Gastrointestinal function following esophagectomy for malignancy

SCIENTIFIC

PAPERS

Gastrointestinal Esophagectomy

Function Following for Malignancy

Richard J. Finley, MD, Andrk Lamy, MD, Joanne Clifton, BA, Kenneth G. Evans, MD, Guy Fradet, MD, Bill Nelems, MD, Vancouver, ~unadu

BACKGROUND: The frequency and causes of gastrointestinal complications following esophagectomy for malignancy are unknown. PATIENTS AND METHODS: We reviewed 295 esophagectomies performed for malignancy hetween January 1980 and September 1994 in order to determine the frequency and causes of early and late gastrointestinal complications. RESULTS: Compared to transhiatal and left thoracoabdominal esophagectomies, esophagectomies carried out through a right posterolaterd thoracotomy with cervical esophagogastric anastomosis had a higher incidence of delayed gastric emptying (ll%), pneumonia (26%), and hospital death (9%). The same operation had a higher incidence of gastroesophageal reflux (20%) and dysphagia requking esophageal dilatation (53%). We found no independent effect of gastric drainage procedures, feeding, jejunostomy, preoperative radiotherapy, pathology, or age on these outcomes. Women had no operative mortality, but a higher incidence of gastroesophageal reflex and diarrhea following esophagectomy. CONCLUSIONS: SurgicaI techniques aimed at improving gastric emptying following esophagectomy for cancer should improve operative morbidity and mortality.

useof a gastricdrainageprocedureor feedingjejunostomy in the developmentof abnormalgastrointestinalsymptoms. PATIENTS

AND

METHODS

BetweenJanuary 1980and September1994,295 patients underwent esophagogastrectomyfor carcinoma of the esophagusor gastric cardia at Vancouver Hospital. They were 234 menand 61 women with a meanageof 64 years (range 16 to 86). All patientswho had squamouscell carcinoma of the cervical esophagusor adenocarcinomaarising from the fundus or the distal two thirds of the stomach were excluded. Patientswith preoperativeevidence of distant metastasis,poor cardiopulmonary reserve, or local invasion of the airway or great vesselsunderwent palliative treatmentother than resectionandwere alsoexcluded. The study patients presented with dysphagia (84%), weight loss>5% of their ideal body weight (68%), heartbum (24%), and regurgitation (21%). They underwent three different types of esophagealresection. Group I consistedof 219 patientswho underwenta transhiatal esophagogastrectomy with cervical esophagogastric anastomosisvia a narrow gastric tube to the left neck. Seven of thesepatientsdied in hospital. Group II consistedof 54 patientswho had an esophagectomy through a right posterolateral thoracotomy. The stomachwas mobilized through a laparotomy and anastomosedto the cervical esophagusthrough a left neck incision. Five of thesepatientsdied in hospital. arcinoma of the gastric cardia and esophagusis one Group III consistedof 22 patients who underwent a left with anastomosis of the most difficult diseases to cure or palliate. As the thoracoabdominalesophagogastrectomy mediansurvival for thesepatientsis lessthan 1 year, treat- of the distal stomachto the esophagusin the left chest. ment should be short, safe, cost effective, and directed at None of thesepatients died in hospital. restoring normal swallowing and gastrointestinalfunction. Pyloromyotomy (230) or pyloroplasty (19) was carried Although radiation therapy, chemotherapy, laser therapy, out in 249 patients.A feedingjejunostomy wasconstructed and intubation have beenusedto treat localized esophageal in 228 patients using a 12-Fr red rubber catheter inserted into the first loop of the jejunum and sewn to the abdomcancer, surgicalresectionis still the standardtreatment. Symptoms of aspiration, dysphagia,early satiety, heart- inal wall. Ninety-seven patients received preoperative inburn, dumping, and diarrheahave been describedfollow- traluminal brachytherapy (1,500 cGy over 30 minutes)and ing esophagectomy.‘*2 The purposesof our study were (1) external beamradiotherapy (4,000 cGy over 4 weeks). to examine the frequency of early and late gastrointestinal Patients were treated with intravenous fluids and nasocomplicationsfollowing esophagectomy,and (2) to deter- gastric drainage for 1 week after the operation. Patients mine the roles of the type of esophagealresection,site of with feeding jejunostomieswere started on full-strength esophagogastricanastomosis,age, sex, pathology, and the tube feeding at a rate of 25 mL/h on the first postoperative day. This rate was increased25 mL/h per day unless the patient developeddiarrheaor abdominaldistention.On From the Department of Surgery, Vancouver Hospital and Health postoperativeday 7, contrastradiography of the upper gasSciences Centre, and University of British Columbia, Vancouver, British trointestinal tract was obtained to examine gastric emptyColumbia, Canada. Requests for reprints should be addressed to Dr. R. J. Finley, UBC ing and the integrity of the esophagogastricanastomosis. Department of Surgery, Room 3100, 910 West 10th Avenue, Laurel A radiological anastomoticleak was defined as any raStreet Pavilion, VHHSC, Vancouver, British Columbia V5Z 4E3, diological defect at the anastomotic site. A clinical leak Canada. was defined asextravasation of saliva or swallowed fluid Presented at the Slst Annual Meeting of the North Pacific Surgical out the drain site or cervical incision. If an anastomotic Association, Coeur d’ Alene, Idaho, November 10-l 1, 1994.

C

THE AMERICAN

JOURNAL

OF SURGERYa

VOLUME

169

MAY

1995

471

GI ___ FUNCTION ...~

FOLLOWING ._--___-.

ESOPHAGECTOMY~INLEY _.- ..-. -. ET AL

_...-

TABLE I Operative Death

__- .-. Type of resection _THE [n = 219) 7 (3) -rTH (n = 54) 5 (91 TA (n = 22) 0 (0) Site of anastomosis Cervical (n = 273) 12 (4) Thoracic (n = 221 0 (0) Gastric drainage Yes [n = 2491 11 (4) No (n = 46) 1 (2) Jejunostomy Yes (n = 228) 9 (4) No (n = 69) 3 (4) Radiotherapy Yes (n = 97) 4 (4) No (n = 2981 8 (41 Pathology Adeno Cardio (n = 117) 2 (2) Squamous Esoph (n = 1011 6 (6) Adeno Esoph (n = 77) 4 (5) Age r65years fn = 144) 8 (51 ~65 years [n = 1511 4 (3) Sex Male (n = 234) 12 (5)’ Female (n = 61) 0 (0) Numbers in parentheses represent percentages.

Early Postoperative Complications Delayed Gastric Anastomotic Leak Emptying

Early Satiety

Pneumonia 30 (141 14 1261 3 (14)

35 (161 12 (221 3 (14)

9 (41 6 (11) 1 (51

7 13) 3 16) 1 (5)

47 (17) 3 114)

15 (61 1 (5)

10 (4) 1 (5)

44 (16) 3 (14)

40 (161 10 [22)

16 (61

10 (41

42 (171 5 (11)

36 (16) 14 (20)

13 (61 3 (4)

10 (4) 1 (1)

35 (15) 13 119)

14 (141 36 (18)

4 (41 12 161

4 (41 7 (31

11 (11) 37 (181

16 (14) 17 (17) 16 (21)

4 (3) 8 (81 4 15)

5 (4) 3 (3) 3 (4)

13 (11) 23 (22) 11 (14)

23 061 27 (181

6 f41 10 (71

6 (4) 5 (3)

22 (15) 25 (16)

40 (17) IO (16)

13 (61 3 (51

8 (3) 3 (5)

40 (17) 7 (11)

0 (01

1 (21

‘P <0.05. THE = transhlatal esophagogastrectomy; llH = transthoracic resection; TA = thoracoabdominai esophagogastrectomy; Adeno Cardio = adenocarcinoma of the gastric cardta; Squamous Esoph = squamous cell carcinoma of the lower esophagus: Adeno Esoph = adenocarcinoma of the esophagus.

leak was observed,jejunostomy tube or parenteral nutrition was continued until the anastomotic leak healed. Delayed gastric emptying was defined as the retention of barium in the stomachfor more than 15 minutesin the upright position after a barium swallow. If there was no evidence of anastomotic leak or delayed gastric emptying, the patient was given a progressive iso-osmolaroral diet until discharge. Pathological findings included adenocarcinomaof the gastric cardia (117), squamouscell carcinomaof the lower esophagus(lOl), and adenocarcinomaof the esophagus (77). Surgical complications and causesof death were analyzed at the end of the hospital stay. At 3-month follow-up, patients were asked if they had experienced cervical dysphagia, symptoms of gastroesophagealreflux, diarrhea (more than two bowel movements per day or explosive diarrhea), dumping syndrome (postprandiallightheadedness or diarrhea),and hoarseness. Patientswho had developed dysphagiafor solid foods underwent esophagealdilatation.

hoarseness)and different covariates (type of resection, cervical or intrath~racic anastomosis, gastric drainageprocedure, jejunostomy, radiotherapy, pathology, age 1~6.5, 265 years], and gender}. Statistical significance was defined as P ~0.05 for all analyses.

RESULTS Twelve of the 295 patients(4%) died in hospitalas a result of their surgery. Seven patientswho underwent transhiatal esophagectomydied from pulmonary embolism(2), myocardial infarction (2), pneumonia (2), and bleeding from residual tumor (1). Five of 54 patients who were treated with ~ans~oracic resection died from aspiration pneumonia(4) and cerebrovascularaccident (1). None of the 22 patients undergoing thoracoabdominal esophagogastrectomydied. Hospital death correlated significantly with pneumonia (P = 0.001) andpulmonary embolism(P = 0.002). Patients who unde~ent esopha~~tomy by right posterolateralthoracotomy had higher incidencesof pneumoniaanddelayed gastric emptying (Table I). Pneumoniawas significantly STATISTICAL, ANALYSIS correlated with delayed gastric emptying (P = 0.02) but Chi-square and Fisher’s exact testswere used in 2 X 2 not hoarseness (P = 0.56). However, delayed gastric emptables. Logistic regressionwas used to evaluate relation- tying, anastomoticleak, early satiety, andpneumoniawere shipsbetween outcome variables (hospital death, anasto- not s~~ni~c~tly associatedwith the site of ~astomosis, motic leak, delayed.gastric emptying, early satiety, pneu- gastric drainage procedure, jejunostomy, radiotherapy, monia, dysphagia, reflux, diarrhea, dumping, and pathology, age, or sex (Table I). 472

THE AMERICAN

JOURNAL

OF SURGERY”

VOLUME

169

MAY

1995

GI UNCTION

FOLLOWING

~OP~G~CTO~~INLEY

ET AL

TABLE II Three-Month Dysphagia

Follow-up of 283 Surviving Patients Reflux Dumping

Diarrhea

Hoarseness

Type of resection THE (n = 212) 69 (33) 13 (6) 12 (61 37 (17) 9 (41 TTH (n = 49) 26 (53) 10 (20)’ 3 (61 11 (221 4 (81 TA (n = 22) 6 (27) 2 (10) 0 (0) 3 (13) 0 (01 Site of anastomosis Cervical (n = 261) 95 (36) 23 (91 15 (61 48 1181 13 15) Thoracic (n = 221 6 (27) 2 I101 0 (01 3 (14) 0 (0) Gastric drainage Yes (n = 238) 89 (37) 21 (91 13 (5) 44 (18) 11 (5) No (n = 45) 13 (29) 4 (91 2 (4) 7 (16) 2 (4) Jejunostomy Yes b-r = 218) 82 (37) 21 (10) 12 (51 39 (18) 8 (41 No fn = 651 20 (30) 4 (61 4 (6) 13 (20) 5 (81 Radiotherapy Yes (n = 92) 40 (43) 11 (12) 3 (3) 20 (22) 6 (6) No fn = 1911 62 (32) 14 (7) 13 (7) 32 (17) 7 (41 Pathology Adeno Cardia in = 114) 34 (30) 5 (4) 5 (41 18 (16) 4 (31 Adeno Esoph (n = 73) 27 (37) 7 (10) 4 (51 13 (18) 2 (3) Squamous Esoph [n = 96) 40 (41) 13 (131 6 (6) 20 (21) 7 (71 Age 265 years (n = 137) 45 (33) 12 (9) 6 (4) 29 (21) 5 (4) 465 years (n = 146) 56 (38) 13 (9) 9 (6) 22 (15) 8 (5) Sex Male (n = 223) 76 (34) 14 (6) 10 (4) 32 (14) 10 (4) Female (n = 601 25 (40) 11 (18)’ 5 (8) 19 (31)’ 3 (5) Numbers in parentheses represent percentages. ‘P <0.05. THE = transhiatal esophagogastrectomy; TrH = transthoracic resection; TA = thoracoabdominal esophagogasbectomy; Adeno Cardio = adenocarcinoma of the gastric cardia; Squamous Esoph = squamous cell carcinoma of the lower esophagus; Adeno Esoph = adenocarcinoma of the esophagus.

t Two of 228 patients(1%) with a feedingjejunostomy required reoperationfor bowel obstruction at the site of the jejunostomy. Three patients who underwent transhiatal esophagectomydeveloped hemiation of the small bowel through the hiatus requiring operative closureof the defect. All 283 patients who survived surgery were available for 3-month follow-up. Of 101 patients (36%) who required esophagealdilatation, the highestincidence was in 53 patients (53%) undergoing esophagectomythrough a right posterolateral thoracotomy (Table JJ). The same group alsohad a higher incidence of gastroesophageal reflux symptoms (P = 0.05) There were no significant effects of anastomoticsite, gastric drainageprocedure,fecding jejunostomy, preoperative radiotherapy, pathology, or age on the frequency of gastrointestinal symptoms at 3 months. Although no women died as a result of esophagectomy, gastroesophagealreflux and diarrhea were more common among women (P = 0.01, P = 0.005, respectively). Only 4% of the patients alive at 1 year required esophagealdilatation.

Other authorshave identified aspirationpneumo~a as a major sourceof morbidity andmortality.‘” Decreasedpharyngeal contraction, cricopharyngeal discoordination,vocal cord paralysis, anastomoticstricture, and reflux from the gastric interposition may facilitate aspirationpneumonia. Our study, and others, do not support this interpretation. Heitmiller and Jones3observed one or more new oropharyngealswallowing abnormalities1 week after transhiatalesophagectomyin 67% of patients,Laryngeal penetration, occurring asa consequenceof diminishedairway protection with incomplete laryngeal elevation, was the most common. Abnormal swallowing resolved or improved in the first postoperativemonth in all Heit~ller’s patients. None of them developed aspirationpneumonia. Using indirect laryngoscopy before and after transhiatal esophagectomyfor carcinoma of the middle and lower esophagus,Johnsonet al4observeda 34% incidenceof left vocal cord paralysis.Thesepatientsdid not have a significant increasein pulmon~ complications. In our study, 13 (5%) of 283 surviving patientshad evidenceof hoarseness,but this symptom did not correlate with the development of pneumonia. COMMENTS The overall hospital mortality rate for esophagectomy Delayed gastric emptying and pneumoniawere related amongpatientsin this study was4%. Six of the 12 operative in our study, particularly in patients who underwent deathswere due to pn~o~i~ 5 of which wem associated esophagectomythrough a right posterolateral~oracotomy. with a documentedhistory of aspiration.Four of 5 deathsin Delayed gastricemptying hasbeen linked to vagotomy, the the groupundergoingesophagectomy via posterolateraltbo- size of the gastricinterposition5torsion of the stomachinto the posterior gutter of the right chest,6lack of a gastric racotomy c~uned as a result of aspirationpneumonia THE! AMERICAN

JOURNAL

OF SURGERYe

VOLUME

169

MAY

1995

473

61

FUNCTION FOLLOWING

ESOPHAGECTOMY~INLEY_--. .-ET AL

drainage procedure,* and jejunal obstruction secondary to a feeding jejunostomy. All of the patients in our series underwent vagotomy. We observed no correlation between delayed gastric emptying and the presence or absence of a gastric drainage procedure or feeding jejunostomy. Patients who underwent esophagectomy through a right posterolateral ~omcotomy had an increased incidence of delayed gastric emptying and late gastroesophageal reflux compared to those who received tmnshiatal esophagectomy or thoracoabdominal esophagogastrectomy . These results may be specific to the particular ways these procedures were performed, as described in the “Patients and Methods” section. Using computed tomography and nuclear medicine gastric emptying studies, Barber and colleagues’ showed prolonged emptying rates in the intrathoracic stomach compared to controls. The relationship was independent of whether the stomach was located in the anterior or posterior me~~tinum. However, the narrower the gastric substitute, the faster was the gastric emptying rate. This finding may explain the difference between the increased rate of delayed gastric emptying seen in patients in our thoracotomy group, who underwent reconstruction with whole stomach, compared to patients in our transhiatal esophagectomy group, who were reconstructedwith 4-cm plastic tubes. Another possible explanation for the high rates of delayed gastric emptying among our thoracotomy group is the torsion of the gastric interposition. We resectedcarcinomasof the midesophagusvia right thoracotomy in continuity with the m~di~tinal pleura and azygous vein:This procedure allows the interposed greater curvature of the stomachto rotate inlo the posterolateralgutter of the right chest. The resultant partial gastric volvulus delays gastric emptying. Three of the patients in our study who underwent esophagectomyby thoracotomy developedthis complication.6Partial gastric volvulus can be minimized by the useof a narrow gastric tube, proper alignmentof the tube,? mediastinalizationof the interposedstomachwith the azygous vein and lung. Maintenance of goodnasogastricsuction throughout the early postoperative period keeps the stomachdecompresseduntil fixation in the m~i~tinum has been achieved. In a randomized study, Fok andcolleagues*observedthat pyloroplasty reducedthe incidenceof delayed gastricemptying. In our study, pyloromyotomy or pyloroplasty did not decreasethe incidence of delayed gastric emptying, pneumonia, or ~~tomotic leak. Nil1 et al9 have recently shown that erythromycin is more effective than pyloroplasty or cisapridein improving delayed gastric emptying after esophagectomyand gastric interposition. None of the patientswe reviewed died or required a reoperation as a result of an anastomoticleak or stricture. Contrast ~io~aphy demons~t~ ~~tomotic leaks in 17%, with a slightly higher incidence in the transthoracic esophagectomygroup. Dewar and colleaguestoshowed anastomoticleaks were significantly correlatedwith a low preoperative serum albumin level, running suture technique,high interopemtiveblood loss,and the occurrenceof delayed posto~rativ~ gastric emptying. Esophagealdilatation for cervical dysphagiawas required in 35% of the patients in our study, with a slightly higher percentagein the 814

transthoracicesophagectomygroup. Dewar et al observed a significant correlationbetween~~tomotic leaksand intraoperativeblood lossand the developmentof anastomotic stricture. &ringer et al’ and Zieren et al*’ identified rates of radiological leaks and strictures after esophagectomy similarto thosefound in this study. In a randomizedstudy, Zieren andcolleaguesshowedthat ~~tomotic leakagewas not affected by the use of one-layer or two-layer anastomosis,but there were more anastomoticstricturesin patientshaving a two-layer anastomosis. The developmentof anastomoticstrictures reducesthe benefit of esophagectomy, but fortunately only 4% of the patientsin our series required esophage~dila~tion 1 year after their operation. Diarrhea hasbeen ascribedto truncal vagotomy following esophagectomy and usudly resolves with time. Eighteen percent of the patients in the present study suffered from diarrhea. One third of the women had significant diarrheawhich impactedon their ability to regaintheir preoperativeweight. Although 5% of patientshad varying degrees of dumping syndrome (postprandial nausea, cramping, sweating, or diarrhea), these symptoms were usually controlled with dietary manipulation. Small-bowel obstruction developedin 5 patientsfollowing esophagectomy.Three patients who underwent transhiatal esophagectomydeveloped hemiation of the small bowel through the hiatusafter dischargefrom hospital.One patient developed quadriplegiaand a diaphragmatichernia secondaryto a motor vehicle accident and required a laparotomy to reduce the small bowel and repair the hiatus. A 16-year-old patient with cerebral palsy had a transhiatal esophagectomyfor an early adenocarcinomain a columnarlined esophagus.He requiredlaparotomy and repair of the diaphragmatic hiatus 7 months after his esophagectomy.Another patient had spontaneoushemiation of the small bowel into the mediastinumand required reo~rdtion 3 monthsafter esophag~tomy. Two patients developed completebowel obstruction requiring laparotomy due to torsion of the jejunum at the site of the feeding jejunostomy. These casesemphasizedthe importance of sewing a broad baseof jejunum to the abdominal wall at the site of the jejunostomy.

CONCLUSION Aspiration pneumoniaand delayed gastric emptying are significant complications of transthoracicesophagectomy for cancer. Gastric emptying may be improved by using a 4-cm diameter gastric tube fixed in the posterior mediastinum. Although esophagectomyis a safeand effective method of managingpatientswith esophagogastriccancer, gastroesophageal reflux, dumping, and diarrheafrequently complicate this procedure.

REFERENCES 1. OrringerMB, Marshall B, StirlingMC. Transhiatal esophageclomy for benign and malignant disease. J Thorac Cardiovusc Surg. 1993;105:26%276. 2. Matbisen DJ, Grill0 HC, Wilkins EW, et al. Transthoracic

esophagectomy: a safeapproach to carcinoma of theesophagus. Ann TIaorac Swg.

1998;45:

137-143.

3. Heitmiller RF, Jones B. Transient diminished airway protection after transhiatal esophagectomy. Am J Surg. 1991:162:442--246.

THE AMERICAN JOURNAL OF SURGERY” VOLUME 169 MAY 1995

GI FUNCTION

4. Johnson PR, Kanegoanker GS, Bates T. Indirect laryngoscopic evaluation of vocal cord function in patients undergoing transhiatal esophagectomy. J Am Coil Surg. 1994;178:605608. 5. Barber L, Kemen M, Wegener M, et al. Effect of site and width of stomach tube after esophageal resection on gastric emptying. Zentralbl Chir. 1994; 119:204-206. 6. Casson A, Inculet R, Finley R. Volvulus of the intrathoracic stomach after total esophagectomy. J 7’horac Cardiovasc Surg. 1990; 100: 633-634. 7. Inculet RJ, Finley RJ, Cooper JD. A new technique for delivering the stomach or colon to the neck following total esophagectomy. Ann Thorac Surg. 1988;45:451-452.

FOLLOWING

ESOPHAGECTOMY/FINLEY

ET AL

8. Fok M, Cheng SW, Wong J. Pyloroplasty versus no drainage in gastric replacement of the esophagus. Am JSurg. 1991; 162:447452. 9. Hill ADK, Walsh TN, HamiIton D, et al. Erythromycin improves emptying of the denervated stomach after oesophagectomy. Br J Surg. 1993;80:879-881. 10. Dewar L, Gelfand G, Finley RJ, et al. Factors affecting cervical anastomotic leak and stricture formation following esophagogastrectomy and gastric tube interposition. Am J Surg. 1992;163:484-489. 11. Zieren HU, Muller JM, Pichhnaier H. Prospective randomized study of one- or two-layer anastomosis following oesophageal resection and cervical oesophagogastrostomy. BrJSurg. 1993;80:608-611.

THE AMERICAN JOURNAL OF SURGERY’

VOLUME 169 MAY 1995

475