Cytornetry (Communications in Clinical Cytornetry) 22:75-79 (1995)

Urinary Sediment Analyzed by Flow Cytometry

Yoshinori Yasui, Noriyuki Tatsumi, Keunsik Park, and Takuzo Koezuka
Department of Laboratory Medicine (Y.Y., N.T.), Department of Medical Information (RP.), Central Clinical Laboratory
(T.K.), Osaka City University Medical School, Osaka City, 545, Japan

A flow cytometer for the automated analysis of urinary sediment was designed, and its performance was
examined by the evaluation of 821 specimens. Auramine 0, a dye for DNA and RNA, was used for the staining
of the sediment. Urine (5 ml or more) was processed by the instrument for sediment analysis. Conventional
microscopic analysis was done for comparison. The RBC count, the WBC count, and the number of bacterial
cells, epithelial cells, and casts found by the flow cytometer and by microscopy were compared. Correlation
was high for all these results. The overall sensitivity, specificity, and efficiency (accuracy) in the items
analyzed were 84.7%, 57.8%, and 67.2%, respectively. One hundred specimens could be analyzed by the
instrument per hour. The instrument seemed useful for screening for urinary tract disorders to identify
specimens that should be analyzed microscopically in routine laboratories. o 1995 Wiley-Liss, Inc.

Key terms: Urine, sediment, flow cytometry, laboratories, hospital, autoanalysis, urinary tract disorders

Automation in the routine clinical laboratory has been

making rapid development, especially in the fields of
clinical chemistry and hematology. However, the examination of urinary sediment is still done by the traditional
method of microscopy. Results can be of critical importance in the diagnosis of urinary tract disorders and others (4).
Urinalysis is noninvasive and is widely used in almost
all clinical fields. Most laboratories do a semiquantitative
paper test of the urine first. Further examination, including sediment analysis, is often omitted, even when the
first test shows abnormal findings, because sediment analysis lacks accuracy and precision and, in addition, the
work is biohazardous and laborious (54,s).We designed
an automated instrument for use in urinary sediment
analysis and examined its performance in a routine laboratory.

MATERIALS AND METHODS

Materials
Fresh urine specimens were obtained from patients or
healthy volunteers. The clinical diagnoses of the patients
were conveyed to us by the physician after our tests were
completed.

2-30 ml of urine (usually more than 5 ml) was put into

the apparatus. The urine was agitated and 0.6 ml of urine
was aspirated through a nozzle. The urine was stained
automatically with auramine 0,a fluorescent dye of DNA
and RNA (1,9), passed through a sheath to flow into a
detector, and then exposed to argon laser light ( 6 ) . The
fluorescence intensity and forward light scatter were analyzed by a data processing unit (Fig. 1) (7).

Microscopic Analysis
Ten milliliters of fresh urine was centrifuged at 500
X g for 5 min, and its supernatant was removed by aspiration, leaving 0.2 ml of urine, which contained the sediment. After gentle mixing of the sediment, it was
smeared onto a slide and observed under a light microscope at 400 X (high power field; HPF) except for casts.
Casts were observed at 100X (low power field; LPF)

(3,4).
Statistics
Statistical analysis was done with Pearsons method for
correlation studies and with Wilcoxons test for nonparametric group studies (2). Differences for which P was
less than 0.005 were taken to be significant.

Efficiency Test

Instrument
A flow cytometer used for reticulocyte analysis (Sysmex R-1000, Toa Medical Electronics Co., Kobe, Japan)
was modified for urinary sediment analysis by the addition of devices for urine transport and software for signal
analysis, both designed with the technical cooperation of
the manufacturer. In testing, a sample cup containing
0 1 9 9 5 Wiley-Liss, Inc.

Sensitivity, specificity, and efficiency (accuracy) were

calculated as described in reference 7. Whether the re-

Received for publication October 5, 1993; accepted August 28,

1994.

76

YASUI ET AL.
Forward scatter signal

REC

tional cells, and small round cells, were displayed in

roughly the same part of the scattergram.

Fluorescencesignal

A Forward scatter intensity

E Fluorescence intensity
C Scatter pulse width
D Fluorescence pulse width
E Cell length
F Nuclear length

Epithelial cell

Bacteria

Reproducibility and Carryover of Results Obtained by the

Automated Method
When five specimens were tested 10 times, the coefficients of variation (CV) for RBC, WBC, bacteria, epithelial
cells, and casts, were 1.9-34.1%, 5.4-26.6%, 2.6-9.3%,
7.3-14.1%, and 55.8-194.4%, respectively. The CV of
the casts was very high because few casts were found
(Table 1).
Carryover studies were done with the five components. Four samples with each component were tested.
All tests showed no carryover (9).

Correlation Between Results by the Automated and

Microscopic Methods

FIG. 1. Forward scatter signals and fluorescence signals of components of urinary sediment.

As part of the study of the correlation of results by the

two methods, 821 urine specimens from 371 male subjects and 450 female subjects (mean age k SD, 45 2 43
years) were analyzed by an Aution Analyzer (Kyoto Daiichi Kagaku Co., Kyoto, Japan). The mean results of urinalysis were a pH of 6.19 1.70, protein of 58.2 616.3
mg/dl, glucose of 205.5 t 2495.6 mg/dl, and hemoglobin
of 0.18 0.83 mg/dl. Correlations of analysis by the automated and microscopic methods for RBC, WBC, and
epithelial cells are shown in Figure 3. Of the five items
examined, RBC, WBC, and epithelial cells showed a significant correlation (P< 0.000 1), but the correlation for
the number of bacteria was not high.

sults were true positives was judged by microscopic analysis of the urinary sediment.

RESULTS
Scattergrams of Components of Sediment
Figure 2 shows a negative sample and a positive sample
with all five main components (REK, WBC, bacteria, epithelial cells, and casts). The positive sample shows typical patterns when each of the five components, which
may be found in abnormal urinary sediment, were increased to more than that of a negative specimen. All
types of epithelial cells, such as squamous cells, transi-

Significant Differences in Values Between Sexes for

Five Items
By both methods all the items tested, except for casts,
were found significantly more often in specimens from
female subjects (Table 2).

Positive Sample

Negative Sample

Casts

_-CL

P I

LL

FIG.2. Scattergrams of negative sample and positive sample with all components. FLS, forward light scatter;
FI, fluorescence intensity; NL, nuclear length; CL, cell length.

77

URINALYSIS BY FLOW CYTOMETRY

R BC

Table 1
Reproducibility Test (Performed 10 Times)
Sample

Mean(/$)

SD

3.3
6.1
1098.8
4.1
97.2

1.0
0.6
20.5
1.4
6.5

29.9
9.5
1.9
34.1
6.7

34.9
18.9
87.5
16.5
72.5

4.4
5.0
4.7
1.8
4.1

12.6
26.6
5.4
11.1
5.7

43.9
276.0
149.1
105.4
276.8

4.1
16.3
3.8
8.2
13.1

9.3
5.9
2.6
7.8
4.7

37.2
31.3
19.5
15.9
21.3

2.7
4.4
2.4
1.5
1.7

7.3
14.1
12.2
9.7
8.1

0.05
0.05
0.19
0.10
0.10

29.1
61.0
55.8
94.4
29.1

RBC

1
2
3
4
5
WBC

1
2
3
4
5

y = 0.363~
+ 0.946. r = 0.765, P ~0.0001

CN%)

Epithelial cells

3
4
5
Casts

1
2
3
4

200

400

(MPF)

Microscopy

WBC

Bacteria

1
2
3
4
5

0.04
0.03
0.34
0.05

y = 0 . 6 7 6 ~- 0.075, r = 0.892, P <0.0001

Microscopy

Epithelial cells
y = 0.308~
+ 0.638, r = 0.619, P ~0.0001

Results of Evaluation of Automated Analysis

In all the items analyzed, sensitivity, specificity, and
efficiency were more than 60%; the specificity was
greater than the sensitivity (Table 3 ) . The overall efficiency was examined, as shown in Table 4. Sensitivity
was 84.7%. The false-negative rate calculated from the
percentage of samples was found to be 5.3% by microscopy (8).

20

40 (MPF)

Microscopy

Renal Disorders
The numbers of epithelial cells and casts in patients
with renal disorders were significantly higher than in the
other subjects (Table 5). Renal disorders were judged
from the total results after analysis of urinary sediment.

Speed
The automated method can analyze 100 specimenshr.

DISCUSSION
A medical technologist first scans the whole field by

microscope on a slide at low and middle magnifications,

and then moves on to observation at high magnifications
if the specimen shows abnormally increased numbers of
certain components. (Technologists screened specimens
first to find if they require further study.) The automated
device being evaluated was designed for use in this kind
of screening in routine laboratories. This is a fully auto-

FIG. 3. Correlation of results from RBC, WBC, and epithelial cells

analyzed by automated analysis and microscopy.

mated urinalysis instrument that can analyze down to the

five main components of urine. It is not necessary to
handle specimens for centrifugation and slide preparation.
RBC, WBC, bacteria, epithelial cells, and casts occupied different areas on the scattergram. RBC and WBC
were more deformed in urine than in blood. Cells were
often found to be slightly shifted from their expected
location on the scattergram; these cells were probably
deformed. Due to the operating system we could interpret such changes, so that pathological changes could be
identified readily by inspection of the scattergram. The
built-in computer provided information about the kind of
abnormality that was present in the specimen, and this

78

YASUI ET AL

Table 2
Significant Differences in Values Between Sexes for Five Items

RBC
WBC
Bacteriab
Epithelial cells
Casts

Total
(/H PF)
9.0 f 70.0
7.3 f 55.6
0.5 f 1.5
5.0 t 15.8
2.0 i_ 23.1

Microscopy
Men
Women
(IHPF)"
(IHPF)a
10.6 t 7 9 . 2
7.2 t 56.9
10.0 2 70.3
4.0 2 27.3
0.2 2 0.5
0.7 t 1.6
1.9 +- 5.8
7.6k19.1
2.9 k 31.6'
1.3 s 12.3'

P
0.0004
<0.0001
<0.0001
<0.0001
0.7005

Automated
Men
(Iul)a
3 . 1 t 29.7
2.8
15.0
0.7 t 2.1
0.7k2.1
0.5 k 4.7

Total
Uul)
4.1 k 3 1 . 8
4.9
42.2
0.9 2.3
2.2t7.8
0.4 k 3.8

*
*

analysis
Women
Uul)
4.6 t 32.4
54.9
6.6
1.1 f 2 . 6
1.122.6
0.5 2 3.2

<0.0001
<0.0001

<0.0001
<0.0001
0.0056

aResults were expressed as the mean t 2 SD.

bBy microscopy, 0-<50 bacteriaIHPF were scored as 0, 50-<150IHPF were scored as + 1, 150-<250/HPF were scored as +2,
and 250-</HPF were scored as +3. By automated analysis, 0-<5 bacteriaipl were scored as 0, 5-<20/p\ were scored as tl,
20-<30/pI were scored as 2, and >30Ipl were scored as + 3
'Per low power field.

Table 4
Overall Efficiency of Sediment Analysis of Five Items"

Table 3
Results of Evaluation of Automated Analysis"

Sensitivity (Yo)
Specificity (%)
Efficiencv (Yo)

RBC
68.5
76.1
74.6

WBC
84.9
90.7
90.1

Bacteria
65.8
85.2
83.3

Epithelial
cells
74.5
89.0
86.3

Casts
61.7
72.7
71.5

aResults by automated analysis were taken to be true positives

if by microscopy (1) there were five or more RBCIHPF; (2) there
were 10 or more WBCIHPF; (3) there were 100 or more bacteria/
HPF; (4) there were eight or more epithelial cells/HPF; (5) there
were 1.6 or more casts/LPF.

informationwas useful in identification of what should be

looked for by microscopy.
One advantage of this automated method was high reproducibility, which is not possible with microscopic
analysis.Correlation between the automated method and
the microscopic method was high. Differences between
all pairs of items had P values of less than 0.0001.
Urine samples from girls and women have more RBC,
WBC, bacteria, and epithelial cells, and our results
showed the same tendency. The tests of each item
showed high sensitivity, specificity, and efficiency, suggesting that automated analysis of urine is practical. Overall efficiency analysis also produced high sensitivity,

Positive
sarnDlesC
222
206
428

Positive sa m piesb
Negative Sam plesb
Total samdes

Negative
SamDles'
40
282
322

Total
samoles
262
488
750

aSensitivity = 84.7%, specificity = 57.8%, efficiency =

67.2%, false-negative rate = 5.3%, false-positive rate =
27.5%.
bResults from microscopic analysis. If at least one of the following conditions is detected, the sample is judged to be positive: six or more RBCIHPF, 10 or more WBCIHPF, > l o 0 bacteriaIHPF, eight or more epithelial cellsIHPF, 1 . 6 or more casts/
LPF.
'Results from automated analysis. If at least one of the following conditions is detected, the sample is judged to be positive: one or more RBC/pI, five or more WBCIpI, 2 0 or more
bacteriaipl, 2.5 more than epithelial cellsipl, more than zero
castsipl.

specificity, and efficiency results. False-negative results

were few. The false-positive rate was relatively high, because the rate of cast identification was high. It may be
difficult to detect a low concentration of casts by microscopy. In our evaluation specimens giving positive results
by automated analysis were reexamined by microscopy.

Table 5
Results i n Patients with Renal Disorders

RBC
WBC
Bacteriab
Epithelial cells
Casts

Renal disordersa
(/HPF)
13.86 I109.27
7.76 2 38.83
0 . 6 3 -t 1.44
6.17 t 16.92
4.46 ? 36.68'

Microscopy
Othera
(IHPF)
8.42 t 62.8
7.14 t 57.69
0.50 2 1.46
4.79 f 15.39
1.69
20.46'

Automated analysis
Renal disordersa
Othera
P
0.2412
0.9637
0.0365
0.0290
0.0016

(/PI)

4.39
5.41
1.05
2.98
0.92

k
k
k
k
k

35.28
26.28
2.63
8.57
7.57

(IpC

3.90 t 30.75
4.70 k 43.81
0.90 ? 2 . 3 4
2.06 k 7.66
0.40 +- 3.08

P
0.0269
0.1998
0.3937
0.0081
0.0357

aResults were expressed as the mean 2 2 SD.

bBy microscopy, 0-<50 bacteriaIHPF were scored as 0, 50-<150/HPF were scored as + 1, 150-<250/HPF were scored as + 2 ,
and 250-</HPF were scored as +3. By automated analysis, 0-<5 bacteriaipl were scored as 0, 5-<20/pl were scored as +1,
20-<30Ipl were scored as + 2, and >3O/pl were scored as + 3.
'Per low power field.

URINALYSIS BY FLOW CYTOMETRY

A low false-negativerate is desirable in all forms of testing

done for diagnosis, particularly screening.

When specimens with crystals visible to the naked eye
were tested in the automated apparatus, the crystals were
interpreted by the machine to be an increase in KBC or
bacteria. Smaller amounts of crystals caused no interference. Mucus, flocculation, and sperm did not interfere
with the test results. Thus the automated analysis of urinary sediment seems to be reliable, and it might be useful
for screening for abnormal specimens during routine urinary examinations.

ACKNOWLEDGMENTS
We thank Dr. Mikio Okamura, First Department of Internal Medicine, Osaka City University Medical School,
for providing information about his patients with kidney
disease.
LITERATURE CITED
1. Davis BH, Bigelow NC: Clinical flow cytometric reticulocyte analysis. Pathobiology 58:99-106, 1990.

79

2. Dwass M: Probability and Statistics. W.A. Benjamin, New York, 1970.

3. Elin RJ, Hosseini JM, Kestner J, Rawe M, Ruddel M, Nishi HH: Comparison of automated and manual methods for urinalysis. Am J Clin
Pathol 86:731-737, 1986.
4. Free HM, Harber MH, Kaufman HW, Lanphear B, Mullins C ,
Schweitzer S, Webster S: Routine urinalysis and collection, transportation, and preservation of urine specimens. NCC1.S Pubkdtion
GPl6-T. Villanova, PA: NCCLS, 1992.
5. International Committee for Standaritation in Haematology: Protocol for evaluation of automated blood cell counters. Clin Lab Haematol 669-84, 1984.
6. Kachel V, Fellner-Feldegg H, Mcnke E: Hydrodynamic properties of
flow cytometry instruments. In
(eds): Flow Cytometry and
Sorting, Ed 2. 1990, p p 27-44.
7. National Committee for Clinical Laboratory Standards: Assessmcnt
of clinical sensitivity and specificity of laboratory tests; proposed
guideline. NCCLS publication GP10-P. Villanova, PA: NCCLS, 1987.
8. Roe CE, Carlson DA, Daigneault RW, Statland BE: Evaluation of the
yellow IRIS. Am J Clin Pathol 86:661-665, 1986.
9. Stockert JC, Del Castillo P, Llorente AR, Rasskin DM, Romero JB,
Gornez A: New fluorescence reactions in DNA cytochemistry. Analyt
Quant Cytol Histol 12:l-10, 1990.
10. Wheeless L, Jr: Slit-scanning. In
(eds): Flow Cytometry and Sorting, Ed 2. 1990, pp 109-125.