Peeling back the onion on ultrasound abnormalities
Article first published online: 29 SEP 2009
Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound in Obstetrics & Gynecology
Volume 34, Issue 4, pages 371–372, October 2009
How to Cite
Evans, M. I. and Andriole, S. (2009), Peeling back the onion on ultrasound abnormalities. Ultrasound Obstet Gynecol, 34: 371–372. doi: 10.1002/uog.7353
- Issue published online: 29 SEP 2009
- Article first published online: 29 SEP 2009
In the three or more decades since the first appreciation of congenital abnormalities on ultrasound, the attempt to associate structural abnormalities with underlying genetic diseases has become routine1, 2. Essentially, the two partially overlap, some genetic diseases having sonographically detectable manifestations, and some structural abnormalities having a clear genetic cause, a genetic predisposition, or an underlying genetic mechanism that was not apparent when the structural abnormality was first observed.
With improvements in ultrasound technology and increasing numbers of clinicians who are skilled in its use, more and more anomalies have become observable. For many of these, the first cases that were diagnosed were associated with other anomalies and, partly as a result, had poor clinical outcomes. The general principle is that the more primitive the technology, the worse a situation has to be before it can be appreciated. Clear cell adenocarcinoma of the vagina and other uterine anomalies associated with diethystilbestrol exposure are a classic example, in which the incidence of cancer/uterine anomalies was thought to be much higher at the beginning and fell with time3. Pertinent sonographic examples of anomalies first thought to have a very high positive predictive value for underlying aneuploidy include choroid plexus cysts, single umbilical artery and increased nuchal translucency thickness4. Reports of these abnormal ultrasound findings have led to there being a body of numerators without the benefit of denominators to gain the perspective needed on the true positive predictive value of these ultrasound findings being a marker for a specific genetic disorder. How good a test is, and this relies on there being a denominator, can be defined in the standard statistical terms of sensitivity, specificity and positive and negative predictive values.5
Two papers in this issue of the White Journal look at the associations of cleft lip with or without cleft palate with underlying genetic disorders6, 7. Gabrielli et al.6 show the discriminatory power of a flat fetal face vs. one with a premaxillary protrusion. They found that all five fetuses with a flat face had underlying aneuploidy, while only one of nine with protrusion did. Thus, a flat face in combination with cleft lip/palate has a sensitivity of 83% and a positive predictive value of 100% for predicting the presence of aneuploidy. Conversely, premaxillary protrusion has a specificity of 100% and a negative predictive value of 88% for predicting the absence of aneuploidy.
Gillham et al.7 reported 570 cases of cleft lip/palate and broke down the association of anomalies with the type of defect. Associated structural anomalies were seen in about 10% of cases with unilateral clefts and 25% of those with bilateral clefts and in 100% of cases with midline clefts. No isolated cleft palates were identified prenatally, but of the cases with isolated cleft palate, 6% had karyotypic or structural abnormalities and over 20% had a genetic syndrome. However, they then state that with cleft lip there is no need to karyotype, but we have grave concerns about that given the uncertainties of ultrasound discrimination of isolated cleft lip versus lip with/without palate. Even a 1% aneuploidy rate would be equivalent to the age risk of a 38-year-old, which clearly is a sufficiently high risk as to warrant cytogenetic evaluation, at least in United States (US) medical practice.
Over the past several years, there has been a trend among sonographers to downplay the use of diagnostic procedures in favor of visualization8. This approach is fraught with danger, however, as the variability of skill of sonographers appears to be significantly greater than is the risk of laboratory or procedure errors9–11. More than a decade ago, we reported that ‘just looking isn't enough’, with over 20% of visualized anomalies having other findings eventually discerned on autopsy or laboratory analysis9. Certainly, improved ultrasound equipment and better-skilled providers have lowered the residual risks but they have not eliminated them. For example, we recently observed10 that amniocentesis utilization in Colorado has plummeted over the past two decades, but the rate of Down syndrome births in women over 35 years of age has doubled. Our analysis found no adequate sociological explanations other than the possibility that false reassurance from seemingly normal scans was leading patients to forego diagnostic (and definitive) testing. Likewise, the first-trimester biochemistry-based detection rates of trisomies 13 and 18 are virtually identical in the UK and the US. However, when nuchal translucency thickness measurements are added into the algorithm, the false-positive rate in the UK is substantially reduced, with a higher sensitivity than that in the US, suggesting better quality control in the UK leading to measurable differences in clinical performance11.
There are two general conclusions derivable from these interesting papers. Firstly, cleft lip with or without cleft palate is a complex finding with enormous variation in its likelihood of serious underlying pathology that can now be categorized in part by its type and association with other findings. Such sub-categorization is required to counsel patients properly. Diagnostic tests for karyotype and eventually other molecular abnormalities (non-aneuploidy) should be increasing, not decreasing.
The more generic issue is that ultrasound is a screening test and, like all screening tests, it should be thought of in terms of its sensitivity, specificity and predictive values for the underlying pathology. In some cases, the problem will be an isolated structural abnormality, which of course can be very serious in its own right, for example in the case of anencephaly. In other cases, however, an anatomical finding, for example endocardial cushion defect, is a true anatomical defect but also has a very high predictive value in screening for underlying aneuploidy. There is still too much confusion between ultrasound as a diagnostic vs. screening test. Hopefully, more studies such as the ones discussed here can help elucidate the differences in implications of the same ultrasound finding.
- 2Prenatal Diagnosis of Congenital Anomalies. Appleton and Lange: Norwalk, CT, 1988., , , , .
- 4First trimester ultrasound screening with nuchal translucency. In Prenatal Diagnosis, EvansMI, JohnsonMP, YaronY, DruganA (eds). McGraw Hill: New York, 2006; 289–308., .
- 5Biochemical screening. In Prenatal Diagnosis, EvansMI, JohnsonMP, YaronY, DruganA (eds). McGraw Hill: New York, 2006; 277–288., , .