Repair of 104 Failed Anti-Reflux Operations

Abstract

Objective:

To assess whether reoperative surgery for failed Nissen fundoplication is beneficial and to classify all mechanisms of failure recognized.

Summary Background Data:

Antireflux surgery is often necessary, but a 10% failure rate is commonplace. We report results for patients undergoing reoperative surgery and present a nomenclature of mechanisms of failure.

Methods:

A total of 104 patients, who had a previous fundoplication for gastroesophageal reflux disease (GERD), underwent reoperative surgery. Manometry (n = 86), endoscopy (n = 101), pH monitoring (n = 27), upright esophagram (n = 90), gastric emptying (n = 26), and symptom assessment (n = 104) were performed prior to reoperative surgery. Patients were also assessed before and during reoperation for mechanism of failure using a newly proposed classification. The operative approach was laparoscopic in 58 patients, via open laparotomy in 12, and a thoracotomy in 34 patients. Follow-up was conducted by phone interview and was completed in 97 patients (97%; 3 were deceased) with a mean follow-up of 32 months (range, 1–146 months).

Results:

The conversion rate to laparotomy for laparoscopic patients was 8%. The perioperative complication rate was 32%. One patient died of respiratory insufficiency after a laparotomy. Seven patients required additional surgery for correction of persistent or recurrent symptoms. The short and long-term complication rate was similar for the different operative approachs. Symptom resolution (rare or absent) occurred in 74% of patients with dysphagia, 75% with heartburn, 85% with regurgitation, and 94% with chest pain. The overall post-reoperative patient satisfaction was 7 on a scale of 1 to 10 and 3 on a scale of 1 to 4 when patients were asked to grade the operative result. There was no difference in the symptom resolution for patients operated upon by the laparoscopic approach as compared with laparotomy, but those patients undergoing a Collis gastroplasty had poorer results. The preoperative accuracy of assessment for mechanism of failure was 78%. A nomenclature of mechanisms of failure is included to aide reoperative assessment and new mechanisms of failure are described.

Conclusion:

Reoperative surgery results for GERD are satisfactory. A variety of operative approaches proved equally effective. Poorer results were observed in patients with more advanced disease.

A total of 104 patients underwent reoperative surgery for failed Nissen fundoplication. Complications, symptom resolution in relation to esophageal disease and operation performed, and a new classification of mechanisms of failure are reported.

Gastroesophageal reflux disease (GERD) is a common disorder accounting for approximately 75% of esophageal pathology.¹ Although medication is often effective, long-term therapy has become commonplace.² However, prolonged medical therapy often includes escalating dosages, which can be an important financial burden on patients, especially as insurance coverage for pharmaceuticals continues to diminish.^2,3 Surgical therapy for GERD has improved with a better understanding of the underlying pathophysiology of GERD and improved patient selection.^4,5

A controlled randomized trial of severe GERD patients demonstrated surgery to be superior to medical therapy in terms of symptom resolution.⁶ Other investigators have provided evidence to favor antireflux surgery over medical therapy,^7,8 and Nissen fundoplication remains the operation of choice in the management of patients with severe GERD and its associated complications. The number of operations is increasing annually. Within the last decade, there has been almost an 8-fold procedural increase. This is attributed to the increased use of esophageal pH and motility testing, greater awareness of GERD-related quality of life issues, the advent of minimally invasive Nissen fundoplication,^9–11 functional results of laparoscopic antireflux procedures that are equal to that of open surgery, significantly less postoperative morbidity, and a shorter hospital stay.^12–14

The surgical management of GERD, whether performed open or laparoscopically, does have a failure rate and reoperation may be required for optimal results. Failure of open fundoplication occurs in 9% to 30% of patients,^4,15,16 whereas published failure rates of laparoscopic Nissen fundoplication are 2% to 17%.^6,17–20 These lower published rates probably reflect shorter follow-up rather than an intrinsically better operation.

Reoperations for failed or recurrent GERD are technically more demanding due to adhesions from previous surgery, obscured anatomy, and more advanced disease. A morbidity of 4% to 40% and a mortality rate of 0% to 4.9% can be expected with reoperation,^15,21,22 and the overall symptom resolution rates as compared with those of primary surgery are compromised.

Large case series are needed to better understand the benefit of reoperative surgery.^23–26 To accomplish this extensive primary surgery experience, a strong referral base and a good esophageal physiology laboratory is necessary. This study adds to accumulated knowledge for prognosis of these complex patients by reviewing our results of 104 failed fundoplications. We also provide a comprehensive nomenclature for fundoplication failure, which includes morphologic, functional, and therapeutic mechanisms in addition to diagnostic errors.

MATERIALS AND METHODS

An Institutional Review Board exemption for informed consent was given as some patients had been followed annually for more than 10 years. Between 1993 and 2004, 850 patients underwent primary laparoscopic Nissen fundoplication (LNF) at Creighton University Medical Center, 28 patients of which underwent both primary and reoperative fundoplication by coauthors (C.J.F. or S.K.M.). An additional 76 patients (73%) were referred for reoperation from other centers for a total of 104 patients. The inclusion and exclusion criteria for this study are shown in Table 1. Patients with reoperative antireflux surgery without a fundoplication in either the primary or reoperative procedure were not included, eg, those with a primary antireflux prosthesis, hiatal hernia repair without fundoplication and those undergoing a gastrectomy for reflux control or a complete fundoplication takedown at reoperation.

TABLE 1. Inclusion and Exclusion Criteria for Study

Preoperative Evaluation

Patients presenting to the esophageal center with persistent or recurrent symptoms after fundoplication were evaluated with a standardized esophageal questionnaire followed by esophagogastroduodenoscopy (n = 101), manometry (n = 86), an upright barium esophagogram (n = 90), 24-hour pH monitoring (n = 27), and/or a gastric emptying study (n = 26). The questionnaire used is shown in Table 2. In addition, timing of symptom recurrence, number of dilations after reoperative surgery, a subjective grading system, and an overall rating of the procedure were included.

TABLE 2. The Symptom Grading System Administered to Patients

Esophagram

The esophagram was performed using both single and double contrast. After a standard upright esophagram was completed, an upright anterior-posterior and lateral chest x-ray was obtained with barium in the distal esophagus. A centimeter ruler was superimposed on the anterior projection and the size of the hiatal hernia was determined by measuring against the ruler from the proximal extent of the gastric tissue to the extrapolated hemi-diaphragm lines at the midline (Fig. 1). A short esophagus was suspected if the gastroesophageal junction (GEJ) was 5 or more centimeters above the hiatus.

FIGURE 1. An upright esophagram shows an incarcerated hiatal hernia with the gastroesophageal junction above the diaphragm. The hemidiaphragm lines are extrapolated to the midline. From that point, the gastroesophageal junction and the proximal extent of the gastric tissue are measured; values a and b. (“a” should not exceed 5 cm and “b” should not exceed 2 cm for the laparoscopic approach).

Upper Endoscopy

Upper endoscopy was performed to determine the presence or absence of cancer, Barrett esophagus by direct inspection and biopsy, a peptic stricture, the size of a hiatal hernia, the presence of food within the stomach, the position of the fundoplication on retroflexion, the length of gastric tissue above the fundoplication, the location of the fundoplication, and the tightness of the fundoplication (also determined by the maximum size dilator that could be safely used). The depth of the fundoplication anterior and posterior tucks and the apposition of tissue to the retroflexed endoscope were used to determine the presence or absence of fundoplication disruption.

Manometry

Manometry was conducted as per standard protocol, but in addition, the fundoplication pressure and relaxation were assessed using 2 to 3 wet swallows at each centimeter level (normal fundoplication pressure is 20–35 mm Hg in our laboratory). Relaxation below 85% at any one level was considered abnormal. Esophageal body dysmotility and wave amplitudes were also noted.

pH Monitoring

Twenty-four-hour pH monitoring was performed selectively on the basis of previous completed positive studies, the response to proton pump inhibitors (PPIs) and presence or absence of esophagitis on endoscopy. A standard probe was used for the majority of patients; however, the BRAVO capsule was used later in the study period.

Gastric Emptying Studies

This study was performed if a previous vagotomy had been performed, there was food within the stomach at upper endoscopy after an appropriate fast, the patient gave a history of vomiting old food, or the patient had intractable vomiting. A half-time of greater than 180 minutes was considered abnormal and an indication for a pyloroplasty.

Mechanism of Failure

A nomenclature for mechanisms postfundoplication failure was developed and is shown in Table 3. The preoperative mechanism was determined by a combination of tests including endoscopy, pH monitoring, manometry, an upright esophagram, and a gastric emptying study, when indicated. The crus closure failures were readily apparent at endoscopy. The stenotic crus closure was difficult to differentiate preoperatively but was suspected if a large dilator could not easily traverse the hiatus and manometry did not show a hypertensive or poorly relaxing fundoplication. The criteria for suspecting a short esophagus were the presence of Barrett esophagus, a peptic stricture, short manometric length, and the GEJ being 5 or more cm above the confluence of the hemidiaphragms on upright esophagram.²⁷ The final determination of esophageal length, however, was at the time of surgery when after esophageal mobilization the GEJ had to rest without tension 3 cm below the arch of the diaphragmatic hiatus if the laparoscopic approach was used, and 2 cm for open surgery.

TABLE 3. Mechanisms of Post-Fundoplication Failure

Indications and Operative Approach

Reoperative fundoplication was advised when the evaluation revealed a surgically correctable disorder that had not or could not respond to aggressive medical management. Patients with an asymptomatic small recurrent hiatal hernia were not offered operative intervention but those with chest pain were operated upon if a cardiac evaluation was negative and the chest pain was repetitive.

An open thoracotomy was performed when more than 2 cm of gastric tissue was within the mediastinum on upright esophagram, a short esophagus was suspected, or the patient had a previous mediastinal dissection. A laparotomy was chosen if the patient had 2 or more prior failed operations. A laparoscopic approach was used if the patient did not meet the above criterion. In patients with a short esophagus, a Collis gastroplasty was performed.

Reoperative Laparoscopic Technique

The undersurface of the lateral segment of the left lobe of the liver was separated from adherent viscera by sharp dissection. The dissection was started posteriorly and caudad where the adhesions were less dense. Care was taken to stay immediately adjacent to the liver capsule and the previous operative note was referenced to determine if an aberrant left hepatic artery was present. The proximal short gastric vessels were ligated in all cases if still present and the pancreatic gastric adhesions/ligament was divided.

After separation of the anterior and posterior wings the fundoplication was dismantled. Final attachments of the right posterior limb to the proximal stomach could be best seen by retracting the anterior fundus to the patient’s right side and using a 45° laparoscope to view the left side of the hiatus. Complete dismantling of the fundoplication, when indicated, was confirmed by intraoperative endoscopy. Sharp dissection was used throughout. Perforations were closed by defining the edges and placing interrupted 3–0 silk sutures, being sure to include the mucosa. Knots were tied intracorporeally. Intraoperative endoscopy was always performed to assure a secure closure and to determine if other perforations were present.

Nissen fundoplication redundancy was tailored to manometric findings. Conversion to a Toupet fundoplication was used if the patient had a too tight Nissen fundoplication and the initial fundoplication was well formed without evidence of twisting or a 2-compartment stomach. A Dor fundoplication was constructed if the patient had no esophageal peristalsis. Crus closure, if necessary, was completed with interrupted O Ethibond nonabsorbable sutures (Ethicon Inc., Sommerville, NJ). Mesh reinforcement was used (n = 4) when crus closure sutures tore muscular tissue or the patient was required to lift heavy objects during employment.

Reoperative Thoracotomy Technique

After exploration, the esophagus was mobilized up to the level of the left mainstem bronchus. Any hernia present was mobilized and the hernia sac entered. If the adhesions on the peritoneal side were difficult to visualize, a diaphragm counterincision was made 2 to 3 cm below the rib cage. The fundoplication was dismantled and a Collis gastroplasty (4–5 cm long) was performed if the skeletonized GEJ could not be easily reduced 2 cm into the peritoneal cavity. A partial fundoplication was used only in patients with very poor esophageal motility. Diaphragm closure was performed with interrupted 0 Ethibond sutures.

Reoperative Laparotomy Technique

The laparotomies were performed through an upper midline incision. Dissection and mobilization were performed as described for the laparoscopic technique although palpation was used liberally to identify the vagus nerves. Dismantling the fundoplication was easier as exposure to the posterior esophagus could be achieved readily. The choice of fundoplication was similar to that used for the laparoscopic approach.

Follow-up

Three patients were lost to follow-up and 4 patients died, 1 of which was a postoperative death. Ninety-seven patients (97% follow-up) were successfully contacted by telephone. The symptom questionnaire (Table 2) was administered to determine symptom recurrence. In addition, medication use, subjective grading of results on a scale of 1 to 10, and overall satisfaction grade (excellent, good, fair, or poor) was determined. The mean length of follow-up was 32 months, and the range was 1 to 146 months. The criteria for failure after reoperation are the following: 1) need for further surgery, 2) daily H2 blocker or PPI for GERD, or 3) grade 3 symptoms.

Data Analysis

An Excel (Microsoft) spreadsheet was used to store data. Symptom results were compared with the mechanism of failure, the type of reoperative procedure performed, the presence or absence of Barrett esophagus, a short esophagus, and complications. To determine the difference in symptom resolution between groups, the χ² with Fisher exact test modification was used. A paired t test was used to compare pre- and postoperative symptoms. A multivariate analysis was conducted to determine risk factors for poor patient outcomes.

RESULTS

A total of 104 patients with a mean age of 50 years (range, 22–80 years), a mean weight of 185 pounds (range, 88–262 lb), a mean BMI of 30 (range, 15–56 kg/m²) underwent reoperative fundoplication at our center during an 11-year period. The baseline patient characteristics are shown in Table 4. The mean duration between the primary procedure and reappearance of symptoms was 22 months (range, 0–120 months) and the mean time elapsed between procedures was 38 months (range, 0–672 months). Ninety patients (87%) had their primary fundoplication performed by laparoscopy, and 14 (13%) by laparotomy. None of the primary fundoplications were performed transthoracically. The most common surgical indication was dysphagia (38.4%). Thirty-eight patients were operated upon for recurrent GERD (37%), 22% chest pain, 1% for a hiatal hernia with a Cameron’s ulcer and anemia, and 1% for an intrathoracic stomach. Surgical indications for those patients with recurrent GERD were positive 24 hour pH monitoring (n = 13) and/or uncontrolled symptoms on PPIs (n = 21), a slipped Nissen with concomitant dysphagia (n = 2), and irreversible esophagitis with symptoms (n = 2). pH testing was not performed if the patient had a good response to PPI initially but required escalating doses or if they had severe esophagitis. Of the 104 patients, 58 were reoperated laparoscopically (56%), 34 by open thoracotomy (33%), and 12 by laparotomy (11%).

TABLE 4. Patient Demographics

Reoperative Surgery

The laparoscopic reoperative procedures were 33 Nissen fundoplications, 22 Toupet fundoplications, 1 Dor fundoplication, 1 Thal fundoplication with Heller myotomy, and 1 Collis Nissen fundoplication. Using the transthoracic approach, we performed 16 Nissen fundoplications, 11 Collis Nissen fundoplications, 3 Collis Belsey fundoplications, 2 Dor fundoplications, and 2 Belsey Mark IV fundoplications. By laparotomy, 8 Nissen fundoplications, 3 Toupet fundoplications, and 1 Collis Nissen fundoplication plus antrectomy and truncal vagotomy were performed. The mean operating time was 4 hours (range, 1–7 hours) and the mean hospital stay for the entire series was 4 days (range, 1–23 days). Hospitalization for laparoscopic surgery was 3 days and 8.4 days for open procedures. The average estimated blood loss during procedures was 280 mL (range, 25–1500 mL). Intraoperative dilation was performed in 83 patients (80%) and the median size of the dilator was 60 Fr. A 48-Fr dilator was used to form the neo-esophagus for patients undergoing a Collis gastroplasty.

Conversion From Laparoscopy to Laparotomy

Five laparoscopic patients were converted to open laparotomy for a conversion rate of 7.9% (5 of 63). The causes of conversion were extensive adhesions in 2 patients, of which one had been operated upon 2 times before, uncontrollable bleeding (n = 2) and inability to form a satisfactory fundoplication due to a funnel-shaped gastric fundus (n = 1). The mean operative time for conversion cases was 5 hours, the mean BMI was 33 kg/m², the mean blood loss was 750 mL, and the mean length of hospitalization was 10 days. One conversion patient developed a postoperative bile fistula secondary to upper arm retractor liver folding. This problem resolved spontaneously.

Perioperative Complications

Intraoperative and early postoperative complications were encountered in 33 patients (32%) and included; pneumothorax (n = 7), intraoperative perforation (n = 11), and a variety of other problems including a subdiaphragmatic abscess (n = 1) probably secondary to an intraoperative perforation that was recognized and sutured. No leak was demonstrated on radiographic studies. All complications are shown in Table 5. One death occurred in a mentally challenged patient who vomited 5 days postoperatively, herniating his stomach into the chest. Intense adhesions were encountered at open reoperative surgery. Postoperatively, the patient developed respiratory failure and after 2 weeks the family withdrew care. Significant intraoperative bleeding was encountered in 3 patients. The sites of bleeding were a posterior short gastric vessel 1300 mL requiring conversion from laparoscopy to laparotomy, a fundoplication division site (750 mL), and a phrenoesophageal artery requiring conversion (500 mL). Eleven patients had intraoperative perforations recognized at the time of surgery; 10 gastric perforations (91%) and one within the neo-esophagus. Complication rates were not significantly different for the 3 operative approaches. The statistical analysis included immediate complications, long-term complications, and perforations alone.

TABLE 5. Complications

Symptom Outcome

Significant symptom improvement and personal satisfaction were seen postoperatively. Symptom resolution (grade 0 or 1) was observed postoperatively in 74% of patients with dysphagia, 75% with heartburn, 85% with regurgitation, and 94% with chest pain. Twenty patients were on PPI therapy when questioned. The reasons for reinstituting medical therapy were not obtained. Table 6 shows comparisons of preoperative and postoperative symptom scores and the overall satisfaction grade and personal grading of the operation. Symptom results in relation to mechanisms of failure and various other parameters are shown in Table 7. Although the differences in symptoms, overall satisfaction, and grading do not reach statistical significance, presumably because of insufficient sample size, a downward trend is apparent with advanced esophageal disease (Barrett esophagus and short esophagus) and its treatment, Collis gastroplasty. Patients undergoing conversion from laparoscopy to laparotomy had a satisfaction mean value of 8 on a scale of 1 to 10 and gave an overall grade of 3.2 on a scale of 0 to 4. Symptom outcomes and personal satisfaction scores in patients undergoing laparoscopy (n = 58) versus open surgery (n = 46) are shown in Tables 6 and 7.

TABLE 6. Symptoms

TABLE 7. Mechanisms of Failure

Adverse Events at Prolonged Follow-up

Twenty-eight patients (27%) encountered complications at prolonged follow-up. These included persistent dysphagia (n = 13), the gas bloat syndrome (n = 5), a disrupted fundoplication (n = 4), intrathoracic fundoplication (n = 2), persistent reflux (n = 2), an incarcerated trocar site hernia (n = 1), and suture erosion within the distal esophagus (n = 1). Twenty-two patients did undergo esophageal dilation after reoperation and 7 patients required additional surgery for correction of continued or recurrent symptoms. Patient failures as previously defined are shown in relation to mechanism of failure in Table 8.

TABLE 8. In Relation to Mechanism of Failure

A total of 3 patients died due to causes unrelated to reoperation. One patient had esophageal cancer and died within 6 months of reoperation. An experienced gastroenterologist endoscoped the patient prior to our operation but did not detect cancer. Subsequently, the operating surgeon has performed 100% of the preoperative endoscopies. Of the remaining patients, 1 died of chronic obstructive pulmonary disease and the other a probable pulmonary embolus secondary to knee surgery.

Onset of Symptoms

Patients with different mechanisms of failure did not significantly differ with regard to timing of symptom recurrence except for the hypertensive fundoplication, which recurred in 11 months with a range of 23 to 37 months. There was an earlier recurrence of symptoms for the laparoscopic group after primary surgery (16 months) as compared with the laparotomy patients (57 months) (P < 0.00001). Patients who experienced recurrence of symptoms after reoperative surgery had earlier recurrences when compared with primary surgery (6 months and 10 months, respectively), but this did not achieve statistical significance (P = 0.65).

Mechanisms of Failure

The underlying mechanisms of failure for primary antireflux surgery are shown in Table 3. The predominant combinations included crus closure failure with partial fundoplication disruption (n = 14), crus closure failure with complete fundoplication disruption (n = 3), crus closure failure with a short esophagus (n = 3), and crus closure failure with partial fundoplication disruption and a short esophagus (n = 4). The accuracy of preoperative mechanism assessment was 78%. Those patients with Barrett esophagus had crus closure disruption (n = 10) 62.5% of the time, and those with severe esophagitis (grade C or D) had crus closure disruption (n = 6) 100% of the time. The accuracy of prediction for each mechanism is shown in Table 9. The short esophagus proved to be most difficult to predict. Hiatal stenosis can only be determined at surgery.

TABLE 9. Predictability of Mechanism of Failure

Predictors of Failure or Success

Various preoperative variables including age, gender BMI, both or either preoperative Barrett esophagus and severe esophagitis (grade III or IV), Collis gastroplasty, laparoscopic primary surgery, open primary surgery, laparoscopic versus open reoperative surgery, and a short esophagus were studied as outcome predictors by multivariate analysis. Barrett esophagus alone and Barrett esophagus with esophagitis were predictive of improved regurgitation scores (P = 0.03; odds ratio, 3.6) and dysphagia control (P = 0.04; odds ratio, 0.3); however, patients after a Collis gastroplasty continued to have significantly greater regurgitation (mean score 0.86 vs. 0.3, P = 0.013).

DISCUSSION

We report a large series of reoperations for failed antireflux surgery and demonstrate that significant and sustained symptom improvement can be achieved. Symptom resolution ranged from 75% for dysphagia to 93% for chest pain at a mean follow-up of 38 months. Laparoscopic revisions are comparable to results attained when operating through an open incision. Patients undergoing a laparoscopic repair of a too tight fundoplication had better results than other subsets of laparoscopic reoperative patients. As expected, results trended downward for patients suffering with more advanced disease. The perioperative complication rate was 33%, including one death. The complication rate of the thoracotomies did not exceed that of patients undergoing an abdominal approach. We were able to predict the mechanism of failure, as found at surgery 78% of the time, and results as related to the mechanisms of failure demonstrated trends that may be important.

These results are encouraging and consistent with previous reports.^23–27 Our series does, however, include patients requiring reoperative surgery for more advanced disease; thus, 16 (15%) patients required an esophageal lengthening procedure. This operation is less than ideal in terms of heartburn and acid regurgitation control as the newly formed neo-esophagus may secrete acid.²⁸ Thirty-four thoracotomies were performed because of advanced disease. The overall complication rate was higher primarily because intraoperative perforations were included, only one of which resulted in postoperative sequelae. Careful attention to operative planes will reduce the perforation rate and that was apparent as our experience increased. The postoperative death due to respiratory insufficiency was probably unavoidable, but the patient that developed metastases from a missed adenocarcinoma demonstrates the need for the surgeon to perform all preoperative endoscopies and that if dissection is more difficult than expected, conversion to an open operation is necessary.

Avoidable technical pitfalls at primary surgery are numerous, but the following are highlighted. Inadequate fundus size; a funnel-shaped fundus is difficult to recognize at laparoscopy. In this circumstance, the temptation is to error by choosing a fundoplication lead point too low on the gastric body creating a 2-compartment deformity. This results in a poorly relaxing fundoplication and unremitting dysphagia. The inadequate fundus can be recognized when calibrating the fundoplication over a 60-Fr bougie and the antidote is thorough mobilization of the gastric fundus. The literature does show this is not routinely required and we agree. However, if one is to become comfortable with short gastric vessel division, routine mobilization is helpful. If inadequate fundic size remains, a Toupet or even a Dor fundoplication is appropriate. Another pitfall is the missed short esophagus that often results in an intrathoracic fundoplication with fundoplication disruption.

The controversy concerning the short esophagus continues, however, all agree that normally the distal esophagus lies within the abdominal cavity. Most agree that the point at which the gastric serosa meets the longitudinal muscle fibers of the esophagus is the GEJ. The left transthoracic approach provides several advantages, including complete skeletonization of the esophagus to the level of the left main stem bronchus, easier preservation of the vagal nerves, fresh dissection planes, and accurate assessment of esophageal length. An excellent view of GEJ after fundoplication dismantling and fat pad resection allows the surgeon to determine exactly where the GEJ lies in relation to the diaphragm. If the GEJ was still proximal to the diaphragm or we were unable to reduce it 2 cm into the abdominal cavity without tension, we then performed a gastroplasty. This anatomic finding was repeatedly seen and has been observed by the senior coauthors in patients at primary surgery as well.

The reoperation failure rate is 20% to 30% usually because of crus closure failure and/or fundoplication disruption. We must ask: is the patient collagen deficient (a skin test is needed), does the patient play a hidden role in their disease (by controlled or uncontrolled vomiting, retching, lifting), or was the operation technically deficient? The later can occur if the fundus size is inadequate, as mentioned at the primary operation, or especially if a short esophagus is not recognized. At reoperation, there is often fibrosis involving the GEJ and the esophageal longitudinal smooth muscle gastric serosa interface may not be easy to distinguish laparoscopically or even by direct vision. If doubt remains conversion to an open procedure is appropriate after which a lengthening procedure may be indicated.

A short esophagus is defined in this series as the GEJ being less than 3 cm below the arch of the crus after laparoscopic esophageal mobilization. A 2 cm measurement was acceptable for open surgery. This may result in crus closure failure, fundoplication disruption, or a slipped repair. This mechanism was most frequently overlooked preoperatively in our series, although we specifically get an upright esophagram to assist in its diagnosis. Further investigation into more accurate predictors of this important entity is needed.

Postoperative pain, the risk of pulmonary complications, inexperience, and lack of thoracic privileges have discouraged surgeons from performing thoracotomies for recurrent benign esophageal disease. Resolution of preoperative chest pain was not compromised after thoracotomy in this series, and no thoracotomy patient required prolonged ventilation. Careful preoperative assessment is necessary to assure that the patient can tolerate this approach and perioperative epidural analgesia was routinely used. On occasion, a diaphragmatic counterincision is necessary if dense abdominal adhesions are encountered. Abdominal surgeons will not find this difficult and closure of the diaphragm is safe and effective. Training and experience with the transthoracic approach are available for general and thoracic surgeons alike. An esophageal fellowship or prolonged preceptorship will make hospital privileges attainable in most circumstances.

Previous attempts have been made to classify fundoplication failure mechanisms. These systems have used numerical designations, anatomic/functional designations, and diagnostic errors in describing specific failures.22–24,29–31 The previous classifications emphasized anatomic malformations, but Skinner³¹ used 5 functional categories in describing the cause of symptoms for fundoplication failures. We attempted to use a mechanism of failure classification but found that most patients had several mechanisms of failure, thus making each patient difficult to categorize. A consensus conference could develop a classification of fundoplication failure, thereby improving communication and eventually prognostication.

The stenotic hiatus is a new phenomenon and has been reported previously by our group.³² It is probably related to cautery dissection of the hiatus and, although rare, is an important new mechanism for patients suffering with dysphagia. It can be easily handled at operation if recognized. The 2-compartment stomach has also been recently reported²⁶ and is an important new mechanism that affected 5 of our patients. We also experienced a too loose fundoplication created at the time of surgery. These patients never experience improvement in their heartburn and regurgitation. To our knowledge, this mechanism, although inherently obvious and expected, has never been reported previously.

Remaining questions include the role of mesh reinforcement, collagen metabolism and its effect on fundoplication failure and recurrent hiatal hernia formation, a better alternative than the Collis gastroplasty, when esophagectomy is appropriate, and the true role of diaphragm stressors. Long-term randomized mesh trials with objective hernia recurrence follow-up are underway. Initial data for mesh reinforcement are encouraging.^33–35 Klinge et al have shown that the collagen I to III ratio is related to ventral hernia recurrence and perhaps this is true for recurrent hiatal hernias as well.³⁶ The role of esophagectomy for patients with advanced benign disease requires further study. Finally, diaphragm stress factors have been analyzed,^37,38 but a more comprehensive series with a validated questionnaire is needed.

CONCLUSION

Reoperative surgery for GERD is safe and effective. The laparoscopic approach appears to be as effective as laparotomy for reoperative surgery, and the percentage of patients requiring a Collis gastroplasty for a short esophagus was high in this series. The transthoracic approach does not result in excessive postoperative morbidities.