PRENATAL DIAGNOSIS
Prenat Diagn 2007; 27: 18–22. Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/pd.1609
False positives in the prenatal ultrasound screening of fetal structural anomalies M. Angels Martinez-Zamora1 , Antoni Borrell1 *, Virginia Borobio1 , Anna Gonce1 , Marimer Perez1 , Francesc Botet1 , Alfons Nadal2 , Asteria Albert3 , Bienvenido Puerto1 and Albert Fortuny1 1
Institute of Gynecology, Obstetrics and Neonatology, Hospital Clinic, University of Barcelona Medical School, Barcelona, Catalonia, Spain 2 Pathology Service Biomedical Diagnosis Center, Hospital Clinic, University of Barcelona Medical School, Barcelona, Catalonia, Spain 3 Pediatric Surgery Service, Hospital Clinic, Hospital de Sant Joan de Dev, Barcelona, Catalonia, Spain Objective To describe the false-positive diagnoses of prenatal ultrasound screening of fetal structural anomalies. Methods Pregnancies with fetal structural anomalies either detected prenatally in our center or referred to us, were registered, evaluated, and followed-up prospectively by a multidisciplinary Congenital Defects Committee. After postnatal follow-up was completed, cases were assigned as true positives, false positives or false negatives and categorized by anatomical systems. Pregnancies referred with a nonconfirmed suspicion of anomaly were not included. The false-positive diagnoses were analyzed. Results From 1994 to 2004, 903 new registry entries of fetuses structurally abnormal at ultrasound with a complete follow-up were included in the Committee database. There were 76 false positives, accounting for 9.3% of all the prenatally established diagnoses. The urinary tract anomalies were the most frequent false-positive diagnoses found (n = 25; accounting for 8.7% of the urinary tract defects), but the genital anomalies showed the higher rate of no confirmation (n = 5; 15.2%). The specific anomalies most commonly not confirmed were renal pyelectasis (n = 9), cerebral ventriculomegaly (n = 9), abdominal cysts (n = 7) and short limbs (n = 7). Conclusion Several prenatally diagnosed anomalies would benefit from prudent counseling, because they may be normal variants or transient findings. Copyright 2007 John Wiley & Sons, Ltd. KEY WORDS:
prenatal ultrasound; screening; routine ultrasound; false positives; renal pyelectasis
INTRODUCTION One of the main goals of obstetric ultrasound is the prenatal detection of congenital anomalies, and although its effectiveness is still debated, a substantial improvement has been observed from 1993 (Ewigman et al., 1993) to date (Garne et al., 2005). In most European countries ultrasound screening has been established as a part of the routine prenatal care, although provision, uptake and quality of prenatal ultrasound vary widely throughout the different European regions. Population-based registries of congenital defects have shown that prenatal detection of fetal congenital anomalies as a screening program for the whole pregnant population is feasible and effective, although reported detection rates range from 14 to 90% (Lys et al., 1989; Grandjean et al., 1999; Levi, 2002; Garne et al., 2005). Moreover, the screening has led to lower perinatal mortality (Saari-Kemppainen et al., 1990). Inaccuracies in ultrasound prenatal diagnosis with reference to the false-positive diagnoses of fetal *Correspondence to: Antoni Borrell, Prenatal Diagnosis Unit, Institute of Gynecology, Obstetrics and Neonatology, Hospital Clinic—Seu Maternitat, Sabino de Arana 1, Barcelona 08028, Catalonia, Spain. E-mail:
[email protected]
Copyright 2007 John Wiley & Sons, Ltd.
structural anomalies pertaining to the assignment of false-positive diagnoses by anatomical systems have rarely been addressed. (Grandjean et al., 1999; Richmond and Atkins, 2005). The aim of this study was to assess the false-positive diagnoses of prenatal ultrasound at a tertiary referral center, considering all pregnancies routinely scanned with one examination in each trimester and postnatally. METHODS Pregnancies in which fetal structural anomalies were prenatally detected, including women who were either booked into our hospital or had been referred from district hospitals, were registered, evaluated and followedup prospectively by a Congenital Defects Committee. The multidisciplinary committee includes obstetricians, neonatologists, pediatric surgeons, pathologists, medical geneticists and cytogeneticists. All pregnancies were followed until the final report from the pediatrician or the pathologist establishing the diagnosis either after birth, fetal demise or termination of pregnancy. In liveborns with urinary tract, cerebral or cardiac defects, postnatal ultrasound was always performed followed by more specific examinations (renography, voiding cystourethogram, and magnetic resonance imaging) when Received: 26 July 2006 Revised: 10 October 2006 Accepted: 11 October 2006
19
FALSE POSITIVES
indicated. When the infants were delivered in a different center, a final report was also obtained from the pediatrician or pathologist. Otherwise the information was obtained from the women, usually by phone inquiry. Additionally, all the congenital anomalies missed by the prenatal scan and diagnosed after birth or in the postmortem examination were also registered in the database. If ultrasound was normal after referral then the pregnancy was not included in the database. False negatives were captured either after complete neonatal follow up of all infants delivered in our center or after referral to our Neonatology Unit owing to nonprenatally diagnosed birth defects. Pregnancies assessed by the Congenital Defects Committee only for a risk factor (such as maternal drug intake, premature rupture of membranes, or severe polyhydramnios) but without a suspected structural anomaly were excluded. After postnatal assessment, structural defects were classified as: 1. True positive when the anomaly was confirmed after birth, fetal demise or termination of pregnancy (clinical or postmortem findings) 2. False positive if the suspected defect was not confirmed postnatally or at postmortem examination 3. False negative if the anomaly was observed after birth or fetal demise but was not prenatally detected. For each of the fetuses, it was noted whether the structural anomalies were single or multiple, and if they were associated with chromosomal or single-gene defects. The number of false positive diagnoses was assessed for each of the anatomical systems, and the proportion of all the prenatal diagnoses was also calculated (false positives/true positives + false positives). Referral of suspected but not prenatally confirmed anomalies in our center were not included. In our center, pyelactasis was defined as dilated anterior–posterior diameter of the renal pelvis measuring at least 4 mm in the second trimester or 7 mm in the third, but less than 10 mm, in the absence of caliceal or ureteral dilatation. Ventriculomegaly was diagnosed when lateral ventricular width was observed to be 10 mm or more, involving only one or both lateral ventricles without an increase of the head circumference. Short limbs were defined by long bone length below the third centile for gestational age.
RESULTS From 1994 to 2004, 1384 new entries were included in the Congenital Defects Committee database. At least one structural anomaly was diagnosed in 1043 of these registered fetuses. The remaining 341 pregnancies were excluded because diagnosis of a structural anomaly was never established, although an extreme high risk recommended a stricter prenatal follow-up. In 903 fetuses structurally abnormal at ultrasound, a complete follow-up was obtained but it was not obtained in the remaining 140 (13%) fetuses. The average number of scans per anomaly case was 2.4, and the mean gestational age at which the prenatal diagnosis of the anomaly was established in our center was 22 weeks, distributed as follows: 20% in the first, 55% in the second and 25% in the third trimester. Chromosomal or single-gene defects were associated in 35% (69/205) of the fetuses with more than one defect and in 9% (62/698) of those with single structural anomalies (Table 1). Anomalies of the skeleton (n = 330), urinary tract (n = 300) and the cardiovascular system (n = 221) were the most commonly observed (Table 2). Prenatal diagnosis was not confirmed postnatally or at postmortem examination in 76 of the 820 fetuses with a structural anomaly, therefore the false positives accounted for 9.3% of all the prenatally established diagnoses (Table 1). The mean gestational age at which the anomaly was falsely diagnosed in our center was 26 weeks, distributed as follows: 9% in the first, 32% in the second and 59% in the third trimester. Urinary tract anomalies showed the highest absolute number of false-positive diagnoses (n = 25), despite the fact that those accounted only for 8.7% (25/260) of all the urinary tract anomalies diagnosed prenatally. The genital (15.2%: 5/28) and the gastro-intestinal systems (11.6%: 15/114) were the systems showing the highest proportion of false positives in relation to all prenatal diagnoses (Table 2). The specific anomalies most frequently over-diagnosed were renal pyelectasis (n = 9), cerebral ventriculomegaly (n = 9), abdominal cyst (n = 7) and short limbs (n = 7) (Table 3). Termination of pregnancy due to fetal structural defects was elected on the basis of the information provided in seven out of 76 pregnancies (9.2%) in which a false-positive diagnosis was eventually established. However, all these fetuses had concomitant
Table 1—Distribution of true positives, false positives and false negatives in registry entries of structurally abnormal fetuses, according to the presence of multiple or single anomalies and whether they were associated with chromosomal or known single-gene disorders and proportions of false-positive diagnoses among all prenatal diagnoses True positives n Fetuses with multiple structural anomalies Chromosomal or single-gene defects No chromosomal or single-gene defects Fetuses with a single structural anomalies Chromosomal or single-gene defects No chromosomal or single-gene defects Total of fetuses Copyright 2007 John Wiley & Sons, Ltd.
184 56 128 560 45 515 744
False positives n (%) 2 1 1 74 1 73 76
(1.1) (3.5) (0.8) (11.7) (2.2) (12.4) (9.3)
False negatives n
Total n
19 12 7 64 16 48 83
205 69 136 698 62 636 903
Prenat Diagn 2007; 27: 18–22. DOI: 10.1002/pd
20
M. A. MARTINEZ-ZAMORA ET AL.
Table 2—Distribution in true positives, false positives and false negatives in the registry entries of fetal anomalies grouped by systems, whether the anomalies were single or multiple and proportion of false positive diagnoses among all prenatal diagnoses True-positives n Anatomic system Skeleton Multiple structural anomalies Single structural anomalies Urinary tract Multiple structural anomalies Single structural anomalies Cardiovascular Multiple structural anomalies Single structural anomalies Central nervous system Multiple structural anomalies Single structural anomalies Gastro-intestinal Multiple structural anomalies Single structural anomalies Facial Multiple structural anomalies Single structural anomalies Respiratory Multiple structural anomalies Single structural anomalies Genital Multiple structural anomalies Single structural anomalies
294 127 167 260 68 192 177 84 93 158 56 102 114 58 56 62 30 32 53 20 33 28 12 14
severe defects, either a chromosomal anomaly (n = 4) or other confirmed defects (n = 3). Two fetuses died prenatally owing to other severe structural defects (bilateral hydronephrosis in one and Fanconi anemia in the other) and the remaining 65 fetuses had a satisfactory outcome. DISCUSSION The present study reports the 11-year experience of our multidisciplinary team in the prenatal assessment of fetal structural defects in a referral center, but these results cannot be considered as the real performance of routine ultrasound in the prenatal detection of congenital anomalies in the general population. Hospital-based studies are prone to ascertainment biases to determine the true falsenegative rates that would require a population-based study and, for this reason, our study is aimed to analyze the false positives, which are not subjected to this limitation. The Committee’s registry gathers cases primarily observed in our own center and those referred, and includes true-positive, false-positive and false-negative cases. A recent report from the population-based registry of the city of Barcelona for the period 1996–1999 shows 55% detection rate for all structural defects (50% for single and 61% for multiple defects) (Salvador et al., 2005). In this registry, pregnancy terminations are properly included and the follow-up of cases is carefully sought as in other Eurocat congenital anomaly registries, (Garne et al., 2005), but false positives are not recorded. The false positives in our series accounted for 9.3% of all the prenatal diagnoses included in the registry. Copyright 2007 John Wiley & Sons, Ltd.
False positives n (%) 14 0 14 25 2 23 8 1 7 15 0 15 15 1 14 0 0 0 2 1 1 5 0 5
(4.5) (8.7) (4.3) (8.7) (11.6) (0) (6.1) (15.2)
False negatives n
Total n
22 12 10 15 15 0 36 12 24 7 4 3 18 1 17 21 5 16 4 1 3 2 1 1
330 139 191 300 85 215 221 97 124 180 60 120 147 60 87 83 35 48 59 22 37 35 13 22
In the Eurofetus study, a multicenter hospital-based study, a similar proportion of false-positive diagnoses was observed (9.9%), and an additional 6% of diagnoses were considered ‘false alarms’ because they could not be confirmed on subsequent prenatal ultrasound examinations (Grandjean et al., 1999). In the population registry of Bas-Rhin (Strasbourg), the false-positive rate was estimated to be 0.5% of the normal pregnancies in which at least one ultrasound examination was performed (Stoll et al., 1995). In a United States hospital-based registry, this rate was estimated to be 0.7% (Gon¸calves et al., 1994). In absolute terms, in our series, false positives were more frequently found for the urinary tract anomalies (n = 25), but accounted for a higher proportion among overall positive diagnoses of the genital (15.2%) and gastro-intestinal (11.6%) systems (as compared to 8.7% of the urinary tract diagnoses). In the Eurofetus study, the most common false-positive diagnoses in both absolute and relative figures were the urinary tract (16.5%), and the gastro-intestinal showed comparable high rates regarding only relative figures (15.2%) (Grandjean et al., 1999). In two hospital-based series, the vast majority of false positives were urinary tract anomalies (Rosendahl and Kivinen 1989; Gon¸calves et al., 1994). In our series, the most common specific false-positive diagnoses were renal pyelectasis, cerebral ventriculomegaly, abdominal cyst and short limbs. The Strasbourg (population-based) registry coincidentally showed that the more frequent unconfirmed anomalies were hydrocephalus, hydronephrosis, and cysts (renal, pulmonar and cerebral) (Stoll et al., 1995). Prenat Diagn 2007; 27: 18–22. DOI: 10.1002/pd
21
FALSE POSITIVES
Table 3—84 false-positive diagnoses in 76 fetuses grouped by systems Urinary tract: 25 —Pyelectasis: 9 —Dilated bladder/ureters: 5 —Hydronephrosis: 5 —Multicystic/Dysplastic kidneys:3 —Hyperechogenic kidneys: 2 —Bladder cyst: 1 Central nervous system: 15 —Ventriculomegaly:9 —Interhemispheric cyst: 2 —Neural tube defects: 1 —Dacrocephaly: 1 —Hydrocephaly: 1 —Agenesis of corpus callosum:1 Gastro-intestinal: 15 —Abdominal cyst: 7 —Hyperechogenic bowel: 2 —Exomphalos:2 —Esophagial atresia:2 —Ileal atresia: 1 —Intestinal dilatation:1 Skeleton: 14 —Short limbs: 7 —Foot malposition: 2 —Polydactily: 1 —Malar cyst: 1 —Sacral tumor: 1 —Craniosynostosis: 1 —Abdominal-wall hypotony: 1 Cardiovascular: 8 —Hypoplastic left heart:3 —Pulmonary atresia: 2 —Atrioventricular septal defect: 1 —Ventricular septal defect: 1 —Narrow aorta: 1 Genital: 5 —Ovarian cyst: 4 —Ambiguous genitalia: 1 Respiratory: 2 —Cystic lung lesion: 2
Pyelactasis, however, should no longer be considered as a real anomaly when isolated, but rather an alarm sign deserving reevaluation in the third trimester and, if still present, a postnatal renal scan is warranted. It is widely accepted that isolated pyelectasis may be physiological or due to a mild reflux or to an uretero-pelvic junction obstruction. A recent German study found that with the use of the same thresholds as in the present series (4 mm in the second and 7 mm in the third trimester) all critically relevant postnatal pielo-caliceal dilatation were predicted (100% sensitivity) but the specificity was low (19% for the second and 48% for the third trimester) (John et al., 2004). It appears that dilatation disappearing in the third trimester would not require further investigation (Cohen-Overbeek et al., 2005). The presence of mild unilateral or bilateral ventriculomegaly is still a subject of controversial interpretation, but it usually carries a favorable prognosis when other associated anomalies, such as abnormal karyotype and fetal infections are ruled out and ventricular width is below 13 mm. A recent United States population-based Copyright 2007 John Wiley & Sons, Ltd.
study on the trends in accuracy of prenatal diagnosis from 1985 to 2000 found a reduction in falsepositive diagnoses for central nervous system anomalies, apart from isolated hydrocephalus displaying one of the highest odds of having no significant abnormality giving a positive prenatal report (1 : 3) (Richmond and Atkins, 2005). Mild ventriculomegaly has been classically defined as a ventricular width of 10 to 15 mm, and around 40% of cases become subsequently normal in utero, although some of them will show some degree of neurodevelopmental delay (Breeze et al., 2005; Goldstein et al., 2005; Parilla et al., 2006). However, recent studies found a favorable outcome when the width is less than 13 mm (Gaglioti et al., 2005). Counseling should be cautiously carried out after the diagnosis of abdominal or ovarian cyst considering that it usually disappears before birth in around half of the cases when followed prenatally (Heling et al., 2002), and postnatally in 17% of those remaining (Borsellino et al., 2006). It is well known that ovarian cysts may be the result of fetal gonadotrophin stimulation owing to the increase in the production of gonadotrophins until midgestation and decline after 30 weeks. Finally, short limbs, as defined by long bone length below the third centile, are expected to be found in 3% of the normal population. Reduced parental height is a reassuring finding when reduced long bones are found; otherwise molecular testing in fetal blood or amniotic fluid is advisable in case of achondroplasia and hypochondroplasia. Sometimes, the false-positive diagnosis may appear as a clear-cut mistake in the case of hypoplastic left heart or a neural tube defect, because when present they display well-defined patterns of severe anomalies. However, even the postmortem examination, gold standard for precise diagnosis in dead fetuses, can be inaccurate. With the increasing ability of ultrasound in the detection of first-trimester anomalies, the methodology for pathology studies of tiny organs has not evolved at the same pace. As an example, first-trimester heart defects are the most difficult anomalies to confirm after termination in chromosomally abnormal fetuses. Thus, the role of the pathology examination as the gold standard of prenatal diagnosis in nonviable fetuses deserves reevaluation, since some anomalies may be missed and wrongly ruled out, labeling the case as false positive, with a crucial psychological and legal implications if the pregnancy was terminated. To conclude, the most common false-positive diagnoses are variants of normality deserving only further evaluation to rule out evolution to real pathologies (hydronephrosis, hydropcephalus or micromelia). Parents may be seriously distressed by a wrong diagnosis of an absent malformation, and the thresholds between normal and abnormal findings need to be reevaluated. ACKNOWLEDGMENTS
We would like to acknowledge the contributions of the remaining members of the Congenital Defects Committee: J.M. Jimenez, V. Penalba, J.M. Martinez, Prenat Diagn 2007; 27: 18–22. DOI: 10.1002/pd
22
M. A. MARTINEZ-ZAMORA ET AL.
V. Cararach (Fetal Medicine), E. Sanchez (Neonatology), C. Aguilar (Pediatric Surgery), N. Bargallo (Neuroimaging), E. Casals (Biochemistry), A. Soler (Cytogenetics), A. Sanchez, A. Seres (Medical Genetics). REFERENCES Borsellino A, Zaccara A, Nahom A, et al. 2006. False-positive rate in prenatal diagnosis of surgical anomalies. J Pediatr Surg 41: 826–829. Breeze AC, Dey PK, Lees CC, Hackett GA, Smith GC, Murdoch EM. 2005. Obstetric and neonatal outcomes in apparently isolated mild fetal ventriculomegaly. J Perinat Med 33: 236–240. Cohen-Overbeek TE, Wijngaard-Boom P, Ursem NT, Hop WC, Wladimiroff JW, Wolffenbuttel KP. 2005. Mild renal pyelectasis in the second trimester: determination of cut-off levels for postnatal referral. Ultrasound Obstet Gynecol 25: 378–383. Ewigman B, Crane J, Frigoletto F, Lefevre M, Bain R, Mcnellis D, The RADIUS Study Group. 1993. Effect of prenatal ultrasound screening on perinatal outcome. N Engl J Med 329: 821–827. Gaglioti P, Danelon D, Bontempo S, Mombro M, Cardaropoli S, Todros T. 2005. Fetal cerebral ventriculomegaly: outcome in 176 cases. Ultrasound Obstet Gynecol 25: 372–377. Garne E, Loane M, Dolk H, et al. 2005. Prenatal diagnosis of severe structural congenital malformations in EUROPE. Ultrasound Obstet Gynecol 25: 6–11. Goldstein I, Copel JA, Makhoul IR. 2005. Mild cerebral ventriculomegaly in fetuses: characteristics and outcome. Fetal Diagn Ther 20: 281–284. Gon¸calves LF, Jeanty P, Piper J. 1994. The accuracy of prenatal ultrasonography in detecting congenital anomalies. Am J Obstet Gynecol 171: 1606–1612. Grandjean H, Larroque D, Levi S, The Eurofetus Study Group. 1999. The performance of routine ultrasonographic screening of
Copyright 2007 John Wiley & Sons, Ltd.
pregnancies in the eurofetus study. Am J Obstet Gynecol 181: 446–454. Heling KS, Chaoui R, kirchmair F, Stadie S, Bollmann R. 2002. Fetal ovarian cysts: prenatal diagnosis, management and postnatal outcome. Ultrasound Obstet Gynecol 20: 47–50. John U, Kahler C, Schulz S, Mentzel HJ, Vogt S, Misselwitz J. 2004. The impact of fetal renal pelvic diameter on postnatal outcome. Prenat Diagn 24: 591–595. Levi S. 2002. Ultrasound in prenatal diagnosis: polemics around routine ultrasound screening for second trimestre fetal malformations. Prenat Diagn 22: 285–295. Lys F, De Was P, Borlee-Grimee I, Billiet A, Vincotte-Mols M, Levi S. 1989. Evaluation of routine ultrasound examination for the prenatal diagnosis of malformation. Eur J Obstet Gynecol Reprod Biol 30: 101–109. Parilla BV, Endres LK, Dinsmoor MJ, Curran L. 2006. In utero progression of mild fetal ventriculomegaly. Int J Gynaecol Obstet 93: 106–109. Richmond S, Atkins J. 2005. A population-based study of the prenatal diagnosis of congenital malformation over 16 years. Br J Obstet Gynaecol 112: 1349–1357. Rosendahl H, Kivinen S. 1989. Antenatal detection of congenital malformations by routine ultrasonography. Obstet Gynecol 73: 947–951. Saari-Kemppainen A, Karjalainen O, Ylostalo P, Heinonen OP. 1990. Ultrasound screening and perinatal mortality: controlled trial of systematic one-stage screening in pregnancy. The Helsinki ultrasound trial. Lancet 336: 387–391. Salvador J, Borrell A, Lladonosa A. 2005. Increasing detection rates of birth defects by prenatal ultrasound leading to apparent increasing prevalences. Lessons learned from the population-based registry of birth defects of Barcelona. Prenat Diagn 25: 991–996. Stoll C, Dott B, Alembik Y, Roth P. 1995. Evaluation of routine prenatal diagnosis by a registry of congenital anomalies. Prenat Diagn 15: 791–800.
Prenat Diagn 2007; 27: 18–22. DOI: 10.1002/pd