Principles of fetal–surgical intervention

Maternal–fetal surgery is a specialty directed towards the goal of improving the natural history of selected congenital disorders by their correction before birth. Observation of irreversible end-organ damage resulting from anatomic malformations led to the compelling theory that correction in utero might permit normal development; rigorous research in small and large animal models has supported this hypothesis. The first report of open maternal–fetal surgery in 1982 by Harrison et al. described the creation of a vesicostomy in a fetus with congenital bilateral hydronephrosis. Since that initial report, the field of fetal intervention has evolved, and the concept of the fetus as a surgical patient has gained acceptance.

The prerequisites for maternal–fetal surgery formulated 3 decades ago continue to apply to current practice, with some minor modifications and require the following: (1) the ability to establish an accurate prenatal diagnosis; (2) a well-defined natural history of the disorder; (3) the presence of a correctable lesion which, if untreated, will lead to fetal demise, irreversible organ dysfunction before birth, or severe postnatal morbidity; (4) the absence of severe associated anomalies; and (5) an acceptable risk-to-benefit ratio for both the mother and fetus. The identification of complex chromosomal or associated anatomic abnormalities, or maternal risk factors such as co-morbid conditions, placentomegaly, and maternal mirror syndrome, are considered to be contraindications to most maternal–fetal surgical interventions.

A number of anatomic anomalies currently meet the criteria for consideration of fetal intervention. The conditions most widely studied are shown in Table 1 . The natural history, pathophysiology, and current evidence supporting fetal therapy, will be discussed for each of these disorders. In the case of fetal lung lesions, myelomeningocele, sacrococcygeal teratoma, and twin–twin transfusion, the efficacy of fetal intervention has been shown with compelling evidence from numerous clinical studies. The role for fetal intervention in congenital diaphragmatic hernia continues to be controversial and the subject of clinical trials.

Congenital malformations treated by maternal–fetal surgery

Congenital malformations treated by maternal–fetal surgery

Pathophysiology and natural history

Congenital cystic adenomatoid malformations (hamartomatous tumours) are thought to arise from aberrant events during lung branching morphogenesis, and are histologically characterised by an overgrowth of one or more components of lung tissue with epithelial and bronchial elements. The ultrasonographic features of these masses vary considerably, from solid-appearing (microcystic) to grossly cystic (macrocystic) tumours. Postnatally, CCAMs may present with recurrent pulmonary infection resistant to antibiotic therapy, pneumothorax, and risk of malignant transformation. Malignancies consist mainly of pleuropulmonary blastoma in infants and young children, and bronchioalveolar carcinoma in older children and adults. We, therefore, recommend resection of symptomatic CCAM lesions within the first few days of life and all prenatally diagnosed CCAM lesions, even if asymptomatic, before 3 months of age to minimise the potential for infection or other complications and permit optimal compensatory lung growth.

Bronchopulmonary sequestration lesions arise as an aberrant protrusion from the developing foregut, and occur in two anatomically distinct subtypes; intralobar sequestrations are located within the lung and are covered by the visceral pleura, and extralobar sequestrations are located outside the normal lung. These lesions characteristically contain non-functional lung tissue and have no communicating bronchi. As a result of the inflow of blood from the systemic vasculature, these tumours may form a high-flow, low-resistance circuit precipitating cardiac failure postnatally. Postnatally, intralobar lesions pose a high risk for infection and cardiac failure, and should therefore be resected. In contrast, small non-cystic extralobar sequestrations may be followed if the diagnosis is not in question; however, the lesion should be resected if there is mass effect, lymphatic congestion with associated pleural effusion, or potential for the development of cardiac failure.

Most fetal lung lesions have a favourable outcome without prenatal intervention, and spontaneous regression is commonly observed. Congenital cystic adenomatoid malformations and bronchopulmonary sequestration growth generally peaks at 26–28 weeks’ gestation, with rates of spontaneous regression ranging from 15–65% in CCAM and as high as 68% in BPS. Importantly, despite a relative or absolute decrease in size and echogenicity during the third trimester, few if any of these lesions regress completely, and a postnatal chest computed tomography scan with intravenous contrast will confirm persistence of the lesion.

Conversely, some lesions may continue to grow at unpredictable rates, and physiologic derangements secondary to mass effect on pulmonary and mediastinal structures may occur. Oesophageal compression and interference with fetal swallowing may cause polyhydramnios, compression of adjacent lung tissue may result in pulmonary hypoplasia, and obstruction of the vena cava with impairment of venous return may precipitate cardiac failure and the ultimate development of hydrops fetalis. Serial ultrasonographic assessment is important, as the development of hydrops is a hallmark of impending fetal demise and warrants immediate intervention. A prognostic tool has been developed using sonographic measurement of the CCAM volume ratio (CVR), obtained by dividing CCAM volume (length × width × height × 0.52) by head circumference. In a retrospective study of 32 fetuses, and a subsequent prospective series of 58 women with prenatally diagnosed CCAM lesions, we observed that a CVR greater than 1.6 was associated with a 75% likelihood of developing hydrops, whereas a CVR of 1.6 or less predicts a low likelihood of hydrops developing. The growth patterns of CCAM lesions with a predominantly macrocystic pattern are less predictable and have been observed to grow rapidly beyond 28 weeks, requiring more frequent monitoring.

Myelomeningocele

Despite efforts to prevent the development of open neural tube defects by folic acid supplementation, myelomeningocele (MMC) remains one of the most common congenital defects of the central nervous system, with an estimated incidence of 3.4 per 10,000 live births. The condition is characterised by protrusion of the meninges and neural elements through the open vertebral arches, and is associated with varying degrees of paralysis of the lower extremities, developmental delay, and bowel, bladder dysfunction, or both. The severe spectrum of comorbidities associated with this disorder has prompted extensive investigation of fetal interventions, culminating in the recent publication of the Management of Myelomeningocele Study.

Pathophysiology and natural history

In addition to the primary open neural defect characterising spinda bifida, almost all fetuses with MMC display a constellation of neuroanatomic abnormalities referred to as the Chiari II malformation, with caudal displacement of the posterior fossa contents through the foramen magnum resulting in inferior displacement of the cerebellar vermis, elongation and displacement of the medulla, and brainstem compression. Hindbrain herniation impairs the normal circulation of cerebrospinal fluid, resulting in the development of hydrocephalus that will require shunt placement in 80–90% of cases. Placement of a shunt results in significant patient morbidity, with about 64% of individuals developing shunt failure caused by obstruction or infection within the first postoperative year, and 95% of individuals requiring at least one surgical shunt revision during the first 25 years of life. Shunt complications are significantly associated with both cognitive outcomes and long-term survival. Severe Chiari II malformation is also associated with brainstem dysfunction, including central apnoea, stridor, difficulty in swallowing and lower cranial nerve dysfunction, and represents the leading cause of mortality in the first 5 years of life.

Although the pathophysiology of this defect continues to be studied in experimental models, the currently accepted hypothesis proposes a ‘two-hit’ model, with primary failure of neural tube closure resulting in an open defect, and resultant exposure of the neural elements to the intrauterine environment. Interestingly, fetal lambs, in which an open neural defect is created and immediately covered, do not show significant neurologic morbidity, whereas those with uncovered defects display defects in lower extremity function and incontinence of urine and stool. These findings led to the hypothesis that neuronal injury occurs as a result of the second ‘hit’, or exposure of the neural elements to the amniotic fluid, and support a role for early repair. Indeed, fetal lambs undergoing MMC repair 4 weeks after creation of the defect demonstrate continence of stool and urine, intact sensation, and preservation of gross neuromuscular function of the lower extremities. Furthermore, in-utero closure of the defect was shown to reverse hindbrain herniation, suggesting that the development of hydrocephalus and brainstem dysfunction might also be prevented by early repair.

Fetal intervention for myelomeningocele

Consideration of prenatal surgical intervention for the repair of MMC requires re-examination of the paradigm that open maternal–fetal surgery be restricted to those instances in which fetal or neonatal demise would be imminent in the absence of surgical intervention. Although morbidity rates are high and life expectancy is shortened in this patient population, mortality rates are relatively low. The rationale for open fetal MMC repair is based upon the prevention of neurologic deficits and associated morbidities. With evidence from animal models supporting the likelihood of improved functional outcomes, the first attempts at human MMC repair were reported in the late 1990s. The first endoscopic attempts at MMC repair were reported in 1997, with no apparent improvement in clinical outcome. A subsequent report comparing outcomes between endoscopic and open repair demonstrated superior results with the open approach and, thus far, attempts at endoscopic repair have not been proven effective in a well-designed study. The first successful open fetal surgery for MMC showing improved neurologic function was carried out at CHOP in 1998, and was followed by two studies of open MMC repairs reported in 1999 ( n = 10) and 2003 ( n = 50) at the same institution. In these studies, overall survival was 90 and 94% respectively, with reversal of hindbrain herniation in 100% of fetuses within 3 weeks of surgery ( Fig. 1 ), and ventriculoperitonea shunt rates were significantly lower than expected on the basis of historical controls, with rates of 10 and 46%, respectively.

Myelomeningocele repair: (a) preoperative magnetic resonance imaging scan showing the Arnold Chiari II malformation; (b) 3 weeks after prenatal myelomeningocele repair showing reversal of the Arnold Chiari II malformation and restoration of cerebrospinal fluid around the hindbrain; (c) pre-repair appearance of a 23-week gestation myelomeningocele after exposure by hysterotomy; (d) post-repair appearance with primary closure of the skin.

On the basis of these promising initial results for open MMC repair, a group of three fetal surgery centres in the USA undertook a randomised-controlled trial (RCT) in 2003 (The Management of Myelomeningocele Study) comparing prenatal open MMC repair with postnatal surgery. The study was carried out at CHOP, Vanderbilt University and University of California, San Francisco (UCSF), with an independent Data and Study Coordinating Center at George Washington University Biostatistics Center. While the trial was being conducted, a moratorium on open fetal MMC repair outside the trial was agreed upon by other fetal centres in the USA. The study was powered to recruit 200 participants, but was halted in December 2010 when a planned interim analysis showed clear benefit for prenatal surgery after randomisation of 183 participants. The results were reported by Adzick et al. in 2011 and included 158 women randomised before July 1, 2009, with 78 in the prenatal group and 80 in the postnatal repair group. The inclusion criteria required that the fetus be 19–25.9 weeks of gestation, with MMC located between T1-S1, with evidence of hindbrain herniation, and normal karyotype without evidence of other abnormalities. The first primary end point of the study was a composite of fetal or neonatal death and the need for a cerebrospinal fluid shunt at 12 months of age. Sixty-eight per cent of the prenatal surgery group compared with 98% of the postnatal group fulfilled the primary end point, with actual shunt rates of 40% compared with 82%, respectively. At 12 months of age, rates of moderate or severe hindbrain herniation were significantly lower in the prenatal group (25%) compared with the postnatal group (67%). The secondary outcome was a score derived from the Bayley Mental Development Index and the difference between the functional and anatomic level of the lesion at 30 months of age, and was significantly better in the prenatal group compared with postnatal repair. Children in the prenatal group were significantly more likely to walk without the use of orthotics or devices (42% v 21%), and scored significantly higher in parent-reported self-care and mobility scores. Importantly, prenatal surgery was associated with higher rates of prematurity and maternal morbidity, although infant mortality rates were equal and no maternal mortalities occurred.

Long-term follow up on neurodevelopmental outcomes will be a fundamental component of evaluation of the overall efficacy of prenatal therapy for MMC. Results of a 5-year follow up on the cohort of women treated at CHOP before The Management of Myelomeningocele Study trial were recently reported. Most children achieved complete independence in cognitive (84%) and mobility (68%) scores, but continued to require significant assistance in self-care. Improved understanding of long-term functional limitations will allow for more effective interventions to maximise clinical outcomes after MMC repair.

The persistence of residual deficits in many women after maternal–fetal MMC repair has led to the hypothesis that open maternal–fetal surgery is either carried out too late to prevent damage to the neural elements, or may itself induce neural damage. As a potential alternative to open maternal–fetal surgical repair, a tissue-engineering approach may represent a less invasive method for covering the defect that could be applied earlier in gestation. The development of a mechanically-stable and amniotic fluid-impermeable tissue layer for MMC repair is a focus of active research at our institution.

Sacrococcygeal teratoma

Teratomas are germ-cell tumours originating from abnormal development and proliferation of pluripotent embryonal cells. Sacrococcygeal teratoma (SCT) arises from the primitive streak and contains tissues originating from all three embryologic layers. It is the most common tumour of the fetus, with an incidence of 1 in 40,000 births and a female-to-male ratio of 4:1.

Prenatal diagnosis is commonly made by routine screening ultrasonography, and magnetic resonance imaging may be useful to assess the anatomic extent and effect of the mass. The American Academy of Pediatrics Surgical Section classification system describes SCTs according to their relation to the sacral area. Type I tumours are predominantly external, with a small presacral component, and may be attached to the body by only a narrow stalk; type II tumours are largely external with significant intrapelvic extension; type III tumours are predominantly internal with significant intrapelvic and intra-abdominal extension and a small external component; and type IV tumours are entirely presacral, with no external component and no significant intrapelvic extension.