Introduction

Vesicoureteral reflux (VUR) is a relatively commoncondition with important sequelae if not detected, in-cluding reflux nephropathy, hypertension and renalfailure [1, 2]. Micturating cystography is indicated in all

infants with suspected VUR, despite normal postnatalultrasound [1, 3, 4, 5].

In accordance with standard practice, all neonates atour institution with antenatally detected renal dilatationare referred for micturating cystourethrography (MCU)and sonography at 6–8 weeks of age [3, 6]. Neonates

ORIGINAL ARTICLEPediatr Radiol (2002) 32: 853–858DOI 10.1007/s00247-002-0812-6

Rachael L. McEwing

Nigel G. Anderson

Sandra Hellewell

Julie Mitchell

Comparison of echo-enhanced ultrasoundwith fluoroscopic MCU for the detectionof vesicoureteral reflux in neonates

Received: 28 March 2002Accepted: 22 July 2002Published online: 25 September 2002� Springer-Verlag 2002

R.L. McEwing (&) Æ N.G. AndersonS. Hellewell Æ J. MitchellDepartment of Radiology, ChristchurchHospital, Christchurch, New ZealandE-mail: rachaelmcewing@hotmail.comTel.: +64-3-3513461Fax: +64-3-3552483

Abstract Background: Fluoroscopicmicturating cystourethrography(MCU) is used for screening andgrading of vesicoureteral reflux(VUR). It involves ionizing radia-tion. This study was designed toassess the efficacy of contrast-en-hanced sonography in predicting thepresence or absence of VUR.Objective: To compare an ultra-sound contrast agent for detection ofVUR in at-risk infants, and tocompare these findings with fluoro-scopic MCU with the aim of deter-mining whether echo-enhancedsonography could be used instead offluoroscopic MCU to identifyneonates who do not have VUR,thus avoiding the use of radiation inthis group. Materials andmethods: From August 1999 toAugust 2000, 97 neonates (69 male,31 female), aged 28–90 days (mean48 days), referred for MCU andrenal ultrasonography for investi-gation of VUR were recruitedconsecutively. Echo-enhancedsonography using stabilized micro-bubbles was followed immediately

by fluoroscopic MCU. VUR wasdiagnosed if transient hyperechoge-nicity appeared within the pelvical-yceal system or ureter. The meannumber of micturitions was 2.7(range 1–6). Results: Reflux wasdetected in 19 kidneys (14 babies) byone or other technique. The findingswere concordant in 181 kidneys(94.2%). Echo-enhanced sonographyhad a sensitivity of 64% (95%CI 35–87%), a specificity of 100%(95–100%), a positive predictive val-ue of 100% (66–100%), and a nega-tive predictive value of 94%(87–98%). Conclusions: The role ofecho-enhanced sonography is limitedat present in our neonatal populationas a screening examination. Its abilityto detect cases of high-grade refluxmay make it an attractive alternativein follow-up of known cases of VUR,and may help to reduce radiationexposure in this group.

Keywords Vesicoureteral reflux ÆCystourethrography Æ Kidney ÆUltrasound contrast agent

with additional risk factors for reflux, such as history ofreflux in a sibling or urinary tract infection are alsoevaluated. Direct radionuclide voiding cystography hasbeen advocated because of its lower radiation dose, butat the cost of poor definition of urinary morphology,and limited ability to grade VUR [1, 7]. There are ob-vious implications in exposing large numbers of neo-nates to ionizing radiation, and avoidance of exposure ina screening test for VUR is an important objective. Anyreplacement test for fluoroscopic MCU must be cost-effective both in price and time taken to perform thestudy, and have diagnostic accuracy similar to or betterthan fluoroscopic MCU at a lower radiation dose. Inparticular, it should have a high sensitivity and negativepredictive value.

The purpose of this study was to compare a prag-matic method of echo-enhanced sonography with fluo-roscopic MCU for the detection of VUR, and toevaluate whether it could be used to determine whichinfants need to proceed to fluoroscopic MCU so thationizing radiation exposure might be minimized in thisvulnerable population. As far as the authors are aware,this is the first such study to evaluate echo-enhancedsonography in a true screening population, in the neo-natal age group. We aimed to perform the examinationin as technically simple a manner as possible, so that itwould be accepted as a screening test, and could bereadily adapted to other busy paediatric radiology de-partments. Thus our aim was to use a pragmatic, re-producible, and economically realistic test.

Materials and methods

Enrolled into this prospective trial were 100 infants (69 male and 31female) ranging in age from 28 to 90 days (mean 48 days). Thestudy period was 1 year (August 1999 to August 2000). The indi-cations for the MCU in these neonates were fetal renal dilatationdetected on an antenatal scan (87; 63 male, 24 female), familyhistory of VUR (12; 5 male, 7 female) and first urinary tract in-fection (1 male). Infants older than 3 months were excluded, aswere infants with a prior radiological diagnosis of reflux, as wewished to study a cohort of infants presenting for their firstscreening test for VUR. At our institution, postnatal screening forVUR is performed if the fetal renal pelvis measures greater than orequal to 4 mm prior to 24 weeks gestation, or greater than or equalto 5 mm in the third trimester.

Three infants (two male, one female) were excluded after thevideotaped echo-contrast sonography was considered nondiag-nostic by the two principal investigators, who were blinded tothe fluoroscopic MCU result. The study was defined as nondi-agnostic if there was suboptimal visualization of one or bothkidneys, because of obscuring bowel gas or excessive crying andmovement of the patient. These problems were compounded ifthe collecting system was not visible. Our study group comprised97 neonates (67 male, 30 female) with 192 kidneys, as 2 hadunilateral renal agenesis. Two kidneys had duplex collectingsystems.

Routine ultrasound of the renal tract was initially performedaccording to our departmental protocol, with supine examinationof the bladder and distal ureters, and supine and prone views of

both kidneys and proximal ureters. Measurements of renal lengthand AP renal pelvis diameter were made. An Aspen ultrasoundmachine (Acuson, Mountain View, Calif.) with a 7-MHz paediatrictransducer was used in all cases. The bladder was then catheterizedtransurethrally by one of the authors (R.M.) with a 5F catheterusing aseptic technique and a urine specimen was obtained formicrobiological analysis. The sonographic echo-enhancing contrastwas prepared and instilled into the bladder.

Levovist (Schering, Berlin, Germany) is a preparation ofgranules initially designed for use in the vascular compartment,consisting of 99.9% galactose and 0.1% palmitic acid. The granulesare suspended in sterile water prior to use, shaken vigorously forabout 10 s, and then allowed to equilibrate for 2 min. During thisprocess some of the particles disaggregate, forming microbubbles,which are stabilized by the layer of palmitic acid [8, 9]. Levovist isavailable in 2.5-g and 4-g vials. In this study, largely for economicreasons, the 2.5-g vial was used. The contents were diluted to aconcentration of 300 mg/ml with sterile water, making 7 ml ofecho-contrast suspension of which 3.5 ml was injected gently intothe bladder, inducing a burst of hyperechogenicity which initiallyobscured the bladder. Prewarmed saline, elevated by a drip-standat a height of 1 m so as not to exceed physiological bladder pres-sures, was then infused into the bladder. This technique demon-strated the bladder well.

Scanning was commenced with the neonate in the supine po-sition, alternating between both kidneys, concentrating on the re-nal pelves during bladder filling and micturition. We aimed towitness at least two episodes of micturition in each child, but thiswas not possible in two (one episode of micturition only visual-ized). The examination was recorded on videotape. The processwas repeated for as long as the contrast remained well concen-trated within the bladder, allowing multiple micturitions in somecases (mean 2.7, range 1–6). After the first emptying of the bladder,the remaining 3.5 ml of Levovist was injected and the bladder re-filled with saline, until a further episode of micturition was ob-served. The mean time taken for the sonographic study followingbladder catheterization was 6.35 min (range 3.10–12.40 min). VURwas visualized as increased echogenicity within the pelvicalycealsystem (Fig. 1).

The ultrasound contrast agent was effective for a variable pe-riod of time in each case (Fig. 2), depending on the length of time ittook for micturition and the degree of bladder filling. In caseswhere micturition was delayed, the contrast agent tended to be-come quite dilute within the bladder, increasing the possibility ofmissing reflux in these cases.

Fig. 1 Vesicoureteral reflux is visualized as a markedly increasedhyperechogenicity within the pelvicalyceal system in a neonatepresenting for investigation of a urinary tract infection

854

After the sonographic contrast study, the child was carried intothe next room with the bladder catheter in situ, where a routinefluoroscopic MCU was performed immediately by different per-sonnel. Ionic contrast (Urografin 76%, Schering), 25 ml dilutedwith normal saline to 100 ml, was instilled into the bladder by dripinfusion from a height of 1 m. The videotaped sonographic ex-amination was reviewed at a later time by the two principal in-vestigators who remained unaware of the results of the fluoroscopicstudy, and the presence or absence of VUR was evaluated by

consensus. We did not attempt to identify grade 1 reflux in thisstudy, nor to grade degrees of VUR. Statistical analysis was per-formed after data collection.

Ethics committee consent was obtained for the study. Parentswere sent information about the procedure by mail, and telephonedprior to their appointment, during which the study was explained indetail. Written informed consent was obtained from all parents onthe day of the examination.

Statistical methods

Echo-enhanced sonography and MCU were compared for diag-nostic accuracy by the McNemar test. Kappa was used to establishagreement between the two methods. Sensitivity and specificitywere calculated treating MCU as the gold standard.

For analysis of left and right kidneys combined, standardconfidence intervals could not be used because the results for refluxwere clustered within babies and therefore kidneys could not beconsidered independent units. The jack-knife method was thereforeused, with babies as the clustering unit in the jack-knife. (The jack-knife method primarily acts to estimate standard errors and bias ofestimators [10].)

Results

After exclusion of three babies in whom the ultrasoundwas considered nondiagnostic, 97 infants were includedin the study. Two babies had unilateral renal agenesis;therefore, 192 kidneys were evaluated. The total numberof babies with reflux detected by either technique was 14(19 kidneys). Equivocal results were obtained on so-nography in three infants (3 kidneys), with the MCUbeing negative in all cases. The results were concordantfor reflux on both sonography and MCU in 8 kidneys,one of which was duplex (Fig. 3), and concordant for noreflux in 173 kidneys (Table 1). Of the 14 babies withVUR detected at fluoroscopic MCU, the echo-enhancedsonogram detected 9 (64%). Of the five babies withVUR missed by echo-enhanced ultrasound, three hadgrade 2 and two had grade 3 VUR.

Fig. 2a–c Dilution of echocontrast is demonstrated over timewithin the dilated pelvicalyceal system of a neonate beinginvestigated for antenatal renal dilatation

Fig. 3 Vesicoureteral reflux is evident within the dilated lowermoiety of a duplex kidney

855

The echo-contrast sonogram detected reflux in 11kidneys. Fluoroscopic MCU detected reflux in 16 kid-neys. Discordant results were obtained in 11 kidneys,with MCU detecting reflux not demonstrated on so-nography in eight kidneys, and sonography showingreflux not detected on MCU in three kidneys. The rea-son why concordance was in fewer kidneys than babiesis because one baby had unilateral VUR demonstratedin different kidneys by each test. Echo-enhanced ultra-sound detected all six kidneys with grades 4 or 5 reflux(Table 2).

We determined P values using binomial probability inMcNemar’s test on/off diagonals (two-tailed) for leftand right kidneys. For 96 left kidneys, P=0.37, kap-pa=0.59; for 96 right kidneys, P=0.68, kappa 0.54; andfor 192 combined kidneys, P=0.22 (probability ob-tained by jack-knifing by babies), kappa=0.75 (95% CI0.55–0.96%). This indicates moderate agreement be-tween fluoroscopic MCU and echo-enhanced sonogra-phy. Of the 14 babies with VUR, echo-enhancedsonography detected 9. McNemar’s test showed thatthere was a discrepancy between the two methods(P=0.06). Taking MCU as the gold standard, the sen-sitivity of echo-enhanced sonography was 64% (95% CI35–87%) and specificity 100% (95% CI 95–100%). Thepositive predictive value was 100% (95% CI 66–100%)and negative predictive value 94% (95% CI 87–98%).

Discussion

Our study group, consisting mainly of neonates withantenatally detected renal dilatation (84%), is typical ofthe cohort of infants presenting for investigation ofpossible VUR. The 15% prevalence of VUR within ourstudy group reflects this. It has been estimated that 10–15% of cases of antenatal hydronephrosis are caused byprimary VUR, with a higher incidence of VUR (20–50%) in children with a proven urinary tract infection [4,11, 12, 13]. In other studies evaluating echo-contrastsonography, much higher detection rates of VUR, up to90%, have been obtained, suggesting that a highlyselected population was being examined [14, 15](Table 3). Children with a prior diagnosis of VUR were

deliberately excluded in our series. Our results are verysimilar to those obtained by Escape et al. [11] (sensitivity69% and specificity 94%) using MCU as the goldstandard. Most of their patients were neonates.

Why has our study shown a lower sensitivity thanother studies (79–100%) [7, 13, 14, 15] for the detectionof VUR using echo-enhanced sonography (Table 3)? Atour institution, a relatively low cut-off rate for fetal renaldilatation is used (4 mm or more at 18–20 weeks, and5 mm in the third trimester), which may help to accountfor the relatively low rate of VUR. Moreover, olderchildren were deliberately excluded from this study, asthe vast majority of examinations performed at our in-stitution are in the neonatal age group, and predomi-nantly those being followed for antenatalhydronephrosis, rather than older children presentingwith urinary tract infection (the single neonate in ourseries referred for MCU because of first urinary tractinfection did demonstrate VUR). Older children cangenerally void spontaneously, and this may facilitate thedetection of VUR. In studies in which a very high sen-sitivity (up to 100%) has been found for the detection ofVUR at echo-enhanced sonography older children wereexamined. In some of the studies the echo-contrasttechniques used would be too complicated for use in theaverage department on a day-to-day basis [13, 14]. Inothers, children were over 6 months of age with a veryhigh prevalence of VUR (77–86%) [14, 15] (Table 3).

Additionally, despite claims of longevity of up to30 min in the bladder, we and other workers found that

MCU-positive MCU-negative Total

US-positive US-negative US-positive US-negative

Babies (n) 9 5 0 83 97Kidneys (n) 8a 8 3 173 192

aVUR seen in contralateral kidney of baby at echo-enhanced sonography

Table 1 Comparison of echo-enhanced ultrasound and fluoro-scopic MCU for detection of VUR in 97 babies (192 kidneys) usingthe fluoroscopic MCU as the gold standard (MCU-positive grade 2or higher VUR at fluoroscopic MCU, MCU-negative fluoroscopic

MCU shows no VUR or grade 1 VUR, US-positive echo-enhancedsonography shows transient hyperechogenicity in renal collectingsystem, US-negative no evidence of increased echogenicity in renalcollecting system)

Table 2 Grade of reflux in kidneys with VUR detected at echo-enhanced sonography

Grade ofVUR (by MCU)

VUR detectedby US

VUR missedby US

0 31 0 1a

2 3 33 2 24 4 05 2 0

aWe did not attempt to detect grade 1 reflux with echo-enhancedUS

856

the echo-contrast microbubbles only persisted for 10 to15 min before becoming excessively dilute. Thus delayedvoiding can cause difficulties in interpretation [1]. Ad-ministration of more contrast would obviously alleviatethis problem.

We simplified the technique for administration ofcontrast as much as possible in this study, in order tomake it more attractive if it were to be introduced asstandard practice in our department. The minimumavailable amount of contrast (2.5 g) was used, a practicealso dictated by reason of cost. If the technique de-scribed by other authors had been used and if we hadused more contrast, we may possibly have detected morecases of reflux with sonography. The mean time for thevoiding sonogram in the study of Farina et al. [15] (ex-cluding catheterization) was 15 min, with 33 min for theentire examination, compared to 6.35 min (mean) post-catheterization in our study. As we and other authorshave noted, the concordance between the two techniqueswas improved when higher grades of reflux were present,and lower grades of reflux were more frequently missedwith echo-enhanced sonography [1, 7, 11] althoughcolour Doppler can improve detection of low-gradeVUR [16].

The disadvantages of contrast-enhanced sonographyinclude operator-dependence, rapid disintegration of themicrobubbles, difficulty in grading VUR (although someauthors claim variably good results) [11, 14], and diffi-culty in examination of the male urethra [14]. Advan-tages of the technique are that it is safe, and moreacceptable to parents, as they are able to stay close to oreven feed the infant, facilitating the procedure for all.Above all, it avoids the use of radiation in very smallchildren.

Our study has a number of strengths. It was aprospective study in a nonselected group of neonatespresenting for screening for VUR. The method of echo-enhanced sonography could be readily adapted by anybusy radiology department. The disadvantages of ourstudy are that the sequence of the MCU technique wasnot randomized, we did not assess for intraobserver orinterobserver variability, and the assessment of VURusing echo-enhanced sonography was based on a con-sensus from the observers. The power of the study islimited by the small number of infants with VUR.

It is unlikely that echo-enhanced sonography can re-place the fluoroscopic MCU as grading of reflux is moredifficult and inaccurate. Rather, the sonographic tech-nique may allow more targeted use of fluoroscopic MCUfor those infants who have or are highly likely to haveVUR. In order for echo-enhanced sonographic MCU tobe introduced into clinical practice, several improve-ments are required: development of a contrast agent thatpersists for longer periods in the bladder without be-coming excessively dilute, a reduction in cost, and im-proved sensitivity. If these can be achieved then theT

able

3Comparisonofstudiesofecho-enhancedsonographyfordetectionofVUR

Reference

No.of

patients

Age

No.(%

)ofchildren

withVUR

No.(%

)of

renalunits

withVUR

Sensitivity

(%)

Specificity

(%)

Positive

predictive

value(%

)

Negative

predictive

value(%

)

Concor-

dance

(%)

Durationof

examination(post-

catheterization)(m

in)

120

0.4–4.9

years

(mean3years)

20(100%

)100a

15–20

2188

5daysto

20years

(median5years)

57(52%

)80(35%

)100

97

14±

7(m

ean)

799

1.1–12.3

years

(mean4.6

years)

63(32%

)79

92

11

49

Birth

to5years

(median10months)

18(18%

)69

94

64

95

90.8

13

46

3weeksto

14years

(median6months)

17(20.2%

)91.7

93.1

68.8

98.5

92.9

20(m

edian)

14

58

0–10years

50(86%

)76

100

86

15

18

3monthsto

10years

13(77.2%

)15

16

74

3weeksto

16years

(mean5years)

42(36%

)81(95%

CI

75–89%

)95(95%

CI

91–99%

)89(95%

CI

84–95%

)92(95%

CI

85–95%

)90%

b

17

169

3daysto

18years

89(30.3%

)90.5

91.4

18

53monthsto

10years

5(100%

)100

Present

study

97

28–90days

(mean48days)

14

19kidneys

64(95%

CI

35–87%

)100(95%

CI

95–100%

)100(95%

CI

66–100%

)94(95%

CI

87–98%

)94.2

6.35(m

ean).

3.10–12.40(range)

aForgrades

III-V

VUR

(36%

sensitivityforgrades

I-II

VUR)

b96%

withadditionofcolourDoppler

857

number of fluoroscopic procedures may be able to bereduced by introducing this as a screening technique [19].

In conclusion, the echo-contrast cystogram may po-tentially be useful as a routine screening examination forVUR in the future. The number of fluoroscopic proce-dures may be able to be reduced by introducing this as ascreening technique [19]. Development of a contrastagent that persists for longer periods in the bladderwithout becoming excessively dilute, and a reduction in

cost would certainly make it an attractive alternative tofluoroscopic MCU. However, given the low sensitivityfor detection of VUR in our population at the presenttime, the authors do not believe its use can be justifiedroutinely at present.

Acknowledgements We thank Dr Elisabeth Wells for helping toplan the size of the study, and for statistical analysis of the results,and Tedi McDonald for assistance with typing and layout.

References

1. Ascenti G, Chimenz R, Zimbaro G,et al (2000) Potential role of colour-Doppler cystosonography withechocontrast in the screening andfollow-up of vesicoureteral reflux. ActaPaediatr 89:1336–1339

2. Darge K, Troeger J, Duetting T, et al(1999) Reflux in young patients: com-parison of voiding US of the bladderand retrovesical space with echo en-hancement versus voiding cystoureth-rography for diagnosis. Radiology201:201–207

3. Najmaldin A, Burge DM, Atwell JD(1990) Fetal vesicoureteric reflux. Br JUrol 65:403–406

4. Tibballs JM, De Bruyn R (1996) Pri-mary vesicoureteric reflux – how usefulis postnatal ultrasound? Arch Dis Child75:444–447

5. Jaswon MS, Dibble L, Puri S, et al(1999) Prospective study of outcome inantenatally diagnosed renal pelvis dila-tation. Arch Dis Child Fetal NeonatalEd 80:F135–138

6. Lepercq J, Beaudoin S, Bargy F (1998)Outcome of 116 moderate renal pelvisdilatations at prenatal ultrasonography.Fetal Diagn Ther 13:79–81

7. Kenda R, Novljan G, Kenig A, et al(2000) Echo-enhanced ultrasound void-ing cystography in children: a new ap-proach. Pediatr Nephrol 14:297–300

8. CampaniR,CalliadaF,Bottinelli O, et al(1998) Contrast enhancing agents inultrasonography: clinical applications.Eur J Radiol 27 [Suppl 2]:S161–170

9. Blomley MJK, Cooke JC, Unger EC,et al (2001) Microbubble contrastagents: a new era in ultrasound. BMJ322:1222–1225

10. Armitage P, Colton T (eds) (1998) En-cyclopedia of biostatistics, John Wileyand Sons, Chichester

11. Escape I, Martinez J, Bastart F, et al(2001) Usefulness of echocystography inthe study of vesicoureteral reflux. J Ul-trasound Med 20:145–149

12. Olbing H, Hirche H, Kosimies O, et al(2000) Renal growth in children withsevere vesicoureteral reflux: 10-yearprospective study of medical and surgi-cal treatment. Radiology 216:731–737

13. Mentzel H-J, Vogt S, Patzer L, et al(1999) Contrast-enhanced sonographyof vesicoureterorenal reflux in children:preliminary results. AJR Am J Roent-genol 173:737–740

14. Bosio M (1998) Cystosonography withechocontrast: a new imaging modalityto detect vesicoureteric reflux in chil-dren. Pediatr Radiol 28:250–255

15. Farina R, Arena C, Pennisi F, et al(2000) Vesico-ureteral reflux: diagnosisand staging with voiding color DopplerUS: preliminary experience (abstract).Eur J Radiol 35:49–53

16. Valentini AL, Salvaggio E, Manzoni C,et al (2001) Contrast-enhanced gray-scale and color Doppler voiding uroso-nography versus voiding cystoureth-rography in the diagnosis and gradingof vesicoureteral reflux. J Clin Ultra-sound 29:65–71

17. Berrocal F, Gaya M, Gomez L, et al(2000) Cystosonography with echoenh-ancer. A new imaging technique for thediagnosis of vesicoureteral reflux (ab-stract, in Spanish). An Esp Pediatr53:422–430

18. Farina R, Arena C, Pennisi F, et al(1999) Retrograde echocystography: anew ultrasonographic technique for thediagnosis and staging of vesicoureteralreflux (abstract, in Italian). Radiol Med97:360–364

19. Darge K, Ghods S, Zieger B, et al(1999) Reduction in voiding cystou-rethrographies after introduction ofecho-enhanced voiding urosonographyin a department of pediatric radiology(abstract). Pediatr Radiol 29:401

858