25 THE PRENATAL MANAGEMENT OF THE FETUS WITH A CORRECTABLE DEFECT
The potential for the field of fetal surgery would never even have been considered without the existence of prenatal ultrasound diagnosis. This makes the discussion of these interventions particularly pertinent to a textbook dedicated to ultrasound in obstetrics and gynecology.
Most malformations considered for fetal surgery may be initially detected as abnormal on routine screening but may not be accurately identified until a targeted examination is performed. It is self-evident that an accurate diagnosis is critical to determining the appropriateness of a fetal surgical intervention. Fetal malformations that are not isolated or are associated with aneuploidy or a syndrome have always been excluded as candidates for intervention. This is because the risk-benefit ratio, given that all procedures involve some maternal risk, is not acceptable. Therefore, great effort should be made to confirm both the accurate diagnosis of a malformation and the fact that it is truly isolated. Ancillary techniques to ensure full and accurate assessment include fetal echocardiography, magnetic resonance imaging (MRI), and karyotyping, depending on what additional problems the fetus is most at risk. Regardless, one must always counsel patients regarding the fallibility of the process and that abnormalities initially that are believed to be isolated may not ultimately prove to be so in the final analysis.
The indications for fetal surgical interventions have expanded, as has the total number of procedures performed, the sites at which they are performed, and the number of physicians performing them. However, overall these procedures remain very limited when compared with the number of pregnancies and even the number of fetuses with malformations. One of the responsibilities of physicians with an interest in prenatal diagnosis and intervention, in the future, is to determine training needs and oversight for operators and centers involved in this field. It is unclear how many centers would be needed given the rarity of malformations and the even smaller proportion of those with a malformation that may need fetal intervention. A decision will need to be made if ease of accessibility is a substitute for volume of procedures performed. Certainly with access to the World Wide Web, patients in our experience have become savvy consumers who are readily able to establish contact and provide information necessary to make a preliminary assessment regarding their fetus’s diagnosis. This, together with the increasing ease of large volume electronic transfer of information, has made geographic proximity less important.
Over the past decade technologic advances have allowed a transition toward less invasive procedures. Initial fetal surgical procedures at the University of California, San Francisco (UCSF) involved maternal laparotomy and hys terotomy. This approach evolved into laparotomy and uterine endoscopy, and most recently, into percutaneous procedures using laparoscopic devices with diameters at times of 3 mm or less. Our experience suggests that the less invasive approaches are associated with a less complicated postoperative recovery for the mothers but does not entirely eliminate morbidity1 (Table 25-1). Each one of these approaches is discussed in detail.
The feasibility of performing a hysterotomy and subsequent closure of the gravid human uterus was tested in the primate (Fig. 25-1). The reason for using the primate model is that other gravid large animal models such as the sheep, are much more forgiving. The sheep uterus is more thin walled and has a multicotyledonary placenta. Closure is more straightforward, and preterm labor is rarely a significant problem. The safety profile in this series of primate fetal surgeries was reassuring, including subsequent fertility.2 The human experience is now extensive, both from UCSF and others,1,3,4 primarily associated with the large numbers of fetal spina bifida repairs. We currently reserve maternal laparotomy/hysterotomy procedures for repair of spina bifida, resection of sacrococcygeal teratoma and other tumors, and lobectomy for congenital cystic adenomatoid malformation (CCAM) of the lung.
FIGURE 25-1 Photograph of operative laparotomy and hysterotomy in a Rhesus monkey. This early work paved the way for the ability to perform such procedures safely in the mother carrying a fetus with a variety of malformations.
We have recently reviewed our experience at UCSF with maternal hysterotomy1 (see Table 25-1). Eighty-seven hysterotomies were performed between 1989 and 2003. There were significant immediate postoperative complications. In the early experience, pulmonary edema related to multiple tocolytic use, particularly nitroglycerin, and aggressive fluid management was a significant problem.5 Thirteen percent required transfusion for intraoperative blood loss. Pregnancy outcomes were also significantly affected by a preterm premature rupture of membrane (PPROM) rate of 52%, and 33% having preterm labor refractory to maximal tocolytic management leading to delivery. The mean time from hysterotomy to delivery was 4.9 weeks (range 0 to 16 weeks). The mean gestational age at the time of delivery was 30.1 weeks (range 21.6 to 36.7 weeks). Others6,7 have similar experiences with respect to an increased risk of preterm delivery following hysterotomy. Most of the morbidity associated with hysterotomy has decreased with experience. Significant pulmonary edema or blood loss is rare, and the mean gestational age at the time of delivery for repair of myelomeningocele (MMC) is now around 34 weeks.
The practical aspects of hysterotomy and postoperative management have evolved since the initial years of experience. The following is a description of our current approach. Lengthy discussions regarding the risks, benefits, and alternatives of the procedure are important, including the experimental nature of the surgery. We generally differentiate the risks to the mother, the fetus, and the pregnancy in our counseling. The risks to the mother are similar to any major abdominal surgery, although in this case, there is no direct physical benefit to her. In addition there are the risks associated with aggressive tocolytic therapy and bed rest in a hypercoagulable state. The risks to the fetus are primarily vascular instability and hypoperfusion intraoperatively leading to organ injury or death, and prematurity due to postoperative complications. The risks to the pregnancy are primarily preterm labor and premature rupture of membranes and preterm delivery. Infectious complications are rare, except when premature rupture leads to prolonged latency. An important additional discussion point is that all subsequent deliveries, including the index pregnancy, must be by cesarean section. Data regarding future fertility are reassuring, with no increased incidence of infertility in the UCSF experience in those patients attempting pregnancy.8 Experience from Children’s Hospital of Pennsylvania (CHOP) suggests a concerning risk for uterine rupture ehiscence in subsequent pregnancies that may be as high as 6% to 12%,9 which would be considerably higher than the risk after previous low transverse cesarean section (1% or less)10 or classic cesarean section (5% to 10%).11 Another potential risk in subsequent pregnancies is placenta accreta. The reason for this risk is that the site of a hysterotomy performed in the second trimester is never in the same area as a cesarean section entry. There is an increased risk of placenta accreta in any case in which implantation is in an area of uterine scarring. Multiple incisions increase the likelihood of implantation in such an area. To our knowledge, there has not been a case of accreta in a fetal surgical patient of ours in a subsequent pregnancy.
Ultrasound is used for guidance before the hysterotomy procedure. Once the patient has had general anesthesia and intubation initiated, and the sterile field is established, ultrasound is used to establish fetal lie, its location within the uterus, and the position of the placenta. Transabdominal manipulation with ultrasound guidance is used to position the fetus such that the fetal surgical site is near the fundus. Depending on the maternal body habitus and fetal size and position, this may be challenging. Laparotomy is then performed, and the ultrasound transducer, covered in a sterile sleeve, is placed directly on the surface of the uterus. The edge of the placenta is identified, because this information is critical in deciding where to perform the uterine entry. Generally, one wants to make the uterine incision centered as far from the placental edge as possible. The reason for this is that once the incision is made, and the amniotic fluid leaves the confines of the uterus, the uterus shrinks. One is generally surprised at how close the incision is to the edge of the placenta regardless of one’s efforts. The reason proximity to the placental edge is so critical is because of the risk of bleeding and abruption, which cannot readily be controlled and, if significant, will usually necessitate immediate delivery for maternal safety.
Ultrasound is also used to identify the specific position of the fetus within the uterus. Generally the incision in the uterus will have been made so as to have optimal access to the fetal part to be surgically addressed. Ultrasound is also used for transuterine monitoring of the fetal heart during the procedure. Following completion of the fetal intervention, the membranes and myometrium are closed with several layers of suture. A catheter is left in the uterine cavity to allow lactated Ringer’s solution to be infused together with antibiotics. Ultrasound determines the volume of “amniotic” fluid, which is generally at a low-normal level to minimize stress on the suture line.
Post-operative management generally involves 24 hours of intravenous tocolysis with magnesium sulfate as well as maintenance with oral indomethacin for a total of 48 hours. Long-term tocolytic maintenance with nifedipine is continued basically until delivery. Antibiotic “prophylaxis” is continued for 24 hours. Ultrasound monitoring to evaluate fetal health, amniotic fluid volume, cervical length, and ductal patency is performed at least daily. Hospital discharge generally occurs 4 to 5 days after surgery. If all goes well, long-term monitoring with ultrasound continues at a weekly frequency.
The growing popularity of videoendoscopic surgery in the 1990s, combined with earlier experience with fetoscopy, paved the way for the concept of endoscopic fetal surgery. The rationale was that a tiny puncture of the amniotic cavity would overcome some of the limiting steps in fetal surgery: (1) preterm labor, which was believed to be triggered by the large uterine incision of open fetal surgery; and (2) significant maternal morbidity associated with a large laparotomy. The ultimate hope was that fetoscopic interventions would be possible by a percutaneous approach. Animal models were developed to test the new instruments and techniques. The ovine model is very resistant to postoperative preterm contractions, as mentioned earlier, and therefore not ideal to study the hypothesis that endoscopic access to the uterine activity was associated with less uterine activity than after hysterotomy. We studied that aspect in midtrimester Rhesus monkeys (Macaca mulatta) who underwent triple cannulation of the amniotic cavity during 60 minutes. No significant postoperative premature contractions could be demonstrated, in contrast to uterine irritability following hysterotomy.12 Use of this higher species is limited by ethical and financial constraints, but today, the model is still in use for study on healing of the fetal membranes after fetoscopic cannulation.
The risks of fetoscopy are associated with the uterine puncture as well as the specific procedure that is being treated. In some cases, adverse outcomes may be inherent to the severity of the disease being treated, such as in twin-to-twin transfusion syndrome (TTTS). In the UCSF experience,1 the morbidities were in some cases similar to the more invasive procedure of hysterotomy and in some cases much more akin to the pattern seen in the so-called fetal image–guided surgery (FIGS). One reason for this is that the initial UCSF approach was “macroinvasive” endoscopy including laparotomy, uterine exteriorization and general anesthesia. Current risk profiles with a percutaneous approach and smaller instruments demonstrate less morbidity.13 This includes a much lower rate of preterm labor and even PPROM. Length of hospitalization and time of return to normal maternal activity are also much improved.
Patients are generally premedicated with a tocolytic agent, often indomethacin, and given prophylactic intravenous (IV) antibiotics. The procedures are performed under local or regional anesthesia. Depending on the gestational age and the tradition of the center, the surgery may be performed in the surgical operating rooms, labor and delivery, or the ultrasound suite. Cannulas, instruments, and particularly, endoscopes have undergone a tremendous evolution in the past 10 years, based on prototypes developed in animal models. Operative fetoscopy is a sonoendoscopic enterprise that has evolved so that the surgical team can see the ultrasound and fetoscopic images simultaneously. Purpose-designed embryo- or fetoscopes typically have remote eyepieces, to reduce weight and facilitate precise movements. Nearly all are bendable fiber-endoscopes rather than conventional rod lens scopes, and as the number of pixels increases over time, image quality improves. Typical diameters are between 1.0 and 2.0 mm. Thin-walled semiflexible plastic cannulas (10f Check-flo introducers, Cook Medical, Bloomington, IN) are used to create amniotic access, so that instrument changes are possible. Sharp trocars have been developed to accommodate the wide range of diameters used for different operations (Karl Storz, Germany). Alternatively, one introduces the endoscope sheath loaded with a sharp obturator (Fig. 25-2). Specifics can be reviewed in several texts.14,15 Basically, however, ultrasound is used to identify an appropriate entry point and then is used to direct the trochar into the amniotic cavity, avoiding the placenta and the fetus, and obviously maternal organs such as bowel and bladder. One group has documented the safety, in their hands, of a transplacental approach.16 Despite this experience, most operators still attempt to avoid the placenta. The trochar is then replaced by the fetoscope. Ultrasound is still used to direct the scope within the uterus, because its field and depth of view can be relatively limited. These procedures are “sono-endoscopic.” For cases of TTTS, the endoscope is in the sac of the recipient twin, the one with the polyhydramnios. The entry point is determined by the factors noted earlier, but also to allow good visualization of the vascular equator of the twins. The whole equator is then explored, and any unpaired vessels consistent with abnormal communications are ablated using the laser fiber that is advanced through the operating channel of the endoscope sleeve. After successful ablation, the endoscope is withdrawn and the polyhydramnios drained through the cannula under ultrasound guidance. Once the fluid has reached a normal level (deepest vertical pocket of around 5 to 6 cm) the cannula is removed. This amnioreduction reduces the risk of port-site leaking and amniotic fluid irritation of the peritoneal cavity, which may be painful. It may also improve placental perfusion, but regardless, makes the patient more comfortable. In many cases, little or no tocolytic medication is needed, and patients are generally discharged within 24 hours or less of the procedure.
(Reprinted with permission from Group TE: The Eurofoetus Group. In Deprest J, Ville Y, Barki G, et al [eds]: Endoscopy in Fetal Medicine. Tüttlingen, Germany, Endopress, 2004, pp 1–58.) Inserts, clockwise from top left: Percutanously inserted 10 Fr cannula; fetoscopic view of arteriovenous anastomosis being coagulated. Bottom: injected placenta after procedure, demonstrating ablated vessels.
Shunts are used for chronic drainage of fluid-filled fetal cavities, organs, and cysts. The first shunt was developed by Harrison at UCSF in the early 1980s.17 This is basically a double pigtail shunt that is introduced through a 14-gauge introducer (Cook Medical, Bloomington, IN). The Rodeck shunt was developed during a similar time period in the United Kingdom. It is also a double pigtail shunt but is longer and has a greater diameter (Rocket Medical, UK). Obviously it uses a larger diameter introducer.18 These catheters are used are for draining obstructed bladders, pleural effusions, and large macrocystic CCAMs.
Radiofrequency ablation (RFA) is most commonly used in the nonobstetric patient for the destruction of tumor tissue in solid organs such as the liver. The group at the UCSF Fetal Treatment Center was first to use it for the application of localized cautery of vascular communications. Initially, it was used for ablating the feeding vessels to the anomalous fetus in twin-reversed-arterial perfusion (TRAP).19 Since that time, we have used it for selective reduction in monochorionic twin gestations discordant for severe anomalies and in patients with severe TTTS without hope for salvage of one of the twins.
The risks of complication of shunt placement and RFA are lower than for the more invasive fetal surgical interventions such as hysterotomy. Obviously by definition, all invasive procedures involve a risk of hemorrhage and infection. However, in our experience, these adverse events occur at a much lower frequency (see Table 25-1).1 The triggering of preterm delivery by these procedures is also quite unusual, although the risk of PPROM remains. There is also the risk of fetal injury, which in cases of monochorionic twins, generally is related to hypotension from acute hypovolemia in the normal co-twin secondary to exsanguination into the placental vascular bed and the other fetus.
As we have become more comfortable with these procedures, the length of hospitalization, complexity of perioperative management, and type of anesthesia have changed. In many cases, they can be performed as an outpatient procedure. Tocolytic prophylaxis is given as a single dose of indomethacin. Routine preoperative antibiotic prophylaxis is given. These procedures are performed completely under ultrasound guidance (FIGS). We perform these procedures using either spinal or local anesthesia. For shunts, a small incision is made in the maternal skin, and then the introducer with the trochar in place is advanced into the amniotic cavity. Care is used to evaluate the myometrium that will be traversed, with a high frequency transducer and color flow Doppler with low flow settings, to avoid large veins. We generally avoid a transplacental approach as well. The trochar and introducer are then advanced into the area where the shunt will be placed. Once in position, the trochar is removed and care taken to not allow the fluid to be drained and escape by placing one’s finger over the end. The shunt is then loaded into the introducer and advanced using a internal pushing device. These pushers either are of a certain length or have marks on them to advance just the internal coil of the catheter stent out of the introducer. It is critical to image this with ultrasound as well. Once the inner coils of the catheter stent are appropriately positioned, the introducer is carefully withdrawn, while at the same time advancing the shunt further, so that the outer coil is positioned on the skin of the fetus, within the amniotic cavity. Care must be taken to have enough of a fluid gap between the fetus and the wall of the uterus to minimize the risk of the outer end of the shunt being stuck in the myometrium or maternal abdominal wall, with the risk of an amnioperitoneal shunt. Given the ready access to these shunts, and the fact that physicians feel capable of performing the procedure because of their comfort with image-guided procedures such as amniocentesis, this may be the most frequently and widely performed fetal surgical intervention. Unfortunately, this experience and associated outcomes are not documented in the scientific literature.
Postoperative management involves maternal and fetal monitoring. Further tocolytic management is tailored based on contraction activity. Frequently, no further medication is necessary. Maternal vital signs should be followed carefully because direct observation of the uterine puncture is not possible to determine hemostasis because of the percutaneous approach. One benefit of not needing tocolysis is that the hemostatic mechanism of the uterus in response to a puncture is a localized contraction. Ultrasound is obviously critical to fetal follow-up.
The RFA device that we currently use at UCSF includes a 17-g needle device (RITA Medical, Fremont, CA). The perioperative management is identical to that discussed earlier with shunts, except that local anesthesia alone is generally used. In the case of TRAP, the instrument is guided into the tissue of the acardiac twin at the level of the cord insertion. The prongs are deployed, and energy transmission to the device initiated. Because of the heat generation, there is outgassing that is readily visible with ultrasound. The procedure is considered completed when there is no evidence of flow in the acardiac twin or the cord leading to it on color and pulsed Doppler imaging. The prongs are then retracted and the device withdrawn. Postoperative monitoring is similar to shunt placement, and further tocolytic management is rarely necessary. The patients can generally be discharged within hours of the procedure.
In an estimated 15% of monochorionic (MC) twins, there is a chronic imbalance in the net flow of blood across the vascular anastomoses, resulting in TTTS. The condition is usually explained by the combination of transfusion from the donor, leading to hypovolemia, oliguria, and oligohydramnios in the donor twin (so-called stuck twin) and hypervolemia, polyuria, and polyhydramnios in the recipient twin. The recipient twin often develops circulatory volume overload and hydrops. The condition coincides with specific patterns of vascular anastomoses. This forms the basis for the surgical approach to this condition. Fetal vessels run on the surface of the placenta, where they can be documented by fetoscopic inspection, and ablated (see Fig. 25-2). Although several types of anastomoses exist, knowledge about their nature (either by Doppler studies or by fetoscopy) does not play a role in TTTS: all intertwin anastomoses should be ablated as to functionally “bichorionize” the placenta (see Fig. 25-2).
The diagnosis of TTTS is based on ultrasound documentation of very specifically discrepant amniotic fluid levels. It requires the presence of oligouric oligohydramnios (deepest vertical pool [DVP] <2 cm), together with polyuric polyhydramnios (8 cm DVP < 20 weeks; ≥10 cm ≥20 weeks; criteria based on the Eurofoetus trial). Intrauterine growth restriction may be part of the clinical picture but is not a criterion for establishing the diagnosis or an indication for therapy. The natural history of TTTS is not clearly defined yet, but in a number of cases, it may be progressive, with development of abnormal Doppler patterns in the umbilical artery of the donor fetus or abnormal venous Doppler patterns in the recipient, and ultimately fetal hydrops or fetal death. The latter signs form the basis of the ultrasound-based Quintero stages20 (Table 25-2). However, cases may occasionally progress acutely from having discrepant fluids (stage I or II) directly to the stage of fetal death (stage V). This staging also does not necessarily determine management based on currently available data.
If left untreated, perinatal loss is more than 80% in severe TTTS; therefore, therapy is mandatory and fetoscopic laser has the best pathophysiologic basis. In addition, the Eurofoetus randomized trial demonstrated better outcomes after fetoscopic laser than after serial amnioreduction.21 The median gestational age at delivery was significantly higher in the laser-treated group compared with the amnioreduction group. Survival of at least one twin to 6 months was higher following laser therapy. Amnioreduction more often leads to demise of both twins. Survival chances were similar for donors and recipients. It is important to stress that laser coagulation is more effective than amnioreduction for a given stage. It is true that survival decreases according to stage; however, even for stage IV, survival of the hydropic recipient is 50%.22 Therefore, this does not justify promoting selective fetocide based purely on the stage of the disease at presentation. Preoperative measurement of the cervix can assess the risk for premature delivery.23 With a cervix less than 30 mm, the risk for delivery before 34 weeks is about 74%. If the cervix is less than 20 mm, the vast majority of patients miscarry. The risk for PPROM is estimated to be around 10% or less; the risk for abruption is 1% to 2% but relates to the amnioreduction part of the procedure. Other uncommon complications are chorioamnionitis and hemorrhage. Post-laser fetal anemia and recurrent or persistent TTTS should be screened for and may suggest incomplete division of the placenta. Recent injection studies have demonstrated that this might be more often the case than expected, and that it does not always lead to clinical problems.24 Cardiac abnormalities can develop as well and require postnatal follow-up by a pediatric cardiologist. Therefore, it should be clear that compulsive postoperative ultrasound follow-up by experienced fetal medicine specialists is a critical component of successful therapy.
In the Eurofoetus trial,21 a higher percentage of infants were alive without major neurologic morbidity at age 6 months or greater in the laser group (Table 25-3). The long-term follow-up of these patients is now pending. Until that time, counseling can be based on the detailed follow-up studies of Hecher’s group,25 with a 6% rate of severe handicap compared with 20% to 25%, which is the rate commonly associated with amnioreduction.