a new automated system for urine analysis: a simple, cost-effective and reliable method for...

BJU International (1999), 84, 454–460

A new automated system for urine analysis: a simple, cost-eVective and reliable method for distinguishing betweenglomerular and nonglomerular sources of haematuriaR.N. KORE, C.S. DOW* and K.M. DES AIDepartment of Urology, Walsgrave Hospitals NHS Trust, and *Department of Biological Sciences, University of Warwick, Coventry,UK

Objective To determine the ability, reliability and accu- to have any detectable pathology in the urinary tracton clinical examination and investigations. Theracy of a new automated system of urine analysis in

diCerentiating glomerular from nonglomerular bleed- remaining 30 patients were diagnosed to have uro-logical or nephrological conditions with no haema-ing in the initial investigation of haematuria, and

compare its eBcacy with conventional phase-contrast turia. In the positive group, 20 (29%) patients werefrom the glomerular group, with a mean (range) redmicroscopy (PCM).

Patients and methods One hundred and six urine samples blood cell size of 4.25 (4–5.1) mm, and 49 (71%) fromthe nonglomerular group, with red blood cells offrom patients in whom the final diagnosis was avail-

able were analysed using electrical flow impedance to 5.47 (4.67–5.70) mm. These ranges overlapped at4.67–5.1 mm; at the decision threshold of 4.75 mm,detect, enumerate and size red blood cells in a conduc-

tive fluid (the cellfacts analyser, Microbial Systems Ltd, the distribution of dysmorphic and eumorphic redblood cells for the glomerular group was 18 (90%)Coventry, UK). All the samples were also tested using

a dipstick method and PCM was carried out for and two (10%), respectively, and for the nonglomeru-lar group was 2 (4%) and 47 (96%), respectively. Thecomparison on 45 of the 106 urine specimens. The

results of cellfacts analysis were correlated with the sensitivity, specificity, PPV and NPV were 90%, 96%,90% and 96%, respectively. Consumable and labourfinal diagnoses to assess sensitivity, specificity, positive

predictive value (PPV), and negative predictive value costs were very low.Conclusions Cellfacts analysis is a simple, rapid, objective(NPV) of this method; the costs were also analysed.

Results Sixty-nine urine samples tested positive for blood and cost-eCective method for diCerentiating glomerularfrom nonglomerular urinary red blood cells, especiallyon dipstick urine analysis and all these were confirmed

to have red blood cells on cellfacts analysis. The when few such cells are present.Keywords Glomerular, nonglomerular, haematuria,remaining 37 samples were negative for blood on

dipstick testing and cellfacts analysis, although seven cellfacts analysis, phase contrast microscopy, red bloodcell sizepatients had been referred with previously detected

microscopic haematuria, none of whom were found

of red blood cells in urine has previously been investi-Introduction

gated by phase contrast microscopy (PCM) and bright-field microscopy [4,5]. However, the accuracy andAsymptomatic microscopic haematuria is common; com-

munity-based studies have found prevalence rates of reliability of these methods have been challenged [6].Even with many red blood cells, the results have been2–5% in middle-aged men undergoing private health

assessment [1], and up to 22% in those aged over equivocal [7]. These techniques are essentially operator-dependent and therefore open to potential errors in60 years [2]. A recent study showed that 26% of such

men may have significant pathology [3]. The source of interpretation. Furthermore, they are labour intensiveand time-consuming.red blood cells in urine could be either glomerular or

not and the ability to distinguish these initially would Glomerular red blood cells are known to be smallerthan their nonglomerular counterparts and this propertyobviously aid in selecting further investigations to ident-

ify the exact cause of the haematuria. The morphology was first used by Shichiri et al. [8] as an objectiveparameter of comparison when they used the Coultercounter S-plus II method to measure red cell volume.Accepted for publication 6 May 1999

454 © 1999 BJU International

AUTOMATED URINE ANALYSIS 455

Other investigators [9] also used this principle success- negative predictive value (NPV) of this method, andcompared its overall eBcacy with PCM.fully, obtaining encouraging results. Although the pres-

ence of debris in urine was considered to undermine theaccuracy of red cell volume analysis [10], it was shown

Patients and methodsthat glomerular and nonglomerular patterns were dis-tinct from that of debris [7]. Nevertheless, the technical The study included 106 patients (65 men and 41

women, mean age 63.5 years, range 20–94) who pro-diBculties associated with Coulter counter analysis insamples with few red cells have prevented this technique vided urine specimens. The size of any red blood cells

were determined using an automated cellfacts urinefrom being used widely in clinical practice.We describe a new method of identifying the type of screening analyser (Microbial Systems Ltd, Coventry,

UK). Fresh urine samples were collected from patientsred cells in urine, using an automated instrument thatuses the principle of electrical flow impedance to detect, attending The Walsgrave Hospitals NHS Trust, either as

outpatients at the urology and nephrology clinics, or asenumerate and size particles in a conductive fluid.As particles pass through a 30×80 mm orifice they dis- inpatients on the urology wards.

Dipstick urine analysis and cellfacts analysis wereplace their volume of electrolyte, thus producing avoltage pulse between electrodes (Fig. 1). This voltage carried out on each specimen, while PCM was carried

out on 45 samples. After an initial dipstick examinationpulse is directly proportional to the volume of particle,irrespective of its shape. The principal advantage is of each sample, they were transported in individual

bottles containing boric acid to the Department ofthat in this instrument the orifice does not becomeblocked. The entire analytical range of the orifice is Biological Sciences at the University of Warwick for

cellfacts analysis. Where appropriate, the samples werecovered, i.e. 0.4–12 mm (equivalent spherical diameter).Consequently it is possible to analyse urine samples for divided into two aliquots, one of which was sent for

PCM, the latter being carried out by staC with thethe presence of bacteria, erythrocytes and leukocytes ina single analysis (Fig. 2). To ensure the validity of the required experience in the technique of PCM. The

investigators were unaware of the diagnosis at thedata, the instrument performs a self-checking procedurebefore and after each analysis. The data (flow rate, outset.

For red cell analysis, 25 mL of urine was aspirated intocalibration, orifice current and temperature), togetherwith operator identification and time of analysis, are a pipette and dispensed into the cellfacts analyser, which

then initiated a testing cycle. The analyser measured allautomatically stored and available for quality control/assurance purposes. In this study, we analysed the the particles in the sample by automatically diluting it

in a conductive solution and passing it through a smallsensitivity, specificity, positive predictive value (PPV) and

80mm

30mm

Samplepluselectrolyte

Fluidflow

Volume

Nu

mb

er

Fig. 1. A diagrammatic representation of the principle of electrical flow impedance. The resultant data is plotted as particle numberagainst particle volume.

© 1999 BJU International 84, 454–460

456 R.N. KORE et al.

Fig. 2. Particle distribution profiles fromfour clinical urine samples. Profiles in therange 0.4–3.5 mm represent bacteria.Samples 1432096A (green) and H132296(light red) are bacteriologically negative;sample H632296 (red) and H132096 (lightgreen) are bacteriologically positive. Profilesin the range 4.0–10.0 mm representerythrocytes (left arrow) and leukocytes(right arrow). All of these data are from a

Diameter (µm)

5.35

Co

un

ts (

mL)

5.00E + 03

4.50E + 04

0.00E + 00 4

6.24

6.92

4.77

5.83 6.6

7.22 7.5

7.75

7.99

8.22

8.43

8.64

8.83

9.02 9.2

9.37

1.00E + 04

1.50E + 04

2.00E + 04

2.50E + 04

3.00E + 04

3.50E + 04

4.00E + 04

Diameter (µm)

1.52

Co

un

ts (

mL)

5.00E + 05

3.50E + 06

0.00E + 00 0

1.92

2.19

1.21

1.74

2.06

2.31

2.41

2.51 2.6

2.68

2.76

2.84

2.91

2.98 3.1

3.16

1.00E + 06

1.50E + 06

2.00E + 06

2.50E + 06

3.00E + 06

3.04

b

a

single analysis of each sample.

orifice in a test cell. All particles were counted and removed and the pellet re-suspended in PBS to give a10-fold concentration. This concentrated sample wastheir size measured. The distribution of the cells and

other particles was displayed in diCerent channels then re-analysed; such centrifugation was required in51 samples.(0.4–10.0 mm equivalent spherical diameter) rep-

resenting increasing size along the abscissa and the The clinical diagnosis was then obtained for all thepatients and they were subsequently categorized asnumber of particles on the ordinate. The raw data were

analysed by computer and the results displayed in a having glomerular or nonglomerular haematuria. Thepatients’ diagnoses were correlated with the size of thecolour-coded graphical format for each sample of urine

analysed (Fig. 3). The peak of the curve represented the red blood cells in their urinary sediment. The ranges ofcell size for the two groups were defined; five decisionmodal value of the size of the particle under consider-

ation. At the end of the test, an automatic wash-cycle thresholds were selected for this purpose. All sampleswith red blood cells smaller than the decision thresholdswas run to prepare for the next sample. If the first test

failed to produce any conclusive results, the sample was were termed dysmorphic and all those larger were termedeumorphic. Based on the selected decision levels, thefurther concentrated in the following way: the sample

was centrifuged at 1600 g for 10 min, the supernatant distribution of dysmorphic and eumorphic samples was



allow epithelial cells to pass through and they weredetected as a characteristic noise signal. The range forred blood cell size was 4.0–5.7 mm. Figure 4 shows thedistribution of dysmorphic and eumorphic red bloodcells; 20 (29%) of patients were in the glomerular group,with a mean (range) red blood cell size of 4.25(4–5.1) mm, and 49 (71%) in the nonglomerular groupwith red blood cells of 5.47 (4.67–5.70) mm. The overlapin range was 4.67–5.1 mm. The decision levels chosenwere 4.6, 4.75, 4.9, 5.0 and 5.15 mm. A ROC curve wasconstructed for these values (Fig. 5). At the decision levelof 4.75 mm the distribution of dysmorphic and eumorphicred blood cells for the two groups is given in Table 2,with the sensitivity, specificity, PPV and NPV at thisdecision level.

The instrument can be configured (in software) toscreen urine specimens for the presence of bacteria, redand white blood cells; the cellular content of cerebro-spinal fluid can similarly be analysed. The capital costFig. 3. The cellfacts urine screening analyser (courtesy of Microbialof the instrumentation is £18 000–20 000, dependingSystems Ltd, UK).

on instrument configuration, and the value wouldnormally be expected to depreciate over 5 years.Service/maintenance costs are as expected for industrialassessed. A ROC curve was constructed and the sensi-

tivity, specificity, PPV and NPV calculated, with their instrumentation, i.e. 7.5–10% of the purchase price.The operational characteristics are simple and do not95% CIs calculated using the method of Conover [11].

The costs of the system were also analysed. require highly skilled medical laboratory staC. It istherefore cost-eCective, not only in absolute timerequired per sample but also in the cost of the operator.

ResultsThe overall cost per analysis was £0.04 (£0.02 for adisposable pipette tip and £0.02 for electrolyte per test).Sixty-nine (65%) patients tested positive for blood on

dipstick urine analysis and all 69 were confirmed to Although the initial capital cost of the instrument mayappear prohibitive, there would be considerable long-have red blood cells on subsequent cellfacts analysis; 37

patients tested negative by dipstick and none of these term savings in terms of operators (skilled labour) andconsumable costs. Even greater savings may be envis-had detectable red blood cells on cellfacts analysis. Of

these 37 patients, seven had been referred with pre- aged by avoiding unnecessary cystoscopies in thoseidentified as having a glomerular source of bleeding atviously detected dipstick haematuria and all were eventu-

ally found to have no abnormality either on clinical the outset.examination or appropriate investigations. The remain-ing 30 patients had been referred for other reasons and

Discussionhad a negative urine analysis. As the principal aim ofthis study was to examine the utility of the instrument Haematuria is one of the most common conditions seen

by urologists and nephrologists. Some studies estimatein diCerentiating glomerular from nonglomerular cells,only those samples testing positive for haematuria were that haematuria accounts for 20% [12] of all urological

visits and 4–13.6% [12,13] of inpatient admissions. Thisevaluated further. Of the 45 specimens assessed usingPCM, only 22 (48%) had red blood cells and in only clinical presentation is caused by many conditions span-

ning a wide spectrum of genitourinary pathology. Thethree of these could significant dysmorphic cells beconfidently identified. The mean (range) time taken for definition of significant haematuria remains vague.

Estimates of the normal red blood cell count vary signifi-a PCM study was 10 (3–21) min.Table 1 shows the clinical diagnoses, sex distribution, cantly. In an earlier study, a total count of 1 million

was reported as normal (the Addis number) [14]. Bothcellfacts analysis of red cell size (modal value) andpositive results, i.e. presence of dysmorphic cells on PCM, bright-field microscopy and PCM used for cell counting

are labour-intensive, time-consuming and subjective.in both groups, respectively. The modal position ofbacteria and white blood cells was 0.5–2.0 mm and Currently used reagent strips are highly sensitive in

detecting occult haematuria and their use in screening6.0–10.0 mm, respectively (Fig. 2). The orifice did not



Table 1 Distribution of patients from the glomerular and nonglomerular groups

Diagnosis No. Sex (M/F) Cell size (mm) PCM +ve/tested*

GlomerularIgA nephropathy 5 3/2 4.05–4.36 2/4Mesangial proliferative 3 2/1 4.23–4.50 0/3

glomerulonephritisMembranous nephropathy 3 2/1 4.12–4.56 0/2Vasculitis 3 2/1 4.16–4.90 0/2Focal glomerulosclerosis 2 1/1 4.18,5.1 0/2Streptococcal glomerulonephritis 2 1/1 4.0,4.17 0/2Minimal change nephritis 2 1/1 4.08 0/2

NonglomerularCarcinoma of the bladder 17 11/6 5.65–5.7Urinary tract infection 8 2/4 4.81–5.41Calculus 7 5/2 5.65–5.7After TURP 4 4/- 5.44–5.69Stricture 4 4/- 5.0–5.61RCC 5 3/2 4.79–5.46Long-term catheter 3 3/0 4.67–5.35Vesicocolic fistula 1 0/1 5.49

*PCM+ve, glomerular (dysmorphic) cells diagnosed; tested, number of specimens examined.

Fig. 4. The distribution of dysmorphic andeumorphic red blood cells. SampleV01056A (green) has a red blood cell size of4.1 mm (dysmorphic) and sample 59987A(red) a red blood cell size of 5.6 mm3.81

Equivalent spherical diameter (µm)4.31 4.71 5.06 5.36 5.64 5.88 6.11 6.33 6.433.50

Co

un

ts

50

100

150

200

250

300

350

400

450

04.07 4.52 4.89 5.22 5.50 5.76 6.00 6.22

(eumorphic).

has resulted in an ever more referrals to urologists and of investigations may outweigh the potential benefit ofearly diagnosis, particularly in younger patients.nephrologists [15].

In the present study, 69 urine specimens were positive Typically, investigations for renal parenchymal diseaseinclude serology for antinuclear cytoplasmic antibody,for haematuria on initial dipstick testing, confirmed in

all by subsequent cellfacts analysis. As 2–22% of patients antiglomerular basement membrane antibody, C3 andC4 complements, radiological assessment with ultrason-with asymptomatic occult haematuria may have signifi-

cant nephro-urological disease [16,17], it has been ography and not infrequently, a renal biopsy. Urologicalinvestigations would normally include urine culture andargued that all such patients should be comprehensively

investigated. Equally, however, the cost and morbidity cytology, IVU and/or ultrasonography, followed by



were positive for red blood cells by PCM, suggesting ithas low sensitivity. Only in 6% of the samples coulddysmorphic cells be confidently diagnosed, giving a verylow diagnostic yield.

Encouraging results have been reported using flowcytometry as a means of localising the source of urinaryerythrocytes [22], but specimens with few red cells werediscounted, therefore raising doubt about the applica-bility of this method in routine practice, especially aseven minor degrees of haematuria may be associatedwith significant pathology. In the present study, sampleswith few red cells were not excluded.

Shichiri et al. [8], in a small study of 12 patients, usedthe Coulter Counter S-plus II as an autoanalyser andshowed large diCerences in erythrocyte size distribution,

5

Specificity

10 15 20 25 30 35 40 45 500

Sen

siti

vity

10

20

30

40

50

60

70

80

90

100

0

4.75 µm

with glomerular cells being of smaller volume. AnotherFig. 5. The ROC curve of the decision thresholds (4.75 mm gave study [7] found red cell volume analysis to be completelythe best sensitivity and specificity).

accurate in localizing haematuria when compared withPCM, which had a 10% false-negative rate. Debris wasnot found to be a confounding variable, unlike in some

Table 2 The test results, sensitivity, specificity, PPV, NPV, using studies [10]. Using cellfacts analysis, the results werea decision threshold of 4.75 mm comparable.

The sensitivity, specificity, PPV and NPV were optimalDiagnostic group/test Glomerular Nonglomerular

at a decision threshold cell size of 4.75 mm, as determinedby the ROC. Other criteria may be applied [23] (e.g. toDysmorphic 18 2minimize the number of false-negatives) to select aEumorphic 2 47decision level. The present level was selected to produceTest (%, 95% CI)

Sensitivity 90 (68.3, 98.8) the fewest false-positive and false-negative results, andSpecificity 96 (86.0, 99.5) that maximized the overall accuracy of the test.PPV 90 (68.3, 98.8) In using the present technique, two potential sourcesNPV 96 (86.0, 99.5)

of variability need to be considered. The first is theinfluence of sample osmolality and pH; as all the presentsamples were rapidly transferred into boric-acid con-tainers after the initial dipstick examination, this wouldcystoscopy. Before subjecting patients to the more invas-

ive investigations, it would clearly be advantageous to have eliminated any potential bias through diCerencesin pH. Furthermore, urinary erythrocyte morphologyknow the source of the red blood cells in urine at an

early stage. has been shown to change to a glomerular pattern onlywhen immersed in a hyperosmolar saline solution, pro-Although a careful search for casts in the urinary

sediment may provide evidence of a glomerular lesion, ducing osmolalities of > 800 mosm/L [7], a situationthat would be rare in clinical practice. In the same study,these particles usually occur in low numbers [4] and are

absent in 20% of cases [18]. PCM detects morphological immersion in urea, which is usually the main osmoticallyactive urinary solute, did not significantly change redchanges in red blood cells as a way of distinguishing a

glomerular from a nonglomerular origin. However, some cell morphology, even at osmolalities of > 400 mosm/L.The second factor is whether the time elapsed betweeninvestigators have not found a close correlation between

red cell morphology and diagnosis; an interobserver sample collection and cellfacts analysis is critical. Mostof the present samples were analysed within an hour ofdiCerence in interpretation was reported on 38% of

occasions in a blinded controlled trial of PCM [9]. collection, and delays of up to 24 h should have no eCectproviding the specimens are stored in boric acid.Furthermore the morphological appearances were not

specific, as dysmorphic cells could also be found in renal However, we have not formally addressed this questionin the present study and it requires further evaluation.tumours [19] and after renal biopsy, in patients with no

parenchymal disease [20]. In a recent study, the renal Although the capital cost of the equipment was rela-tively high, the analysis of the costs showed that consum-origin of the urinary red blood cells was found to

correlate with histological findings in only 33% of cases able and labour costs were minimal. The othercharacteristics, i.e. speed, simplicity, the ability to archive[21]. Only 22 samples (from 45) in the present study



in the diagnosis of glomerular haematuria. Lancet 1986;results and objectivity with minimal transcriptional error2: 781–2made this method a cost-eBcient and attractive option.

9 Banks R, Raynaulds S, Hanbury D. Identification of theDetecting the site of bleeding in the urinary tract issource of haematuria by automated measurement ofessential in selecting appropriate steps in the investig-urinary red cell volume. Abstracts of the XXV Congress ofative protocol of haematuria. Cellfacts analysis providesThe European Dialysis and Transplant Association, 5–8 Sept

a suitable screening test, correlating closely with dipstick1988, Madrid, Spain, 1988: 15

analysis for detecting microscopic haematuria; it is 10 Gibbs DD, Lynn KL. Red cell volume distribution curves insimple, fast and cost-eCective. The characteristics of this the diagnosis of glomerular and non-glomerular haema-method and its ease of use make it worth considering as turia. Clin Nephrol 1990; 33: 143–7an important first step in the investigation of haematuria, 11 Conover WJ. Practical Nonparametric Statistics, 2nd edn.

New York: Wiley, 1980especially when there are few red blood cells. However,12 Doss AK. Haematuria. Urol Cutan Rev 1947; 51: 676we emphasize that the present results require confir-13 Carter WC III, Rous SN. Gross haematuria in 110 adultmation in larger scale studies and that the information

urologic hospital patients. Urology 1981; 18: 342–4obtained from such screening should not be used in14 Addis T. The number of formed elements in urinaryisolation. If validated, this technique oCers considerable

sediments of normal individuals. J Clin Invest 1926;potential, not only in the initial investigation of haema-2: 409–15

turia but also, especially from a urological perspective,15 Woolhandler S, Pels RJ, Bor DH, Himmelstein DU, Lawrence

in deciding which patients have persistent or recurrent RS. Dipstick urinalysis screening of asymptomatic adultshaematuria in the absence of any identifiable ‘urological’ for urinary tract disorders I. Haematuria and proteinuria.disease, and thus confidently direct them to the JAMA 1989; 262: 1214–9nephrologist. 16 Thompson IM. The evaluation of microscopic haematuria:

population based study. J Urol 1987; 138: 1189–9017 Carson CC, Segura JW, Greene LF. Clinical importance of

Acknowledgement microhaematuria. JAMA 1979; 241: 149–5018 Fasset RG, Horgan BA, Mathew TH. Detection of glomerularWe thank Mr A.R.E. Blacklock, Mr M.I. Wills, Drs D.C.

bleeding by phase contrast microscopy. Lancet 1982;Dukes, M. Edmunds and R. Higgins for allowing us to1: 1432–4include their patients in this study.

19 Gyory AZ. Diagnostic value of dysmorphic urinary red cells(letter). New Eng J Med 1996; 335: 1323

20 Pollock C, Liu PL, Gyory AZ et al. Dysmorphism of urinaryReferences red blood cells value in diagnosis. Kidney Int 1989;

1 Ritchie CD, Bevan EA, Collier SJ. Importance of occult 36: 1045–9haematuria found at screening. Br Med J 1986; 292: 681–3 21 Favro S, Bonfante L, D’Angelo A et al. Is the red cell

2 Britton JP, Dowell AC, Whelan P. Dipstick haematuria and morphology really useful to detect the source of haema-bladder cancer in men over 60; results of a community turia? Am J Nephrol 1997; 17: 172–5study. Br Med J 1989; 299: 1010–2 22 Apeland T. Flow cytometry of urinary erythrocytes for

3 Messing EM, Young TB, Hunt VB et al. The significance of evaluating the source of haematuria. Scand J Urol Nephrolmicrohaematuria in men 50 or more years old: findings of 1995; 29: 33–7a home screening study using urinary dipsticks. J Urol 23 Pillsworth TJ, Haver VM, Abrass CK, Delaney1987; 137: 919–22 CJ. DiCerentiation of renal from non-renal haematuria by

4 Kesson AM, Talbott JM, Gyory AZ. Microscopic examination microscopic examination of erythrocytes in urine. Clinof urine. Lancet 1978; 2: 809–12 Chem 1987; 33: 1791–5

5 Birch DF, Fairley KF. Haematuria: glomerular or non-glomerular? (letter). Lancet 1979; 2: 845–6

Authors6 Venkat Raman G, Pead L, Lee HA, Maskell R. A blindcontrolled trial of phase contrast microscopy by two R.N. Kore, MS, MCh(Urol), FRCS(Urol), Formerly Senior

Registrar and Research Fellow, presently Fellow inobservers for evaluating the source of haematuria. Nephron1986; 44: 304–8 Endourology, Episcopal Hospital, Philadelphia, USA.

C.S. Dow, PhD, Senior Reader in Microbiology.7 de Caestecker MP, Hall CL, Basterfield PT, Smith JG.Localisation of haematuria by red cell analysers and phase K.M. Desai, ChM, FRCS, Consultant Urologist.

Correspondence: Mr K.M. Desai, Department of Urology,contrast microscopy. Nephron 1989; 52: 170–38 Shichiri M, Oowada A, Nishio Y, Tomita K, Shigai T. Use Walsgrave Hospital NHS Trust, CliCord Bridge Road, Coventry

CV2 2DX, UK..of auto analyser to examine urinary red cell morphology


a new automated system for urine analysis: a simple, cost-effective and reliable method for...

Documents