Thoracoscopic Versus Open Congenital Diaphragmatic Hernia Repair: Single Tertiary Center Review

Introduction

Congenital diaphragmatic hernia (CDH) is a congenital anomaly diagnosed in 1 in 2600–3700 births.^1–3 Developmentally, CDH is caused by a failure of the diaphragm to close during the 9th and 10th weeks of gestation, allowing intra-abdominal contents to herniate into the thoracic cavity. Mass effect from abdominal contents in the chest can severely compromise lung growth and maturation. At birth, many neonates with CDH suffer from underlying pulmonary hypoplasia and pulmonary hypertension, which can lead to respiratory failure. ¹ In the most severe cases, infants may require extracorporeal membrane oxygenation (ECMO). Given the severe pulmonary disease associated with CDH, mortality has traditionally been 50%–60%,^1,2,4 although more recent studies have demonstrated survival as high as 80%–90% in select patients at specialized centers.^4–7 Survivors often have significant long-term morbidity.^1,2,7

CDH was generally repaired by open laparotomy until 1995, when the first minimally invasive surgery for CDH was described. ⁸ Although some CDHs can be repaired laparoscopically, minimally invasive repair generally refers to a thoracoscopic approach. Since the advent of thoracoscopic CDH repair, some studies have reported increased survival, decreased length of postoperative mechanical ventilation and hospitalization, and improved cosmesis compared to open CDH repair.^9,10 However, two main potential drawbacks of thoracoscopic CDH repair have limited widespread acceptance of this technique. First, several studies have demonstrated an increased recurrence rate from thoracoscopic CDH repair compared to open repair.^1,4,10–16 Whether this is due to an inherent technical failure of thoracoscopic approach or due to limited experience with a novel technique remains unclear. Second, thoracoscopic CDH repair has been associated with increased intraoperative physiologic derangements due to carbon dioxide insufflation.^1,9 The physiologic effects of carbon dioxide insufflation on intraoperative arterial blood gas are well documented, but the clinical significance of these findings is not well understood. Currently, only a few centers routinely perform thoracoscopic CDH repair and even fewer report their results. ¹³ Thus, thoracoscopic versus open CDH repair remains an area of active debate within pediatric surgery.

During 2007–2015, both open and thoracoscopic CDH repairs were performed at Levine Children's Hospital at Carolinas Medical Center in Charlotte, North Carolina. With our growing experience with minimally invasive surgery, we hypothesize that thoracoscopic CDH repair will have equivalent or lower recurrence rates and similar outcomes compared to open CDH repair.

Materials and Methods

We performed a retrospective review of a single pediatric surgical group's outcomes from open versus thoracoscopic CDH repair from January 1, 2007 to August 31, 2015. All infants with a Bochdalek-type CDH born at the institution or transferred to the neonatal care unit immediately after birth and who underwent surgical repair at the institution were included in the study. Infants who died before CDH repair and infants with Morgagni-type diaphragmatic hernias were excluded from the bivariate analysis. We reviewed the medical records of all included patients to obtain demographics, perinatal course and comorbidities, operative details such as operative time and type of repair, perioperative morbidity and mortality, overall survival, and recurrence rates. The primary outcome of interest was hernia recurrence rate. Secondary outcomes included intraoperative hypercarbia or acidosis, hospital length of stay, duration of postoperative mechanical ventilation and oxygen dependence, morbidity, and overall mortality. A total of 54 patients were included in the bivariate analysis.

Open CDH repair was generally performed using a left subcostal incision (for left-sided CDH repair) with the patient in the supine position. The viscera were manually reduced from the chest, and the diaphragm was mobilized. Small diaphragmatic defects were closed primarily using nonabsorbable horizontal mattress sutures with or without pledgets. If mesh was needed to close a larger defect, polytetrafluoroethylene (ePTFE) mesh with biomaterial (GORE^® DUALMESH^®) or polypropylene mesh was affixed with nonabsorbable horizontal mattress sutures.

Thoracoscopic CDH repair was performed in the right lateral decubitus position with an axillary roll (for left-sided CDH repair). A 3 mm trocar was inserted medial to the tip of the scapula, and the chest was insufflated gently with low pressure CO₂ insufflation (4–8 mmHg). Two additional 3 mm ports were placed under direct visualization medially and laterally for working ports. When possible, the insufflation pressure was decreased once the viscera were reduced back into the abdomen. The edges of the diaphragm were then mobilized. As in the open repair, small defects were repaired primarily using nonabsorbable horizontal mattress sutures, and larger defect was repaired with ePTFE biomesh. Pledgets were used to reinforce the diaphragmatic sutures when needed. For mesh repairs with minimal native diaphragm, the posterior-lateral sutures were placed around the ribs.

Conversion from minimally invasive to open surgery was at the discretion of the operating surgeon based on factors such as hemodynamic instability and size of the defect. This was done either by converting to laparotomy or thoracotomy. For the purpose of data analysis, minimally invasive converted to open procedures were considered open.

All statistical analyses were performed using standard statistical methods. The study population was described using percentages for categorical variables, means and standard deviations for normally distributed variables, and medians and ranges for non-normally distributed continuous variables. Bivariate analysis was used to compare outcomes between the thoracoscopic and open cohorts, and subgroup analysis was used to stratify patients based on need for mesh repair, as a surrogate for defect size. Due to the small numbers in bivariate and subgroup analysis, we used Fisher's exact test for categorical variables. We used Student's t-test for normally distributed continuous variables and Wilcoxon Rank-Sum test for skewed continuous variables. A P value of <.05 was considered significant. All statistical analyses were performed using Stata/I.C. 14.1, StataCorp LP (College Station, TX) and reviewed with a biostatistician. The study was approved by the Institutional Review Board of the Carolinas Healthcare System.

Results

A total of 65 infants with CDH were treated at the institution during the study period. Of these, 11 were excluded: 6 neonates died before surgical repair, 2 had Morgagni-type hernias, and 3 patients were diagnosed later in infancy. The remaining 54 patients were included in the analysis (Fig. 1). One patient was repaired more than 30 days after birth, but during the initial admission. All other patients were repaired during the neonatal period.

OPEN IN VIEWER

FIG. 1.

Patient inclusion.

All patients were treated primarily in the neonatal intensive care unit at Carolinas Medical Center and were operated on by 1 of 6 surgeons in a pediatric surgical group. In most cases, the primary surgeon was assisted by an upper-level general surgical resident or another attending surgeon.

Approximately 72% of the patients were male and 65% were Caucasian. The average estimated gestational age (EGA) was 39 weeks (range 31–42 weeks), and the average birth weight was 3.2 kg (range 1.41–4.57 kg). Most of the mothers had received prenatal care (93%), although only 65% had a prenatal diagnosis of CDH (Table 1).

Over 95% of the patients were intubated shortly after birth, before operative repair. The duration of mechanical ventilation was widely variable, dependent on the severity of the underlying pulmonary hypoplasia and resultant lung disease. The median duration of mechanical ventilation was 8.5 days (range 1–125 days) and 2 patients required tracheostomy and mechanical ventilation at discharge. Nineteen patients (35%) required high-flow oscillating ventilation (HFOV) preoperatively and 10 patients required ECMO. All patients requiring ECMO were placed on ECMO before surgery. One patient was repaired while on ECMO, while the rest were weaned off of ECMO before surgical repair. The median duration of ECMO circuit was 8.5 days (range 5–17). Measurements of the severity of pulmonary hypertension were not standardized throughout the study period; however, 26 patients (48%) required inhaled nitric oxide. All of these patients were weaned off of nitric oxide before surgery (Table 1).

Approximately 50% of the patients had a concomitant congenital heart defect, most commonly atrial septal defects, ventricular septal defects, patent foramen ovale, and patent ductus arteriosus (N = 21). Other cardiac defects included Ebstein's anomaly (N = 1), dextrocardia (N = 1), hypoplastic aortic arch (N = 3), balanced atrioventricular canal defect (N = 1), and unspecified cardiomyopathy (N = 1). Sixteen patients had other noncardiac congenital anomalies, including genitourinary anomalies (N = 3), hydrocephalus or other intracranial anomalies (N = 2), tracheo- or laryngomalacia (N = 3), pulmonary sequestration (N = 1), Trisomy 21 (N = 1), situs inversus (N = 1), or limb anomalies (N = 1).

Of the 54 patients included in the analysis, 35 underwent attempted thoracoscopic repair. Ten thoracoscopic cases were converted to open, resulting in a conversion rate of 29%. Of these, 3 were converted to thoracotomy, and 7 were converted to laparotomy. As a result, 25 cases were successfully completed using thoracoscopy and 29 were completed open.

Sex and race did not differ between the thoracoscopic versus open CDH repair groups (Table 2). Median estimated gestational age was 39 weeks for patients who underwent both open and thoracoscopic repair. Weight at time of repair did not differ between the groups (3.07 ± 0.64 kg in open repair cohort versus 3.34 ± 0.54 kg in thoracoscopic repair cohort, P = .1).

More patients in the open repair group required HFOV (57% versus 12%, P = .001) and inhaled nitric oxide (67% versus 33%, P = .03), possibly reflecting the increased severity of pulmonary hypertension and hypoplasia in this group. However, although preoperative ECMO requirement tended to be higher in the open repair cohort compared to the thoracoscopic repair cohort (26% versus 12%), this difference was not statistically significant (P = .3). Almost all patients required preoperative mechanical ventilation, with a longer median duration of ventilation in the open cohort compared to the thoracoscopic cohort (12 versus 6 days, respectively, P = .02). Importantly, patients who underwent thoracoscopic repair had a significantly shorter course of postoperative mechanical ventilation compared to patients who underwent open repair (median 3 days [range 0–9] versus 4.5 days [range 1–105], P = .004) (Table 2).

As expected, patients who underwent open repair tended to have larger hernia defects with a higher percentage requiring mesh repair (75% versus 20%, P < .001). Despite this, herniation of the stomach and/or liver into the chest was not a contraindication for thoracoscopic repair (Table 2). Operative time was similar between the thoracoscopic and open cohorts (155 ± 58 minutes versus 171 ± 48 minutes, respectively, P = .4). In fact, operative time tended to be longer in the open cohort, possibly owing to the larger size of the hernia defect or the increased time attributable to conversion from thoracoscopic to open. In contrast to previous studies, there was no difference in intraoperative pCO₂ or pH measurements between groups. However, 2 of the thoracoscopic cases were converted to open due to rising intraoperative partial pressure of carbon dioxide (pCO₂) (Table 2).

Five patients experienced intraoperative complications. Two patients sustained splenic injuries, 1 in the patient undergoing open repair and 1 in a patient undergoing thoracoscopic repair, necessitating conversion to open. One patient in the thoracoscopic cohort developed a left tension pneumothorax requiring chest tube placement in the operating room. Another was noted to have colonic volvulus with ischemia which necessitated conversion to open. The last patient developed supraventricular tachycardia during reduction of the liver in a right-sided CDH. This patient's procedure was converted to open and subsequently aborted. This patient later died of severe pulmonary hypoplasia and respiratory failure.

Two patients in the open repair cohort developed recurrent CDH, 1 at age 10 months and 1 at 3 years. The first patient underwent attempted thoracoscopic repair, which was converted to thoracotomy for mesh placement. He gradually developed a recurrent left-sided hernia with a significant amount of herniated organs in the chest. He was taken back for repair at 10 months. This repair was started thoracoscopically, but converted to thoracotomy due to dense adhesions. The other patient underwent primary open repair of a large right-sided hernia with mesh. He was noted to have a recurrence on routine surveillance at age 3 and was followed clinically. He remained asymptomatic, but the hernia increased in size and he underwent thoracoscopic repair of his recurrent hernia at age 5. Both of these patients had required HFOV and ECMO postnatally before their original repairs. None of the patients repaired thoracoscopically recurred.

Postoperative complications varied as expected, based on the initial surgery. Four patients in the open cohort developed ventral incisional hernias (14%) and 5 developed bowel obstructions requiring operative lysis of adhesions (17%). Three patients in the thoracoscopic cohort developed physiologically significant pneumothoraces requiring chest tube placement (12%). Other minor complications included superficial wound infections, urinary tract infections, and pneumonia. One patient, who had an open repair with mesh performed at 4 weeks of age, subsequently developed an esophageal perforation, suspected to be due to mesh erosion. This patient was taken back to the operating room at ∼2 months of age for mesh removal and wide drainage. The patient then underwent staged repair of the diaphragmatic hernia with a reverse latissimus dorsi flap reconstruction at 4 months of age. This was not counted as a recurrence as the patient had complete agenesis of the left diaphragm, and no attempt was made at primary closure at the time of mesh removal.

Of the patients who underwent CDH repair, 1 died in the postoperative period during the initial hospital admission (hospital mortality 1.9%). Median length of stay was 37 days (range 9–292 days) and patients who underwent thoracoscopic repair had a statistically shorter length of stay compared to patients who underwent open repair (median 18 days [range 10–88] versus 47 days [range 9–292], P = .006) (Table 2). Most patients did not require supplemental oxygen (78%), although 4 patients were discharged on nasal cannula and 2 were discharged with home ventilators (Table 1). One patient died from respiratory failure during a subsequent hospital admission, resulting in an overall mortality of 3.7% in the study population.

Much of the decision for thoracoscopic versus open repair depends on the degree of preoperative respiratory compromise, which can be related to the size of the hernia defect. This was difficult to assess accurately based on the available records. Therefore, we performed a subgroup analysis using need for mesh repair as a surrogate for defect size.

Within the subgroups, we examined the differences in postoperative mechanical ventilation and overall length of stay between patients in the thoracoscopic and open cohorts. Out of the 26 patients who required mesh repair, 5 were completed thoracoscopically and 21 were completed open. Patients in the thoracoscopic cohort trended toward shorter duration of total and postoperative mechanical ventilation, although the difference was not statistically significant (Table 3). Overall length of stay was essentially unchanged between the open and thoracoscopic cohorts (median 70 versus 66 days, P = .4). Twenty-seven patients did not require mesh repair. Of these, 7 underwent open repair and 20 underwent thoracoscopic repair. Postoperative duration of mechanical ventilation and total length of stay were similar between the 2 cohorts (Table 4).

Discussion

In this study of 54 patients, roughly one-half underwent successful thoracoscopic surgery. This is one of the largest cohorts reported from a single institution. In contrast to several prior studies, outcomes in our cohort were not inferior in the thoracoscopic group, with no recurrences and statistically similar operative times, intraoperative pCO₂ and pH levels, and mortality. Duration of postoperative mechanical ventilation and total length of stay were statistically lower in the thoracoscopic group, although this was likely related to defect size and underlying pulmonary function, rather than method of repair. Patients who underwent thoracoscopic repair had smaller hernia defects and, based on parameters such as need for HFOV and ECMO, likely had less severe pulmonary hypoplasia and pulmonary hypertension. This difference inevitably translates into longer need for mechanical ventilation, longer length of stay, and higher mortality in the open repair cohort. Unsurprisingly, the difference in postoperative mechanical ventilation and length of stay disappeared in subgroup analysis when controlling for defect size. That being said, even for patients with large defects and the need for significant pulmonary support, thoracoscopic repair is a viable option once the patients have been clinically stabilized. In addition, patients who underwent thoracoscopic repair avoided the potential complications of laparotomy, such as ventral hernia and bowel obstruction.

Management of CDH in neonates has posed a challenge to pediatric surgeons and neonatologists since the early part of the 20th century. The first successful CDH repair was performed in 1929, ¹⁷ but mortality remained high over the next half century due to the poor options for treatment of respiratory failure in infants. Improvement in neonatal critical care and respiratory support has increased survival from CDH from 20% to over 50% ¹⁷ by providing a method to stabilize critically ill neonates and improve oxygenation. However, the pathophysiology of CDH was still not well understood. Until the 1980s, immediate repair of the hernia was encouraged, as the compression of the lung by the herniated viscera was thought to be the main driver of pulmonary hypertension.^1,17 More recently, we have gained appreciation for the underlying pulmonary hypoplasia of both the ipsilateral and contralateral lung, which results in pulmonary hypertension and subsequent morbidity and mortality. As such, surgical care can safely be delayed until the neonate has stabilized and the labile pulmonary hypertension has resolved. ¹

Surgical repair of CDH can be performed through multiple modalities. Traditionally, left-sided CDHs were repaired through a left subcostal or left thoracotomy incision. Since the first thoracoscopic CDH repair in 1995, numerous pediatric surgeons have described their experience with minimally invasive approaches. Laparoscopic repair of CDH has been reported, particularly for Morgagni hernias. However, the use of laparoscopy is limited in the repair of large hernias as reduction of the viscera can substantially limit the working space and visualization. Thoracoscopy, however, offers the advantage of a natural working space in the chest created by the hypoplastic lung.

Thoracoscopic CDH repair offers several potential benefits over open repair, including less postoperative narcotic use, faster wean from mechanical ventilation, and avoiding the complications of laparotomy.^9,11,18 However, several concerns about thoracoscopic repair have been raised in the literature. The three most common concerns are increased operative time, increased recurrence rate, and hypercarbia resulting from CO₂ insufflation. As a result, thoracoscopic versus open repair has remained a topic of debate in the pediatric surgical community. This study addresses these concerns and demonstrated no difference in all three measures in thoracoscopic compared to open repair.

Thoracoscopic repair is often criticized for requiring longer operative times.^11,16,18 This is a common criticism of all minimally invasive techniques, and, as seen with other minimally invasive procedures, the difference decreases as surgeons become more familiar with the technique.^3,12 This study showed no significant difference in operative time between open and thoracoscopic procedures.

CDH registries from the 1990s and early 2000s demonstrate the early results of thoracoscopic CDH repair. ¹³ Multiple early studies reported a high rate of recurrence following thoracoscopic repair compared to open repair, ranging from 5% to 40% for thoracoscopic repair^{4,10,11,15,16,19} and 2% to 10% for open repair.^2,4,11,13,19 CDH recurrence is often thought to be due to technical failure of the repair, resulting from loose or widely spaced sutures, repair under tension, or inadequate mobilization of the diaphragm rim. ¹¹ It is not clear whether these technical failures are inherently more likely to occur during thoracoscopic technique or whether the early high recurrence was due to unfamiliarity with a novel technique. Later studies, including this study, have not shown a significant increase in recurrence rates between thoracoscopic and open technique.^3,12,20 This study reports a low overall recurrence rate of 3.7%, with no recurrences noted in the thoracoscopic cohort after a median follow-up of over 2 years.

Several studies have suggested that CO₂ insufflation may result in hypercarbia that could be detrimental in already fragile infants.^4,9,21 Although elevated pCO₂ and acidosis are well documented in thoracoscopic repair compared to open repair, it is not clear whether this has any clinical significance. ¹⁸ Our study did show a slight increase in intraoperative pCO₂ and associated decrease in intraoperative pH, but no significant difference between these measures in the thoracoscopic and open cohorts.

Not all patients with CDH are appropriate candidates for thoracoscopic repair. Patients with minimal pulmonary hypertension and respiratory function that rapidly stabilizes are often considered good candidates for thoracoscopic repair, while patients with severe respiratory failure requiring ECMO and inhaled nitric oxide are often excluded. Several studies have examined anatomic and physiologic criteria to predict successful thoracoscopic repair. These include presence of the stomach in the abdomen on chest X-ray (indicating intact hiatus), minimal ventilator settings (peak inspiratory pressures in the low 20s), and no pulmonary hypertension.^7,10 In addition, premature infants or infants with significant cardiac defects are often not considered good candidates for thoracoscopic repair.^5,11,15,19 Using these criteria, prior studies estimate that 25%–30% of patients would be candidates for thoracoscopic repair. ²⁰ In this retrospective review, thoracoscopic repair was possible in almost 50% of patients and had equivalent outcomes compared to open repair, even in patients requiring ECMO and mesh repair for large defects.

This study has the usual limitations of a retrospective review. Incomplete documentation resulted in some missing variables in certain records. In addition, the institution changed from paper to electronic medical records during the study course, and some of the records from the earliest patients are missing details of their hospital and operative course. These missing data are documented in the results and should be taken into consideration when interpreting the results.

Previous studies have examined the effect of thoracoscopic surgery on the intraoperative pCO₂ out of concern that CO₂ insufflation may lead to hypercarbia and acidosis. We did attempt to address this in our study; however, we must acknowledge the weaknesses of these results. Measurements of preoperative, intraoperative, and postoperative pCO₂ and pH were not standardized. Preoperative and postoperative laboratories were not drawn at the same time in relation to when the surgery took place and, as such, cannot easily be compared. Most patients had at least one blood gas drawn during surgery, which allows for slightly more accurate comparison of intraoperative blood gases; however, the timing of these laboratories was not standardized based on duration of CO₂ insufflation. We can use these measurements to assess trends in levels of hypercarbia and acidosis, but should not give too much weight to these findings.

The small sample size of this study may have affected our ability to accurately detect statistically significant differences between relatively uncommon events, such as hernia recurrence, between the two cohorts. The subgroup analysis, stratifying patients by need for mesh repair, required even smaller sample sizes. As such, our interpretation of the results of the subgroup analysis is limited. Although there appears to be no difference in outcomes such as duration of postoperative ventilation and length of stay between the thoracoscopic and open cohorts, we cannot determine whether this is because the modes of repair are actually equivalent or whether the subgroups are too small to detect a difference. Based on prior studies, we believe that thoracoscopic repair may offer a benefit in terms of decreased need for postoperative mechanical ventilation and subsequent decreased length of stay, but repeat studies with a larger population would be required to confirm this.

The major pathology underlying CDH is the underlying pulmonary hypoplasia and pulmonary hypertension. Pulmonary hypertension can be difficult to quantify in neonates, although markers such as need for increased ventilator support with or without ECMO or evidence of right heart strain on echocardiogram can be used as surrogates. Although the diagnosis of pulmonary hypertension is often included in studies examining the outcome of CDH repair, these studies do not always qualify how the diagnosis was made or the severity of the pulmonary hypertension. The same problem exists with this study. In the future, we would like to see studies involving a more comprehensive database with prospectively collected data focusing on assessing the degree of pulmonary hypertension. This would allow for better matching of risk factors between cohorts and, as a result, a more accurate comparison.

Conclusion

The treatment of CDH remains a challenge in neonatology and pediatric surgery. The timing and method of surgical correction of the defect can be variable, and the underlying pulmonary status of the infant often is a stronger predictor of length of stay and outcome. This study suggests that selective use of thoracoscopy for CDH repair can be performed safely in neonates even in the setting of large-sized defects, herniation of the stomach and/or liver into the chest, or the need for preoperative ECMO. This review of thoracoscopic CDH repair showed no recurrences, no increase in operative time, and no significant difference in perioperative parameters compared to open repair, while demonstrating that these patients may avoid the long-term complications of laparotomy. As a result, we suggest that as minimally invasive technique improves, thoracoscopic repair has the potential to be superior to traditional open repair.