Professional Documents
Culture Documents
Jooyoung Cho, Kyeong Jin Oh, Beom Chan Jeon, Sang-Guk Lee* and Jeong-Ho Kim
Urine sediment analyzers can be divided into two dif- between the measurements. In accordance with the CLSI guideline
ferent systems based on their principle of function; one GP16-A3, all samples were analyzed within 2 h of their receipt in our
laboratory [22]. There was no significant contamination (less than
is a digital image-based system, and the other is flow
1%) from previous samples in carry-over study that used high- and
cytometry-based system [2, 5, 9]. Digital image-based low-level pooled samples of red blood cells (RBCs) and white blood
systems analyze urine sediments based on several images cells (WBCs) (data not shown).
taken by a built-in digital camera, followed by automatic Parameters that were common for all five urine analyzer plat-
particle recognition and sorting [7, 8, 11, 12, 15, 20]. Flow forms were tested: RBCs, WBCs, squamous epithelial cells (SQEPs),
non-squamous epithelial cells (NECs), crystals, hyaline casts, patho-
cytometry-based systems analyze electrical impedance,
logic casts, yeast, sperm and mucous threads. Bacteria were not eval-
forward-scattered light (FSC), side-scattered light (SSC),
uated in this study because urine culture is a reference method and
and side fluorescence light (SFL) of urine sediments in not a manual microscopy technique.
order to distinguish between the particles [6, 9, 10, 14, 17, This study was performed with authorization from the Insti-
19]. With generational increments in platform instrumen- tutional Review Board (IRB) of Severance Hospital (IRB no. 1-2017-
tation, both of these systems have improved in analytical 0038).
parameters detected by the LPF lens. Finally, the images are analyzed reported as “positive” for the presence of either crystal, pathologic
by a built-in automated analyzer. In this study, we used a prototype cast, yeast or sperm. If a case met the criteria of image review, a flag
of a YD URiSCAN® PlusScope, as it was not yet officially launched. was displayed. Because the UF-5000 is a flow cytometry-based sys-
tem and cannot provide digital camera images, the UD-10 was used
in the image review process. For other devices, we could review their
own camera screen. The microscopic review rate is the percentage of
Manual microscopic examination
uninterpretable cases (as determined by image review) due to being
out of focus, low image quality or involving crowded particles. There-
Sediment examination using manual microscopy was performed fore, such cases were confirmed by manual microscopy.
according to CLSI guideline GP16-A3 [22]. It has been established that
the gold standard chamber for the microscopic examination is the
Fuchs-Rosenthal chamber. However, we used the standardized KOVA Interference study
cell chamber system [KOVA® Systemic Super Pac 1000 w Counting
Grids (KOVA International Inc., Garden Grove, CA, USA)] for micro-
We evaluated the effect of interference on RBC, WBC and bacterial
scopic counting because it provides a standardized quantitation [23].
counts by crystals, pathologic casts (non-hyaline casts) and yeast.
Each 12 mL urine sample was centrifuged at 400 g for 5 min, and
The differences between semi-quantitative grades of RBC, WBC and
then approximately 11 mL of supernatant was removed. The mean
bacteria in total and specific samples (“positive” for each interfer-
final volume of precipitate that remained after centrifugation was
ent, as determined by manual microscopy) were analyzed [25]. The
1.10 ± 0.05 mL (mean ± standard deviation, n = 50). The concentra-
results for RBC and WBC counts were categorized into six ranges as
tion fold for sediment from total urinary volume used was 10.91. Each
follows: 0–2/HPF, 3–5/HPF, 6–10/HPF, 11–20/HPF, 21–50/HPF and
chamber consisted of 9 × 9 small grids. The total area under the grids
>50/HPF. These categories were then converted to the corresponding
(Agrids) was 9 mm2 (3 × 3 mm) and total volume of these grids (Vgrids)
numbers (from 0 to 5). For bacteria, the result was expressed as ‘neg-
was 0.9 μL (chamber depth: 0.1 mm). Thus, the volume of one small
ative’, ‘1 + ’, ‘2 + ’, and ‘3 + ’, and these semi-quantitative grades were
grid was 0.0111 μL (≈0.9 μL/81) (Supplementary Table 1). Therefore,
converted to the corresponding numbers (from 0 to 3). Median grades
when converting particles/grid to particles/μL, the value was multi-
of RBC, WBC and bacteria in the total samples and each selected
plied by 90 (1 μL = 90 grids). According to the instructions posted on
group were calculated. The differences between these median grades
the website of KOVA International Inc. [24], we estimated the con-
were compared using the Mann-Whitney test, and the difference was
version factor from particles/grid to particles/μL using the following
considered significant if p-value < 0.05.
formula:
particles/µL
= (particles/grid) × 90 grid/µL/fold concentration (= 10.91)
Statistical analysis
= (particles/grid) × 8.25 grid/µL
Passing-Bablok regression was used for comparison of quantitative
Two well-trained laboratory technicians counted each type of cells parameters (RBC, WBC and SQEP). For qualitative parameters (crys-
in the sediments in ten small grids at HPF, and the average number tal, hyaline cast, pathologic cast, NEC, yeast, sperm and mucous
of particles/grid was converted to particles/μL, using the formula thread), we evaluated clinical performance using agreement rate,
above. This value was then used as reference for each sample. Inter- sensitivity, specificity, positive predictive value (PPV), negative pre-
observer coefficient of variation (CV) between the two technologists dictive value (NPV) and Cohen’s kappa value. Statistical analysis was
was 3.70% for RBCs and 3.56% for WBCs. performed using SPSS version 20.0 (IBM Corp., Armonk, NY, USA)
and Microsoft Excel 2010 (Microsoft Corp, Redmond, WA, USA) with
Analyse-it version 3.90.7 (Analyse-it Software, Ltd., Leeds, UK).
Comparative study of urine sediment analyzers and
manual microscopy
For RBCs, WBCs and SQEPs, the mean number of cells counted by
Results
two technologists were compared with numbers obtained from
each automated urine sediment analyzers using the Passing-Bablok
regression analysis. For other parameters (crystal, hyaline cast,
Comparison of quantitative parameters
pathologic cast, NEC, yeast, sperm and mucous thread), the result of urine sediment analyzers with manual
was expressed as either “positive” or “negative”. The cut-off values microscopy
for bacteria and each of the qualitative parameters were determined
using values suggested by the platform manufacturers (Supplemen- Numeric data for RBC, WBC and SQEP were analyzed using
tary Table 2).
Passing-Bablok regression. Manual microscopic counts
using KOVA chamber were used as reference (X axis) and
Image and microscopic review rate counts from each of automated urine sediment analyzers
were used for comparison (Y axis). The Passing-Bablok
Image review rate was determined using the percentage of the follow- equations for the five urine sediment analyzers are pre-
ing cases: cases reported as “invalid” (cannot be analyzed) and cases sented in Table 1. The proportional biases between RBCs
–
0.64
0.74
0.61
0.76
measured using the UF-5000, Cobas® u 701, and UAS800
RBC, red blood cell; WBC, white blood cell; SQEP, squamous epithelial cell; p/μL, particles per microliter; CI, confidence interval. aThe Cobas® u 701 reports the SQEP as semi-quantitative value
instead of quantitative value. Manual microscopic counts using the standard KOVA chamber were used as reference (X axis) and counts from each of automated urine sediment analyzers were
compared with values measured using the manual micro-
scope were within ±20% of each other. The slopes for RBC
1.78 (0.86–3.07)
0.00 (0.00–0.00)
Iris iQ®200SPRINT were 0.99, 0.87, 0.88, and 0.84, respec-
Not availablea tively. For WBC, the UF-5000, Cobas® u 701, UAS800, and
SQEP, p/μL
1.16 (1.07–1.31)
0.96 (0.71–1.25)
0.96 (0.78–1.24)
Slope (95% CI)
0.92
0.71
0.76
0.91
0.77
0.00 (0.00–0.00)
1.14 (1.04–1.24)
1.14 (1.02–1.25)
0.91 (0.85–1.00)
1.28 (1.12–1.35)
0.80 (0.63–1.05)
0.74
0.63
0.50
0.68
0.73
0.99 (0.74–1.32)
0.87 (0.63–1.37)
0.88 (0.64–1.39)
0.84 (0.72–0.96)
0.73 (0.52–1.01)
UAS800
Table 2: Diagnostic accuracy of qualitative parameters of the five urine sediment analyzers compared with manual microscopy.
Crystal (n = 111)
UF-5000a
Overall 91.7% 29.7% 99.8% 94.4% 91.6% 0.42 (0.37–0.47)
With the exception of amorphous crystal 96.0% 46.8% 99.8% 93.8% 96.0% 0.59 (0.53–0.65)
Cobas® u 701 91.2% 62.2% 95.1% 61.7% 95.0% 0.57 (0.52–0.61)
UAS800 92.3% 49.5% 97.9% 75.0% 93.7% 0.55 (0.51–0.60)
Iris iQ®200SPRINT 87.9% 68.5% 90.4% 47.9% 95.7% 0.49 (0.46–0.53)
URiSCAN® PlusScope 87.7% 18.0% 96.7% 41.2% 90.2% 0.20 (0.15–0.24)
Hyaline cast (n = 26)
UF-5000b 95.3% 23.1% 97.2% 17.6% 98.0% 0.18 (0.11–0.25)
Cobas® u 701 92.3% 19.2% 94.2% 8.1% 97.8% 0.08 (0.03–0.13)
UAS800 86.0% 65.4% 86.6% 11.3% 99.0% 0.16 (0.12–0.19)
Iris iQ®200SPRINT 96.4% 3.8% 98.8% 7.7% 97.5% 0.04 (−0.02 to 0.09)
URiSCAN® PlusScope 96.4% 3.8% 98.8% 7.7% 97.5% 0.04 (−0.02 to 0.09)
Pathologic cast (n = 53)
UF-5000 92.0% 23.6% 95.9% 25.0% 95.6% 0.20 (0.14–0.26)
Cobas® u 701 89.4% 73.6% 90.2% 30.4% 98.4% 0.38 (0.34–0.43)
UAS800 84.6% 81.1% 84.8% 23.6% 98.8% 0.31 (0.27–0.35)
Iris iQ®200SPRINT 94.1% 0.0% 99.5% 0.0% 94.6% –
URiSCAN® PlusScope 94.6% 0.0% – 0.0% – –
Non-squamous epithelial cell (n = 211)
UF-5000 86.6% 44.4% 88.2% 12.1% 97.7% 0.14 (0.10–0.18)
Cobas® u 701 80.1% 44.4% 81.4% 8.1% 97.6% 0.08 (0.05–0.11)
UAS800 94.1% 19.4% 96.8% 18.4% 97.0% 0.16 (0.10–0.22)
Iris iQ®200SPRINT 96.0% 8.3% 99.2% 27.3% 96.7% 0.11 (0.05–0.18)
URiSCAN® PlusScopec – – – – – –
Yeast (n = 36)
UF-5000 95.3% 44.4% 97.1% 36.4% 97.9% 0.38 (0.31–0.45)
Cobas® u 701 94.4% 58.3% 95.7% 33.3% 98.4% 0.39 (0.33–0.46)
UAS800 94.9% 50.0% 96.5% 34.6% 98.1% 0.38 (0.32–0.45)
Iris iQ®200SPRINT 91.8% 33.3% 94.0% 16.9% 97.5% 0.19 (0.13–0.24)
URiSCAN® PlusScope 96.7% 25.0% 99.3% 56.3% 97.3% 0.33 (0.25–0.42)
Sperm (n = 5)
UF-5000 99.6% 20.0% 100.0% 100.0% 99.6% 0.33 (0.09–0.58)
Cobas® u 701 96.3% 100.0% 96.2% 11.6% 100.0% –
UAS800 97.9% 0.0% 98.4% 0.0% 99.5% –
Iris iQ®200SPRINT 99.5% 0.0% – 0.0% – –
URiSCAN® PlusScope 99.5% 0.0% – 0.0% – –
Mucous thread (n = 145)
UF-5000 88.8% 39.3% 97.0% 68.7% 90.6% 0.44 (0.40–0.49)
Cobas® u 701 89.6% 89.7% 89.6% 58.8% 98.1% 0.65 (0.62–0.68)
UAS800 91.7% 81.4% 93.5% 67.4% 96.8% 0.69 (0.66–0.72)
Iris iQ®200SPRINT 88.3% 75.2% 90.5% 56.8% 95.6% 0.58 (0.54–0.61)
URiSCAN® PlusScope 92.5% 70.3% 96.2% 75.6% 95.1% 0.69 (0.65–0.72)
PPV, positive predictive value; NPV, negative predictive value; CI, confidence interval. aThe UF-5000 dissolves amorphous crystals using a
special reagent, so amorphous urate and amorphous phosphate could not be detected. bThe UF-5000 reports total casts instead of hyaline
casts. cThe URiSCAN® PlusScope has no parameters for non-squamous epithelial cells.
Image and microscopic review rate were relatively high in the Cobas® u 701 (13.3%) and UAS800
(18.8%). The rates of “invalid” calls were relatively higher
Image review rates and microscopic review rates are in the Cobas® u 701 (3.2%) and UAS800 (4.2%). Consider-
shown in Table 5. We obtained relatively high flagging ing these factors together, the overall image review rates
rates for crystals using the Cobas® u 701 (11.3%) and Iris for the UF-5000, Cobas® u 701, UAS800, Iris iQ®200SPRINT
iQ®200SPRINT (16.2%). Flagging rates for pathologic casts and URiSCAN® PlusScope were 9.0%, 24.6%, 25.2%, 17.7%
Table 3: Detection rates (sensitivity) of specific crystals of the five urine sediment analyzers.
Crystal
UF-5000
Overall 4.9% 25.0% 71.9% 0.0% 7.7% 31.3% 29.7%
With the exception of amorphous crystal – – 71.9% 0.0% 7.7% 31.3% 46.8%
Cobas® u 701 65.9% 75.0% 71.9% 0.0% 46.2% 43.8% 62.2%
UAS800 58.5% 62.5% 50.0% 0.0% 38.5% 31.3% 49.5%
Iris iQ®200SPRINT 78.0% 75.0% 75.0% 0.0% 53.8% 43.8% 68.5%
URiSCAN® PlusScope 0.0% 12.5% 50.0% 100.0% 0.0% 12.5% 18.0%
Table 4: Detection rate (sensitivity) of specific pathologic casts of in the Iris iQ®200SPRINT, and by yeast in the UAS800.
the five urine sediment analyzers. The median level of bacteria was significantly influenced
by crystals and pathologic casts in the UAS800 and Iris
Pathologic cast
iQ®200SPRINT (Table 6).
Cellular Granular Total
(n = 8) (n = 45) (n = 53)
UF-5000
Cobas® u 701
12.5%
50.0%
26.7%
77.8%
24.5%
73.6%
Discussion
UAS800 87.5% 80.0% 81.1%
Iris iQ®200SPRINT 0.0% 0.0% 0.0% In this study, we evaluated the diagnostic performance of
URiSCAN® PlusScope 0.0% 0.0% 0.0% five recently introduced automated urine sediment ana-
lyzers. For quantitative comparisons of RBC, WBC and
SQEP, urine sediment analyzers showed that the propor-
and 6.5%, respectively. Microscopic review rates were tional bias was ±20% of values obtained using the manual
similar among the five systems (2.3%–4.5%). microscopic with the standard KOVA chamber. However,
the Iris iQ®200SPRINT for WBC and URiSCAN® PlusScope
for RBC showed proportional bias greater than 20%. The
Interference study reasons for these biases are unclear, but there are several
possible disadvantages of the manual methods such as
The median of RBC count is significantly influenced the inaccuracies associated with manual counting, the
by crystals in the Iris iQ®200SPRINT and URiSCAN® problems in sample processing and storage, and cell loss
PlusScope. In addition, the RBC count was also influenced during centrifugation. Furthermore, correct identification
by yeast in the UF-5000, Iris iQ®200SPRINT and URiSCAN® of RBC is difficult because of the presence of dysmorphic
PlusScope. However, the RBC count was not significantly RBCs, ghost RBCs, or misclassification with yeasts or other
influenced by pathologic casts. The median of WBC possible interferents [21]. In addition, other possible error
count was influenced by crystals and pathologic casts sources including focused field variation, recognition of
Table 5: Image review rates and microscopic review rates of the five automated urine sediment analyzers.
All samples Crystal positive samples Pathologic cast positive samples Yeast positive
samples
Amorphous Amorphous Calcium oxalate Triple Uric acid (+) Total Cellular Granular Total
phosphate (+) urate (+) (+) phosphate (+) cast (+) cast (+)
Median Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value Median p-Value
RBC (gradinga)
UF-5000 0.255 – – – – 0.625 0.006 0.538 0.138 0.188 0.344 0.382 0.089 0.375 0.701 0.378 0.050 0.264 0.099 0.639 0.002
Cobas® u 701 0.231 0.951 0.159 1.250 0.333 0.125 0.749 0.154 0.318 0.491 0.085 0.491 0.975 0.125 0.872 0.444 0.277 0.396 0.348 0.250 0.065
UAS800 0.254 0.561 0.876 1.250 0.341 0.438 0.513 0.308 0.690 0.509 0.991 0.509 0.723 0.000 0.358 0.356 0.134 0.302 0.301 0.444 0.115
Iris iQ®200SPRINT 0.114 0.024 0.885 1.000 0.039 1.031 0.000 0.385 0.018 0.455 0.655 0.455 0.000 0.250 0.045 0.267 0.138 0.189 0.547 0.417 0.015
URiSCAN® PlusScope 0.167 0.171 0.306 0.500 0.829 1.031 0.000 0.154 0.469 0.473 0.204 0.473 0.002 0.000 0.484 0.200 0.873 0.170 0.906 0.667 0.000
WBC (gradinga)
UF-5000 0.216 – – – – 0.313 0.729 0.308 0.868 0.375 0.329 0.355 0.115 −0.250 0.020 0.200 0.653 0.132 0.645 0.333 0.246
Cobas® u 701 0.253 0.512 0.581 0.000 0.253 0.438 0.816 0.154 0.305 0.063 0.120 0.327 0.158 0.375 0.890 0.244 0.611 0.264 0.604 0.500 0.083
UAS800 0.216 0.537 0.970 0.000 0.315 0.344 0.484 0.385 0.774 0.063 0.359 0.355 0.727 0.000 0.052 0.200 0.271 0.170 0.081 0.583 0.007
Iris iQ®200SPRINT 0.230 0.902 0.000 1.125 0.020 0.719 0.000 0.385 0.380 0.625 0.266 0.764 0.000 0.500 0.466 0.511 0.006 0.509 0.005 0.333 0.222
URiSCAN® PlusScope 0.088 0.341 0.054 0.375 0.536 0.281 0.526 0.154 0.790 0.188 0.069 0.227 0.295 0.125 0.037 0.267 0.130 0.208 0.551 0.444 0.094
Bacteria (gradingb)
UF-5000 0.037 – – – – 0.063 0.668 0.077 0.288 0.000 0.668 0.018 0.453 0.000 0.762 0.022 0.276 0.019 0.263 0.083 0.397
Cobas® u 701 0.404 1.780 0.000 1.125 0.027 0.625 0.098 1.154 0.005 0.438 0.934 1.127 0.000 0.875 0.018 1.044 0.000 1.019 0.000 0.361 0.931
UAS800 0.548 1.780 0.000 1.125 0.198 0.844 0.061 1.231 0.013 0.688 0.524 1.236 0.000 1.250 0.008 1.156 0.000 1.170 0.000 0.722 0.136
Iris iQ®200SPRINT 0.045 0.585 0.000 0.125 0.929 0.156 0.259 0.462 0.037 −0.063 0.393 0.318 0.001 0.000 0.878 0.333 0.080 0.283 0.120 0.333 0.113
URiSCAN® PlusScope 0.048 0.122 0.111 0.125 0.459 0.000 0.264 0.000 0.715 0.125 0.297 0.073 0.416 0.250 0.016 0.111 0.148 0.057 0.681 0.028 0.851
RBC, red blood cell; WBC, white blood cell; p/HPF, particles per high-power field. Bold font indicates statistical significance (p-value < 0.05). aThe results for RBC and WBC counts were
categorized into 6 ranges as follows: 0–2/HPF, 3–5/HPF, 6–10/HPF, 11–20/HPF, 21–50/HPF and >50/HPF. These categories were then converted to the corresponding numbers (from 0 to 5).
b
For bacteria, the results were expressed as ‘Negative’, ‘1 + ’, ‘2 + ’, and ‘3 + ’, and these semi-quantitative grades were converted to the corresponding numbers (from 0 to 3). A Mann-Whitney
test was performed to compare the median grading values between total samples and specific samples with possible interferents.
Cho et al.: Comparison of five automated urine analyzers
clumped cells or misidentification of particles could the 701 showed relatively higher sensitivity for casts than the
cause of bias [3, 12, 21, 26]. There were several limitations other three analyzers. However, these analyzers showed
in the microscopic quantitative results of RBC, WBC and low positive predictive values (PPVs) and kappa values
SQEP in our study. First, we did not consider the differ- due to high false-positive rates [3, 9, 27]. In contrast, the
ences of fields between a digital camera and manual UF-5000 showed relatively lower sensitivity, although the
microscopy [4]. Second, we counted the cells in ten small sensitivity for casts can be altered depending on cut-off
grids of KOVA chamber according to instructions from values. The universal cut-off for casts has not yet been
KOVA International Inc. However, counting the entire established for the UF-5000. For UF-1000i, a previous
chamber would have been better for quantitation accu- model from Sysmex Corp., many laboratories use 1.0/μL
racy. Additionally, we did not verify the conversion factors or 1.5/μL as cut-off values. We used 1.5/μL for hyaline
(from particles/HPF to particles/μL), but rather used the casts and 1.0/μL for pathologic casts as cut-off values for
values provided by each manufacturers. So, further study UF-5000 in this study. Additionally, upon applying the
may be needed in order to verify the conversion factors. cut-off of 0.53/μL for hyaline casts and 0.23/μL for patho-
We evaluated the concordance of qualitative results logic casts as suggested by a recent study [21], the sensitiv-
from five urine sediment analyzers with results from ity increased from 23.1% to 57.7% for hyaline casts (data
manual microscopy for crystals, hyaline casts, pathologic not shown), and from 23.6% to 56.4% for pathologic casts
casts, yeast, sperm and mucous threads. Although the (Supplementary Table 3). However, increased sensitiv-
clinical importance of these sediments is negligible when ity was accompanied with a rise image review rates and
present in small amounts, this study focused on analyti- false-positive rates. For NECs, the UF-5000 and Cobas®
cal performance according to their individual cut-offs. For u 701 showed relatively high sensitivity, but their PPV
crystals, the Cobas® u 701 and Iris iQ®200SPRINT showed and kappa values were low. A previous study reported
high sensitivity (over 60%) but with relatively low speci- that renal tubular epithelial cells look like WBCs, which
ficity. It has been reported that some of the false-positive may have caused confusion [11]. For yeast, the UF-5000,
results for crystals found in image based analyzers may be UAS800 and Cobas® u 701 identified the hyphae well. For
due to their misclassification as other types of particles, sperm, the Cobas® u 701 showed the highest sensitivity.
such as dysmorphic erythrocytes and yeast, and some The UF-5000 and UAS800 also showed numeric results
of the false-negative results may be due to specific crys- over 0, but still below the cut-off value, while sperms were
tals such as calcium oxalate monohydrate being similar observed using manual microscopy. Therefore, adjust-
to erythrocytes [7, 21, 27]. Sysmex UF-5000 showed rela- ment of cut-offs for sperm is recommended. For mucous
tively low detection rates for crystals because UF-5000 threads, four out of the urine sediment analyzers (except
uses a special reagent to intentionally dissolve amor- for the UF-5000), showed 70% ~ 90% sensitivity. This may
phous crystals. Therefore, when we analyzed diagnostic be due to the difference in principle between flow cytom-
performance (excluding amorphous urate and amorphous etry-based methods and digital image-based approaches.
phosphate), we found that sensitivity for crystals was The performance of the Cobas® u 701 and
increased, which was comparable with other image-based UAS800 showed high sensitivity for crystals and casts
analyzers. However, sensitivity for triple phosphate crys- and was usually also accompanied with high-false posi-
tals was lower in UF-5000 than image-based analyzers, tive rates. Previous studies have also reported that digital
while detection of calcium oxalate and uric acid crystals image-based systems show higher false-positive rates than
showed similar sensitivity between UF5000 and image- flow cytometry-based systems [3, 5, 27]. Contrary to flow
based analyzers. Our study is the first to evaluate detec- cytometry-based systems, digital image-based systems
tion rates of each specific kind of crystal. We found that have shown a greater tendency to make false positive calls
there were some differences in the sensitivity for each in the presence of mucus, fibers or other contaminants
crystal among the five sediment analyzers. Sensitivity [3, 21]. In general laboratory practice for reporting urine
ranged from 50.0% to 75.0% for calcium oxalate, from sediment results, laboratory technicians confirm results
0.0% to 53.8% for triple phosphates, and from 12.5% to by additional image review, or by microscopic review,
43.8% for uric acid crystals. when urine sediment analyzers identify the existence of
It is difficult to accurately distinguish casts from crystals and casts. Therefore, false positive identifications
other sediments. Aggregation of other sediments might can be filtered and/or confirmed with time and effort, by
make them look like casts, or existing casts may be lysed either medical technologists, or by clinical pathologists,
or dissolved so as to be below the detectable range in and reliable results can eventually be reported. From this
sample processing process [3]. The UAS800 and Cobas® u point of view, methods with high sensitivity for crystals
and casts might be preferred. However, the preference the accurate identification of small particles are needed,
can vary, depending on perspectives on the efficiency and which would facilitate the achievement of high sensitivity
clinical importance of urinary crystals and casts. and specificity for each sediment.
Among the parameters measured using urine sedi- In conclusion, for RBCs and WBCs, all five analyzers
ment analyzer, bacteria were excluded in the diagnostic showed excellent performance, and use of automated
performance analysis such as sensitivity and specific- urine chemistry and sediment analyzers can replace tra-
ity. As urine culture is the reference method rather than ditional manual microscopy. However, the Cobas® u 701
manual microscopy, additional study comparing bacteria and UAS800, which are based on digital image-based
result from urine sediment analyzers with conventional systems, showed high false positive rates in detecting bac-
urine culture is needed. In our following study, we are teria, while the UF-5000, based on flow cytometry, showed
planning to compare diagnostic performance for bacteria better performance than any of the other platforms for
in these five automated urine sediment analyzers using bacteria. In contrast, the sensitivity of the Cobas® u 701
urine bacterial culture as a reference method. and UAS800 for pathologic casts and crystals was high,
In addition, we evaluated image review and micro- although image review rates were also high. The detection
scopic review rates. The Cobas® u 701 and UAS800 showed rate for crystals and casts, and review rates can be changed
higher review rates than other instruments, which reciprocally according to cut-offs for the UF-5000. Auto-
resulted mostly from flagging frequencies for crystals mated urine analyzers are expected to reduce the burden
and casts. The UF-5000 showed relatively lower review of manual processing, with reliable results. However, each
rates when higher cut-off values were applied for casts. automated urine sediment analyzer has certain distinct
However, after applying a cut-off of 0.23/μL, the flagging features. Therefore, laboratory directors are encouraged
rate for the pathologic casts increased from 5.1% to 17.9%, to understand these features, and use each system in an
and total image review rates rose from 8.6% to 23.0% (data appropriate way, considering clinical algorithms and lab-
not shown), similar to those of the UAS800 and Cobas® oratory workflows.
u 701. The Iris iQ®200SPRINT, the APR™ system auto-
matically sorts and classifies each of the images taken. Author contributions: All authors have accepted respon-
However, in some cases, sediments identifiable by image sibility for the entire content of this manuscript and have
review were classified as “UNCC” or “UNCX”. If the opera- approved its submission.
tor reviews the results screen and marks each particle to Research funding: This work was supported by five com-
be verified or moved to other parameter category, sensitiv- panies – Sysmex Corporation (Kobe, Japan), Roche Diag-
ity for crystals and casts can be significantly enhanced. nostics International (Rotkreuz, Switzerland), Siemens
However, the Iris iQ®200SPRINT had higher microscopic Healthineers (Erlangen, Germany), Beckman Coulter
review rates because the operator occasionally could not (Brea, CA, USA), and YD diagnostics (Yongin, Korea).
identify the particle on the screen from the image review. Employment or leadership: None declared.
Likewise, the accuracy for the Cobas® u 701 and UAS 800 Honorarium: None declared.
can be increased by modifying (correcting) images of Competing interests: The funding organizations played
the recognized particles, but we evaluated only unmodi- no role in designing the study, collection, analysis and
fied results which were automatically reported by the interpretation of data, writing of the report, or in the deci-
manufacturer’s image processing software. And, because sion to submit the findings for publication.
we evaluated a prototype of the URiSCAN® PlusScope,
the automatic image processing software was not the
final version. Therefore, the performance of URiSCAN®
PlusScope should be confirmed in future studies using the
References
final released product.
1. Carlson DA, Statland BE. Automated urinalysis. Clin Lab Med
In interference studies, the levels of RBC counts were 1988;8:449–61.
influenced by crystals, especially by calcium oxalate 2. Riley RS, McPherson RA. Basic examination of urine. In:
monohydrate crystals, and by yeast with single or small McPherson RA, Pincus MR, editors. Henry’s clinical diagnosis
hyphae. The level of bacteria detected is also influenced and management by laboratory methods. St. Louis, MO: Elsevier
Saunders, 2017:442–80.
by the presence of crystals, especially by amorphous phos-
3. Lee W, Ha JS, Ryoo NH. Comparison of the automated cobas u
phates and urates, and by casts. As mentioned already, 701 urine microscopy and UF-1000i flow cytometry systems and
digital imaged-based systems confuse small particles manual microscopy in the examination of urine sediments. J Clin
with bacteria. Therefore, further technical advances for Lab Anal 2016;30:663–71.
4. Cho E-J, Ko D-H, Lee W, Chun S, Lee HK, Min W-K. The efficient 16. Ko DH, Ji M, Kim S, Cho EJ, Lee W, Yun YM, et al. An approach
workflow to decrease the manual microscopic examination of to standardization of urine sediment analysis via sugges-
urine sediment using on-screen review of images. Clin Biochem tion of a common manual protocol. Scand J Clin Lab Invest
2018;56:70–4. 2016;76:256–63.
5. Haber MH, Blomberg D, Galagan KA, Glassy EF, Ward PC. Color 17. Jiang T, Chen P, Ouyang J, Zhang S, Cai D. Urine particles
atlas of the urinary sediment: an illustrated field guide based on analysis: performance evaluation of Sysmex UF-1000i and
proficiency testing. Northfield, IL: College of American Patholo- comparison among urine flow cytometer, dipstick, and visual
gists, 2010. microscopic examination. Scand J Clin Lab Invest 2011;71:30–7.
6. Ben-Ezra J, Bork L, McPherson RA. Evaluation of the Sysmex 18. Delanghe J, Speeckaert M. Preanalytical requirements of urinaly-
UF-100 automated urinalysis analyzer. Clin Chem 1998;44:92–5. sis. Biochem Med (Zagreb) 2014;24:89–104.
7. Altekin E, Kadicesme O, Akan P, Kume T, Vupa O, Ergor G, et al. 19. Langlois MR, Delanghe JR, Steyaert SR, Everaert KC, De Buyzere
New generation IQ-200 automated urine microscopy analyzer ML. Automated flow cytometry compared with an automated
compared with KOVA cell chamber. J Clin Lab Anal 2010;24:67–71. dipstick reader for urinalysis. Clin Chem 1999;45:118–22.
8. Lamchiagdhase P, Preechaborisutkul K, Lomsomboon P, 20. Bakan E, Ozturk N, Baygutalp NK, Polat E, Akpinar K, Dorman E,
Srisuchart P, Tantiniti P, Khan-u-Ra N, et al. Urine sediment et al. Comparison of Cobas 6500 and Iris IQ200 fully-automated
examination: a comparison between the manual method and urine analyzers to manual urine microscopy. Biochem Med
the iQ200 automated urine microscopy analyzer. Clin Chim Acta (Zagreb) 2016;26:365–75.
2005;358:167–74. 21. Previtali G, Ravasio R, Seghezzi M, Buoro S, Alessio MG. Per-
9. Chien TI, Kao JT, Liu HL, Lin PC, Hong JS, Hsieh HP, et al. Urine formance evaluation of the new fully automated urine particle
sediment examination: a comparison of automated urinalysis sys- analyser UF-5000 compared to the reference method of the
tems and manual microscopy. Clin Chim Acta 2007;384:28–34. Fuchs-Rosenthal chamber. Clin Chim Acta 2017;472:123–30.
10. Delanghe JR, Kouri TT, Huber AR, Hannemann-Pohl K, Guder WG, 22. Lewis MA, editor. CLSI Document GP16-A3. Urinalysis; approved
Lun A, et al. The role of automated urine particle flow cytometry guideline, 3rd ed. Wayne, PA: Clinical and Laboratory Institute
in clinical practice. Clin Chim Acta 2000;301:1–18. (CLSI), 2009.
11. Zaman Z, Fogazzi GB, Garigali G, Croci MD, Bayer G, Kranicz T. 23. Ottiger C, Huber AR. Quantitative urine particle analysis:
Urine sediment analysis: analytical and diagnostic performance integrative approach for the optimal combination of automation
of sediMAX – a new automated microscopy image-based urine with UF-100 and microscopic review with KOVA cell chamber.
sediment analyser. Clin Chim Acta 2010;411:147–54. Clin Chem 2003;49:617–23.
12. Wah DT, Wises PK, Butch AW. Analytic performance of the iQ200 24. KOVA International. KOVA® GLASSTIC® SLIDE 10 WITH GRIDS
automated urine microscopy analyzer and comparison with – Instructions for use. https://www.kovaintl.com/downloads/
manual counts using Fuchs-Rosenthal cell chambers. Am J Clin DI-91064-17-final.pdf. Accessed: 19 Dec 2018.
Pathol 2005;123:290–6. 25. Wang J, Zhang Y, Xu D, Shao W, Lu Y. Evaluation of the Sysmex
13. Chien TI, Lu JY, Kao JT, Lee TF, Ho SY, Chang CY, et al. Compari- UF-1000i for the diagnosis of urinary tract infection. Am J Clin
son of three automated urinalysis systems – Bayer Clinitek Pathol 2010;133:577–82.
Atlas, Roche Urisys 2400 and Arkray Aution Max for testing 26. Budak YU, Huysal K. Comparison of three automated systems
urine chemistry and detection of bacteriuria. Clin Chim Acta for urine chemistry and sediment analysis in routine laboratory
2007;377:98–102. practice. Clin Lab 2011;57:47–52.
14. Hannemann-Pohl K, Kampf SC. Automation of urine sediment 27. Ince FD, Ellidag HY, Koseoglu M, Simsek N, Yalcin H, Zengin
examination: a comparison of the Sysmex UF-100 automated flow MO. The comparison of automated urine analyzers with manual
cytometer with routine manual diagnosis (microscopy, test strips, microscopic examination for urinalysis automated urine analyz-
and bacterial culture). Clin Chem Lab Med 1999;37:753–64. ers and manual urinalysis. Pract Lab Med 2016;5:14–20.
15. Wesarachkitti B, Khejonnit V, Pratumvinit B, Reesukumal K,
Meepanya S, Pattanavin C, et al. Performance evaluation and
comparison of the fully automated urinalysis analyzers UX-2000 Supplementary Material: The online version of this article offers
and Cobas 6500. Lab Med 2016;47:124–33. supplementary material (https://doi.org/10.1515/cclm-2019-0211).