Chapter 14: FETAL ANOMALIES: OVERVIEW

The detection of anatomic congenital anomalies is one of the goals of prenatal care.¹ The information required for diagnosis and management of the obstetrical patient with a fetal congenital anomaly demands knowledge in a variety of disciplines, including embryology, anatomy, genetics, obstetrics, pediatric surgery, teratology, and, certainly, diagnostic imaging. Many health care professionals involved in diagnostic imaging have had limited exposure to the other fields. The purpose of this chapter is to introduce the reader to congenital anomalies, and provide an overview of the definition and magnitude of the problem and pathogenic mechanisms of gross congenital anomalies. The principles of prenatal diagnosis with ultrasound, the use of ultrasound as a screening tool for the detection of congenital anomalies, and the management options once a congenital anomaly has been detected are also discussed.

A congenital anomaly consists of a departure from the normal anatomic architecture of an organ or system. Anomalies may result from an intrinsically abnormal primordium or anlage of an organ, or from a normal primordium that is affected during development by extrinsic forces.^{2,3, and 4} Growing interest in prenatal development, coupled with the need for a uniform nomenclature to refer to errors in morphogenesis, led an international working group to propose a set of terms useful in the classification of anatomic congenital anomalies. Individual alterations of form or structure can therefore be classified as malformations, deformations, and disruptions. However, it should be stressed that it is not always possible to assign an anomaly to a specific class. In fact, malformations, deformations, and disruptions may overlap (see section on sequences).

Malformation

A malformation is a morphologic defect of an organ, part of an organ, or a larger area of the body resulting from an intrinsically abnormal developmental process. The term intrinsically abnormal developmental process refers to an abnormality in the primordium (anlage) of the organ. This abnormality may not be identifiable in early stages of development. The typical example is a limb bud that appears normal in early embryonic life, but later develops an extra digit. Malformations can be considered the result of a developmental arrest of the primordium (incomplete morphogenesis), redundant morphogenesis, or aberrant morphogenesis (Figure 14-1). Examples of these types of malformations are listed in Table 14-1. Although malformations often occur during the embryonic period (until the ninth postmenstrual week),⁵ some may also arise during later stages of development. A general principle is that the earlier the malformation is initiated, the more complex the resulting anomaly (or anomalies).

Deformation

Deformation refers to an abnormal form, shape, or position of part of the body caused by nondisruptive mechanical forces. The primordium of the organ is normal, but development is affected by mechanical forces that are extrinsic or intrinsic to the fetus. For example, a club foot deformity (Figure 14-2) may be the result of intrauterine constraint due to oligohydramnios (extrinsic force) or lack of movement due to the neural defect associated with spina bifida (intrinsic force). Table 14-2 presents common forces leading to deformations. Four main factors influence the pathogenesis of deformations: pressure, fetal plasticity, fetal mobility, and the rate of fetal growth.^{6,7,8, and 9} Deformations tend to occur late in gestation, because during this time there is rapid fetal growth in a potentially constraining intrauterine environment.^{6,7,8,9,10, and 11} Removal of the mechanical force responsible for the deformation results in normalization or improvement of the anomaly. Spontaneous resolution after birth occurs in approximately 90% of deformations.¹² Table 14-3 compares malformations and deformations. Therefore, in general, the term malformation describes defects that are likely to have arisen during organogenesis, and the term deformation is reserved for defects arising after the embryonic period.

Disruption

A disruption is a morphologic defect of an organ, part of an organ, or a larger region of the body resulting from a breakdown or interference with an originally normal developmental process. A typical example of this type of anomaly is digital amputation associated with amniotic band syndrome (Figure 14-3).^{13,14, and 15} Disruptions are sporadic events.

Dysplasia

Another concept frequently used by dysmorphologists is that of dysplasia, a term referring to abnormal organization of cells into tissue(s) and its morphologic result(s). The term dysplasia in pathology refers to an anaplastic process. Its use in dysmorphology is broader and refers to any type of tissue disorganization. Osteogenesis imperfecta is a dysplasia in which the primary disorder affects collagen; accordingly, structures containing significant quantities of the particular type of defective collagen are affected.

The Fetus with Multiple Anomalies

A fetus may have multiple congenital anomalies. This association may occur simply by chance, or may be part of a pathogenetically related event. A set of terms has been coined to describe the relationship between coexistent anomalies: polytopic field defect, sequence, syndrome, and association.

Polytopic Field Defect

A polytopic field defect is a group of anomalies derived from the disturbance of a single developmental field. A developmental field is a region or part of an embryo that responds as a unit to embryonic interactions, and results in complex or multiple anatomic structures. Opitz discussed in detail the meaning and implications of the concept of developmental field defects.^{4,16,17,18, and 19} The embryo is omnipotent (the primary field) up to a certain time, during which further organization and differentiation occurs in a number of different developmental autonomous areas (secondary fields). Disturbances in a developmental field may result in multiple, and usually contiguous, anomalies (monotopic field defects) or multiple, distantly located anomalies (polytopic field defects). An example of a monotopic field defect is holoprosencephaly, where there are coexistent abnormalities of the central nervous system and face (Figures 14-4 and 14-5). Abnormalities in the acrorenal field are frequently cited as illustrative examples of polytopic field defects. There are at least 24 different genetic conditions in which both kidneys and limbs are involved, a fact that has been explained by invoking a relationship between the mesonephros and limb buds during embryogenesis. This concept is supported by the proximity between these 2 structures in early life, and by experiments showing an inductive effect of mesonephros on proliferation and differentiation of limb bud cartilage. A complete map of the developmental fields of the human embryo is not available, but could be constructed by classifying all human malformations and searching for causal heterogeneity among them. Opitz¹⁷ proposed that each time a certain malformation is seen in at least 2 causally different conditions, a developmental field has been identified, because identical structure means identical development, independent of differences in causal mechanisms. Mammalian primordia have only a limited number of responses to various dysmorphogenetic insults. The existence of developmental fields limits the possibility of independent responses from different structures of the organism.

Sequence

A sequence is a pattern of multiple anomalies derived from a single known or presumed prior anomaly or mechanical factor. Thus, there are malformations, deformations, and disruption sequences. The spectrum of anomalies of holoprosencephaly is an example of a malformation sequence. The precordal mesoderm is responsible for cleavage of the prosencephalon (forebrain) and normal development of the median facial structures. A primary defect of the precordal mesoderm leads to defects in both the brain and the face. In the brain, there is incomplete division of the cerebral hemispheres and underlying structures. Anomalies of the face include different degrees of hypoplasia of the median central structures (cyclopia, ethmocephaly, cebocephaly, and median cleft lip) (Figures 14-6 and 14-7).

Severe oligohydramnios may lead to intrauterine constraint and a typical deformation sequence that includes abnormal positioning of limbs (club foot) (Figure 14-8), breech presentation, Potter facies (Figure 14-9), growth deficiency, amnion nodosum, and pulmonary hypoplasia.^{20,21,22,23, and 24}

Early amniotic rupture with the formation of amniotic bands may lead to a disruption sequence including amputations of fingers, bizarre facial clefts, and asymmetric encephalo-celes (Figure 14-10).^{15,25,26,27,28,29,30,31, and 32}

Syndrome

A syndrome is a pattern of multiple anomalies thought to be pathogenetically related and not known to represent a single sequence or a polytopic field defect. The term syndrome is frequently employed to refer to a single cause, such as Down syndrome. The difference between a malformation sequence and a malformation syndrome is best understood with an example. Isolated holoprosencephaly is a malformation sequence. However, if holoprosencephaly occurs with other congenital anomalies in an infant with trisomy 13, the condition is a malformation syndrome.

Association

An association refers to a nonrandom occurrence in two or more individuals, of multiple anomalies not known to represent a polytopic field defect, a sequence, or a syndrome. The term association carries a purely statistical connotation, and not a pathogenetic or causal implication. With increasing knowledge, associations may become syndromes or polytopic field defects.

Familiarity with all of these terms is important because correct classification of an anomaly has implications for clinical diagnosis, management, and genetic counseling.^2,10 Disruptions are sporadic events, and therefore do not tend to recur in future pregnancies. The recurrence rate of deformations depends on the cause of the mechanical force leading to the anomaly. If the etiology of the abnormal mechanical force is a fetal malformation (eg, spina bifida), the recurrence risk is different than if the abnormal force were related to a leiomyoma.¹¹ The diagnosis of a malformation suggests a chromosomal abnormality, a monogenic defect, or a multifactorial disorder.^{4,10,33,34, and 35} Table 14-4³⁶ presents a glossary of terms used to describe congenital anomalies, including some of the terms discussed previously above.

Although there are several systems to classify congenital anomalies, an easy and practical method is to divide them into major and minor. A major anomaly is one with medical, surgical, or cosmetic importance, and with impact on morbidity and mortality. A minor anomaly is one that does not have a serious surgical, medical, or cosmetic significance, and does not affect normal life expectancy or lifestyle. Obviously, this classification is subjective and arbitrary. In addition, there is an overlap between minor anomalies and normal anatomic or phenotypic variants. A phenotypic variant occurs with a frequency of more than 4% in the general population, whereas minor anomalies occur at a rate of less than 4%. Clearly, this is also an arbitrary definition. Some ageneses are common enough to be considered "normal" anatomic variants (ie, absence of the muscle palmaris longus or upper lateral incisor). The importance of minor anomalies, however, is that they may serve as indicators of altered morphogenesis and more serious defects. Impor-tantly, 90% of infants with 3 or more minor anomalies will have a major anomaly.³⁷ Thus, the identification of several minor anomalies requires a careful search for hidden anomalies, in particular cardiac, renal, and vertebral disorders. Tables 14-5 and 14-6 have been compiled from the pediatric and genetic literature, and display the most common major and minor anomalies, respectively.^{10,37,38,39,40,41,42,43, and 44}

A practical problem in clinical genetics is the classification of a "funny-looking kid" who has several dysmorphologic features that may correspond to multiple minor anomalies, or multiple normal developmental variants. The 2 approaches that have been considered helpful are the analysis of family resemblance, and analysis of associated anomalies. On close inspection, the child with multiple phenotypic variants will resemble some other family members. By contrast, the child with multiple minor anomalies due to aneuploidy, for example, should not look like other family members. The child with multiple phenotypic anomalies also should not have other major external or internal anomalies.

Ultrasonographic Markers

Fetal anatomic defects observed by ultrasound may raise the suspicion of a more complex set of anomalies or chromosomal disorders. For example, a lemon-shaped head is commonly associated with spina bifida, whereas an atrioventricular canal defect is a major cardiac anomaly associated with Down syndrome.

Some minor anomalies may also carry a higher risk of chromosomal disorders. They are collectively called markers. Sonographic markers are variations in normal anatomy that, except for their relationship to aneuploidy (especially trisomy 21 or Down syndrome), are unlikely to be clinically significant. Compared with structural anomalies, markers are often transient (eg, choroid plexus cysts) and nonspecific findings, which can also occur frequently in euploid fetuses. Some of the most common sonographic markers seen in the second trimester include echogenic intracardiac focus, choroid plexus cysts, shortened long bones, hyperechoic bowel, nuchal fold thickening, renal pyelectasis, clinodactyly, and hypoplastic or absent nasal bone.

A few markers, however, have such a strong association with chromosomal anomalies that their identification per se is an indication for fetal karyotyping (eg, increased nuchal translucency thickness in the first trimester).⁴⁵ Other markers can be very common, and are usually not associated with any handicap. For example, when echogenic intracardiac focus or choroid plexus cysts are found in isolation in a low-risk patient, they do not carry such a high risk for chromosomal anomalies to warrant the performance of an invasive procedure. It is important to remember, however, that the presence of these markers should prompt a thorough evaluation of the fetus for other anomalies, before considering the need for fetal karyotyping. As a general statement, as the number of identified markers increases, the risk for aneuploidy in the fetus increases as well. Table 14-7 presents a list of major anatomic defects and sonographic markers for chromosomal anomalies detectable by prenatal ultrasound.⁴⁶

The precise incidence of congenital anomalies is difficult to determine. Accurate documentation depends on many factors, including (1) age at examination (prenatal period, newborn period, infancy, or later in life)^38,39,43; (2) the experience of the observer (eg, general pediatrician versus dysmorphologist)¹⁷; (3) the definition of an anomaly (major, minor, normal phenotypic variation)^37,47,48; (4) the type of examination (body surface examination, extensive examination including evaluation of internal organs); and (5) ethnic, geographic, and social variations in the incidence of individual malformations.^{46,47,48,49,50, and 51} Follow-up of infants is also extremely important, because only one-third of congenital anomalies are recognized in the newborn period.⁴⁸

One of the first attempts to determine the incidence of congenital major and minor anomalies at birth was conducted by Marden et al.³⁷ These investigators, particularly interested in dysmorphology, examined 4412 newborns during the first 2 days of life. A body surface examination, which did not include auscultation of the heart and abdominal palpation as standard procedures, was performed in all infants. A buccal smear for sex chromatin was also obtained. The incidence of major and minor anomalies was 2.04% and 14.7%, respectively. Importantly, among the 20 newborns having 2 or more minor defects, 90% had 1 or more major anomalies. Chromosomal aneuploidy was detected by buccal smear and phenotype examination in 4 of 1000 infants. Later, using data from the Collaborative Perinatal Project of the National Institute of Neurological and Communicative Disorders and Stroke, Chung and Myrianthopoulos reported the incidence of anomalies in a population of 52,332 liveborns.⁵² Table 14-8 shows the incidence of major, minor, and multiple anomalies, sequences, and syndromes in their study. Table 14-9 presents the incidence of congenital anomalies in live births, fetal deaths, and neonatal deaths.⁴⁸ When all types of morphologic anomalies were considered, the incidence was 15% among liveborn infants. Males were affected more frequently than females. The excess was attributed to a differential incidence of major anomalies, because the prevalence of minor anomalies was not different between the genders.

A substantial fall in maternal and infant mortality rates was achieved during the 20th century. Environmental interventions, improvements in nutrition, advances in clinical medicine, wider access to health care, increased surveillance and monitoring of disease, better education, and higher living standards contributed to this accomplishment. From 1915 through 1997, while the United States experienced a 93% drop in infant mortality (from approximately 100 per 1000 to 7.2 per 1000 live births),⁵³ the relative contribution of congenital anomalies to the perinatal death rate increased. In 1995, according to the Centers for Disease Control and Prevention, birth defects were the leading cause of infant mortality in the United States.^54,55 From 1968 to 1995, the proportion of infant deaths attributable to birth defects increased from 15% to 22%.^56,57 Similarly, in Scotland, the overall perinatal mortality declined by 75% between 1939 and 1941 and between 1974 and 1976, but over the same 37-year time span, the contribution of congenital anomalies to perinatal mortality increased from 10% to 25%.⁵⁸

There is also an increased awareness about the role of congenital disease in determining morbidity. For instance, it has been estimated that at least 1% of all diseases requiring hospital admissions have a genetic basis or genetic contribution; as many as 1 of every 4 hospitalized children has a disease that is at least partly genetically determined, and approximately 1 of every 20 children is affected by a disorder that is completely genetic in origin.⁵⁹

The birth of a congenitally malformed baby has implications for the use of health care resources. However, from a purely clinical and economic point of view, the birth of a baby with a lethal congenital abnormality will not place as large a burden on the health care system as would the birth of a baby with a condition associated with possible survival and long-term handicap.⁶⁰ The burden to the individual and society imposed by congenital anomalies has been examined by Chung and Myrianthopoulos.⁵² Table 14-10 illustrates the rates for postnatal mortality, neurologic abnormality at 7 years of follow-up, psychological deficit at age 7 years, and the requirement for major surgery for infants with different types of congenital anomalies. Infants with anomalies detected within the first year have a significant increase in the risk of death, and in all parameters of evaluated postnatal morbidity. The authors estimated that if major congenital anomalies did not occur, a reduction of 16% in postnatal mortality to age 7 years could be achieved. Infants with congenital anomalies also impose an economic burden to society and contribute stress to the family nucleus. For example, the incidence of divorce and sibling social maladjustment is greater in families of children with spina bifida than in families of infants without congenital anomalies.^61,62

Causative factors for congenital malformations may be identified in approximately 40% of the cases, and are usually divided into 4 major groups: (1) single gene disorders, (2) chromosome abnormalities, (3) multifactorial conditions (involving both environmental and genetic components), and (4) environmental factors.⁶³

Approximately 7.5% of all congenital malformations are caused by a single gene mutation.⁶⁴ Autosomal mutations occur when the gene is located in a non–sex chromosome, and may either be dominant (eg, adult polycystic kidney disease, achondroplasia, hypochondroplasia, Treacher Collins syndrome, Holt-Oram syndrome) or recessive (eg, Meckel-Gruber syndrome, achondrogenesis, short-rib polydactyly syndrome, camptomelic dysplasia). Autosomal dominant conditions have a recurrence risk of 50% for the subsequent offspring, whereas autosomal recessive conditions have a recurrence risk of 25%. The term X-linked disorder is reserved for single gene mutations in the X chromosome (eg, fragile X syndrome, otopalatodigital syndrome). In this case, women are symptomless carriers, and the disease is usually manifested only in males.

Chromosomal anomalies are responsible for about 6% of all serious congenital malformations among liveborn infants. They may be numerical or structural in nature. Polyploidy is an example of a numerical chromosomal anomaly, with 1 or more extra sets of 23 chromosomes added to the normal diploid karyotype (eg, triploidy: 69,XXX, 69,XXY, or 69,XYY). Approximately 20% of all abortuses have a polyploid karyotype.⁶⁵ Trisomies occur in approximately 55% of abortuses and 98% of newborns with chromosomal abnormalities, and are characterized by the presence of an extra chromosome (eg, trisomy 21, trisomy 18, and trisomy 13).⁶⁵ When 1 whole chromosome is missing, the abnormality is known as monosomy (eg, Turner syndrome: 45,X0). Deletions are examples of structural chromosomal derangements, in which part of a chromosome arm is lost (eg, 5p–, or cri du chat syndrome). Exchange of chromosome segments may occur in the absence of gene loss, and may be either translocations or inversions. Although carriers of balanced translocations are phenotypically normal, they are at increased risk of producing genetically unbalanced gametes and abnormal offspring.⁶⁶

Multifactorial conditions are responsible for 20% of malformations in liveborn fetuses. Examples of malformations with a multifactorial inheritance include spina bifida and congenital dislocation of the hip. These anomalies are the result of interactions between a relatively large number of genes with similar effects and nongenetic, usually undefined, factors.^64,66,67

A study conducted in Norway attempted to sort out environmental and genetic factors as causes of congenital anomalies.⁶⁸ A total of 371,933 mothers whose first and second births were recorded in a population-based registry were studied. The investigators attempted to estimate the recurrence risk of congenital anomalies in the second pregnancy of women whose first baby was affected by a congenital anomaly, in order to dissect the contribution of environmental factors. They also examined the effects of changing partner or city of residence between the first and second completed pregnancies in the recurrence risk of congenital anomalies. The results of this study showed that the recurrence risk of any congenital anomaly in the second pregnancy was at least 2.4 times higher for those who had a previous anomalous baby. Six different defect categories accounted for more than two-thirds of all the defects (Table 14-11). Mothers of babies affected by cleft lip, cardiac defects, limb defects, clubfoot, and genital defects had a significantly higher risk of having a second child with a similar defect. Changing partners between the first and second pregnancies did not affect the recurrence risk. However, those who remained in the same municipality during the second pregnancy had an 11.6 times higher risk of having a baby with the same birth defect as the previous malformed one (27.3 per 1000 [115 of 4200] versus 2.3 per 1000 [454 of 192,999]; relative risk = 11.6, 95% CI = 9.3 to 14.0]. These observations indicate that currently unidentified environmental factors may be responsible for birth defects, and that identification of potential teratogens may lead to the development of preventive measures. On average, it is estimated that 60% to 70% of pregnant women use from 3 to 10 medications during pregnancy, and that 20% to 30% of pregnant women abuse some chemical substance during pregnancy.⁶⁹

Currently, 2% to 3% of the spectrum of congenital malformations is attributed to teratogens,⁷⁰ with most malformations resulting from exposures during days 18 to 40 postconception (except for the palate, central nervous system, and genital structures, which can be affected at later stages of development).⁶⁷ Table 14-12 lists known and potential human teratogens, and the respective anomalies associated with prenatal exposure to them. The Food and Drug Administration lists 5 categories of labeling for drug use in pregnancy. They are as follows:

• Category A. Adequate and well-controlled studies in pregnant women have failed to demonstrate a risk to the fetus in the first trimester of pregnancy (and there is no evidence of a risk in later trimesters).

• Category B. Animal reproduction studies have failed to demonstrate a risk to the fetus, and there are no adequate and well-controlled studies in pregnant women.

• Category C. Animal reproduction studies have shown an adverse effect on the fetus, there are no adequate and well-controlled studies in humans, and the benefits from the use of the drug in pregnant women may be acceptable despite its potential risks.

• Category D. Positive evidence of human fetal risk based on adverse reaction data from investigational or marketing experience or studies in humans, but the potential benefits from the use of the drug in pregnant women may be acceptable despite its potential risks (eg, if the drug is needed in a life-threatening situation or serious disease for which safer drugs cannot be used or are ineffective).

• Category X. Studies in animals or humans have demonstrated fetal abnormalities, or there is positive evidence of fetal risk based on adverse reaction reports from investigational or marketing experience, or both, and the risk of the use of the drug in a pregnant woman clearly outweighs any possible benefit (eg, safer drugs or other forms of therapy are available).

It is our belief that all pregnant women should have access to expert obstetrical sonography, and the examination should be both thorough and systematic. The successful sonographic recognition of congenital anomalies depends on knowledge of several factors: embryology, fetal anatomy, ultrasound resolution, and natural history of the disorder (Figures 14-11 and 14-12). Ultrasound resolution has improved dramatically, and a detailed examination of the fetus for structural anomalies and markers of aneuploidy is possible. This, along with three-dimensional imaging, high frequency transvaginal probes, and color Doppler technology have led to an ever-increasing recognition of congenital anomalies. Factors that may limit visualization include equipment quality, expertise of the sonologist and ultrasound center, fetal positioning, gestational age, fluid abnormalities, maternal body habitus or tissue density, and other scanning characteristics. Techniques that can be done to improve visualization include altering the maternal position (to effectively change the fetal position), and using various scanning probes (including transvaginal). It is also important to remember that while repeating the sonographic exam later may aid in the resolution of anatomic detail of some structures (eg, the heart), this may also impair the visualization of others (eg, extremities, face). Scanning late in gestation, however, can still provide important information and be highly sensitive for fetal anomalies, and may even impact pregnancy management.

The prenatal diagnosis of a fetal abnormality has many advantages. For instance, it may trigger additional testing to refine the diagnosis or prognosis, prepare the parents for an adverse pregnancy outcome, alter pregnancy management (eg, termination, intrauterine therapy, early delivery, delivery at a tertiary care center), lead to counseling and planning by subspecialists, and alert the clinician to the possibility of other malformations. In addition, for many patients, the benefits of sparing psychological and social costs are also very important. Moreover, there is a further advantage to the prenatal diagnosis of fetal anomalies. We examined the value, from a patient's perspective, of a targeted ultrasound examination performed once an abnormal karyotype was discovered.⁷¹ All women valued the scan because it helped them to actually visualize the anomalies and accept the diagnosis of aneuploidy. This, in turn, affected their plans for pregnancy management. All patients considered the effect of the ultrasound greater than that of the chromosomal diagnosis alone, and all believed that ultrasound should be used for patients in similar clinical situations.

Most patients will choose anomaly screening if given a choice and counseled on the current diagnostic capability of sonography. Pregnant women usually do not perceive ultrasound as a test, but instead as a tool to evaluate their fetus as a patient. Because inaccuracy is intrinsic to the diagnostic process in medicine and cannot be completely eliminated, it is imperative to inform patients of what to expect from the available diagnostic tools in the diagnosis of congenital anomalies. Because the ability to detect congenital anomalies is directly and clearly dependent on operator skills, patients who desire prenatal diagnosis of congenital anomalies should be referred to centers and sonologists that are capable of performing these exams appropriately.

The first reports of prenatal recognition of congenital anomalies with ultrasound were published in 1961 by Donald and Brown,⁷² and in 1964 by Sunden.⁷³ Sunden documented identification of 3 cases of "acrania." Subsequently, the first prenatal diagnosis of a congenital anomaly with ultrasound that altered obstetric management was reported by Campbell et al in 1972.⁷⁴

The cardinal principle behind the prenatal diagnosis of congenital anomalies by using ultrasonography is recognition of a departure from normal fetal anatomy. Congenital anomalies are generally recognized with ultrasound by one of the following means: (1) absence of a normal anatomic structure; (2) a disruption of the contour, shape, location, sonographic texture, or size of a normal anatomic structure; (3) presence of an abnormal structure; (4) abnormal fetal biometry; and (5) abnormal fetal motion.

For example, the inability to identify a normal structure such as the fetal stomach or calvarium may suggest esophageal atresia, and ancencephaly or acrania, respectively (Figure 14-13). A localized defect in the calvarium indicates the presence of a cephalocele (Figure 14-14). A displaced stomach into the chest is diagnostic of a diaphragmatic hernia (Figure 14-15). This condition is often first suspected because of displacement of the heart within the chest, even before visualization of the stomach or intestine in an abnormal location is identified. Duodenal atresia is diagnosed by identifying a "double-bubble" sign (abnormal morphology of the stomach). Fetal tumors are typically identified when an additional structure is visualized that alters fetal anatomy (Figure 14-16). Abnormal biometry is used for the recognition of disorders characterized by fetal disproportion, such as skeletal dysplasias and microcephaly. Finally, absence of motion also identifies infants at risk for arthrogryposis multiplex congenita or other neuromuscular congenital disorders (Figure 14-17).

Although differences in maternal body habitus, fetal positioning, and sonographic expertise among centers performing obstetric ultrasound may preclude visualization of minor abnormalities in every case, prenatal diagnosis of small defects such as simian crease, skin tags, and epicanthal folds⁷⁵ have been reported. We performed a study to determine the sensitivity of using a complete anatomic sonographic survey in detecting fetal abnormalities (via correlation with perinatal autopsy results).⁷⁶ Of the 73 minor abnormalities found on autopsy, the sensitivity for minor abnormalities by sonography was poor (18%). However, the prenatal detection of minor anomalies, usually recognized only by a body surface examination of the newborn, is expected to improve with technologic advancements such as three-dimensional ultrasound.^{77,78,79,80,81,82, and 83}

In addition, prenatal diagnosis of congenital malformations during the first trimester should improve, as a result of implementation of nuchal translucency screening programs for the detection of chromosomal anomalies.^{45,84,85,86,87,88,89, and 90} The reader should bear in mind, however, the potential limitations of prenatal diagnosis that exist at this stage. Because of the embryologic timetable, some defects may not be diagnosed during the first trimester. For example, physiologic midgut herniation might preclude the prenatal diagnosis of small omphaloceles before 12 weeks (Figures 14-18 and 14-19).^{91,92, and 93} Another example of limitation with early prenatal diagnosis is the case of infantile polycystic kidney disease. This defect is recognized by signs of in utero renal failure, such as oligohydramnios and a nonvisualized bladder, coupled with enlarged hyperechogenic kidneys. However, sonographic observations have demonstrated that kidney measurements and function may be normal in early fetal life, even though infantile polycystic kidney disease is present. Therefore, the diagnosis of this condition is not always possible before the 24th week of gestation.^94,95

Second Trimester Ultrasound

There are many accepted indications for ultrasonographic examination during pregnancy, and 60% to 70% of all pregnant women in the United States undergo an examination at some stage during pregnancy.⁹⁶ In other industrialized countries, this rate is much higher. The accuracy of ultrasound to detect congenital anomalies has been the subject of considerable debate in the literature. For optimal diagnostic accuracy of fetal abnormalities, sonographer training must be targeted to ensure that detailed examinations are done appropriately. In addition, the following conditions should be met: (1) sonographers must be familiar with normal fetal anatomy, sonographic landmarks, and normal variants; (2) proper ultrasound equipment should be used; (3) there needs to be adequate scanning time; and (4) the highest frequency transducer is used to maximize fetal anatomic resolution.

One of the first studies to quantify the diagnostic accuracy of ultrasound in high-risk pregnancies was performed in the United Kingdom in the late 1970s and early 1980s; 95% of malformations were correctly diagnosed.⁹⁷ Subsequently, many studies have examined the efficacy of detection rates for congenital anomalies from routine ultrasound screening programs (Table 14-13).^{98,99,100,101,102,103,104,105,106,107,108,109,110, and 111} Detection rates for populations at high risk are shown in Table 14-14.^{112,113,114,115,116,117,118, and 119} The sensitivity (ie, ability to detect an anomaly when it is actually present at autopsy or at birth) ranged from 16% to 92% for pregnancies undergoing routine ultrasound screening. This wide range of sensitivity is a significant point because 75% of patients with abnormal fetuses are considered low risk.¹²⁰ We believe that much of the variation in results reflects differences in the quality of training. This is evident in that lower detection rates (16% and 21%) have been reported from earlier multicenter trials including laboratories with different levels of ultrasonographic expertise. Most studies from specialized centers have reported sensitivities in the range of 60% to 80% for major anomalies. It is our opinion that this should be the minimum acceptable standard for centers that offer prenatal diagnosis.

In light of these wide variations in the sensitivity of prenatal sonography for congenital anomalies, it is understandable why the reliability and utility of sonographic screening for fetal malformations has been challenged. Even among those who recommend routine sonographic screening, the number and timing of scans are debated. Many biases and problems may explain much of the variation in sensitivity,¹²⁰ and it is important to review this now (Table 14-15). Some explanations for the disparity include technical issues (eg, maternal habitus, fetal positioning), lesions that resolve over time, and the undetectability of certain abnormalities by sonography. However, for differences in large studies, these factors are unlikely explanations. Studies enrolling only pregnancies at high risk (those with a previous ultrasound suggestive of congenital anomalies, positive family history, or teratogen exposure) have reported higher detection rates, because of selection bias. Screening high-risk women is likely to be more effective, because sonologists become more focused since they expect to find an anomaly. One group noted an average sensitivity of 55% in a low-risk population, compared to 92% in a high-risk population.¹²⁰ In addition, selection bias regarding the type of setting may also play a substantial role, depending on whether the patient source is an office practice or hospital. As stated previously, the most important bias is the variation in sonographic skills, capability, and experience of the individual performing the sonographic examination. This was shown in the Helsinki study, which incorporated patient populations from two hospitals.⁹⁸ The sonographic sensitivity for detecting anomalies was found to be 2-fold higher in the university hospital, versus the city hospital (77% versus 36%). Similarly, Levi et al demonstrated that improvements in operator skill can result in better performance in the detection of abnormalities.¹²¹ Between 1984 and 1989, detection of anomalous fetuses by routine second and third trimester scans was 40.4%. However, following this, a program of education and training was introduced, which resulted in an 11% improvement in detection to 51%.

For each sonographic examination, a systematic and intensive approach is necessary, with full attention paid to all structures. Naturally, each center has differing criteria as to what constitutes a detailed sonographic examination. However, the less detailed the examination, the lower the sensitivity rate will be. Therefore, the quality of the sonographic exam will clearly affect the detection rate for anomalies. The type of anomaly plays a role as well in one's ability to diagnose congenital anomalies on sonography. For example, anencephaly is more likely to be detected than a small ventricular septal defect in the heart. The gestational age at screening will also influence sensitivity, since various anomalies may appear at different gestational ages. Some, for instance, are seen only later in gestataion (eg, duodenal atresia). The higher the frequency of sonographic exams performed, the greater the anomaly detection rate. Levi et al reported that the average sensitivity of studies that included only scans performed once before 20 weeks was 45%.¹²⁰ However, the sensitivity increased to 60% if the scanning was performed several times during the pregnancy. The prevalence of specific abnormalities within a population can also markedly affect the sensitivity of ultrasound screening. Studies that exclude certain abnormalities (because they are thought to be either minor or undetectable) can also dramatically change the sensitivity, compared with a calculation that includes all abnormalities. A natural extension of this is that there is a difference in the definition of major versus minor anomalies among some studies. For example, one study discovered that the sensitivity of sonography for diagnosing both minor and major anomalies was only 8.7%; however, if only major anomalies detectable by ultrasound were included, the sensitivity dramatically rose to 75%.¹⁰⁶ Studies that report data by number of anomalies (versus the number of fetuses) will overestimate the sensitivity. For instance, one study reporting a high sensitivity of 85.3% reported in this manner, and in fact, there is no mention of the number of anomalous fetuses.¹⁰⁴ Finally, other factors that may affect the fetal anomaly detection rate include the postnatal ascertainment of anomalies, such as inadequate length of neonatal follow-up (some anomalies may not express themselves immediately after birth), suboptimal neonatal evaluation at birth (may not be complete, particularly if the anomalies are not externally visible, such as cardiac septal defects), and lack of autopsies (which is considered the "gold standard").¹²⁰

Therefore, it seems that research into the efficacy of routine ultrasound screening for fetal abnormalities should ensure that comprehensive postnatal ascertainment is undertaken. This includes the following recommendations: (1) in the cases of fetuses aborted spontaneously or iatrogenically, or in stillbirths, that an autopsy is undertaken, whenever possible, by a recognized perinatal pathologist; (2) all newborn babies are examined by a pediatrician before discharge; and (3) population-based fetal anomaly registers are established and methods of reporting to them are accessible.⁶⁰

In keeping with the recommendations of postnatal ascertainment, we conducted a study to determine the sensitivity of using a complete anatomic sonographic survey in detecting fetal abnormalities via correlation with perinatal autopsy results.⁷⁶ All autopsies were performed by a single perinatal pathologist. The sensitivity of sonography in detecting anomalous fetuses was 97% (85 of 88), and the sensitivity of detecting major fetal abnormalities was 75% (224 of 299). There was either complete agreement (or only minor differences) between sonographic and autopsy findings in 65% (57 of 88).

Because the sensitivity for prenatal sonographic detection of fetal anomalies is clearly highly dependent on all of the aforementioned factors, there is insufficient evidence to list a single estimate of the sensitivity of routine ultrasound as a screening tool for fetal anomalies. Although a false-negative diagnosis may leave the family with the emotional, medical, social, and economic burdens imposed by a child born with a congenital anomaly, false-positive diagnoses may be equally ominous, because they may lead to termination of a pregnancy with a normal fetus. Specificity is the statistical parameter that measures the proportion of negatives that are correctly identified (ie, the percentage of truly normal fetuses who are identified as normal on ultrasound). This parameter is inversely related to the rate of false-positive diagnoses. Contrary to the various sonographic detection rates (sensitivity) reported for congenital anomalies mentioned above, specificity was invariably high in all studies (typically 99% or greater), no matter the type of population scanned, or the ultrasonographic expertise of those performing the examinations (see Tables 14-13 and 14-14). Therefore, ultrasound is helpful in excluding anomalies and detecting normal features, but may not be equally reliable in detecting anomalies. With continued technological advances in sonography and more uniform and detailed training of sonologists, it is anticipated that the sensitivity of screening programs may continue to improve over time.

Besides the variation in overall sensitivity of prenatal sonography for congenital anomalies, there is also considerable disparity in the detection of anomalies for various body systems. Lower sensitivities have been found for facial, limb, skeletal, and cardiovascular anomalies.^{99,102,103,118,121,122, and 123} Similarly, in 2000, a systematic review by Bricker et al. was published that examined the clinical effectiveness of ultrasound screening in pregnancy.⁶⁰ They found that the detection rates of major fetal abnormalities by screening second trimester sonographic examination varied with the organ system affected, with generally high sensitivity for central nervous system abnormalities (76.4%) and low rates for skeletal and cardiac abnormalities (23.8% and 17.4%, respectively). By contrast, our study found that the sensitivity of sonography in identifying abnormalities (correlated with perinatal autopsy) was greater than 70% in the following systems: central nervous, cardiac, urinary, extremities, genitalia, ribs, and hydrops.⁷⁶ Importantly, the variation of detection within anatomical systems themselves is also wide (eg, hypoplastic left heart versus atrial septal defect).

The specific prenatal detection rate for congenital heart anomalies is another important issue. The live birth incidence of congenital heart disease is estimated to be 0.8% to 0.9%, and 20% to 30% of perinatal deaths may be attributed to congenital heart disease.¹²⁴ As discussed later in this chapter, prenatal diagnosis followed by delivery in specialized centers may significantly improve perinatal outcome. Most of the studies with low detection rates performed unsatisfactorily in the detection of congenital heart defects. Table 14-16 summarizes studies on the accuracy of ultrasound for detecting congenital heart anomalies.^{125,126,127,128,129,130,131,132,133,134,135,136,137,138, and 139} As expected, sensitivity was higher in single centers that implemented intensive training programs for the examination of the fetal heart, with systematic inclusion of the 4-chamber view and outflow tracts as part of the routine examination, as well as analysis of blood flow by color Doppler imaging. A recent large, prospective, multicenter study aimed to evaluate the diagnostic accuracy of ultrasound screening based on the combination of the 4-chamber and outflow-tract views.¹⁴⁰ All operators had received specific training. Over 6300 pregnant women had complete follow-up; 58 cases of congenital heart disease were observed at birth or postmortem. The sensitivity, specificity, and positive and negative predictive values were 65.5%, 99.7%, and 70.4% and 99.7%, respectively. This sensitivity was significantly higher than that in a similar study (15%) performed 10 years earlier in the same setting.¹⁴¹ The authors attributed this improvement in part to extension of the examination to the outflow-tract view, but also to technological developments and better training of the operators.

Most cases of ventricular hypoplasia, tetralogy of Fallot, transposition of the great arteries, and endocardial cushion defects should be accurately diagnosed. However, cardiac anomalies that are still missed with relative frequency include atrial septal defects, ventricular septal defects, valvular stenoses, and anomalous pulmonary venous return. In some series, false-positive diagnoses were high, but they corresponded to small ventricular septal defects, which are likely to close spontaneously before or after birth.

First Trimester Ultrasound

Improvements in transvaginal sonographic technology and the discovery of a strong association between increased nuchal translucency thickness and chromosomal anomalies have fueled an interest in first trimester prenatal diagnosis. Early diagnosis is a desirable goal of prenatal care, because it allows detection of nonviable pregnancies, accurate determination of chorionicity in twin pregnancies, early pregnancy termination for selected anomalies, and the possibility of introducing more effective therapeutic strategies for treatable conditions.

Since the introduction of transvaginal probes and the first reports of anomaly detection in the first trimester,^{142,143,144, and 145} several centers have reported their experience with prenatal diagnosis at this gestational period.^{143,146,147,148,149,150,151,152,153,154,155,156, and 157} It has been shown that 6%, 75%, 96%, 98%, and 98% of the fetal structures can be imaged by transvaginal ultrasound by 10, 11, 12, 13, and 14 weeks, respectively. Accordingly, 13 weeks has been proposed as the optimal time to perform nuchal translucency measurements and conduct the first trimester anatomic evaluation.¹⁵⁸ Table 14-17 summarizes studies that have evaluated the diagnostic accuracy of first trimester ultrasound for the detection of congenital anomalies. Examinations performed in the first trimester can detect about 60% of the major congenital anomalies, and another 20% can be picked up during the second trimester scan, bringing the combined detection rate for congenital anomalies to approximately 80%. The most frequently detected anomalies diagnosed during first trimester screening are central nervous system defects (such as anencephaly, holoprosencephaly, and encephalocele). Additionally, megacystis, gastroschisis, and exomphalos are also commonly diagnosed in the first trimester.

Although there are potential advantages in detecting serious congenital anomalies in early pregnancy, these should be balanced against the possible drawbacks. Pitfalls include the identification of malformed fetuses destined to miscarry spontaneously,^{159,160, and 161} and an increased rate of false-positive diagnoses, because abnormalities such as isolated cystic hygroma and hydronephrosis may regress spontaneously over the remainder of the pregnancy.¹⁵⁴ A number of abnormal fetuses will die in utero in the late first or early second trimester. First trimester screening may preempt this, and in pregnancies where the fetus is destined to die in utero, parents may be unnecessarily exposed to the burden and long-term consequences of having to consciously decide whether to terminate the pregnancy.⁶⁰ Further limitations of first trimester screening include the fact that not all congenital anomalies are apparent by this time (eg, cranial findings associated with neural tube defects), and expertise in early transvaginal sonography is not necessarily widespread. Therefore, in many cases, a targeted scan later in gestation is still warranted. Moreover, first trimester anomaly screening may still be affected by the same factors that are problematic in the second trimester, such as technical difficulties and variation in sonographic experience and skills. This is illustrated in a study by Taipale et al.¹⁶² More than half of major fetal abnormalities were identified during first trimester screening in 20,465 unselected pregnant women who were screened at 13 to 14 weeks. However, sensitivity was lowest in the first year (22%), and highest (79%) in the sixth year.

Legal Implications of Accuracy Studies

Malpractice suits related to fetal anomalies are the most common type of litigation involving ultrasound (75% of the cases) in the United States.¹⁶³ Other frequent claims are related to gynecologic ultrasound (15%) and abdominal ultrasound (10%). The most frequent claim is missing an anomaly on a sonogram performed for a standard indication, such as dating.

From prior discussions, it should be clear that a normal ultrasound scan does not guarantee that a fetus will be born normal, there may be anomalies which are missed, and not all congenital anomalies can be detected on prenatal sonography. Patients should understand that the diagnostic efficacy of prenatal sonography for fetal anomalies may involve both false-positive and false-negative diagnoses. The implications of these potential diagnostic errors must also be discussed.

Physicians performing obstetrical ultrasound should inform their patients about these facts, preferably in writing, as a visible statement on the front page of the ultrasonographic report. As stated previously, several technical difficulties may impair the adequate anatomic study of a fetus: maternal habitus and depth of penetration, abnormalities in amniotic fluid volume, and fetal positioning during the examination; however, this is certainly not an inclusive list. For instance, diaphragmatic hernia (a major structural abnormality with life-threatening implications) is frequently missed in scans performed before 24 weeks. The same holds true for a number of other anomalies, including duodenal atresia, esophageal atresia, coarctation of the aorta, hydronephrosis, clubfeet, and hydrocephaly. Structures that have not been satisfactorily imaged during the examination should also be clearly described in the report.

The following issues need to be considered in a program of prenatal diagnosis of fetal congenital anomalies.

Performance of Prenatal Diagnosis

When prenatal diagnosis is performed, knowledge of the spectrum of the disease and its implications, associated anomalies, and natural history of the disorder is required. In addition, the possibility of genetic syndromes should always be kept in mind, particularly if there are multiple anomalies identified (eg, trisomy 18, Meckel-Gruber syndrome, etc). A detailed, comprehensive, and targeted sonogram should be performed to carefully evaluate the fetus, along with evaluation of the placenta, fluid volume, etc. Other modalities such as three- and four-dimensional sonography, Doppler imaging, and transvaginal sonography may also be utilized to complement the two-dimensional imaging.

Workup of an Abnormal Finding

One of every five newborns affected with a birth defect has more than 1 major anatomic abnormality, and this association is often of critical prognostic importance.^48,52 Therefore, the identification of any anomaly in a fetus must prompt a careful search for other associated abnormalities. Echocardiography and fetal karyotype determinations must be considered, depending on the findings. The association between congenital heart disease and extracardiac anomalies has been demonstrated in a recent series, showing that 23% of fetuses referred for echocardiography (because of an extracardiac anomaly) had congenital heart disease. The relationship between different types of congenital heart disease and various extracardiac anomalies has been reviewed elsewhere.^164,165

Chromosome abnormalities have been previously documented in the pediatric literature in infants with congenital anomalies. Approximately one-third of fetuses with structural anomalies have a chromosome disorder.^{35,93,94,95,96,97,98,166,167,168,169,170,171,172,173,174,175,176,177, and 168} The information derived from chorionic villus sampling (CVS) or amniocentesis may allow more informed counseling, and influence obstetrical management. CVS and amniocentesis have been the standard methods for obtaining material for karyotype determination, but placental biopsy and percutaneous fetal blood sampling can also be used when a rapid answer is desired, or additional studies from fetal blood are indicated.^169,170 Finally, because many congenital anomalies may be indicative of a genetic disorder or syndrome, genetic counseling should be considered. This is important not only for the current pregnancy, but also for future pregnancies, since there may be recurrence risks, depending on the abnormality.

Pregnancy Termination

In countries where pregnancy termination is available for social, as well as medical reasons, this option is offered to mothers carrying an anomalous fetus. The gestational age limit at which termination of pregnancy can be offered varies across countries, and even across different states in the United States. Some clinicians and ethicists believe that elective termination of pregnancy in the third trimester can be offered to some mothers carrying a fetus with a uniformly lethal condition, for which prenatal diagnosis is certain (eg, anencephaly). Recently, these infants have become a potential source of organs for transplantation, an option that raises serious ethical and practical medical questions.^171,172

Site, Mode, and Timing of Delivery

Ideally, the delivery of fetuses with serious congenital anomalies should occur in a center with a neonatal intensive care unit. For example, some congenital anomalies such as ductal-dependent cardiac lesions and abdominal wall defects may benefit from maternal transport and delivery in tertiary care centers, where pediatric cardiologists and surgeons are available for immediate assessment and intervention. In addition, there may be unexpected complications posed by undiagnosed associated anomalies. An interdisciplinary team composed of specialists in maternal–fetal medicine, diagnostic imaging, neonatology, and human genetics should be available for consultation. Depending on the specific nature of the anomaly, other specialists such as a pediatric surgeon, pediatric cardiologist, cardiovascular surgeon, and neurosurgeon may also be required. Paramedical personnel, such as a social worker, are also important in providing emotional support to the family.

Some anomalies may alter the method of delivery. Conjoined twins, giant omphaloceles, large sacrococcygeal tumors, and severe hydrocephaly are examples of these conditions. The optimal method of delivery for other anomalies such as omphaloceles, gastroschisis, and myelomeningocele has not been determined.^{173,174,175,176, and 177} Some disorders, which are not uniformly lethal, but are frequently associated with neonatal death or serious neurologic handicap, require careful discussion with the parents ahead of time, to define management in the event of fetal distress. This should avoid unprepared, critical decisions regarding the infant to be made in the setting of labor or immediately after delivery.¹⁷⁸ For these conditions, some favor a nonaggressive management approach, because analysis of the risk-to-benefit ratio is slanted toward maternal well-being; therefore, some clinicians would advise against a cesarean section for fetal indication. However, the final decision rests with the parents.

Delivery of an infant with a congenital anomaly should ideally occur at term. This is because surgical corrective procedures may be delayed by respiratory distress syndrome or other problems of prematurity. Earlier delivery, however, may be considered for some rare conditions that worsen in utero.

There has been a continuous debate concerning the routine use of ultrasound in obstetrics. While some, including a consensus conference sponsored by the National Institutes of Health, have favored selective use of ultrasound for a number of indications, others have opposed its routine use during pregnancy.^{96,98,105,179,180,181,182,183,184, and 185}

The RADIUS (Routine Antenatal Diagnostic Imaging with Ultrasound) study was the largest randomized clinical trial of routine ultrasound screening during pregnancy.⁹⁶ The trial was conducted in the United States, and recruited patients from 109 practices (81 private, 15 academic, and 13 health maintenance organizations). Ultrasound examinations were conducted in 28 laboratories equipped with state-of-the-art equipment, and using a rigid, uniform scanning protocol. The examinations were performed by 60 different technicians, and subsequently reviewed by 94 sonologists (including 2 of the trial investigators, 13 maternal–fetal medicine specialists, and 75 radiologists). In-service training, review of the first 25 sonograms performed by each sonographer and sonologist, and additional training and review, if necessary, were provided by one of the senior investigators. The content of the examinations consisted of the following: identification of the number of fetuses, presentation, placental location, amniotic fluid volume, biparietal diameter, head circumference, abdominal circumference, femur length, and an anatomic survey, including intracranial anatomy, 4-chamber view of the heart, demonstration of the spine in transverse and coronal planes, stomach, kidneys, bladder, umbilical cord insertion, and all extremities. A total of 15,530 women were randomly allocated to have either 2 ultrasound exams (at 15 to 22 weeks, and at 31 to 35 weeks) or an ultrasound exam only if indicated. The objective of the first scan was to detect gestational age errors, multiple pregnancies, and major malformations, and the second scan was conducted to detect fetal growth disorders (intrauterine growth retardation, macrosomia) and anomalies that appeared late, or were unrecognized in the early scan.

The study was designed to test the hypothesis that routine ultrasound screening during pregnancy would decrease the perinatal death rate, or the rate of moderate and severe neonatal complications.⁹⁶ The secondary objective of the study was to test whether routine ultrasound screening during pregnancy could improve maternal management and outcome.¹⁸⁰ The results showed no difference in the rate of adverse perinatal outcome between patients allocated to routine ultrasound screening and those in the control group. Similarly, there was no difference in the rate of adverse maternal outcome and rates of induced abortion, amniocentesis, tests of fetal well-being, external version, induction of labor, cesarean section, and duration of hospitalization. Patients allocated to the ultrasound screening group had a lower rate of tocolysis and post-term pregnancies.

Around the same time, a meta-analysis of 4 previously conducted randomized clinical trials^{98,181,182, and 183} was published in the British Medical Journal,¹⁸⁴ and included data from 15,935 pregnancies (7992 allocated to routine sonography versus 7943 allocated to selective scanning). In contrast to the RADIUS study, this meta-analysis showed that women allocated to routine ultrasound screening had a higher detection rate of small-for-gestational-age (SGA) infants, multiple pregnancy, and severe malformations, when compared with women who had ultrasound only if indicated. More importantly, the perinatal mortality rate was significantly lower in patients allocated to routine scanning (Figure 14-20). This effect was largely attributable to the contribution of the Helsinki trial,⁹⁸ which reported a 49.2% reduction in perinatal mortality (from 9.0 per 1000 to 4.6 per 1000) in women who underwent routine sonography. The investigators considered that the lower perinatal mortality "was due to improved early detection of malformations which led to induced abortions."

Why the Discrepancy Between the Results of the Two Studies (RADIUS Compared to Meta-Analysis)?

Many of the conclusions of the RADIUS study are questionable, and we will provide the evidence to support our view in the following paragraphs.^{186,187, and 188} In brief, we believe that the results of the RADIUS study failed to demonstrate a decrease in perinatal or maternal morbidity and mortality because of (1) poor performance in the detection of congenital malformations; (2) lack of statistical power and uniform management protocols to detect a reduction in cesarean section rates for post-term pregnancies; and (3) lack of statistical power to detect an improvement in maternal or perinatal outcome in multiple gestations.

Detection of Congenital Anomalies

Table 14-18 shows that routine scanning was better than ultrasound when indicated, in the detection of congenital anomalies in the RADIUS study.¹⁰⁵ The difference was significant both before and after 24 weeks of gestation (the limit for pregnancy termination in most of the United States). However, the higher detection rate did not translate into improved perinatal outcome. Several factors may be responsible for this lack of improvement. First, the detection rate of congenital anomalies in the study group (16.6%; 31 of 187) was extremely low, and does not represent the current diagnostic capabilities of ultrasound. Table 14-19 is a compilation of published studies in which ultrasound was used as a screening method for the detection of anomalies before 24 weeks of gestation.^{98,99,100,102,103, and 104} Excluding the results of the RADIUS study from this table, 52,295 pregnancies were screened in 6 different European centers, with an overall sensitivity of 50.9% (393 of 772). The difference between the sensitivity of the RADIUS study and that of this combined series is highly significant (16.6% versus 50.9%, 95% CI = 3.40 to 8.04, P < .00001). This low sensitivity may be largely attributed to the extremely poor performance of nontertiary care centers, which conducted 63% of the examinations. Despite the careful study design, detection rates were significantly higher for tertiary care centers, when compared with community hospitals and office-based facilities (35% [19 of 54] versus 13% [8 of 64]; relative detection rate = 2.7, 95% CI = 1.3 to 5.8]. Nontertiary care centers were unable to detect any craniofacial, cardiac, gastrointestinal, and skeletal anomalies.¹⁰⁵

The meta-analysis conducted by Bucher and Schmidt¹⁸⁴ indicated that routine sonography in the Helsinki trial⁹⁸ reduced perinatal mortality, mainly because a substantial proportion of women diagnosed to have fetuses with congenital anomalies chose to terminate their pregnancies. Indeed, the live birth rate (number of live births per pregnancy) was not different between patients allocated to routine scanning and those allocated to ultrasound if indicated.¹⁸⁴ Therefore, routine ultrasound reduced perinatal mortality simply by placing the deaths that would have occurred in the perinatal period in a different category, namely pregnancy terminations.¹⁸⁸

Correction of Dating Errors

The correct assessment of gestational age has 2 potential advantages: (1) reduction of unnecessary administration of tocolysis because of incorrect diagnosis of preterm labor; and (2) improvement of outcome and management of post-term pregnancy.

In the RADIUS study, women who were routinely scanned received tocolysis less frequently than women having ultrasound only if indicated (3.41% versus 4.22%, P < .01).⁹⁶ Because the rate of preterm delivery was similar in the 2 groups, it can be concluded that routine sonography decreased the rate of unnecessary tocolysis. Such a gain is desirable because tocolysis is expensive and has certain risks. Indeed, cardiovascular side effects induced by β-adrenergic agents have been implicated as a cause of iatrogenic maternal death of patients in premature labor.¹⁸⁹

Similarly, women allocated to routine scanning had a lower rate of post-term gestation (10.1% versus 21.4% above 41 weeks, P < .001; and 3.2% versus 4.6% above 42 weeks, P < .001).¹⁸⁰ These findings are consistent with the results of the Helsinki trial, in which ultrasound reduced the rate of post-term gestation (above 42 weeks) from 5.9% to 2.9%, P < .01).⁹⁸ It has been demonstrated by a large, randomized clinical trial that routine induction of labor in post-term gestations with prostaglandin E₂ reduces the cesarean section rate by 3.3 percentage points (from 24.5% to 21.1%), when compared with expectant management.¹⁹⁰ This benefit could not be demonstrated in the context of the RADIUS study, in part because routine induction of labor was not the standard of care for post-term gestations enrolled in the trial. Also, although the rate of adverse perinatal outcome was 40% lower in the patients allocated to routine sonography (1.6% versus 2.6%), this difference was not statistically significant. This is not surprising, because the study would need at least 6848 post-term patients to detect this difference with 80% power. The RADIUS study had only 592 post-term gestations, and thus a power of less than 12%. Similarly, 5202 patients would have been required to detect a significant reduction (24.5% to 21.2%) in the rate of cesarean section for post-term gestations, with an 80% power.^186,188

Detection of Multiple Pregnancies

The RADIUS study found that routine ultrasound was significantly better than ultrasound when indicated in the detection of multiple gestations before 26 weeks (98.5% [67 of 68] versus 62.3% [38 of 61]; odds ratio = 40.5, P < .001].⁹⁶ Moreover, 13.1% (8 of 61) of twins born to patients in the control group did not have the diagnosis made until delivery admission. However, the improved diagnosis did not result in improved perinatal or maternal outcome and management. The rate of adverse perinatal outcome in multiple gestations was 25% (17 of 68) in patients allocated to routine ultrasound, versus 37.7% (23 of 61) in those in the control group. This difference was not statistically significant (relative risk = 0.8, 95% CI = 0.39 to 1.11, P = .13). However, the RADIUS study had less than 36% power to detect this difference; 448 twin gestations would have been required to test the hypothesis, whereas the RADIUS study had only 129.¹⁸⁸

Detection of Growth Disorders

The RADIUS study found no differences in the birth rate of SGA infants (defined as birth weight below the 10th percentile for age) between the screening and control groups (2.3% for each group).⁹⁶ This is not surprising because ultrasound is a diagnostic, and not a therapeutic, tool. There are no reasons to expect that performing a diagnostic ultrasound by itself can correct a fetal growth disorder. To assess the value of routine sonography, we need to know the detection rate of SGA and macrosomia in each study group. However, such information has not been provided in the 3 reports of the RADIUS study. Assessment of the value, or lack thereof, of routine ultrasound in the diagnosis of SGA requires information about the potential consequences of false-negative and false-positive diagnoses of SGA in each of the study groups. The statement that there was no difference in perinatal outcome between patients having routine ultrasound versus patients having ultrasound when indicated does not answer this question.

Cost of a Routine Screening Program

It has been estimated by the authors of the RADIUS study that a policy of universal ultrasound screening during pregnancy would increase health costs by more than $1 billion. They have reasoned that screening 4 million pregnant women in the United States (at a rate of 1.6 scans per patient) at a cost of $200 per scan, would cost more than $1 billion. Even if the number of patients screened were restricted to 40% of the women who would not have a formal indication for routine scanning, it would still cost $500 million.⁹⁶

The available data from the RADIUS study indicate that the majority of pregnant women would have had an ultrasound examination during pregnancy.⁹⁶ First, 60.5% (32,317 of 53,367) of women screened for the trial were considered ineligible to participate because they had indications to be scanned. The authors did not provide information about when these examinations were performed, although we suspect that a large number were conducted in the second and third trimesters, and could have been used to screen for congenital anomalies. Second, an additional 6% of women (3163 of 53,376) declined consent. Although the reasons for this have not been stated, a fraction may not have wanted to give up prenatal diagnosis by ultrasound to participate in a research project. An additional 4.4% (2357 of 53,367) were lost to follow-up before randomization. Among the women randomized to the control group, 45% had an ultrasound examination for established indications. Thus, it can be inferred that 6988 of the 15,530 women (45%) eligible for randomization would have had an ultrasound examination for established indications. Based on this figure, we have estimated that at least 74% of women would have an ultrasound examination during pregnancy, and most of them at a gestational age when they could have been screened for congenital anomalies. Therefore, the argument of universal screening only applies to the remaining patients (approximately 26% to 30%).

In 2000, Vintzileos et al addressed the cost-benefit analysis of routine second trimester sonography.²⁰⁰ This was performed based on the RADIUS trial, comparing routine second trimester sonography in low-risk pregnant women with a policy of not offering screening. They found that routine second trimester sonographic screening is associated with net benefits, only if the sonogram is performed at a tertiary care center. Similarly, Van Dorsten et al assessed the cost-effectiveness of anomaly screening in their patient population (tertiary care center), and concluded that routine screening was cost-effective.¹⁰⁷

Another approach to examine the issue of cost-effectiveness is to compare the cost of detecting a congenital anomaly with ultrasound to other screening programs that are part of standard prenatal care. DeVore²⁰¹ compared the cost of detecting an anomaly with ultrasound screening performed by individuals with different ultrasonographic skills to the cost of maternal serum α-fetoprotein (MSAFP) screening (Table 14-20). If ultrasound screening is performed by individuals who have a diagnostic rate similar to that reported for tertiary care centers participating in the RADIUS study, the cost of detecting a malformed fetus would be $10,805 less than the cost of detecting an anomaly with MSAFP screening. However, if screening is performed by individuals whose diagnostic rates are similar to those reported for nontertiary centers, the cost would be $75,237 higher than the cost of MSAFP screening. To offer a cost-effective program of routine ultrasound screening during pregnancy, DeVore proposed that physicians should be reimbursed for second trimester ultrasound exams based on their reported diagnostic skills.

Vintzileos et al²⁰² evaluated the possible economic impact of offering targeted ultrasonography instead of routine amniocentesis for patients with elevated MSAFP in the United States. Their hypothesis was that the cost of universal amniocentesis in patients with an elevated concentration of MSAFP in the second trimester should be at least equal to the cost of universal targeted ultrasonography, with amniocentesis used only for those with abnormalities on ultrasound. The cost of universal targeted ultrasonography would include the cost of all normal ultrasound examinations, the cost of amniocentesis for those with abnormal ultrasonographic results, and the lifetime cost of caring for all live infants with spina bifida. They estimated the annual cost of genetic amniocentesis in patients with elevated MSAFP to range from US $90 million to US $214 million, if the therapeutic abortion rates were 100% and 50%, respectively. However, the annual cost of targeted ultrasonography for patients with elevated MSAFP (considering a detection rate of 90%, the worst possible scenario given the sensitivities reported in the literature) would be US $54 million and US $165 million, if the therapeutic abortion rates were 100% and 50%, respectively (this estimate includes both the cost of sonograms and the cost of amniocentesis performed because of abnormal ultrasonographic results). Thus, a targeted sonogram in patients with an elevated MSAFP in the United States has the potential for annual savings of approximately $36 million to $49 million. Importantly, in addition to the cost benefit, they estimated that 268 fetal losses would be prevented each year because of avoidance of invasive prenatal diagnostic testing in a substantial number of fetuses. Therefore, they concluded that ultrasound for detection of congenital anomalies is cost-effective, provided that it is performed in tertiary care centers.

Finally, in a systematic review by Bricker et al, they found that routine ultrasound scanning in the second trimester was shown to be relatively cost-effective.⁶⁰ They noted, however, that the skill of sonographers in detecting anomalies and the time taken to perform a scan had a significant effect on the relative cost-effectiveness. Therefore, based on these studies there seems to be little doubt that the cost-effectiveness of routine sonographic screening for fetal anomalies rests not with the machine, but on the caliber of the sonologist. Sonographers who are not properly trained to perform routine anomaly screening may have a detrimental effect on the efficiency, in terms of cost-effectiveness, of any screening program.⁶⁰ This result was also reiterated by 2 studies.^201,203 Therefore, investment in higher levels of training for all sonologists would not only be better for patients and their fetuses, but would also improve cost-effectiveness.

We have briefly outlined the capabilities of ultrasound as a tool to diagnose congenital anomalies. The examination should have high accuracy, provided that it is performed by skilled operators with appropriate equipment. Quality prenatal ultrasound diagnosis is a reality in many institutions, and, in our view, there is no reason it should not be standard practice.

In 2000, a systematic review was published that examined the clinical effectiveness and cost-effectiveness of ultrasound screening in pregnancy.⁶⁰ They reported that routine ultrasound in early pregnancy appears to enable earlier detection of clinically unsuspected fetal malformation at a time when termination of pregnancy is possible. In addition, they reported that routine sonography before 24 weeks reduces perinatal mortality rate if detection of fetal malformations is an important objective and a high level of diagnostic expertise exists, and if termination of pregnancy for fetal abnormality is widely accepted in the population screened.⁶⁰ Similarly, another study concluded that routine screening for anomalies improves perinatal outcome by leading to termination of pregnancy for certain anomalies, and delivery at tertiary care centers for life-threatening malformations.¹⁰⁷

Prior knowledge of the presence of some abnormalities associated with possible survival and long-term handicap may alter the management of pregnancy, with regard to mode of delivery or place of delivery, in an attempt to improve outcome. Maternal transport to centers equipped with neonatal intensive care and surgical capabilities may have a measurable impact on perinatal outcome for selected defects. For example, teratomas, anterior cystic hygromas, and thyroid masses may significantly obstruct the neonatal airway and require tracheal intubation on the mother's abdomen before umbilical cord clamping.²⁰⁴ Evidence that the outcome of conditions, such as gastroschisis²⁰⁵ and cardiac abnormalities,²⁰⁶ may be improved by prenatal detection exists in the literature. Other studies suggest that immediate survival rates are improved when cardiac defects are diagnosed prenatally.^{207,208, and 209} It has been suggested that as many as 200 infants die each year in the United Kingdom with unrecognized congenital heart disease, that approximately 100 die from isolated unsuspected cardiac abnormalities, and that among those fetuses, half have reparable lesions.²¹⁰ Chang et al²¹¹ evaluated prenatal diagnosis and referral of fetuses with critical left ventricular outflow tract obstruction for delivery and postnatal treatment in a tertiary care center. Among 22 fetuses with severe left outflow tract obstruction, fetal echocardiography predicted critical obstruction in all but 1 patient. Seventeen patients underwent cardiac surgery as neonates, and 77% (13 of 17) of them survived. One patient had a successful balloon aortic valvotomy for critical aortic stenosis, but died later of sepsis. Copel et al²¹² also reported improved survival for prenatally diagnosed fetuses undergoing biventricular heart repair in a specialized center (96% versus 76%, P < .05). In addition, many investigators have demonstrated that the prenatal diagnosis of congenital heart defects has a positive impact on the preoperative condition of neonates with severe cardiac malformations.^208,211,213 Prenatal diagnosis has been associated with decreased preoperative morbidity, less acidosis, and evidence of better end-organ perfusion.^208,211,213

There is no clear evidence that prenatal diagnosis with planned delivery in tertiary care centers improves outcome for other anomalies. Skari et al²¹⁴ compared the effect of prenatal diagnosis on management of labor and neonatal outcome in neonates with congenital diaphragmatic hernia, abdominal wall defects, meningomyelocele, and bladder exstrophy. They found that although all prenatally diagnosed fetuses were delivered by cesarean section in the hospital attached to the neonatal surgical unit, there was no obvious difference in neonatal outcome. The neonatal survival rate was 77% (10 of 13) for the prenatally diagnosed, versus 96% (22 of 23) for the postnatally diagnosed fetuses (P = .12). In addition, no differences were observed in the length of hospital stay, or in time spent on ventilatory support or parenteral nutrition. They also found a significantly lower gestational age at birth in the prenatally diagnosed group. They emphasized that there may be a difference in the severity of the condition among diagnosed versus nondiagnosed abnormalities, with the prenatally diagnosed cases being the more severe or serious, thus making comparisons difficult. Of concern is the fact that there may be more intervention, and hence higher risk of maternal morbidity, in pregnancies in which abnormalities are prenatally diagnosed, without a subsequent improvement in long-term outcome.⁶⁰ Dillon and Renwick²¹⁵ reported similar observations in 56 fetuses with gastroschisis and in 43 with omphalocele. Fourteen fetuses with gastroschisis and 10 with omphalocele were not diagnosed prenatally. Delivery of these babies away from the regional pediatric surgery center did not adversely affect perinatal outcome.

Termination of pregnancy is the most prevalent change in management resulting from prenatal diagnosis. Zimmer et al²¹⁶ reported a rise in the number of pregnancy terminations (from 0.35% to 0.86%, P < .01) as a result of improved detection rates of congenital anomalies (from 53.9% to 79.6%, P < .001) over a 5-year period. The improved detection rates followed by action at diagnosis resulted in a significant decrease in the incidence of newborns with congenital anomalies (from 1.95% to 1.34%, P < .01). Data from the Hawaii Birth Defects Program also showed a significant decrease in birth prevalence rates of anencephaly, spina bifida, encephalocele, and trisomies 21, 18, and 13 between 1987 and 1996 as a result of prenatal diagnosis followed by pregnancy termination.²¹⁷ The 1998 report from the International Clearing House for Birth Defects Monitoring Systems (ICBDMS) showed a significant reduction in the birth prevalence of neural tube defects in the majority of participating centers (Atlanta,²¹⁸ Canada, Australia, New Zealand, England, France, Spain, Italy, Norway, Netherlands, Finland, Czech Republic, Hungary, Israel, and Japan). ICBDMS is a nongovernmental association (officially related to the World Health Organization), representing more than 30 birth defect monitoring systems in 34 countries. This reduction was attributed to pregnancy termination of anencephaly and spina bifida after prenatal diagnosis.²¹⁹ The exceptions were centers located in countries where pregnancy termination is presently illegal (Mexico and many South American countries).

A decrease in the prevalence of hypoplastic left heart syndrome was observed in the United Kingdom as a result of routine screening of cardiac anomalies with the 4-chamber view.²²⁰ Similar findings were reported in France, where a decreased prevalence at birth of major congenital cardiac defects was observed, as a result of prenatal diagnosis followed by termination of severe cases.¹³⁸ In Boston (Brigham and Women's Hospital), an active birth defect surveillance system registered a significant drop in live births with cardiovascular malformations as a result of improvements in ultrasound detection rates.²²¹

If detection of congenital anomalies is one of the goals of prenatal care, it seems that routine ultrasound examination is the best available tool to accomplish this objective. We have described the potential advantages of accurate prenatal diagnosis. Innovations in ultrasound technology and research are changing the practice of prenatal diagnosis and will continue to do so, using new technologies. In the future, we anticipate there will be an increasing number of centers performing quality ultrasonographic diagnosis in the first trimester. Ultrasound will continue to occupy a predominant role in screening for chromosomal anomalies because of nuchal translucency, and refinements in syndromic diagnoses will likely happen as a result of three- and four-dimensional technology. It is our hope that more sonologists, through education, will master the skills of examining the fetal heart in order to improve perinatal outcome.

We and others¹⁸⁵ favor routine obstetrical ultrasound for all pregnancies, provided it is performed properly. However, striving to improve the diagnostic accuracy of prenatal sonography should be the goal of all who perform ultrasound. In addition, when fetal sonographic examinations are performed, an attempt should be made for a complete and detailed fetal anatomic survey. The clear evidence that quality of sonographic scanning is not as good in nonspecialist centers also needs to be addressed.

It appears that prenatal diagnosis is here to stay. Parental demands, the more advanced age of motherhood, and the increasing level of maternal education demand the prediction of morbidity when a congenital defect is recognized on prenatal sonography. Therefore, the upcoming challenges of prenatal diagnosis involve the improvement of predictors, and better refinement of the quality of ultrasound.

Acknowledgments

This research was supported (in part) by the Perinatology Research Branch, Division of Intramural Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, and DHHS.

The authors would also like to acknowledge and thank Eli Maymon, Percy Pacora, Katherine Bianco, and Philippe Jeanty for their past contributions to this chapter, which was previously published in the Sixth Edition.

Highlighted References