Prenatal sonographic diagnosis of skeletal dysplasias

Authors


Abstract

Objective

To assess the types and numbers of cases, gestational age at specific prenatal diagnosis and diagnostic accuracy of the diagnosis of skeletal dysplasias in a prenatal population from a single tertiary center.

Methods

This was a retrospective database review of type, prenatal and definitive postnatal diagnoses and gestational age at specific prenatal diagnosis of all cases of skeletal dysplasias from a mixed referral and screening population between 1985 and 2007. Prenatal diagnoses were grouped into ‘correct ultrasound diagnosis’ (complete concordance with postnatal pediatric or pathological findings) or ‘partially correct ultrasound diagnosis’ (skeletal dysplasias found postnatally to be a different one from that diagnosed prenatally).

Results

We included 178 fetuses in this study, of which 176 had a prenatal ultrasound diagnosis of ‘skeletal dysplasia’. In 160 cases the prenatal diagnosis of a skeletal dysplasia was confirmed; two cases with skeletal dysplasias identified postnatally had not been diagnosed prenatally, giving 162 fetuses with skeletal dysplasias in total. There were 23 different classifiable types of skeletal dysplasia. The specific diagnoses based on prenatal ultrasound examination alone were correct in 110/162 (67.9%) cases and partially correct in 50/162 (30.9%) cases, (160/162 overall, 98.8%). In 16 cases, skeletal dysplasia was diagnosed prenatally, but was not confirmed postnatally (n = 12 false positives) or the case was lost to follow-up (n = 4). The following skeletal dysplasias were recorded: thanatophoric dysplasia (35 diagnosed correctly prenatally of 40 overall), osteogenesis imperfecta (lethal and non-lethal, 31/35), short-rib dysplasias (5/10), chondroectodermal dysplasia Ellis-van Creveld (4/9), achondroplasia (7/9), achondrogenesis (7/8), campomelic dysplasia (6/8), asphyxiating thoracic dysplasia Jeune (3/7), hypochondrogenesis (1/6), diastrophic dysplasia (2/5), chondrodysplasia punctata (2/2), hypophosphatasia (0/2) as well as a further 7/21 cases with rare or unclassifiable skeletal dysplasias.

Conclusion

Prenatal diagnosis of skeletal dysplasias can present a considerable diagnostic challenge. However, a meticulous sonographic examination yields high overall detection. In the two most common disorders, thanatophoric dysplasia and osteogenesis imperfecta (25% and 22% of all cases, respectively), typical sonomorphology accounts for the high rates of completely correct prenatal diagnosis (88% and 89%, respectively) at the first diagnostic examination. Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Congenital skeletal disorders comprise skeletal dysplasias, dysostoses and reduction deformities. Skeletal dysplasias (‘chondrodysplasias’ or ‘osteochondrodysplasias’) are developmental disorders of chondro-osseous tissue. Dysostoses are malformations of single bones, alone or in combination (e.g. isolated polydactyly). Reductions are secondary malformations of bones1, 2. Advances in molecular genetics have helped elucidate the biological basis for many of the diseases in the first two groups, making some of them amenable to molecular genetic diagnosis in addition to the classical diagnostic approach of morphological assessment2. However, specific prenatal diagnosis of skeletal dysplasias still presents a considerable diagnostic challenge, not least because of their rarity3, 4.

Prenatal series and diagnostic algorithms, based on sonographically detectable features of skeletal dysplasias, have been described5–7. We analyzed the data of all suspected and confirmed cases of skeletal dysplasias from a mixed screening and referral population seen in our institution between 1985 and 2007. The purpose of this study was to report the types and proportions of skeletal dysplasias in this group and gestational ages at diagnosis, to carry out a detailed analysis of biometric parameters and to describe the characteristic sonographic features of the most common skeletal dysplasias.

Patients and Methods

Between 1985 and 2007, data were collected prospectively from all fetuses with skeletal dysplasias diagnosed in a mixed screening and referral population in one large tertiary center. For this study, we sought and analyzed details of prenatal management and postnatal pediatric or pathological studies from all cases. Wherever possible, molecular diagnosis was sought.

We used the terms ‘correct ultrasound diagnosis’ and ‘partially correct ultrasound diagnosis’ to describe the accuracy of the first diagnostic examination done in our center compared with the final pediatric or pathological diagnosis. Either there was complete concordance of the specific ultrasound diagnosis, or a skeletal dysplasia other than the one diagnosed by ultrasound was present, or the cases were false positive or false negative.

In all cases, the diagnostic ultrasound examination included measurement of the long bones in all segments of all four extremities, examination of the hands, spine and head and assessment of mineralization and bone shapes, in addition to a full fetal anatomical survey5–7. For the purposes of this study, biometric data (femur length (FL), biparietal diameter (BPD), head circumference (HC), thoracic circumference (ThC) and abdominal circumference (AC)) were collected and analyzed in relation to published normal values. The published normal data from Snijders and Nicolaides8 were used to provide centiles for FL, BPD, HC and AC. Those from Laudy and Wladimiroff9 were used to provide centiles for ThC after 20 weeks; prior to this age the centiles were extrapolated from their normal data. In the graphs each fetus was represented only once, at the gestational age of the first diagnostic examination.

For values beyond the typical normal ranges (such as the 5th and 95th centiles), deviation from the gestational age mean can be expressed best as multiples of the SD or Z-scores, enabling numerical quantification for extreme values such as those encountered in skeletal dysplasias. This also allows comparison of different parameters over gestation on the same y-axis10. Z-scores were calculated for FL, ThC and HC for all affected fetuses. Values in mm and Z-scores were plotted across the gestational age range from 12 to 40 weeks' gestation. We assumed that the reported normal populations were normally distributed.

For the 10 most common skeletal dysplasias encountered in our study, we describe diagnostic sonoanatomical features, derived from clinical experience gathered in this study, a genetic textbook on skeletal dysplasias1, a digital dysmorphology database11 and the online catalog of human genes and genetic phenotypes, OMIM12.

Results

Frequencies of skeletal dysplasias, detection rates and gestational ages at diagnosis

From the database, we identified 178 fetuses in which the diagnosis of ‘skeletal dysplasia’ had been made. Of these, 162 cases were confirmed as skeletal dysplasias, there being 23 different classifiable types. Ten types of skeletal dysplasia occurred more than twice and accounted for 137 of these 162 cases. The two most common types were thanatophoric dysplasia (TD, types 1 and 2; 40/162, 24.7%) and osteogenesis imperfecta (OI types 2, 3 and 4; 35/162, 21.6%). Another 13 known types of skeletal dysplasia occurred once or twice, and there were another nine fetuses with unclassifiable skeletal dysplasias. For the 10 types of skeletal dysplasia that occurred more than twice, the type, number of cases, proportion of completely or partially correct diagnoses and gestational age at diagnosis are shown in Table 1.

Table 1. Prenatal diagnosis of the 10 most common skeletal dysplasias in this study
Skeletal dysplasianFetuses (n) with PD:*GA at PD (completed weeks)Fetuses (n) with specific PD at gestational age:
CorrectPartially correctUp to 14 + 615 + 0 to 19 + 620 + 0 to 21 + 622 + 0 to 28 + 629 + 0 to 34 + 635 + 0 to term
  • *

    Cases diagnosed correctly or partially correct at the first diagnostic examination. ATD, asphyxiating thoracic dysplasia; FN, false-negative diagnosis; GA, gestational age; PD, prenatal diagnosis; Recurr., recurrent/familial case.

Thanatophoric dysplasia4035/405/4016–35 7101382
Osteogenesis imperfecta 2, 3 or 43531/354/3514–3911110841
Short rib dysplasias105/105/1016–34 (1 recurr. at 18 weeks) 3232 
Ellis-van Creveld94/94/914–33 (1 recurr. at 14 weeks; 1 FN at 39 weeks)13122 
Achondroplasia97/91/928–37 (1 FN at 31 weeks) 261
Achondrogenesis87/81/812–1717 
Campomelic dysplasia86/82/816–30 2231 
ATD Jeune73/74/719–35 11221
Hypochondrogenesis61/65/621–32 132 
Diastrophic dysplasia52/53/520–25 32 

The overall prenatal detection rate for all cases with confirmed skeletal dysplasias was 98.8% (160/162). The rates of completely and partially correct prenatal sonographic diagnosis at the first diagnostic examination were 67.9% (110/162) and 30.9% (50/162), respectively. There were two false negatives: one case with chondroectodermal dysplasia Ellis-van Creveld (prenatally diagnosed at 39 completed weeks as ‘suspected genetic syndrome, short long bones’) and one with achondroplasia (diagnosed at 31 weeks as ‘structurally normal fetus, suspect familial large head, intrauterine growth restriction (IUGR) affecting mainly extremities’).

The less common types of skeletal dysplasia included two cases with chondrodysplasia punctata (diagnosed correctly at 21 and 23 weeks), two cases of spondyloepimetaphyseal dysplasia (one diagnosed correctly at 12 weeks and one partially correctly at 32 weeks), two cases with lethal hypophosphatasia (diagnosed partially correctly at 16 and 19 weeks), one case each of dyssegmental dysplasia Rolland Desbuquois (diagnosed partially correctly as ‘either OI or campomelic dysplasia’ at 28 weeks), Kniest dysplasia (diagnosed partially correctly at 19 weeks), kyphomelic dysplasia (diagnosed correctly at 31 weeks), opsismodysplasia (misdiagnosed as TD at 33 weeks), osteocraniostenosis (misdiagnosed as TD at 21 weeks), otopalatodigital syndrome type 2 (diagnosed correctly at 22 weeks), platyspondylic chondrodysplasia Shiraz type (diagnosed correctly at 28 weeks), spondyloepiphyseal dysplasia (diagnosed partially correctly at 35 weeks), sponastrimic dysplasia (diagnosed partially correctly at 29 weeks), metaphyseal dysplasia McKusick type (diagnosed partially correctly at 27 weeks) and nine skeletal dysplasias that remained unclear even at pediatric or pathological examination (diagnosed partially correctly at 13–30 weeks).

In addition to the 162 true skeletal dysplasia cases there were another 16 cases in which the diagnosis ‘skeletal dysplasia’ was made prenatally. Of these, four cases were lost to follow-up, and in 12 cases no skeletal dysplasia was found postnatally. However, in seven of these 12 false positives there was significant morbidity: dysostoses (n = 2), IUGR (n = 3), unclear dysmorphic syndrome (n = 1) and IUGR with joint position anomalies and porencephaly (n = 1).

The gestational age (GA) at the first specific prenatal ultrasound diagnosis depended on the type of skeletal dysplasia (Table 1 and Figure 1) and (more weakly) on the year in which the case occurred (Figure 2). Figure 1 shows the numbers of prenatally diagnosed cases according to gestational age group for the 10 most common dysplasias. All fetuses with achondrogenesis, an extremely severe skeletal dysplasia, were diagnosed between 12 and 17 completed weeks. In the two largest groups (TD and OI) there was a decrease in gestational age at diagnosis (Figure 2, solid and dashed lines) to about 20 weeks in the more recent years, while achondroplasia consistently became apparent only in late gestation. Overall, in the years before 1995, the median GA at diagnosis was 24 weeks, with 42% of the cases being diagnosed before 24 weeks. After 1995, 62% of cases were diagnosed before 24 weeks and the median GA at diagnosis was 21 weeks.

Figure 1.

Prenatally diagnosed skeletal dysplasias: numbers of cases by gestational age groups for the 10 most common dysplasias: thanatophoric dysplasia (types 1 and 2) (equation image); osteogenesis imperfecta (all types) (equation image); short rib dysplasias (equation image); Ellis-van Creveld (equation image); achondroplasia (equation image); achondrogenesis (equation image); campomelic dysplasia (equation image); asphyxiating thoracic dysplasia Jeune (equation image); hypochondrogenesis (equation image) and diastrophic dysplasia (equation image). The peak of diagnoses is between 15 and 29 weeks, but some types are often first apparent in the third trimester.

Figure 2.

Gestational age at diagnosis of skeletal dysplasias according to year of diagnosis, for the 10 most common dysplasias (thanatophoric dysplasia (types 1 and 2) (equation image); osteogenesis imperfecta (all types) (equation image); short rib dysplasias (equation image); Ellis-van Creveld (equation image); achondroplasia (equation image); achondrogenesis (equation image); campomelic dysplasia (+); asphyxiating thoracic dysplasia Jeune (equation image); hypochondrogenesis (equation image); and diastrophic dysplasia (equation image)) and all other skeletal dysplasias (equation image). Linear regression lines indicate possible trends of the gestational age at diagnosis over the study period: thanatophoric dysplasia (types 1 and 2) (equation image); osteogenesis imperfecta (all types) (equation image); and achondroplasia (equation image).

Biometric parameters

The FL, BPD, HC and ThC data were available from 177, 176, 167 and 137 fetuses, respectively, at the first diagnostic examination (AC data, which did not contribute to diagnosis, are not included).

The FL was short (by definition (< 5th centile) as well as Z-scores < −2) in all cases of skeletal dysplasia except in four fetuses: one with FL at the 8th centile (−1.4 SD) at 18 weeks, in which the prenatal/final diagnosis was mesomelic campomelia; one with FL at the 81st centile (+0.9 SD) at 27 weeks, and a prenatal/final diagnosis of ‘unclassifiable skeletal dysplasia’; one with FL at the 8th centile (−1.4 SD) at 23 weeks, and a prenatal/final diagnosis of ‘complex syndrome’/hypochondrogenesis; and one with FL at the 9th centile (−1.3 SD) at 31 weeks and a prenatal/final diagnosis of ‘structurally normal fetus, suspected IUGR’/achondroplasia. In Figure 3a the FLs are expressed as absolute measures in mm (and in Figure S1 online they are expressed as Z-scores) according to gestational age. Achondrogenesis and TD showed the most severely abnormal FL.

Figure 3.

Femur length (a), thoracic circumference (b) and head circumference (c) plotted against gestational age in skeletal dysplasias (thanatophoric dysplasia (types 1 and 2) (equation image); osteogenesis imperfecta (all types) (equation image); short rib dysplasias (equation image); Ellis-van Creveld (equation image); achondroplasia (equation image); achondrogenesis (equation image); campomelic dysplasia (equation image); asphyxiating thoracic dysplasia Jeune (equation image); hypochondrogenesis (equation image); and diastrophic dysplasia (equation image)) and for all other skeletal dysplasias (equation image). The 5th and 95th centiles are indicated8, 9; for thoracic circumference, centiles before 20 weeks' gestation were extrapolated from normal data9. Each fetus is plotted only once, at the gestational age of the first diagnostic examination.

Lethality in skeletal dysplasias is typically caused by impaired lung development secondary to a narrow thorax13. In Figure 3b ThCs are expressed as absolute values (and in Figure S2 online they are expressed as Z-scores).

The HC was typically increased in some skeletal dysplasias: in 20 of the 38 cases of TD with HC measurement, in four of the eight such cases of campomelic dysplasia, in two of the nine such cases of achondroplasia and in two of the 10 such cases of short-rib dysplasia. In Figure 3c the HCs are expressed as absolute measurements (and in Figure S3 online they are expressed as Z-scores).

Sonographic assessment of lethality

There were 114 fetuses with lethal skeletal dysplasias. Of these, 113 were correctly diagnosed as lethal (99%) and one with hypophosphatasia (Z-score for FL was −3.1 and for ThC was −0.1) was misdiagnosed as having non-lethal skeletal dysplasia.

There were 17 cases of mostly lethal skeletal dysplasias, of which 10 (59%) had a correct specific diagnosis. Five fetuses with Ellis-van Creveld were misdiagnosed (three as lethal and two as non-lethal skeletal dysplasia), one case with asphyxiating thoracic dysplasia Jeune (ATD Jeune, or Jeune syndrome) was suspected to have a non-lethal skeletal dysplasia (the patient opted for termination) and one with Rolland-Desbuquois syndrome (misdiagnosed at 28 weeks as OI, liveborn).

In 28 of 31 (90%) fetuses affected by non-lethal conditions this status was correctly diagnosed. Three were misdiagnosed as having lethal skeletal dysplasias, but all of these resulted in live births; these included one fetus with achondroplasia first presenting at 37 weeks, one with either spondyloepiphyseal dysplasia or Kniest dysplasia presenting at 35 weeks and one with an unclear skeletal dysplasia presenting at 27 weeks.

Diagnostic sonographic criteria

TD was the most common skeletal dysplasia in our study (40 of 162). Its characteristics include severe micromelia and brachydactyly. In TD1 the long bones are bowed (‘telephone receiver femur’), while they are straight in TD2. In spite of severe platyspondyly, the trunk length is normal. The thorax is narrow, with short ribs and a prominent abdomen. The HC is large in half of cases, with the nasal bridge depressed, and in some (mainly TD2) there is craniosynostosis (‘cloverleaf skull’) (see Figure S4 online for typical sonographic features). Some fetuses with TD have hydrocephaly. In the third trimester, polyhydramnios is common. Figure 4a shows the biometric parameters HC, ThC and FL in cases of TD. Growth in FL slowed down with advancing gestation, leading to steadily decreasing Z-scores with advancing age. Fetuses with TD had ThC measurements markedly below the 5th centile with only three exceptions: one fetus with TD1 at 16 completed weeks had a ThC at the 11th centile, one with TD1 at 21 weeks had a ThC at the 10th centile and one with TD2 at 24 weeks had a ThC at the 14th centile. There were several fetuses with TD diagnosed below the gestational age range of the normal curves we used for ThC (i.e. below 20–40 weeks' gestation, Laudy and Wladimiroff9). In 16 of 32 (50%; no measurement was available in two cases) of the fetuses with TD1 and in four of six of those with TD2, the HC was above the 95th centile.

Figure 4.

Head circumference (filled symbols), thoracic circumference (open symbols) and femur length (+ and − symbols) in thanatophoric dysplasia Type 1 (●, ○, −) and Type 2 (▪, □, +) (a) and in osteogenesis imperfecta lethal (●, ○, −) and non-lethal (▪, □, +) forms (b), with 5th and 95th centiles indicated8, 9. For thoracic circumference, centiles before 20 weeks' gestation were extrapolated from normal data9.

OI was the second most common skeletal dysplasia in our study (35 of 162). Characteristics of the lethal type 2 (29 of 35) include severe micromelia, typically with irregular bending of long bones and ribs due to multiple intrauterine fractures. The affected bones are very short and irregular and the thorax is narrow. The HC is mostly normal; the skull, however, shows an abnormal sonographic translucency and compressibility (see Figure S4 online for typical sonographic features). The spine is sometimes irregular in appearance, with platyspondyly due to fractures of the vertebrae. It may be difficult to distinguish OI type 2 from the lethal form of hypophosphatasia using sonography alone. In the non-lethal forms of OI, usually only single tubular bones are affected and the ThC and HC are normal (Figure 4b). In cases with OI, the FL slowed down with advancing gestation, leading to steadily decreasing Z-scores with advancing age. All fetuses with lethal OI had a FL below the 5th centile or below −2 SD for gestational age. Among the six cases with non-lethal OI, only one fetus had a normal FL (at 25 weeks, with a FL at the 33rd centile and Z-score of −0.5). FL Z-scores decreased markedly with gestational age in lethal OI, and the shortening was more pronounced than in non-lethal forms. The ThC measurements in OI were available for 27 of the 35 cases. In confirmed lethal OI, the ThC (data available on 22 cases) was below the 5th centile in all but one of the fetuses (which had a ThC at the 5.4th centile at 26 weeks). In the five cases with non-lethal forms of OI and data on ThC available, it was normal in two, low normal in two and abnormally small in one.

The short-rib dysplasias are a group of lethal skeletal dysplasias with autosomal recessive inheritance, characterized by micromelia with or without polydactyly (mostly postaxial). The thorax is narrow with short, horizontally oriented ribs and a protuberant abdomen, and there are a variety of other possible malformations of the internal organs. There are four types of short-rib dysplasia, but differentiation is difficult because of overlapping symptoms. Type 1 (Saldino–Noonan) is characterized by polydactyly of the hands and sometimes of the feet, often congenital heart disease, sometimes facial clefts, brain anomalies and anomalies of the kidneys and genital tract. Type 2 (Majewski) is characterized by pre- or postaxial polydactyly and brachydactyly of the hands and sometimes of the feet. The head is large, the nasal bridge is depressed, the mandible is small (micrognathia) and there are often facial clefts. Internal malformations include cystic dysplasia of the kidneys and hypoplasia of the cerebellar vermis. Type 3 (Verma–Naumoff) is characterized predominantly by postaxial polydactyly and brachydactyly of the hands and feet and gastrointestinal, renal and other defects. Types 1, 2 and 3 can exhibit fetal hydrops and situs inversus. Type 4 (Beemer–Langer) is characterized by bowed limbs, pre- and postaxial polydactyly in half of cases, facial clefts in nearly all cases, and, less commonly, cardiac, cerebral, gastrointestinal or renal anomalies and omphalocele. In this study there were two cases of short-rib dysplasia Type 3, one of Type 2, two of Type 4 and a further five other cases including one recurrent case (diagnosed at a targeted scan at 18 weeks).

Chondroectodermal dysplasia Ellis-van Creveld and ATD Jeune are also grouped with the short-rib dysplasias and are mostly lethal. In Ellis-van Creveld, there is an acromesomelic micromelia with polydactyly of the hands in most cases and of the feet in some cases. Half of the fetuses have structural heart disease (atrial or ventricular septal defect) and some have a Dandy–Walker malformation. In ATD, the thorax is long and narrow, the long bones have mild rhizomelia and some have polydactyly. Most affected fetuses have cystic dysplastic kidneys. We diagnosed eight cases of Ellis-van Creveld (at 14–33 weeks plus one false-negative at 39 weeks) and seven cases of ATD Jeune (at 19–35 weeks).

All eight cases of achondrogenesis were diagnosed between 12 and 17 weeks. In achondrogenesis there is early hydrops and a short trunk (short crown–rump length), with a very narrow barrel-shaped thorax and a markedly prominent abdomen. In achondrogenesis types IA/B, both of which are autosomal recessive, there is extreme micromelia, with short hands and feet, poor mineralization, a large head, a flat face and a short neck. Achondrogenesis Type II (autosomal dominant) is less severe, presenting somewhat later in gestation than does Type I and frequently having polyhydramnios. Hypochondrogenesis (autosomal dominant) represents a clinical variant within the achondrogenesis–hypochondrogenesis spectrum, with a small thorax, short limbs, a flat face with micrognathia, a short trunk and macrocephaly. The nose is flat and the nasal bridge is depressed, resembling TD. The six cases of hypochondrogenesis in our study were diagnosed between 21 and 32 weeks. Both achondrogenesis and hypochondrogenesis are lethal.

Campomelic dysplasia (autosomal dominant) is characterized by mesomelia and mild to moderate bowing of the legs and, less commonly, of the arms. Anterior angulation of the tibia (see Figure S4f online) and the short fibula are a pathognomonic feature of campomelic dysplasia. The fingers and toes are short and the feet are clubbed. The face is flat, with micrognathia and low-set ears, and there may be nuchal edema. The thorax is small and bell-shaped, with 11 pairs of ribs, and the scapulae and claviculae are hypoplastic. Occasionally there is hydronephrosis. In some affected chromosomally male fetuses sex reversal occurs, as the mutated gene, SOX9, is also involved in gonadal development. Our eight cases were diagnosed between 16 and 30 weeks.

Diastrophic dysplasia (autosomal recessive) is a non-lethal skeletal dysplasia without mental impairment. The long bones are thick, markedly shortened and straight. The fingers are short, with ulnar deviation and ‘hitchhiker’ thumbs. There are clubfeet and micrognathia. In our cohort there were five cases diagnosed between 20 and 25 weeks.

Achondroplasia (autosomal dominant) is a non-lethal skeletal dysplasia without mental impairment. All nine of our cases were diagnosed after 27 weeks (at 28–37 weeks). At around 20 weeks' gestation, fetuses with achondroplasia had normal biometric parameters, including FL, which became abnormally short only in the third trimester. The micromelia is rhizomelic and the heads tend to be large. Typical facial features include prominent forehead, depressed nasal bridge and mid-face hypoplasia. The phalanges are short; typical gaps between the fingers and digital deviation lead to the appearance of a ‘trident’ hand.

The chondrodysplasia punctata group contains 11 skeletal dysplasias, some of which have been detected prenatally. The mostly lethal type, the rhizomelic form (autosomal recessive), is characterized by a moderate rhizomelic shortening of the otherwise straight long bones, exhibiting typical hyperechogenic areas (premature calcification) in the cartilaginous skeletal parts, i.e. the epiphyses of the long bones (see Figure S4g) or the calcaneus. The face is flat and there is micrognathia and sometimes microcephaly. The thorax is normal in size. The rhizomelic form is associated with severe mental retardation. The two cases in our cohort were diagnosed at 19 and 21 weeks.

Discussion

To the best of our knowledge this is the largest single-center study of sonographically diagnosed fetal skeletal dysplasias: we investigated 162 affected fetuses with confirmed clinical genetic, pathological or molecular diagnosis. Prenatal sonographic diagnosis of fetal skeletal dysplasias is challenging because of their rarity (21–47 per 100 000 deliveries3, 4), the multitude of differential diagnoses with overlapping features and the phenotypic variability. With the discovery of the molecular basis of many of the skeletal dysplasias their classification has evolved, from being based initially on morphological, mostly radiological features only, to being based largely on pathogenesis2. In fact, molecular genetics is the gold standard for an increasing number of skeletal dysplasias, and it may be the only diagnosis if the pregnant woman opts for termination or when pathological diagnosis is not possible4, 7. A revised classification of skeletal dysplasias, incorporating newly recognized disorders, pathogenetic concepts and molecular and biochemical properties was recently provided by Superti-Furga and Unger2. Web resources can aid in the selection of appropriate molecular genetic tests (for an example and selection of links, see Goncalves et al.7 and Superti-Furga and Unger2).

Classification and frequencies of skeletal dysplasias

The most recent classification of genetic skeletal disorders includes 372 different conditions in 37 groups defined by molecular, biochemical and/or radiological criteria2. Of them, 215 have been found to be associated with mutations in 145 genes, but there are still some entities which are not clinically classifiable. Prenatal identification of the specific type of a particular skeletal dysplasia requires a detailed and structured examination, for which protocols have been suggested5, 7. Besides the 23 different types of skeletal dysplasia that we diagnosed in our series, nine of the 162 (5%) cases could not be classified. Rasmussen et al.4, Gaffney et al.14 and Doray et al.15 reported 16%, 20% and 6% unclassifiable skeletal dysplasias, respectively, in their cohorts.

Our cases were from a mixed screening and referral population. We therefore could not calculate the incidence of individual skeletal dysplasias, and the true detection rate was difficult to establish. However, the most common types of skeletal dysplasia in this fetal population and their frequencies were similar to those found in previous studies (Table 2).

Table 2. Frequencies of skeletal dysplasias reported in different prenatal series
Final diagnosisThis study (n = 162)Sharony et al.16 (1993) (n = 172)Goncalves and Jeanty20 (1994) (n = 139)Gaffney et al.14 (1998) (n = 25)Doray et al.15 (2000) (n = 42)
  1. Values are given as n (%) for cases with true skeletal dysplasias. Goncalves and Jeanty20: 127 from multiple centers plus 12 from previous publications; Sharony et al.16: 54 false-positives; Doray et al.15 five false-positives; Gaffney et al.14 three false-positives plus seven lost to follow-up. ATD, asphyxiating thoracic dysplasia.

Thanatophoric dysplasia40 (25)27 (16)43 (31)5 (20)4 (10)
Osteogenesis imperfecta 2, 3 or 435 (22)37 (22)39 (28)8 (32)13 (31)
Short-rib dysplasia10 (6)8 (5)6 (4)2 (8)1
Ellis-van Creveld9 (6)02 (1)03 (7)
Achondroplasia9 (6)4 (2)15 (11)4 (16)7 (17)
Achondrogenesis8 (5)17 (10)9 (6)3 (12)5 (12)
Campomelic dysplasia8 (5)14 (8)4 (3)11
ATD Jeune7 (4)3 (2)002 (5)
Hypochondrogenesis6 (4)0000
Diastrophic dysplasia5 (3)2 (1)3 (2)01

Diagnostic accuracy of prenatal diagnosis

We described the accuracy of our prenatal diagnosis with regard to the final diagnosis, as previously reported14–16, as either ‘correct ultrasound diagnosis’ or ‘partially correct ultrasound diagnosis’. There are few reports with data on the diagnostic accuracy of prenatal diagnosis of skeletal dysplasias14–19 (Table 3), not all of which include numbers of false negatives and false positives. The accuracy of prenatal diagnosis seems to have improved in recent years.

Table 3. Studies reporting the diagnostic accuracy of prenatal ultrasound of skeletal dysplasias
StudySingle/multicenternUltrasound diagnosisFalse positive (n)False negative (n)
Correct (%)Partially correct (%)
  • *

    Included only lethal skeletal dysplasias. ND, not done/data not given.

Sharony et al.16 (1993)Multi172356554ND
Tretter et al.17 (1998)*Single26485210
Gaffney et al.14 (1998)Single2544563ND
Hersh et al.18 (1998)Single265050NDND
Doray et al.15 (2000)Multi4262385ND
Parilla et al.19 (2003)Single20653570
This studySingle1626831122

In the studies listed in Table 3, the false positives were mainly fetuses with dysmorphic syndromes or IUGR. Sharony et al.16, for example, reported that among 54 false positives in their study, 36 had dysmorphic syndromes and 15 were structurally normal IUGR fetuses. In our study, among 12 false positives, there were two fetuses with dysostoses, two with dysmorphic syndromes and eight with IUGR or which were normal small babies.

So far only a few studies have discussed false-negative diagnoses of skeletal dysplasias. Tretter et al.17 and Parilla et al.19 report that they had no false negatives. We had two false-negative diagnoses: one fetus with achondroplasia was wrongly diagnosed as ‘structurally normal fetus, suspect familial large head, IUGR affecting mainly extremities’ and one with Ellis-van Creveld was diagnosed as ‘suspected genetic syndrome, short long bones’. Ascertainment of the false negatives was limited because of incomplete follow-up in ‘suspected normals’. Our overall follow-up for (supposedly) normal cases over the study period was about 80%. Also, some families of affected cases did not respond spontaneously; in our lost-to-follow-up group there was one fetus with a clear skeletal dysplasia whose parents refused to provide any postnatal data to us.

The ability to achieve the correct specific diagnosis by prenatal ultrasound depends on the type of skeletal dysplasia. The two most common skeletal dysplasias in our study, TD and OI, were diagnosed correctly in 88% and 89% of cases, respectively. The data for the other most common skeletal dysplasias and for previous studies are shown in Table 4.

Table 4. Diagnostic accuracy of prenatal ultrasound for the 10 most common skeletal dysplasias in this study
Final diagnosisThis study (n = 162)Sharony et al.16 (1993) (n = 172)Tretter et al.17 (1998) (n = 26)Gaffney et al.14 (1998) (n = 25)Doray et al.15 (2000) (n = 42)
  • Values are given as n (%).

  • *

    Detailed evaluation not possible (figures shown here calculated from Sharony et al.16, as exact numbers were not given by the authors).

  • Figures quoted in the text.

  • Included in short-rib dysplasia. ATD, asphyxiating thoracic dysplasia; NA, data not available.

Overall110/162 (68)60/172 (35)*13/26 (50)11/25 (44)30/42 (71)
Thanatophoric dysplasia35/40 (88)(70)6/12 (50)2/5 (40)3/4 (75)
Osteogenesis imperfecta 2, 3 or 431/35 (89)(50)5/6 (83)5/8 (63)9/13 (69)
Short-rib dysplasia5/10 (50)NA1/3 (33)0/20/1
Ellis-van Creveld4/9 (44)NA003/3 (100)
Achondroplasia7/9 (78)NA02/4 (50)7/7 (100)
Achondrogenesis7/8 (88)NA1/2 (50)1/3 (33)3/5 (60)
Campomelic dysplasia6/8 (75)NA0/20/11/1 (100)
ATD Jeune3/7 (43)NA02/2 (100)
Hypochondrogenesis1/6 (17)NA0/100
Diastrophic dysplasia2/5 (40)NA000/1

Lethality

Lethality in skeletal dysplasias is caused mainly by pulmonary hypoplasia secondary to thoracic hypoplasia13. The discrimination between lethal and non-lethal forms of skeletal dysplasias is, of course, of major clinical importance. In our study, with one exception, all fetuses with lethal disease were correctly classified (113/114, 99%); the misdiagnosis occurred in a fetus with hypophosphatasia and a normal ThC (Z-score, −0.1) examined at 19 weeks.

Among the mostly lethal conditions, Ellis-van Creveld was the most difficult to classify correctly: in five of nine affected fetuses the sonographic assessment of lethality was imprecise. Three of 31 fetuses with non-lethal skeletal dysplasias were misdiagnosed (at 27, 35 and 37 weeks), but all three cases resulted in live births.

Four other studies have addressed the ability to discriminate between lethal and non-lethal skeletal dysplasias: Hersh et al.18 predicted lethality correctly in 92% of cases (23 fetuses with lethal and three with non-lethal conditions). Gaffney et al.14 correctly identified all 20 cases with lethal and two of five with non-lethal skeletal dysplasias. Doray et al.15 identified 17 of 21 lethal cases, five of eight mostly lethal cases and all eight non-lethal cases; in five cases they could not determine lethality. Parilla et al.19 only report on their 16 lethal cases of skeletal dysplasia, all of which were diagnosed correctly.

Morphology

In the two most common skeletal dysplasias, TD and OI, their typical sonomorphological signs accounted for the high rate of overall and correct specific diagnosis. In addition, for some skeletal dysplasias, there are almost always subtle but pathognomonic signs which can be found within the skeletal apparatus, the face or the internal organs. Examples include the angulated tibia in campomelic dysplasia or polydactylies. In the Results section we provide descriptions of the sonographic morphology of the 10 most common skeletal dysplasias.

There are, of course, pitfalls in this sonomorphological approach. For example, in one case in which ‘TD’ was diagnosed at 33 weeks (ThC Z-score, −2.4; FL Z-score, −7.8), the fetus actually had opsismodysplasia (micromelic dwarfism, frontal bossing and short metacarpals).

Gestational ages at diagnosis

For many reasons, including termination laws, a timely specific prenatal diagnosis is important. In Germany, compulsory routine measurement of the fetal FL was introduced in 1995. In our cohort, there was a trend towards earlier prenatal diagnosis over the years, raising the percentage of correct prenatal diagnosis before 24 weeks from 42% to 62% in the last 10 years, mainly due to earlier diagnosis of TD and lethal OI. For France and the UK, both with ultrasound screening programs, the reported percentages of diagnosis before 24 weeks are similar (62% and 71%14, 15). In the USA, where ultrasound screening is not universal, these percentages range from 37%4 to 58%16 or 62%17.

In our study, achondroplasia was the only skeletal dysplasia that was not detected before 24 weeks. In all nine cases, the FLs at 20 weeks had been reported as normal by the referring gynecologists. In most studies achondroplasia is detected late in pregnancy: Gaffney et al.14 reported four cases detected at 31–47 weeks, while Doray et al.15 reported seven fetuses detected at 28–35 weeks. Goncalves and Jeanty20 reported 15 fetuses with achondroplasia: two had femur measurements before 24 weeks and both were normal, while another two measured at 25 and 36 weeks also had normal FL. There is only one recent report, by Tonni et al.21, in which rhizomelic shortening of the long bones (< −2 SD) was diagnosed at 17 weeks in a fetus with achondroplasia.

Biometry

In ultrasound screening, the FL is the best parameter for the detection of skeletal dysplasias20. However, not all cases of skeletal dysplasia have shortening below −2 SD. In our study, there were five cases with a FL greater than −2 SD: one case each of achondroplasia (FL, −1.3 SD at 31 weeks), hypochondrogenesis (FL, −1.4 SD at 23 weeks), OI Type 4 (FL, −0.5 SD at 32 weeks) and short-rib dysplasia (FL, −1.4 SD at 18 weeks) and one with an unclear skeletal dysplasia (FL, −0.9 SD at 27 weeks). Goncalves and Jeanty20 reported nine of 137 fetuses (6.6%) with a FL above the 5th centile (five with achondroplasia, three with hypochondroplasia and one with OI). They commented that the FL is the best parameter to distinguish at least among the five most common skeletal dysplasias. Our data agree with this, the measurements of the other long bones and the head and thorax circumferences contributing mainly to the differential diagnoses.

Extreme deviations of biometric measurements are typical for many skeletal dysplasias, which therefore cannot be expressed meaningfully as centiles or graphically in typical ‘normal ranges’ with centiles. The World Health Organization (WHO) recommends Z-scores for comparison of anthropometric measurements with the normal ranges10. We therefore decided to analyze and display the quantitative data both in the ‘conventional’ way of plotting the affected values on normal ranges and to display them as Z-scores.

In addition to showing the quantitative excess of an affected parameter more clearly, Z-scores allow comparison of multiple biometric parameters over gestational age in the same clear x-y plot. This is evident, for example, in the biometric synopsis of FL, ThC and HC for TD and OI (Figure 4 and Figures S5 and S6 online). Additionally, for some skeletal dysplasias, the biometric measurements deviated progressively from the expected mean as the gestational age advanced. This can be seen particularly well in the Z-score plots, for example in the FL Z-score curves for TD and OI (Figures S5 and S6 online). Thus we chose this approach over the one used in another large biometry study of fetal skeletal dysplasias, which grouped affected fetuses by percentages of the mean20.

Genetic analysis

Defects of both pre- and postnatally expressing genes can lead to skeletal dysplasias. Genetically determined growth changes, mostly symmetrical, are apparent in all skeletal segments in which the particular gene is expressed, but the severity of the effect may vary in the affected segments1, 2. Signalling genes (e.g. mutated fibroblast growth factor receptor FGFR-3 leading to reduced enchondral chondrocyte proliferation in achondroplasia) and genes regulating cell structure and function such as production or removal of matrix components (e.g. faulty collagen COL1A1 and COL1A2 in OI) can be involved.

We performed additional molecular studies for suspected skeletal dysplasias prenatally or postnatally as available at that time (data not shown), but included only the sonographic information in our analysis of detection rates and diagnostic accuracy. Molecular genetics can contribute essential diagnostic information and complement prenatal sonographic assessment4, 7. By how much the diagnostic accuracy can be improved by integrating molecular studies into the prenatal diagnostic approach remains to be determined. In this study, we have shown that a specific diagnosis is possible by prenatal ultrasound alone in the majority of cases. In inconclusive cases ultrasound can be used to narrow the possible diagnoses down in order to select appropriate further tests, including molecular genetic analysis7, which can accurately assess recurrence risks. For example, in a fetus with the sonographic features of OI Type 2, the final molecular diagnosis revealed lethal hypophosphatasia, raising the recurrence risk from around 5% (empirically for OI Type 2) to 25% (autosomal recessive). Table S1 (online) lists the known genetic causes for the most common skeletal dysplasias in our study.

Conclusion

In conclusion, we have analyzed the largest prenatal single-center series of skeletal dysplasias, examining 162 fetuses and at least 23 different types of dysplasia. In the majority of cases the initial ultrasound examination was diagnostic, with either a fully or at least a partially correct diagnosis. In particular, the diagnostic accuracy with regard to lethality was high. Biometric parameters in skeletal dysplasias are best expressed as Z-scores. Typical morphology often allows a specific prenatal ultrasound diagnosis. In the most common fetal skeletal dysplasias ultrasound diagnosis can be confirmed by molecular genetic testing, which should be sought for quality assurance.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article:

Normal ranges were taken from Snijders and Nicolaides8 for femur length, head circumference and abdominal circumference (for 14–40 weeks) and from Laudy and Wladimiroff9 for thoracic circumference after 20 weeks; prior to 20 weeks, the thoracic circumference normal range was extrapolated from Laudy and Wladimiroff9. Each fetus is plotted only once at the gestational age of the first diagnostic examination. TD, thanatophoric dysplasia; OI, osteogenesis imperfecta; AP, achondroplasia; AG, achondrogenesis; EvC, Ellis-van Creveld; Jeune, asphyxiating thoracic dysplasia Jeune; SR, Short rib dysplasia; HG, hypochondrogenesis; CD, campomelic dysplasia.

Figure S1 Femur length in skeletal dysplasias expressed as Z-scores against gestational age.

Figure S2 Thoracic circumference in skeletal dysplasias expressed as Z-scores against gestational age.

Figure S3 Head circumference in skeletal dysplasias expressed in Z-scores against gestational age.

Figure S4 Characteristic sonographic appearances of skeletal dysplasias encountered during the study.

Figure S5 Biometric parameters in thanatophoric dysplasia Types 1 and 2, expressed in Z-scores.

Figure S6 Biometric parameters in osteogenesis imperfecta lethal and non-lethal forms, expressed in Z-scores.

Table S1 Molecular diagnosis in the 10 most common skeletal dysplasias in this study.

Ancillary