Congenital diaphragmatic hernia (CDH), a defect in the diaphragm, is thought to be due to failure of the pleuroperitoneal canal to close by 9 to 10 weeks of gestation and results in varying degrees of pulmonary hypoplasia from compression of the developing lungs by the herniated viscera (Harrison et al., 1993b). The traditional view that the diaphragm forms by fusion of the septum transversum, the esophageal mesentery, the pleuroperitoneal folds, and in growth of musculature from the lateral body wall, is now being questioned (Pober, 2008). In model organisms, recent work suggests that a nonmuscular diaphragmatic anlage first develops, and that the diaphragmatic musculature derives from muscle precursors that migrate through the pleuroperitoneal folds at approximately day 37 of gestation (Babiuk and Greer, 2002; Pober, 2008).

During the development of the diaphragm, the peritoneal cavity is quite small and the midgut is normally present in the umbilical cord as physiologic herniation of the cord. If closure and muscularization of the pleuroperitoneal canal has not occurred by 9 or 10 weeks of gestation, when the midgut returns to the abdomen to undergo its normal 270-degree rotation, the viscera may herniate into the thorax through the posterolateral diaphragmatic defect because of limited intra-abdominal space (Areechon et al., 1963). If herniation occurs before the closure of the pleuroperitoneal canal there is no hernia sac. However, if pleuroperitoneal membrane has formed but is not muscularized, a hernia sac will be present and is observed in 10% to 15% of cases (Areechon et al., 1963). Occasionally, a “transient” herniation may occur later in gestation, with little effect on pulmonary development (Adzick et al., 1985a; Stringer et al., 1995).

The clinical course of an infant with isolated CDH depends entirely on the degree of pulmonary hypoplasia and severity of pulmonary hypertension. The degree of pulmonary hypoplasia depends on the timing of herniation during development, the volume of viscera herniated, and duration of herniation, that is, whether or not the viscera slide in and out of the thorax (Harrison et al., 1993a; Stringer et al., 1995).

An appreciation of the pathophysiology of CDH requires an understanding of the normal growth and development of the tracheobronchial tree and pulmonary vasculature. Reid (1977) has described four overlapping stages of normal histologic development: embryonic (from conception to 5 weeks); pseudoglandular (5–16 weeks); canalicular (16–24 weeks); and terminal sac or alveolar (24 weeks to postnatal life). The bronchial tree is almost completely developed by 16 weeks of gestation, at which time the adult complement of airways is established. However, alveoli continue to develop after birth, increasing in number until about 8 years of age (Boyden, 1977).

The severity of pulmonary developmental abnormalities depends on the time and the extent to which the herniated viscera compress the adjacent lung. A large intrathoracic mass effect that develops during the formation of the conducting airways (pseudoglandular stage) will reduce the number of bronchial divisions, decreasing the thoracic volume available for lung development (Geggel and Reid, 1984). The herniation of viscera in CDH usually occurs during the pseudoglandular stage of lung development (5–16 weeks) (Reid et al., 1977). In fetuses with CDH, the major bronchial buds are already present, but the number of bronchial branches is pruned and greatly reduced (Companale et al., 1955). The number of alveoli per acinus may be normal, but the absolute number of alveoli is decreased because of the reduced number of bronchial divisions. These morphologic changes are more pronounced in the lung ipsilateral to the diaphragmatic hernia, but the contralateral lung is similarly affected by compression from the shifted mediastinum (Figure 37-1) (Harrison et al., 1980a, 1980b, 1981, 1990, 1993a; O’Rourke et al., 1984; Adzick et al., 1985b; Hasegawa et al., 1990). Persistent mass lesions during later stages of lung development (canalicular or alveolar stages) will result in a reduction not only in airway size, but in the number and size of saccules, alveoli, and preacinar and intra-acinar vessels (Areechon et al., 1963). Concomitant with these changes in the fetal tracheobronchial tree is an increase in the thickness of the arterial media and extension of muscle peripherally into the small preacinar arteries (Levin et al., 1978). The pulmonary hypoplasia that is associated with CDH is one of the major determinants of morbidity and mortality (Figure 37-1). Some have suggested a “two-hit” theory in which pulmonary underdevelopment occurs first but is then made worse by subsequent mechanical compression (Keijzer et al., 2000). In addition, the pulmonary vasculature is abnormal with overmuscularized vessels. In addition to the increased muscularization of the preacinar arteries, Geggel et al. have demonstrated that there is a reduction in size of the pulmonary vascular bed in CDH (Geggel et al., 1985). These changes in the pulmonary vascular bed are the histologic correlate of the pulmonary hypertension seen in experimental models of CDH and newborns with pulmonary hypoplasia.

There is evidence that mutations in specific genes that are involved in diaphragmatic development and/or the vitamin A pathway may play a role in the etiology of CDH and pulmonary hypoplasia (Ackerman and Pober, 2007). Key genes that are mutated in teratogenic mouse models of CDH include the transcription factors Fog2, couptf2, wt1, slit3, and GATA4, and molecules involved in cell migration and signaling such as slit3 (Bielinska et al., 2007). It is not clear that this teratogenic model from which these mutations have been identified apply to human CDH. Interestingly, in the human, these genes map to areas that have been consistently shown to have chromosome abnormalities that are associated with CDH. There has also been one case in a human mutation in FOG2 in which there was pulmonary hypoplasia and a diaphragmatic abnormality consistent with eventration, but no CDH (Ackerman et al., 2005).

Approximately 85% to 90% of diaphragmatic hernias occur on the left side, 10% to 15% are on the right side, and a few are bilateral. In 60% of cases, the diaphragmatic hernia is either isolated or associated with malformations that are due to hemodynamic or mechanical consequences of the CDH. In 40% of cases the CDH is nonisolated or part of a syndrome. All studies have shown that infants with syndromic (or “complex”) CDH have higher mortality (Pober, 2007). The incidence of CDH has been estimated at between 1 in 3000 and 1 in 5000 livebirths (Puri and Gorman, 1984). These estimates ignore the significant numbers of intrauterine fetal death, stillbirths, and neonatal deaths that occur before transfer to a tertiary care facility. An accurate incidence of CDH is more likely around 1 in 2200 births (Fitzgerald, 1979; Harrison et al., 1979; Reynolds et al., 1984; Puri, 1989). In years past, the postnatal survival rate of infants with CDH has traditionally been quoted as 50% (Adzick et al., 1981), but this figure represents survival in a favorably selected group of patients who survive not only to term, but also transfer to a referral center for further treatment (Harrison et al., 1979, 1990, 1993a, 1994; O’Rourke et al., 1984; Adzick et al., 1985a). The most severely affected neonates die before they are transferred to a tertiary care center. Harrison has referred to this discrepancy as the “hidden mortality” of CDH (Harrison et al., 1979). A more recent meta-analysis found that the average mortality for prenatally diagnosed cases was 75%, for cases ascertained as part of a population-based study it was 48%, and for cases transferred to a tertiary facility it was 45% (Skari et al., 2000).

The cause of CDH is unknown but it has been reported in association with maternal ingestion of thalidomide, Bendectin, quinine, and antiepileptic drugs (Hobolth, 1962; Kup, 1967; Hill, 1974; Greenwood et al., 1976; Tubinsky et al., 1983). Associated anomalies are seen in 25% to 57% of all cases of CDH, but this figure rises to 95% in stillborn infants (Crane et al., 1979; Tubinsky et al., 1983; Puri, 1984). The associated anomalies may include congenital heart defects, hydronephrosis or renal agenesis, intestinal atresias, extralobar sequestrations, and neurologic defects, including hydrocephalus, anencephaly, and spina bifida (Crane et al., 1979; Tubinsky et al., 1983). CDH has been described as a finding in Fryns, Beckwith–Wiedemann, and Pierre–Robin syndromes as well as in congenital choanal atresia (Thorburn et al., 1970; Evans et al., 1971; Harrison et al., 1991). Chromosomal anomalies are diagnosed in 10% to 20% of cases of CDH diagnosed prenatally. The most common diagnoses include trisomies 21, 18, and 13 (Lesk et al., 1959; Crane et al., 1979; Tubinsky et al., 1983).

Approximately 60% to 90% of cases of CDH are detected prenatally by sonography or MRI depending on the center reporting ascertainment (Pober, 2008). The diagnosis of CDH should be suspected if the stomach bubble is not observed in its normal intra-abdominal location. The fetal chest should be viewed in the true transverse plane, and landmarks such as the inferior margin of the scapula should be used to identify the abdominal viscera in the chest (Lesk et al., 1959). Abdominal viscera that are seen cephalad to the inferior margin of the scapula or at the same level of the four-chamber view of the heart are herniated, confirming a diagnosis of CDH (Figures 37-2 to 37-4). The herniated abdominal viscera associated with a left-sided CDH may be the easiest to detect. The fluid-filled stomach and small bowel contrast strikingly with the more echogenic fetal lung.

Liver Herniation

The position of the fetal liver is one of the most significant and reproducible independent prognostic factors, with liver herniation predictive of poor outcome (Harrison et al., 1990; Cannie et al., 2006; Hedrick et al., 2007; DePrest et al., 2009). Kinking of the sinus venosus is a reliable sign of left-sided CDH with herniated left lobe of the liver (Figure 37-5). In a retrospective review of 16 fetuses with left CDH, Boostaylor et al. (1995) found that bowing of the umbilical segment of the portal vein (the portal sinuses) to the left of midline and coursing of portal vessels to the lateral segment of the left hepatic lobe toward or above the diaphragmatic ridge are the best predictors for liver herniation into the left chest. Another subtle finding is an echodense space between the left border of the heart and the stomach, which is due to interposed herniation of the left lobe of the liver. Sonographic or MRI delineation of the diaphragm is not always possible. Even identifying the diaphragm cannot exclude CDH because only a portion of the diaphragm may be missing.

The location of the gallbladder may also be helpful in diagnosing CDH because it may be herniated in the right chest in right-sided CDH or displaced to the midline or left upper quadrant with left-sided CDH. A large-volume herniation will result in mediastinal shift with polyhydramnios. Mediastinal shift is thought to interfere with swallowing, thus resulting in polyhydramnios (Harrison et al., 1991). Since the stomach is often rotated 180 degrees counterclockwise from its normal anatomic positionupinto the chest, it is more likely that there is partial gastric outlet obstruction due to kinking at the gastroduodenal junction. The stomach position is also a good predictor if observed in a posterior or midthoracic location if the liver is herniated (Boostaylor et al., 1995). CDH has been also reported in association with concomitant bronchopulmonary sequestration cystic adenomatoid malformation, and teratomas. These may be noted as echogenic masses seen in association with the CDH.

The extent of pulmonary hypoplasia is the most important determinant of survival in CDH. Hasegawa et al. (1990) have proposed using a ratio of the cross-sectional area of the lung to thorax (L:T ratio) in sonographic transverse section of the fetal chest at the level of the four-chamber view of the heart to assess the likelihood of pulmonary hypoplasia. They found, in a small series of eight fetuses with CDH, that the L:T ratio was below 2 SD from the mean ratio obtained in 156 normal controls. There was also an inverse correlation between the L:T ratio in the fetus and in the post natal A-aDO₂ (alveolar-to-alveolar oxygen difference) values (Hasegawa et al., 1990).

Metkus et al. (1996) reported the use of the right-lung to head circumference ratio (LHR) as a sonographic predictor of survival in fetal diaphragmatic hernia. The LHR is the two-dimensional area of the right lung taken at the level of the four-chamber view of the heart. This is divided by the head circumference. In a retrospective review of 55 fetuses diagnosed with left-sided CDH, the LHR was found to be predictive at its extremes. At low values (i.e., small right lung), fetuses with LHRs <0.6 did not survive with postnatal therapy. But in fetuses with LHRs >1.35, survival was 100% with conventional postnatal therapies, including ECMO (Cannie et al., 2006; DePrest et al., 2006). The survival of fetuses with LHRs between 0.6 and 1.35 was 61%. At an NIH symposium, Harrison et al. (2003) provided additional data in the group of fetuses with values between 0.6 and 1.35. Survival with an LHR <1.0 was only 11%. The accuracy of the LHR described by Metkus et al. (1996) was validated in two subsequent prospective studies (Flake et al., 2000). The LHR has not been widely adopted due to the difficulty in accurately and reproducibly obtaining the LHR.

There now have been three different techniques reported for obtaining lung:head circumference ratio and a fourth modification in which the observed LHR is normalized to an expected LHR. The only two of these methods have been validated in prospective studies. In the technique first reported by the University of California San Francisco (UCSF) group, the largest transverse width of the right lung is obtained from the cross-sectional view of fetal chest at the level of the four-chamber view of the heart. This transverse measurement is taken parallel to the sternum from the right side of the Ao to the edge of the lung at the right chest wall. The anterior-posterior (AP) measurement is obtained perpendicular to this measurement. A second technique obtains the longest transverse measurement at the level of the four-chamber view of the heart independent of the orientation of the sternum. The third technique captures the image of the cross-sectional view of the chest at the level of the four-chamber view and traces the outline of the right lung to obtain the area and divides by the head circumference. Each of these techniques yields slightly different results that may alter the perceived prognosis. These techniques are not only highly user-dependent, but the prognosis based on these results may not translate from one center that sees a high volume of fetuses with CDH to one that sees only a few cases each year. Case in point, Crombleholme et al. (2009) have reported the Cincinnati Children’s experience with LHR, finding a survival of 100% when LHR was >1.0 and 50% with LHR <1.0. These findings are in contrast to older reports of prognosis based on LHR that indicates the institutional-specific nature of the utility of LHR in predicting survival. The accuracy of the LHR in predicting outcome has been challenged by the Columbia group (Arkovitz et al., 2007), who reported that the LHR in their series was not predictive of outcome. Methodical problems with LHR acquisition may be an issue, but this does point to concern regarding how easily translatable use of LHR is from one center to another.

Between 12 and 32 weeks’ gestation, normal lung area increases four times more than head circumference (DePrest et al., 2009). For this reason, Jani et al. (2007) proposed referencing LHR to gestational age by expressing the observed LHR as a ratio to the expected mean LHR for that gestational age. In a study from the CDH antenatal registry of 354 fetuses with isolated left and right CDH between 18 and 38 weeks, Jani et al. found that observed/predicted LHR predicted postnatal survival. The O/E LHR tended to be more accurate at 32 to 33 weeks than at 22 to 23 weeks’ gestation. The O/E LHR was also found to correlate with short-term morbidity indicated (Jani et al., 2007).

A novel approach was reported by Mahieu-Caputo et al. (2001) using the thoracic volume minus the mediastinal volume to yield an estimate of what the lung volume would be expected to be if there was no CDH and dividing the actual lung volume by this estimate to yield the percent predicted lung volume (PPLV). Mahieu-Caputo et al. (2001) found that the observed/expected fetal lung volume ratio was significantly lower in CDH patients who died with a mean of 26% compared to those who survived with a mean of 46%. This same group reported a larger experience from a 4-year prospective multicenter study of 77 fetuses with isolated CDH diagnosed between 20 and 33 weeks’ gestation (Gorincour et al., 2005). They found that the observed/expected lung volume was significantly lower in fetuses with CDH that died (23%) compared to those that survived (36%). When the observed to expected fetal lung volume ratio was below 25%, there was a significant decrease in postnatal survival to 19% versus 40.3%. While these survival rates are lower than usually reported in the United States, they still support the utility of this prognostic technique.

Using this same technique that she termed PPLV, Barnewolt et al. (2007) reported their preliminary experience in Boston with 14 patients with CDH in which there was a clear break point at a PPLV of 15%. Fetuses with PPLV more than 20% had 100% survival while those with PPLV <15% had a 40% survival and all required prolonged ECMO. However, Crombleholme et al. (2009), in reporting the Cincinnati Children’s experience with PPLV with 28 patients, found that PPLV was not as predictive of outcome as LHR (Crombleholme et al., 2009). In this series, three of the four deaths occurred in patients with PPLV more than 15%. In contrast, survival with LHR >1.0 was 100% and all deaths occurred in patients with LHR <1.0.

Fetal MRI has been also applied to directly measure total lung volumes to predict outcome in CDH. Hubbard et al. (1997) found that fetal lung volumes obtained by MRI at midgestation did not accurately predict postnatal outcome. Kilian et al. (2006) reported a series of fetal MRI-derived lung volumes at 34 to 35 weeks’ gestation. They noted that most of the growth in lung volume occurs in late gestation, as reflected in the later sharp upward sweep of lung volume normograms. They reasoned that in the presence of a large CDH there would not be the normal increase in lung growth. In a series of 38 cases of CDH, both right-sided and left-sided, they correlated lung volume with survival and the need for ECMO. They found that the mean lung volume of survivors was 35 cc, while mean lung volume of nonsurvivors was 9 cc. The mean lung volume of those infants requiring ECMO was 18 cc, while 25 cc was the mean lung volume of those that did not require ECMO.

At the time of 34 weeks’ gestation MRI, measurement of the branch pulmonary artery diameter and the descending Ao allows calculation of the modified McGoon index. Vuletin et al. (2009) have shown that the modified McGoon <1.0 and the prenatal pulmonary hypertensive index (PPHI, branch pulmonary arteries divided by the cerebellum to normalize for age) correlates with severe postnatal pulmonary hypertension at 3 weeks of age.