To compare the diagnostic accuracy of two-dimensional (2D) ultrasound and three-dimensional (3D) computed tomography (CT) for the diagnosis of fetal skeletal anomalies.
To compare the diagnostic accuracy of two-dimensional (2D) ultrasound and three-dimensional (3D) computed tomography (CT) for the diagnosis of fetal skeletal anomalies.
Eleven pregnant women underwent 2D ultrasound and 3D-CT. Ten fetuses presented skeletal anomalies on 2D ultrasound and one fetus had a normal ultrasound exam but a familial history of osteopetrosis. We compared retrospectively the diagnoses established on 2D ultrasound and 3D-CT with the neonatal and/or postmortem work-up, which were used as the gold standard.
2D ultrasound provided the correct diagnosis in only two of the 11 cases. CT yielded the correct diagnosis in eight; in six of these, 2D ultrasound had been inconclusive. 3D-CT was more accurate than was 2D ultrasound in visualizing vertebral anomalies (abnormal shape of the vertebral bodies, abnormal interpedicular distance), pelvic bone malformations (delayed ossification of the pubic bones, abnormal acetabular shape) and enlarged metaphysis or synostoses in long bones. In three cases, neither 2D ultrasound nor CT provided the correct diagnosis.
In this series, which included a variety of skeletal dysplasias, 3D-CT had a better diagnostic yield than did 2D ultrasound. Both imaging techniques are useful in the management of fetal dysplasia; 2D ultrasound is a useful screening test and 3D-CT is a valuable complementary diagnostic tool. Copyright © 2007 ISUOG. Published by John Wiley & Sons, Ltd.
Skeletal dysplasias constitute a heterogeneous and complex group of disorders, which affect bone growth and development and result in abnormal shape and size of the skeleton. Current classification includes more than 150 different entities1, of which many are exceedingly rare. Although sonography has proved reliable for the prenatal detection of skeletal abnormalities (the overall sensitivity of two-dimensional (2D) ultrasound is estimated to be ∼ 60%2, 3), the precise diagnosis of the dysplasia is often difficult to make before delivery. This is a clinically relevant issue, because skeletal dysplasias may be associated with severe disability and may even be lethal.
Recent studies have suggested that three-dimensional (3D) ultrasound and computed tomography (CT) may have a better sensitivity than does 2D ultrasound for the antenatal diagnosis and characterization of skeletal dysplasias4, 5. In one study, which included six cases, the correct diagnosis was made in all six cases by both 3D ultrasound and 3D-CT, and in four cases by 2D ultrasound6. In addition, 3D ultrasound and 3D-CT identified significantly more abnormalities than did 2D ultrasound. The authors concluded that 3D-CT and 3D ultrasound are more accurate for the prenatal diagnosis of fetal bone abnormalities.
In this series of 11 pregnancies with fetal skeletal anomalies suspected on 2D ultrasound, we compared the diagnostic accuracy of 2D ultrasound with that of 3D-CT using the postnatal and/or postmortem diagnosis as the gold standard. This study extends the report of Ruano et al.6 by including a larger number of fetuses and a wider range of skeletal dysplasias.
Between January 2004 and May 2006, 10 pregnant women with fetal skeletal dysplasia suspected on 2D ultrasound underwent 3D-CT. One fetus with a normal ultrasound exam was also studied because of a familial history of osteopetrosis. The fetuses were aged from 26 to 36 weeks. The ultrasound examination was performed by experienced radiologists on a SSD-5500 (Aloka, Wallingford, CT, USA) ultrasound machine. In addition to the standard complete anatomical investigation of the fetus, it included measurements of the long bones and morphological studies of all accessible parts of the skeleton. 3D-CT was performed after 2D ultrasound and the result of the ultrasound examination was known by the radiologist who performed the CT. The acquisition was carried out using a CT multislice 16 scanner (Siemens, Erlangen, Germany) with the following parameters: 40 mAs, 120 KV, 16 slices per rotation, 0.75 pitch and 0.75-mm slice thickness. This corresponded to a mean irradiation dose given to the fetus of 3.12 mGy (CT-dose index weighted). The acquisition lasted about 20 s and was performed during maternal apnea to prevent kinetic artifacts that can mimic fractures or bone deformations. A total of 350–500 images per fetus was stored for further analysis. Post-processing consisted of a 3D reconstruction of the entire fetal skeleton using maximum intensity projection after segmentation and removal of the maternal pelvic bones. This procedure was performed with the inspace software on the Leonardo Workstation (Siemens). The whole process took about 20 min. The 3D-CT exams were analyzed at the time of acquisition, i.e. before the postnatal or postmortem diagnosis was available. The ultrasound and CT diagnoses were compared retrospectively with the postnatal and/or postmortem diagnoses. The neonatal work-up included pathological, radiological and genetic studies. The study protocol was approved by the ethics committee of our institution and all women gave informed consent to participate in the study.
Prenatal imaging data, together with postmortem or postnatal findings, are given in Table 1. Of the 11 fetuses, three presented isolated long-bone shortening on 2D ultrasound. In Cases 1 and 2, possible diagnoses included growth restriction and achondroplasia, but the 2D ultrasound exam could not distinguish between the two; in Case 3, achondroplasia was suspected because the bones looked slightly curved in addition to being shortened. In Case 1, 3D-CT did not demonstrate any additional anomaly, supporting the diagnosis of growth restriction; this diagnosis was eventually confirmed after delivery. In Case 2, 3D-CT demonstrated misshaped ovoid vertebral bodies, and lack of ossification of pubis, tarsal bones and cervical vertebral bodies. These anomalies suggested the diagnosis of spondyloepiphyseal dysplasia, but no precise diagnosis could be made from postmortem radiographs or fetopathology. In Case 3, 3D-CT demonstrated slightly curved tubular bones, a decrease in the interpedicular distance from the upper to the lower spine, and spurs extending downwards from the medial aspect of the acetabular roofs. These findings, in association with long-bone shortening, were highly suggestive of achondroplasia. This diagnosis, which had already been suggested by 2D ultrasound, was established by the postmortem work-up.
|Case||GA at diagnosis (weeks)||Prenatal ultrasound||Prenatal 3D-CT||Final diagnosis|
|1||33||Short long bones (< P3)||IUGR/achondroplasia||Short long bones||IUGR||IUGR|
|2||26||Short long bones (< P3)||IUGR/achondroplasia||Short and broad long bones, wide metaphysis, ovoid vertebral bodies, absence of ossification of the tarsal, pubic and cervical vertebral bodies||Spondyloepiphyseal dysplasia (type ?)||Spondyloepiphyseal dysplasia (type ?)|
|3||36||Short long bones (< P3)||Suspicion of achondroplasia||Short and broad long bones, trident acetabular roofs, reduced interpedicular distance||Achondroplasia||Achondroplasia|
|4||31||Short long bones (< P3), short ribs, anhydramnios, renal hyperechogenicity||Asphyxiating thoracic dysplasia||Short long bones, short ribs, platyspondyly, partial ossification of the sacrum, no pubic ossification, flat acetabular roofs, cupped femoral metaphyses||Spondylometaphyseal dysplasia (type ?)||Spondylometaphyseal dysplasia (type ?)|
|5||30||Short long bones (< P3), short ribs||Thanatophoric dwarphism||Short and broad long bones, short ribs, no vertebral anomalies, trident acetabular roofs||Asphyxiating thoracic dysplasia||Asphyxiating thoracic dysplasia|
|6||33||Suspected vertebral anomaly||Suspected L3 vertebral anomaly (type ?)||Hemivertebra in L3||Isolated vertebral anomaly, hemivertebra in L3||Isolated hemivertebra|
|7||32||Short long bones (P5), micrognathia, vertebral anomaly||?||Fused hemivertebra in D12||Isolated vertebral anomaly, fused hemivertebra in D12||Isolated fused vertebra in D12|
|8||30||Vertebral anomaly (D12-L1)||Isolated vertebral segmentation anomaly (type? localization?)||Multiple vertebral segmentation anomalies (thoracic and lumbar)||Multiple vertebral segmentation anomalies (thoracic and lumbar)||VACTER association|
|9||26||Short long bones, cloverleaf skull||Apert syndrome?||Short long bones, normal hands and feet, radiocubital synostosis||Pfeiffer syndrome||Pfeiffer syndrome|
|10||30 (Reccurent osteopetrosis?)||Normal, no bone fractures||Osteopetrosis not excluded||No bone fractures, suspected increased bone density||Osteopetrosis not excluded||Osteopetrosis|
|11||36||Bilateral femoral incurvation||Osteogenesis imperfecta?||Symmetrical angulation of the femoral diaphysis, wide anterior fontanelle, suspected decreased bone density||Osteogenesis imperfecta not excluded||Normal|
Two other fetuses (Cases 4 and 5) presented short long bones and ribs on 2D ultrasound. In Case 4, these anomalies were associated with hyperechogenic renal parenchyma and anhydramnios. This association suggested the diagnosis of asphyxiating thoracic dysplasia, but this diagnosis was not supported by 3D-CT, which showed several additional skeletal anomalies including platyspondyly, incomplete ossification of the sacral vertebrae, lack of pubic bone ossification and flattening of the acetabular roofs (Figure 1). These findings suggested the diagnosis of spondylometaphyseal dysplasia, but a specific diagnosis could not be made from the postnatal work-up. In Case 5, 2D ultrasound showed shortening of long bones and ribs, which suggested thanatophoric dwarphism. However, the flattened vertebral bodies generally seen in this syndrome were absent on 3D-CT. This examination demonstrated a trident deformity of the acetabular roofs, which suggested the diagnosis of asphyxiating thoracic dysplasia (Figure 2), and was confirmed by postmortem radiographs.
In three cases (Cases 6, 7 and 8), abnormal vertebral segmentation was seen on 2D ultrasound. In Case 6, the anomaly was localized on the third lumbar vertebra but could not be characterized precisely; 3D-CT identified a third lumbar hemivertebra, which was confirmed after delivery. In Case 7, shortened long bones, micrognathia and abnormal vertebral curvature were suspected on 2D ultrasound. 3D-CT confirmed only the vertebral anomaly, which consisted of a fused 12th hemivertebra. This isolated vertebral malformation was confirmed by the postnatal work-up. In Case 8, 2D ultrasound demonstrated an abnormal vertebral angulation at the thoracolumbar junction. 3D-CT showed a complex malformation affecting several vertebrae (Figure 3) and a butterfly-like deformation of the 8th thoracic vertebra. In addition to the skeletal anomalies, the postnatal work-up demonstrated a rectovaginal fistula and a left major vesicoureteral reflux, which established the diagnosis of VACTER association. Neither ultrasound nor CT had suggested this diagnosis before delivery.
In Case 9, 2D ultrasound showed a cloverleaf skull deformity and long-bone shortening. Based on these findings, Crouzon syndrome was suspected. 3D-CT excluded the classical mitten hands and feet malformations usually seen in this syndrome, but demonstrated bilateral radiocubital synostosis, which suggested Pfeiffer syndrome (Figure 4). This diagnosis was confirmed by genetic studies and postmortem radiography, which showed the typical deformity of the distal phalanx of the thumbs (not visible on 3D-CT because of their incomplete ossification at 26 weeks' gestation).
In a family with a history of osteopetrosis, the 2D ultrasound examination of the fetus (Case 10) was normal. 3D-CT was performed to exclude bone fractures that could have been overlooked by 2D ultrasound. No fracture was seen and no definite diagnosis was established antenatally. Osteopetrosis was confirmed after delivery by the skeletal survey and the pancytopenia, deafness and blindness of the newborn.
The final fetus presented bilateral femoral diaphyseal angulation on 2D ultrasound and was suspected of having osteogenesis imperfecta (Case 11) (Figure 5a). 3D-CT demonstrated abnormal and bilateral angulations of the femora associated with a decrease in bone density of the cranial vault and an enlarged anterior fontanelle (Figure 5b). These anomalies were compatible with the diagnosis of osteogenesis imperfecta, but the symmetrical deformations of the bones and the absence of fractures casted doubt on this diagnosis. It was eventually excluded after delivery; the newborn was normal.
The prenatal diagnosis of skeletal dysplasia is often difficult to make, especially in the absence of family history. Currently, the technique used for the prenatal detection of these abnormalities is 2D ultrasound2, which has a sensitivity of ∼ 60%3. 3D ultrasound has been reported to have a somewhat better sensitivity compared with 2D ultrasound and to be particularly useful for the evaluation of facial dysmorphism and anomalies involving the hands and feet4, 5. After 30 weeks' gestation, standard orthogonal X-rays of the maternal abdomen may help to visualize the fetal skeleton and identify possible abnormalities in bone shape and size. However, superposition of fetal and maternal bones often makes it difficult to visualize precisely the fetal skeleton.
Recently, 3D-CT has been tested for cephalometric analysis of maxillary growth in sheep fetuses7 and has been used in an isolated case of suspected fetal hypochondroplasia8. The new CT equipment used both in this study and in that of Ruano et al.6 can provide 3D reconstructions of the whole fetal skeleton in one acquisition. In both studies, the total CT dose index was 3.12 mGy, similar to the irradiation exposure of conventional fetal radiological examinations (3 mGy)9. Ruano et al.6 demonstrated that 3D-CT was more accurate than was 2D ultrasound for the prenatal diagnosis of fetal bone abnormalities. Our study, which included a larger number of patients and a wider range of dysplasias, confirms and extends these results. We performed 3D-CT in 10 fetuses with skeletal anomalies on 2D ultrasound and one fetus with a normal ultrasound examination but a familial history of osteopetrosis. Whereas ultrasound provided the correct diagnosis in only two cases (Cases 3 and 6), 3D-CT yielded the correct diagnosis in eight (Cases 1–7 and 9); in six of these, the diagnosis had not been made by 2D ultrasound. In the last three cases (Cases 8, 10 and 11), neither 2D ultrasound nor 3D-CT provided the correct diagnosis.
There are, however, quantitative differences between our results and those of Ruano et al.6. Two-dimensional US and 3D-CT provided more frequently the correct diagnosis in the study of Ruano et al.6 than they did in our study; 2D ultrasound allowed a correct diagnosis in four of six cases in the study of Ruano et al.6 and in two of 11 of our cases; the corresponding figures for 3D-CT were 6/6 and 8/11, respectively. These differences may be due to the fact that our study included very rare entities that are difficult to identify whichever technique is used (see below). The dysplasias reported by Ruano et al.6 are more frequent and are characterized by bone deformities which are more easily identifiable, for example bone shortening and curvature in achondroplasia or bone fractures in osteogenesis imperfecta; these two diagnoses represented five of the six cases in their study, and so there were only two cases for which the correct diagnosis was made by 3D-CT but not by 2D ultrasound. In contrast, 3D-CT demonstrated additional, and also more specific skeletal findings that were overlooked by ultrasound in six of our 11 cases; it also helped to establish the correct diagnosis in eight of the 11 cases. Mostly, these specific skeletal deformities affected the vertebral column, the pelvic bones and the ossification points.
In three cases, neither ultrasound nor CT provided the correct diagnosis. In Case 8, 3D-CT identified all the skeletal deformities but the final diagnosis was only made after delivery, based on the association with visceral anomalies. In Cases 10 and 11, 3D-CT was inconclusive. In Case 10, bone density was apparently increased but the diagnosis of osteopetrosis could not be established unequivocally by 3D-CT, which did not demonstrate fractures. In Case 11, 3D-CT showed symmetrical femoral deformities, an enlarged anterior fontanelle, and a decrease in bone density, all of which were suggestive of osteogenesis imperfecta. Again, however, the absence of bone fractures made it impossible to establish the diagnosis firmly.
In conclusion, we found 3D-CT to be more accurate than was 2D ultrasound for the diagnosis of fetal skeletal dysplasia. 3D-CT was more precise in depicting the morphology of the spine (vertebral body shape) and pelvic bones, and in detecting bone synostosis. These abnormalities are often inconspicuous on ultrasound, but may be of great importance in establishing a precise diagnosis. It should be stressed, however, that 3D-CT is currently not sufficiently accurate for the analysis of metaphyseal deformities and for the assessment of bone density.
We suggest that 2D ultrasound and 3D-CT have complementary roles in the management of skeletal dysplasia: ultrasound is best suited as a screening test for the detection of these anomalies and CT is a valuable complementary diagnostic tool. Further studies on a wider range of dysplasias and a larger number of fetuses are now required to validate these findings and conclusions.
We would like to acknowledge our colleagues who referred the patients in this study: Dr L. Cobin, Dr P. Colart, Dr C. Donner, Dr N. Florence, Dr A. François, Dr A. Gans and Dr P. Simon.