Reoperation for failed antireflux operations

J

THoRAc CARDIOVASC SURG

91:511-517,1986

Reoperation for failed antireflux operations Experience with gastroesophageal reflux in patients without prior operations has yielded understanding of pathophysiology, surgical techniques, and results. Less is known about patients with failed antireflux operations. This report of 61 patients undergoing repeat antireflux procedures addresses this issues. Not included are patients with gastroesophageal reflux after ulcer operations or with inappropriate antireflux operations for motility disorders. Group A patients (n = 34) had only one previous operation, Group B (n = 19) had two, and Group C (n = 8) had three or more. Group C had significantly (p < 0.05) more dysphagia and less heartburn than Group A. This observation correlated with findings from manometry, pH testing, and endoscopy, which showed progressively worse esophageal body function and a greater incidence of severe esophagitis and esophageal leak, but less gastroesophageal reflux, in Group ethan B and in Group B compared to A. Operative mortality was 4.9 %. Repeat antireflux operations in the 58 survivors were as foUows: Group A included 25 standard antireflux procedures and seven bowel interpositions, and 75% were transthoracic. Group B included 16 antireflux procedures and one bowel interposition, and 82 % were transthoracic. Group C included four antireflux procedures and three interpositions, and aU were transthoracic. Clinical results were exceUent or good in 85 % in Group A, 66 % in Group B, and only 42 % in Group C (A versus C, p < 0.05). Surgical complications increased from 27% in Group A to 75% in Group C (p < 0.05). Conclusions: (1) Patients with one prior operation and recurrent gastroesophageal reflux are similar to patients with no prior operations. (2) Results of repeat antireflux operations deteriorate with increasing operations because of impaired esophageal function and progressive tissue destruction. Therefore, second reoperations must be definitive and resection and reconstruction with healthy tissue considered. (3) A transthoracic approach is preferable for first reoperations and mandatory after multiple antireflux procedures.

Alex G. Little, M.D., Mark K. Ferguson, M.D., and David B. Skinner, M.D., Chicago, Ill.

Experience with patients with gastroesophageal reflux has yielded understanding of pathophysiology, indications for and techniques of surgical intervention, and expected results of treatment. Gastroesophageal reflux is due to diminished competence of the cardia with resultant reflux of gastric contents into the esophagus.':? Treatment is individualized and based partially on endoscopic findings. In the absence of erosive esophagitis or Barrett's epithelial changes, patients should be treated medically with suppression or neutralization of gastric acid. Patients requiring surgical intervention are those who have persistent symptoms despite adequate medical management or who have or develop complica-

tions such as erosive esophgitis, stricture, or Barrett's epithelial changes. Surgical treatment requires establishment of a competent cardia with one of several techniques that have in common restoration of the cardia to a fixed intra-abdominal location and either a partial or complete fundoplication. Results in these patients are very good, with an 85% to 90% long-term success rate." Much less is known about patients who require reevaluation and treatment after a previous antireflux operation. We reviewed our experience within 61 patients who underwent a repeat antireflux operation at The University of Chicago Medical Center to increase the understanding of these patients.

From the University of Chicago Medical Center, Department of Surgery, Chicago, Ill.

Patient population

Read at the Eleventh Annual Meeting of The Western Thoracic Surgical Association, Incline Village, Nev., June 16-20, 1985. Adddrcss for reprints: Alex G. Little, M.D., The University of Chicago, Department of Surgery (Box 168), 5841 S. Maryland Ave., Chicago, Ill. 60637.

This review encompasses 61 patients operated on by the authors at The University of Chicago Medical Center after previous antireflux procedures. We have not included patients in whom gastroesophageal reflux developed after operations directed at peptic ulcer 511

The Journal of Thoracic and Cardiovascular Surgery

5 1 2 little. Ferguson, Skinner

Table I. Prior operations in patients undergoing reoperation Other

Antireflux

Group A Nissen Abdominal Thoracic Hill Collis (plus)* Allison Resection/bowel interposition Angelchik prosthesis Crural/hiatal hernia repair Unknown Totals

13 12 1

I

Group B

I

Group C 15

14 13 1

0 2 I 1 0 13 4 34

9 6 2 0 0 5 0 12 5 38

Type Vagotomy and pyloroplasty Cholecystectomy Repair perforation/ligation cardia

I

No. 12 6 4

0 2 0 I 2 4 3 27

*Collis gastroplasty plus a Nissen or Belscy operation.

Table II. Symptoms at time of reoperation Heartburn

Regurgitation

Dysphagia

Group A Group B Group C

1.4 ± 0.2 2.2 ± 0.2 1.1 ± 0.4

1.3 ± 0.2 1.9 ± 0.2 1.7 ± 0.4

1.2 ± OJ 1.5 ± 0.2 2.1 ± 0.2

A vs B B vs C A vs C

p
p < 0.05 P < 0.05 p < 0.05

p= NS p= NS p < 0.05

Legend: NS. Not significant.

Table m. Classification ofpatients according to cause of recurrence of symptoms Type V

(%) Group A Group B Group C

11.8 5.3 12.5

disease of the stomach or duodenum, without concomitant antireflux operations, or patients who underwent inappropriate antireflux operations for esophageal motility disorders. All prior operations are called "antireflux" if they were designed by the operating surgeon-as nearly as patient history and review of available operative notes can determine-to correct an incompetent cardia. In fact, many were described as either crural or hiatal hernia repairs. There are a total of 27 male patients and 34 female patients. The age range is from 12 to 78 years with a mean of 51.0 years. Five of the prior antireflux operations were performed at The University of Chicago. For purposes of analysis the patients are divided into three groups based on the number of prior operations. A total of 99 prior operations had been performed in the 61

patients. Group A patients had one previous antireflux operation. There are 34 patients in this group, 17 male and 17 female. The age range is from 12 to 78 years with a mean of 49.5 years. Group B patients had two previous operations and there are 19 in this group, five male and 14 female. The age range is 29 to 65 years with a mean of 49.8 years. Group C patients had three or more previous operations. This group includes eight patients, five male and three female. Age range is from 33 to 76 years with an average of 60 years. Previous operation, differentiated between antireflux procedures and other operations, is listed in Table I. To be emphasized is the number of patients having either a crural or anatomic hiatal hernia repair. In many cases an accurate and reliable description of the previous operations is lacking. For some patients no operative note could be explained and, when available, the operative note frequently either lacked detail, e.g., "a standard Nissen was performed," or was even more generally vague, e.g., "we performed a crural repair." As previously described,' patient symptoms of heartburn, regurgitation, and dysphagia are quantified from zero (minimal or nonexistent) to four (maximal or most severe). As shown in Table II, Group C patients have significantly (p < 0.05) more dysphagia and regurgitation, but less heartburn, than Group A patients. Based on a detailed analysis of symptomatic patients seen in our Thoracic Surgery Esophageal Function Laboratory between January, 1975, and December, 1978, a classification of the cause of symptoms after antireflux operations has been established," This classification schema is based on the following: (1) esophageal manometry, which characterizes the lower esophageal high-pressure zone and esophageal body motility; (2) the standard acid reflux test, which evaluates the competence of the cardia"; (3) the acid clearance test,

Volume 91

5 13

Reoperation for failed antireflux operations

Number 4 April, 1986

Table IV. Type of reoperative antireflux and ancillary procedures and operative approach Other

Antireflux Group A

Nissen Abdominal Thoracic Bclsey Hill Resection/interposition Exploration Operative approach Abdominal Thoracic

I

Group B 0 6

5 14 2 7

I 8 (24'}(,) 26 (76%)

2 0 2

7 3 2

3 (16%) 16 (84%)

which measures the ability of the esophagus to clear an instilled acid load'; (4) esophageal pH monitoring, which quantitates acid (pH < 4) and alkaline (pH> 7) reflux and from which the esophageal clearing efficiency can be evaluated by the length of reflux episodes'; and (5) barium radiographic studies of the upper gastrointestinal tract and examination with a flexible endoscope, which define the anatomy and the presence and severity of esophagitis. The causes of symptoms can be separated into the following five categories: type I-loss of competence of the cardia or sphincter mechanism with acid gastroesophageal reflux; type IIfailure of esophageal clearance of physiologic acid reflux episodes; type III-loss of competence of the cardia and failure of esophageal body clearance, i.e. a combined abnormality; type IV-the presence of alkaline gastroesophageal reflux; type V-absence of reflux, the symptoms being due to problems of esophageal or wound healing, disorders of another organ system, or esophageal motility disorders. Our patients, as shown in Table III, have been classified accordingly. Most patients have a type III or combined defect; however, 70.6% (type I plus type III) of Group A patients have gastroesophageal reflux compared to 50% of Group C patients. Clearance failure (type II plus type III) is more common in Group C (75%) than either Group A (61.8%) or Group B (57.9%) patients. Separate from this classification schema, no patients in Group A had actual disruption of or leakage from the esophagus or site of previous gastrointestinal operation, one patient (5.3%) in Group B had a persisting esophageal leak, and three (37.5%) patients in Group Chad leaks continuing from their prior operation. Similarly, six (75%) Group C patients had endoscopic esophagitis compared to nine (26%) of the Group A patients and nine (47%) of the Group B patients. The indications for reoperation varied among the three groups. In Group A the major indication was

Type

Group C

6

10 5

\

I 0 5 0

DIlation Vagotomy Pyloroplasty Cholecystectomy Antrcctorny

Roux-en-Y Remove Angelchik prosthesis

I

No.

5 3 3

I 3 2 2

0 8 (100%)

uncontrollable symptoms, particularly heartburn, in 75% of the patients. Most had been treated medically in the interval since their initial operation but to no avail. Considering they all had a mechanically incompetent cardia, this does not seem surprising. In contrast, in Group C patients the major reason for reoperation was esophagitis or actual esophagogastric discontinuity. The reoperative antireflux procedures performed at The University of Chicago in the 61 patients are shown in Table IV. Findings at operation were diverse. In most patients there was evidence that a standard antireflux operation had been done, or at least attempted. The most frequent determinable cause of failure seemed to be disruption of the original repair, probably caused by sutures pulling out of the esophagus. In others it was evident the hiatus had never been dissected although there were a few, seemingly random, stitches in the crura. Two of our patients had exploration only. In one patient in Group A a contained perforation extending into the right paravertebral gutter was unexpectedly found. Resection was impossible through the left thoracotomy and the patient was unprepared for this possibility. One patient in Group B died intraoperatively of aortic hemorrhage. Unusually intense periesophagitis had fused the esophageal mucosa at the site of a previous myotomy into a single tissue plane with the aorta. At autopsy an occult aortic dissection was found. Most operations were performed via a thoracic approach, though in 24% of Group A patients and 16% of Group B an abdominal route was chosen. The abdominal approach was selected only when a previous thoracic operation had not been done and there was important other intra-abdominal disease, such as gallstones or duodenal ulceration, necessitating an additional abdominal procedure. The best approach for exposure of the cardia was through the chest, particularly after

The Journal of

5 14

little, Ferguson, Skinner

Thoracic and Cardiovascular Surgery

Table V. Morbidity and mortality of reoperation Croup A

Croup B

I

0

0 0

0

0

1

0 2 2 0 I 5 (26(,7,)

Anastomotic leak Perforated diverticulum in colon interposition segment Recurrent leak after emergency resection Loss of interposed bowel Infection Iatrogenic trauma Incomplcte operation Deaths Totals

2

3 I I 9 (27%)

Croup C

0 2

0 0 J

6 (75%)

Table VI. Results of reoperation Croup A

No. Excellent

Good Fair Poor

1

17 10 2

2

Croup B

%

No.

54 31 7 7

6 3

6

3

1

Croup C

%

No.

33 33

2

17 17

1%

J

o 4

multiple operations. Because exposure was better, dissection was safer and more controlled. In Group A patients dissection could usually be done through the hiatus, but in the others the diaphragm was incised peripherally to allow more extensive access to the left upper quadrant. Results of reoperation Reoperation is associated with considerable surgical morbidity, as shown in Table V. The iatrogenic trauma listed in the table includes injuries to the spleen necessitating splenectomy, to the hepatic vein necessitating ligation, and to the left inferior pulmonary vein, also necessitating ligation. There were two anastomotic leaks. In addition, in one patient a diverticulum perforated in the middle of the colonic interposition segment 6 days after the operation, and in another the esophageal stump began leaking after an emergency resection for leakage following a Collis-Nissen operation done elsewhere. Because the latter patient was in septic shock, the esophagus, with its stump decompressed by a sump nasal tube, had to be left in the mediastinum. Her condition never improved and the esophageal stump eventually gave way. The increase in operative morbidity from Group A to Group C is statistically significant (p < 0.05). Overall operative mortality consists of three patients, a 4.9% mortality: One patient died of exsangui-

nation in the operating room from entry into an occult aortic dissection; one patient died 3 months postoperatively of thoracic sepsis and cardiac dysrhythmias after transfer to another hospital; the third patient died 3 months postoperatively of septic sequelae and a recurrent leak following emergency resection because of perforation after a Collis-Nissen procedure done at another hospital. Follow-up is complete in 53 or 90.4% of the 58 patients who survived the operation. Follow-up consists of a combination of clinic visits or direct telephone contact, or both, with either the patient or the family. Median length of follow-up is 3.8 years with a range of 6 months to 10 years. Results of reoperation are shown in Table VI. Symptoms are classified as follows. Patients without any heartburn, regurgitation, or dysphagia, who take no medications, and who required no further treatment after our operation are classified as having excellent symptomatic results. Patients with mild symptoms without medications or with no symptoms while taking medications are considered to have good results. Patients classified as having a fair symptomatic result are those with mild symptoms while taking medications and those who required occasional postoperative dilations but no further operation. Finally, patients are considered to have a poor symptomatic outcome if symptoms were not improved from the preoperative status or if a further operation was required. Eighty-five percent of Group A patients have excellent or good results, compared to only 42% of Group C patients. Discussion Limited information is available about patients requiring reoperation because of an unsuccessful antireflux operation. It can be concluded, based on previous reports, that reoperative antireflux procedures can be performed with a reasonable morbidity and mortality, after one prior operation, although both morbidity and mortality are increased from primary operations. Reported reoperative 30 day hospital mortalities are 3.3% and 0% in two relatively large experiences.v? although one report of a much smaller number of patients listed a mortality of 17.6%.10 Although the morbidity and motality rates for our reoperations are reasonable, they are much higher than for first-time operations, which should carry a mortality of less than 1% and be associated with little morbidity. Antireflux operations are not frequently performed but are common enough that the problem of failure of the primary operation must be addressed. Our experience suggests that these patients must be thought of according to the number of prior operations. Symptom-

Volume 91 Number 4 April, 1986

atic patients requiring an operation after only one prior operation are similar to unoperated patients. Although some, about 25% in our experience, have erosive esophagitis necessitating their reoperation, in the majority it is uncontrollable symptoms that lead to reintervention. Most patients with an incompetent cardia after an unsuccessful operation have such a severe mechanical defect that medical therapy will fail. The reason for the failure of the first operation in these patients is not always clear. It is impossible to attribute recurrence to the technique of anyone of the standard antireflux procedures used in this country. Without knowing the denominator, which should include all procedures, one cannot know the failure rate. Although most of our patients had prior Nissen fundoplications, as this is the most common antireflux procedure it should have the greatest number of recurrences or failures. It is also difficult to address the issue of patient selection for the initial operation. It is not clear whether the primary operation failed to correct reflux or was simply an inappropriate operation that created an iatrogenically incompetent cardia where one had not existed previously. In fact, the operations in most of our patients seemed to have been done for suspected reflux and not as procedures ancillary to other abdominal operations. It is dismaying to note the number of patients in whom either a clearly inappropriate anatomic repair of a hiatal hernia or a crural repair was performed. In these patients failure can be attributed to a lack of understanding of or familiarity with the principles and techniques of antireflux surgery. In most cases the esophagus can be saved at the first reoperation and a standard antireflux procedure performed. This is particularly true when the thoracic route is used so that the esophagus can be extensively mobilized, which permits a tension-free reduction of the repair into the abdomen. This first reoperation morbidity is reasonably low, 3% in our patients. Long-term results are good. Eighty-five percent of our patients had excellent or good symptomatic results. This is in accord with previous reports of an 80% or greater incidence of good or satisfactory long-term clinical results.":" After multiple prior operations, the picture changes. Patients have more damage to the esophagus, manifested by abnormal function, a greater incidence of esophagitis, and a greater incidence of actual esophagogastric discontinuity. The problem is no longer "simple" gastroesophageal reflux but now includes anatomic and functional esophageal impairment. Not surprisingly, surgical results deteriorate. There is more surgical mortality and morbidity and long-term results are considerably worse. It is tempting to try to retain the patient's native

Reoperation for failed antireflux operations

5 15

esophagus and stomach, but our experience suggests that, in this group with multiple prior operations, the third operation, i.e., the second reoperation, must be definitive and resection and replacement with healthy tissue should be strongly considered. The stomach should not be advanced into the chest and used to replace a resected esophagus because of reflux complications. The choices are colon "Or jejunum. Polk! I has reported excellent results using jejunal interposition. Our choice has been an isoperistaltic segment of left colon, which we prefer because of its reliable blood supply, because preparation of the pedicle is technically easier than for the jejunum, and because a greater length can be developed if necessary. Both colon and jejunum serve as satisfactory functional esophageal replacements and neither are prone to reflux complications. The choice should be guided by availability and the surgeon's experience. Although difficult to document, our experience suggests that reoperations are best done through a thoracic approach. After only one prior operation, and in patients with other abdominal problems necessitating surgical attention, the abdominal approach can be chosen. The thoracic approach, however, particularly in patients with multiple previous operative interventions in the left upper quadrant, provides maximum exposure of the hiatus and, with peripheral detachment of the diaphragm, excellent exposure of the left upper quadrant. The esophagus can be fully mobilized to the aortic arch, which establishes the optimal situation for salvage of native esophagus, if possible. If resection and interposition are necessary, the left colon can be easily prepared through this approach and all anastomoses, particularly the esophagocolonic, performed with optimum exposure. Another theoretical option is that of duodenal diversion for treatment of recurrent reflux esophagitis." A few of our patients had peptic ulcer disease which necessitated measures such as vagotomy, pyloroplasty, and antrectomy, some with Roux-en-Y diversions. We do not recommend duodenal diversion alone with the principal indication for operation is reflux symptoms or complications. If the cardia is incompetent, reflux of gastric contents will continue, and even if the contents are so benign that chances of recurrent esophagitis are minimized, the risk of regurgitation and aspiration remains. In addition, the side effects of duodenal diversion are not insignificant. Concluding, our experience shows that first antireflux reoperations can provide satisfactory results with acceptable morbidity and mortality. Results deteriorate with increasing operations because of impaired esophageal

5 1 6 Little, Ferguson, Skinner

function and progressive tissue destruction. Second reoperations must be definitive, and resection and reconstruction must be considered because of poor results of subsequent operations. We believe a transthoracic operative approach is preferable for first reoperations and mandatory after multiple operations. REFERENCES

2

3

4

5

6

7

8 9

10 II

12

Little AG, DeMeester TR, Kirchner PT, O'Sullivan GC, Skinner DB: Pathogenesis of esophagitis in patients with gastroesophageal reflux. Surgery 88: I01-107, 1980 Dodds WJ, Dent J, Hogan WJ, Helm JF, Hauser R, Patel GK, Egide MS: Mechanisms of gastroesophageal reflux in patients with reflux esophagitis. N Engl J Med 307:15471552, 1982 Ellis FH, Crozier RE: Reflux control by fundoplication. A clinical and manometric assessment of the Nissen operation. Ann Thorac Surg 38:387-392, 1984 Skinner DB, Belsey RH: Surgical management of esophageal reflux and hiatus hernia. J THORAC CARDIOVASC SURG 53:33-54, 1967 Russell CO, Hill LD: Gastroesophageal reflux, Current Problems in Surgery, Vol 20, No.4, Chicago, 1983, Year Book Medical Publishers, Inc. Skinner DB, Klewmentschitsch P, Little AG, DeMeester TR, Belsey RH: Assessment of failed antireflux repairs, Esophageal Disorders. Pathophysiology and Therapy, TR DeMeester, DB Skinner, eds., New York, 1985, Raven Press, pp 303-313. Skinner DB, DeMeester TR: Gastroesophageal reflux, Current Problems in Surgery, Vol 23, No. I, Chicago, 1976, Year Book Medical Publishers, Inc. Hill LD: Management of recurrent hiatal hernia. Arch Surg 102:296-302, 1971 Maher JW, Hocking MP, Woodward ER: Reoperations for esophagitis following failed antireflux procedures. Ann Surg 201:723-727,1985 Zucker K, Peskin GW, Saik RP: Recurrent hiatal hernia repair. Arch Surg 117:413-414, 1982 Polk HC: Jejunal interposition for reflux esophagitis and esophageal stricuture unresponsive to valvuloplasty. World J Surg 4:731-736, 1980 Herrington JL, Mody B: Total duodenal diversion for treatment of reflux esophagitis uncontrolled by repeated anti reflux procedures. Ann Surg 183:636-644, 1976

Discussion DR. ROBERT W. JAMPLIS Palo Alto, Calif

I believe that the authors have proven and quantified what we and they predicted and expected-namely, the more operations, the less heartburn and the more dysphagia, because the more operations, the tighter the repair is going to get. This proves the old axiom: Better do it right the first time. This means that the surgeon must (I) establish an intra-

The Journalof Thoracic and Cardiovascular Surgery

abdominal esophagus, (2) recreate the angle of His, and (3) sew up the crura, as hiatal hernia is the cause of 95% of cases of reflux esophagitis. From what we have heard from Dr. Little, and know ourselves, the most popular operation is the Nissen II, not because it is necessarily the best, but because it is easier to do. Even though Dr. Hill has shown how to find the median arcuate ligament and preaortic fascia, most general surgeons cannot find them, and so they do not do it right. I believethere are several reasons for this: I. The short gastrics are not completely divided. As a result, the fundus is not mobilized completely. 2. The sutures are not placed correctly through the stomach and esophagus. 3. The sutures are not snugged tightly enough around at least a 32F dilator inside the esophagus. Just as in parathyroid surgery, where one should err in taking out too much parathyroid tissue rather than too little, because the former condition can be treated with calcium and vitamin 0 whereas the latter requires reoperation, so here one should err on the side of making the repair tight, as it can always be dilated. However, if the repair is too loose, another operation will be necessary. This brings me to the subject of dilation. We used to rely on getting a swallowed string down the esophagus, or, rarely, doing a gastrotomy and retrograde dilatation using Tucker dilators. Now, my colleague, Dr. Walter Cannon, and I have been using a blunt flexible wire put down directly through the stricture via the mouth or via an esophagoscope or a gastroscope. This has been very safe and effective in our hands, and we dilate with Plummer-Vinson dilators over the wire. I agree with Mercer's and Hill's report (J THORAC CARDIOVAse SL:RG 91:371-378, 1986) that rarely are resective operations necessary, because usually antireflux operations and aggressive dilations are adequate. However, I would agree with the authors that, if resection is necessary, the jejunum and/or the colon should be used. I prefer the stomach when dealing with cancer, but only if it is completely brought up into the chest, because bringing up only a portion of the stomach is asking for esophagitis. I also agree with the authors that the transthoracic approach should be used in any reoperation, for the same reason that a traumatic diaphragmatic hernia should be approached transthoracically. Visibility is much better and therefore the adhesions can be dissected more safely, particularly around the aorta. I would like to ask Dr. Little two questions. I. Of the slipped Nissen failures, were there any type III failures as described by Salk and Peskin of San Diego, wherein only the top sutures gave way leading to an anteriorly produced gastric pouch which caused obstruction? 2. If I have the correct figures, your reoperations included 18 Nissen lIs, 22 Belsey Mark IVs, 5 Hills, and 14 interpositions. The interpositions you did for dysphagia and because of poor tissues, but what made you choose among the Nissen, Belsey, and Hill procedures?

Volume 91 Number 4 April, 1986

DR. LUCIUS D. HILL

Reoperation for failed antireflux operations

5 17

Seattle, Wash.

ed to normal. So, I repeat, it is not the number of operations, but the devastating complications that we see.

Over the past two decades our group has operated on 263 patients for failed antireflux operations. The majority of the initial operations were failed Nissen procedures. I was interested to see that the majority of the failed reoperations the authors have had experience with were Nissen procedures. Our conclusions differ from the authors' is that it is not necessarily the number of opreations the patient has had, but the kind of operation and the kind of devastating complications that develop after the Nissen procedure. The common complications are the slipped Nissen, the patulous Nissen, complete disruption, and so on. The kind of complications that have been the greatest challenge to us are devastating complications that appear to be peculiar to the Nissen operation. These include a gastrobronchial fistula into the left lung from a transthoracic Nissen with a gastric ulcer and a gastrobronchial fistula into the right lung from a transabdominal Nissen with an ulcer that bored through the diaphragm into the right lung. Although we are able to isolate these fistulas and take them down, these patients are far more of a challenge to us than the patient who may have been operated on three or four times but in whom the anatomy is not so distorted. A gastroaortic fistula developed in a patient with a failed Nissen procedure and a large gastric ulcer posteriorly. Dr. James placed a clamp on the aorta, sewed up the hole in the arota, and closed the gastric ulcer, anchoring the gastroesophageal junction posteriorly. The patient survived. To our knowledge, this is the first successful repair of a gastroaortic fistula. More recently, we encountered another fascinating complication. A patient with a slipped Nissen had a large gastric ulcer that bored a hole through the diaphragm and the pericardium, so that the base of the ulcer was the myocardium. This patient was in the coronary care unit being treated for a myocardial infarction when it was realized that he had had a Nissen procedure. Dr. Traverso was able to take this down, close the gastric ulcer, anchor the gastroesophageal junction posteriorly, and save this patient. This is the first successful repair of a gastropericardial fistula that we are aware of. This was a case of true heartburn! Heartburn is one of the worst misnomers in medicine because all other types of heartburn have absolutely nothing to do with the heart. This patient is alive and well and his abnormal electrocardiogram has revert-

DR. LITTLE (Closing) I appreciate Dr. Jamplis and Dr. Hill sharing their thoughts. To use Dr. Hill's analogy, the problem is not with the automobile, but with the driver's license. Dr. Jamplis emphasized the importance of doing it right the first time. It is clear that many of these patients were in initially treated by a surgeon who was not familiar enough with the operation he was attempting. Therefore, I would emphasize more the driver and the driver's license than the technique itself. I agree that there are three valid operations in that area available to the surgeon. All three can work if the surgeon has experience with them, but I would decry as much as I can the occasional surgeon dissecting the hiatus or, even more, the concept of placing an Angelchik prosthesis around the cardia. I think the interesting complications of that device have only begun to surface. In reference to Dr. Jamplis' question about how a Nissen fundoplication, for example, fails, we were usually unable to decipher that during the operation. In many patients there was really no concrete evidence that a fundoplication had ever been performed, which is to say there were lots of stitches, but no particular way that the fundus appeared to be wrapped around or attached to the esophagus. There may have been a few random stitches in the crura. Again, I think it is the surgeon and the surgeon's experience that are the most important factors, and not so much the actual selection of operation. The choice of reoperation is an important issue. We try to be selective and to base our choice largely on the findings at operation. We use the preoperative esophageal function tests as much as we can. When the patient has an intact stomach and relatively normal esophageal function, as shown either by motility or some of the radionuclide swallows, either a Nissen fundoplication or a HeiseyMark IV gives good results. If there is an intact stomach and esophageal function is not normal, our choice is frequently a Heisey repair, because that provides a lower pressure sphincter as measured manometrically. When there has been loss of stomach, either because of a previous operation or because we are doing an antrectomy for peptic ulcer disease, our choice has been a Hill procedure.