5 ±08
1.6 ± 0.8
53.3
50.0
429±8.1
36.4 ± 7.6
18.9
20.9
0/(%)
Mean
42.4 ± 6.3
38.6 ± 6.3
14.9
16.3
28.7
302
45.7 ± 3.2
378 ± 5.0
42.4
44.4
41.4 ± 7.2
35.3 ± 6.8
0/(%)
Comparison of iQ<!>200. UriSed and manual WBC countS
CellslHPF
--
UnSed
Q"'2oo
Manual (cellslHPF)
0-5
5-10
11-20
21-50
> 51
Total
O-S
5-10
11-20
21-50
>
0-5
5-10
11-20
21-50
> 51
Total
466
17
I
0
0
484
8
35
I
I
0
45
2
17
16
0
2
J7
J
0
J
6
J
15
0
0
I
I
17
19
478
69
22
8
22
600
469
JJ
2
0
0
504
8
32
8
0
I
49
I
4
11
2
0
18
0
0
I
6
9
16
I
0
0
0
12
IJ
WBC counts using the manual method and the iQ®200
and UriSed analysers are compared in Table 3. The Gammastatistics value was 0.975. When we analysed the data
in relation to clinically positive versus negative results
(5 WBCs/HPF). the iQ®200 and manual methods differed
significantly (McNemar test; P < 0.00 I). Overall. the nonconcordant results could have affected 5.16 % of all clinical
diagnoses (Table 3). 5imilarly. comparison between the
UriSed and manual methods for WBC measurements gave
Gamma statistics of 0.974. and again there was a significant
difference in terms of clinical evaluation (McNemar test;
P < 0.00 I). Among all cases. 7.5 % of the results were nonconcordant (Table 3). Finally. the Gamma statistics for the
comparison of UriSed and iQ®200 was 0.969. and this rate
was established as different in both automated techniques
(McNemar test; P < 0.000 I). Analysis of the data in terms
of clinically positive and negative results (5 WBCs/HPF)
showed non-concordance
in 6.66 % of all cases (Table 4).
Comparison
between
the iQ®200 and manual
methods in respect of RBC measurement
showed a
good correlation (Gamma statistics 0.956). but the two
methods differed significantly in relation to clinically positive
versus negative results (two RBCs/HPF) (McNemar test;
Table 4
51
Total
478
69
22
8
22
600
Comparison of UriSed and iQ<!>200 WBC counts
Uri5ed (cellslHPF)
Q"'2oo
0-5
6-10
11-20
21-50
> 51
Total
(cellslHPF)
0-5
6-10
11-20
21-50
> 51
Total
473
22
7
2
0
503
11
21
0
I
14
2
1
18
0
I
I
8
6
16
0
0
0
I
12
IJ
483
45
37
15
19
600
IS
2
0
49
P < 0.00 I); clinical non-concordance
was 10.2 %, overall
(Table 5). Comparison between the UriSed and manual
methods showed a similar pattern (Gamma statistics. 0.958;
McNemar test P < 0.000 I). with a difference in respect
of the clinical results (Table 5). Similar results were also
obtained from the comparison between iQ®200 and UriSed
in terms of RBC measurement (Gamma statistics. 0.975;
McNemar test, P < 0.000 I and non-concordance
affecting
clinical results in 8.7 % of all cases) (Table 6).
Although measurements of EC counts by the three
methods correlated. there were differences in 9 % of the
cases between manual and iQ®200. 10.5 % of the cases
© 2009 Portland Press Ltd
141
142
0. K. Akin and others
Table 5
Comparison
of iQ~200 and manual RBC counts
CellslHPF
-UnSed
00"'200
Manual (cellslHPF)
G-2
3-5
6-10
11-25
26-50
>
G-2
3-5
6-10
11-25
26-50
> 51
Total
430
42
2
2
I
0
477
IJ
JI
8
0
0
0
52
I
8
13
I
0
0
~3
0
5
2
11
I
0
19
0
0
I
5
2
I
9
0
0
0
3
3
14
20
Table 6
Comparison
51
Total
G-2
J-5
6-10
11-25
26-50
>
444
86
26
22
7
15
600
439
67
11
2
I
0
519
5
13
7
2
0
0
27
0
6
6
2
0
0
14
0
0
2
12
3
I
18
0
0
0
2
2
I
5
0
0
0
2
I
IJ
16
51
444
86
26
22
7
15
600
Table 8 Sensitivity. speciflCity. PPV and NPV results from iQID200.
and KOVA~ systems
of iQ~200 and UriSed for RBC counts
Total
UriSed
Uri5ed (cellslHPF)
00"'200
(cellslHPF)
G-2
3-5
6-10
11-25
26-50
> 51
Total
Table 7 Comparison
EC counts
G-2
3-5
6-10
472
39
7
2
0
0
520
4
12
8
3
0
0
27
0
I
7
4
2
0
14
11-25
I
0
I
8
6
2
18
26-50
> 51
Total
0
0
0
2
I
2
5
0
0
0
0
0
16
16
477
52
23
19
9
20
600
Method
Cells
Sensitivity
SpeciflClty
PPV
NPV
00"'200
RBC,
WBC,
75.8
85.5
96.1
97.4
87.8
90.3
91.5
96.0
Un5ed
RBC,
WBC,
68.7
759
988
97.6
95.8
902
89.2
93.4
KOVA"'
RBC,
WBC,
68.0
85.2
97.0
97.2
89.8
89.3
88.6
96.0
00"'200'
RBC,
WBC,
93.7
936
99.1
99.6
97.6
98.5
97.5
98.1
UriSed'
RBC,
WBC,
91.7
94.2
99.J
98.5
98.1
94.9
96.8
98.3
among iQ~200. UriSed and manual techniques for
Manual (cellslLPF)
G-2
Low
MedIum
High
G-5
00"'200
Uri5ed
467
473
15
9
4
4
0
0
Low
,Q"'200
Un5ed
34
48
42
30
9
8
I
0
Medium
iQ"'200
UriSed
0
4
5
9
10
9
7
0
'Q"'200
Un5ed
I
I
0
0
0
2
High
• Results obtained when assessed together With strip or KOVA(!l.
Table 9 Relationship among non-concordances.
WBCs and RBCs
Parameter
n
WBC,
RBC,
Specific gravity
SW
56"
1.017±0.005
1.017±0.006
1.016± 0.007
1.015 ± 0.006
0.987
0.118
P value.
pH
518'
82"
between manual and UriSed, and 8 % of the cases between
iQ®200 and UriSed (Table 7).
The sensitivity, specificity, PPY and NPY results
obtained using established criteria are given in Table 8.
There were no statistically significant differences in the
specific gravity and pH measurements among cases by any
of the three techniques, irrespective of the concordance
among the results (Table 9).
Discussion
Manual analysis of urine sediment is fraught with methodological problems. Many factors may impair its precision and
~ 2009 Portland Press Ltd
specific gravity and pH for
P value.
7.77 ± 1.08
7.82 ± 0.89
0.648
7.65 ± 1.02
7.83 ± 0.97
0.123
• Concordant results for wacs or RBCs measured by all three methods.
10 Non-concordant
results for WBe or RBC measured by all three methods.
accuracy [I I), ranging from centrifugation to the different
interpretations of cell or cast in a urine sediment by different technicians [12]. In addition, the process requires approx. 5-10 min of technician time per specimen [13). Therefore there have been attempts to automate the process to
improve accuracy and precision and to save technician time.
Comparison
The iQ®200 and UriSed are two analysers currently
used for microscopic evaluation of urine. Since both
analysers yield patient results as cells/HPF and cells/LPF,
as does the manual test, it should be easy to compare
them. The UriSed microscopic analyser is now being used
in an automated form by combining it with LabUMat, which
measures urine chemistry.
The results of the present study show statistically
significant correlations among the manual, iQ®200 and
UriSed methods. Ben-Ezra et al. [6] found that results
from the UF·I 00 automated urine anaiyser correiated with
manual counts. Various researchers
have shown that the
iQ®200 gives results that correlate significantly with manual
counts, but, in the results from specimens with fewer
cells, the reproducibility and concordance
between the
two tests cfecreased, and there were statistically significant
differences (especially at less than ten cells/HPF) [6,8,9,1416]. This non-concordance was attributed to cell lysis during
centrifugation and resuspension or to protein aggregation
[6,14]. However, the same researchers showed that the
measurements
become more concordant when the cell
counts in the urine rise. In our study, the level I sample gave
quite reproducible results, especially in manual analysis, but
the coefficient of variation was < 20 % for pathological specimens. We predicted that non-concordance among the methods stemmed from specific-gravity and pH differences, so
we studied the effects of these variables. However, the nonconcordance was not attributable to these two parameters
(Table 9) and further studies will be needed to explain it.
The results showed non-concordance
between the
methods sufficient to affect clinical diagnoses, particularly
for urine samples with high cell counts close to the limits
of the technique (six to ten cells/HPF for WBC, three
to five cells/HPF for RBC) (Tables 2-6). Ben-Ezra et al.
[6] reported similar results; they found non-concordance
in analytical accuracy for WBC and RBC measurements in
5.6-14.2 % of patients. Although other workers
have
reported similar levels of non-concordance affecting clinical
results, that study [6] was the first to present these data
[14-16]. Table 8 shows PPVs and NPVs obtained through
the evaluation of automated systems and the manual
system, either individually or in unison. Thus, when the
doubtful results were ev~luated by the manual method and
the automated techniques together, PPV and NPV values
were observed to have increased.
In terms of analytical accuracy, a significantly high number of false negative results were obtained, even though their
specificities were adequate (Table 8). As observed in the
present study, analytical accuracy was increased if the results
were evaluated by strip analysis or if a visual assessment was
carried out. We consider that verification of microscopic
analyses of urine by manual microscopic and chemical analyses would decrease the number of clinically significant
of automatic
and manual urine analysis
errors, particularly for cases at the limits of the cell or cast
techniques [14-16].
In addition, even though microscopic urine analyses
are less frequently performed because the use of automated
techniques has increased (especially in higher-capacity
hospitals), the usefulness of the manual method should be
emphasized in educating clinical laboratory staff, bearing in
mind that it is still the gold standard.
In summary, although the two automated techniques,
UriSed and iQ®200, are highly reproducible and are able to
analyse large numbers of urine samples quickly and simultaneously, it is important to confirm the results by manual urine
analysis, especially for pathological cases at the limits of the
techniques, and/or to compare them with urine-strip results.
Acknowledgement
We thank BioDPC, Istanbul, Turkey (the representatives of
Iris Diagnostics in Turkey) and MED-KIM, izmir, Turkey (the
representatives of 77 Elektronika Kft in Turkey) for their
scientific and technical support during the present study.
Funding
This work was supported by the Kecioren Research and
Training Hospital [educational grant no. 20080200 I]. The
funding organization played no role in the design of the study,
review and interpretation of the data, nor in the preparation
or approval of the manuscript.
References
2
3
4
5
6
7
8
Fuller,C. E..Threane, G. A and Henry, J. B. (200 I) In Clinical
DiagnosISand Management by Laboratory Methods, 20th edn
(Henry, j. B.. ed). pp. 367-402. WB. Saunders. Philadelphia
Okada, H.. Sakai, t, Kawabata, G.. Fujisawa,M., Arakawa, 5..
Hamaguchi, Y.and Kamidono, S. (200 I) Am. J. On. Pathol.
115,605-610
Regen~er,A, Haenni, V.. Risch,L.. Kochli,H. P, Colombo, J. P.
Frel, R. and Huber, A R. (200 I) On. Nephrol. 55. 384-392
Ottlger, C. and Huber, A R. (2003) On. Chem. 49. 617-623
Roe. C. E.. Carlos. D. A, Daigneau~. R.Wand Standant. B. E.
(1986) Am. J. On Pathol. 86, 661-665
8en-Ezra, j .. 80rk, L.and McPhearson, R.A (1998) On. Chem.
44, 92-95
Elin.R.J .. Hosselni, J. M.. Kestner, j.. Rawe, M., Ruddel, M. and
NISh"H. H. (1986) Am. J. C1in.Pathol. 86, 731-737 .
Wah, T. D.. Wises. P.K.and Butch, A W (2005) Am. J. On.
Pathol. I23, 290-296
~ 2009 Portland Press Ltd
14]
144
O. K. Akin and others
9
Linko. 5.. Kouri. 1. 1.. Toivonen. E.. Ranta PH .. Chapoulaud. E.
14
10
Ringsrud. K. M. and Lmnea. j. j. (1995) Urinanalysis and Body
15
Fluids. Mosby- Year Book Inc, St Louis, MO
1I
12
Chien,T I.. Kao. J. 1., Liu, H. L.. Lin, P c.. Hong.
J. 5.. Hsieh, H. P
and Chien, M. J. (2007) Clin. Chm Acta 384. 2B- 34
and Lalla M. (2006) Clin. Chm Acta 372. 54-<>4
Shayanfar N .. Tobler U.. von Eckardstein, A. and Bestmann, L.
(2007) Clin Chem. Lab. Med 45, 1251-1256
Carlson, D. E. and Statland, B. E. (1998) Clin. Lab. Med. 8,
16
Lamchiagdhase, P, Preechabonsutkul,
K.. Lomsomboon.
P.
449-461
Srisuchart, P,TantlMi, P, Khan-u-Ra. N. and Pneechaborisutkul.
Winkel. P, Statland, B. E. and jorgenson, J. (1974) Clin. Chem.
B. (2005) Clin. Chim Acta 358,167-174
20,436-439
13
College
of
American
Laboratory Workload
Pathologists
American Pathologists. Skokie. IL
:0 2009 Portland Press Ltd
(19B4)
Manual
for
Recording Method, p. 140. College of
Received 22 September 2008117 November 2008; accepted 21 November 2008
Published as Immediate Publication 21 November 2008. doi: 10.1 0421BA20080 188