Management of Enterovesical Fistulas R. Lawrence Moss, MD, John A. Ryan, Jr., MD, Seattle.Washington
The records of 51 patients diagnosed with enterovesical fistulas at Virginia Mason Medical Center from 1974 to 1988 were reviewed. Diverticulitis ( 4 1 % ) , Crohn's disease ( 1 7 % ) , and colorectal cancer ( 1 6 % ) were the m a j o r causes. In 50 of 51 patients, the diagnosis was made on the basis of the clinical history and the urine culture. Radiologic and endoscopic studies failed to identify the fistula in 20%. though all were confirmed at operation or autopsy. In four of eight patients with fistulas secondary to colorecta] cancer, malignancy was not diagnosed preoperatively. Operation was performed in 84% of the patients. One-stage resection of the bowel was performed in 66% of patients with the intent of removing the fistula. The complication rate was 8% with no deaths. Al] multi-stage procedures were performed for fistulas complicated by abscess or bowel obstruction. There were two postoperative deaths in patients with metastatic cancer undergoing palliative diversion. All eight patients treated by diverting colostomy had persistent fistulas and urinary sepsis. All eight patients treated with antibiotics but without operation were free of complications of the fistula until death from other causes. Enterovesical fistula is a clinical diagnosis. Preoperative studies should be used to delineate the bowel disease and search for malignancy rather than to see the fistula, which is clinically apparent. One-stage resection of the involved bowel is the procedure of choice in the absence of abscess or bowel obstruction. When resection is not feasible, medical management with antibiotics is preferable to colostomy.
1 ~ nterovesical fistula is an uncommon but significant F__~complication of inflammatory and neoplastic disease. Although the existence of this disorder has been known for centuries, its diagnosis and management continue to be controversial. Frequently, patients with enterovesical fistulas are followed months or years before the condition is recognized and treated effectively. This study Fromthe Virginia Mason MedicalCenter. Seattle,Washington. Requestsfor reprintsshouldbe addressedto JohnA. Ryan,Jr, MD, Section of General Surgery,Virginia Mason Clinic, 1100 Ninth Avenue, Seattle, Washington98112. Presentedat the 76th Annual Meetingof the North PacificSurgical Association.Victoria.BritishColumbia,Canada,November10-11. 1989. 514
of enterovesical fistulas was undertaken to answer the following questions: (1) How is the diagnosis best made? (2) Can malignancy be diagnosed preoperatively? (3) What is the best operative plan? (4) What is the role of palliative diverting colostomy? and (5) Is nonoperative treatment effective? PATIENTS AND M E T H O D S A review of the records of Virginia Mason Medical Center, Seattle, Washington, from 1974 to 1988 revealed 51 patients with enterovesical fistulas. The progress records, operative notes, laboratory data, radiographs, and pathology reports were reviewed. Long-term patient status was obtained from the medical record for 33 patients and telephone interview for 18. The mean duration of information was 47 months. RESULTS The 51 patients identified with enterovesical fistulas included 31 men (61%) and 20 women (39%). The causes of fistula are summarized in Table I. Diverticulitis was the most frequent cause, with Crohn's disease and colorectal cancer as the two other major causes. Causes not listed included adenocarcinoma of the bladder, lymphoma, and ovarian carcinoma in one patient each. In one patient, the cause was unknown at the time of the patient's death of congestive heart failure. Eight patients (16%) had received pelvic radiation prior to the development of the fistula. This includes patients with prostatic cancer (four patients), cervical cancer (two), rectal cancer (one), and ovarian carcinoma (one). The mean age at diagnosis was 59 years, with a range of 16 to 85 years. Age was related to cause, with fistulas secondary to Crohn's disease or cervical cancer occurring in younger patients (Table I). Irritative voiding symptoms and recurrent cystitis were the most frequent presenting symptoms. Dysuria was the most common complaint, present in 37 patients (73%). Pneumaturia occurred in 35 patients (69%). Urine cultures were performed in all patients and were positive for enteric organisms in 88%. Escherichia coli was the most common pathogen. Twelve patients (24%) had cultures revealing the presence of multiple organisms in the urine. Significant complaints referable to the gastrointestinal tract (diarrhea, abdominal pain, nausea) were present in half of the patients. Physical examination was generally of little value in identifying the fistula or determining the underlying disease process. An abdominal mass was present in 25%, including 41% of patients with malignant disease and 18% with benign disease. Abdominal tenderness was present in 29% of patients. Overall, the clinical history and the urine culture provided a presumptive diagnosis of the fistula in 50 of 51 patients. In only one patient (2%) was an enterovesical fistula found incidentally at operation. Table II summarizes the 149 diagnostic studies per-
THE AMERICAN JOURNAL OF SURGERY VOLUME159 MAY 1990
ENTEROVESICAL FISTULAS
formed in these patients. A diagnostic study may be helpful in three ways: It may reveal the fistula directly, allow the presumptive diagnosis of the fistula by indirect evidence, or identify the etiologic disease with or without revealing the fistula. Direct confirmation of a fistula may be provided by visualization or catheterization of the fistulous tract at endoscopy or by radiographic demonstration of the tract with contrast. Indirect evidence indudes findings such as an inflammatory papule at cystoscopy or air in the bladder seen radiographically. Important related abnormalities include findings such as diverticuli or strictures, cancer, or inflammatory bowel disease revealed by contrast radiographs or endoscopy. Cystoscopy was the most frequently used study and also the most likely to confirm diagnosis. However, even this most reliable study revealed evidence of fistulization only 67% of the time. Barium enema, though it infrequently revealed the fistula (17%), was abnormal in a great majority of cases (800) and very helpful in delineating a colonic cause. Flexible sigmoidoscopy was frequently abnormal (58%), but proctoscopy was least likely to reveal any abnormality. Computed tomography (CT) was used in only six patients and was abnormal (air seen in the bladder) in two. Intravenous pyelography rarely revealed the fistula; though 39% were abnormal, only two revealed upper tract abnormalities and none changed the operative plan. In 10 patients (20%), radiologic and endoscopic studies failed to provide either direct or indirect evidence of the fistulas, all of which were confirmed at operation or autopsy. The average time from the onset of symptoms to diagnosis was 14 weeks. The range was from immediate diagnosis in some to a 2-year delay in one patient, who was treated for recurrent urinary tract infections ultimately found to be from a sigmoid carcinoma perforating into the bladder. The delay in diagnosis was the same for all causes of enterovesical fistula, benign or malignant. In four of eight patients (50%) with fistulas secondary to colorectal cancer, the diagnosis of malignancy was not made preoperatively. In two of these patients, the cancer was an unexpected finding at stage 2 of a multistage procedure 7 and 16 weeks after the initial operation. In one patient, the colon containing the malignancy was separated from the bladder and then found to contain malignancy after excision. In three of these four patients with unsuspected cancer, diagnostic endoscopy was not performed preoperatively (Table IH). Operation was performed in 43 of 51 patients, while 8 patients were managed medically. The surgical treatment consisted of 35 resections for cure of the fistula and 8 diverting colostomies without resection. Twenty-three resections (66%) were done in 1 stage, 7 (20%) in 2, and 5 (14%) in 3 stages. The choice of operation was made on the basis of the presence or absence of intraperitoneal sepsis, the extent of involved bowel, the degree of intestinal obstruction, and the clinical condition of the patient. Three minor complications and two major complications occurred. Both major complications consisted of adhesive small bowel obstruction and pelvic abscesses requiring reoperation with successful outcomes. Two patients died
TABLE I Causes of Enteroveslcal Fistula in 51 Patients
Cause
Number*
Mean Age at Diagnosis (y)
Diverticulitis Crohn's disease Colorectal cancer Prostate cancer Complication ot previous operation Cervical cancer Other
21 (41) 9 (17) 8 (16) 4 (8) 3 46) 2 (4) 4 (8)
66 36 65 68 56 46 64
* Values in parentheses are percentages.
T A B L E II
Diagnostic Studies in 51 Patients with Enteroveslcal Fistula
Test
Cystoscopy Cystography IVP Barium enema Flexible sigmoldoscopy Proctoscopy CT scan
No. Done
Direct Evidence of Fistula (%)
Any Evidence of Fistula (%)
Abnormal (%)
36 13 31 35 12 16 6
56 46 3 11 0 0 O
67 53 10 17 0 0 33
94 61 39 80 58 6 33
IVP = intravenous pyelography.
T A B L E III
Fistulas Secondary t o C o l o r e c t a l C a n c e r Age Sex 64 82 67 65 43 55 79 68
M M M M M M M F
Cancer Dx Preop
DukesKlrklln Stage
Barium Enema
Endoscopy
Biopsy
Operation
No No No No Yes Yes Yes Yes
B-2 D B-2 C-2 --* D D D
Yes Yes Yes Yes Yes No No No
No No No Yes Yes No Yes Yes
No No No No Yes No Yes No
Yes Yes Yes Yes Yes Yes No No
9 Unknown Dx = diagnosis.
postoperatively; both had undergone palliative procedures for widely metastatic cancer (Table IV). Among 33 patients with benign disease, all but 1 were operated on. One patient with a fistula from diverticular disease had widely metastatic lymphoma and was managed medically. Eleven of 17 patients (65%) with malignant disease had surgery. For diverticulitis, 11 of 17 (65%) resections were done in one stage. For Crohn's disease, eight of nine (89%) resections were done in one stage. Single-stage procedures were performed with only 2 complications in 23 resections (8%). Both complications were minor; one was a wound infection and the other, a
THE AMERICAN JOURNAL OF SURGERY
VOLUME 159
MAY 1990
515
MOSS AND RYAN
T A B L E IV
Results in 43 Surgically Treated Patients with Enterovesical Fislula
Procedure
n
Complications
Death
One-stage Two-stage Three-stage Colostomy
23 7 5 8
2 2 1 2
0
Persistence of Fistula
0 0 2
0 1~ 0 8
* One patient underwent colostomy and reanastomosis without resection.
deep venous thrombosis. No fistulas recurred in this group. All multiple-stage procedures were done for fistulas complicated by intraperitoneal sepsis or bowel obstruction. The complication rate was higher in this group, as was expected, but there were no postoperative deaths. A fistula persisted in a patient who underwent a two-stage procedure without resection of the involved bowel. Diverting colostomy was done without an attempt to resect the involved bowel in eight patients: two with diverticulitis, three with colon cancer, and one each with prostate, cervical, and ovarian carcinoma. As noted above, both postoperative deaths in our series occurred in this group. All patients in this group had persistent fistulas and urinary tract infections. One patient with distal urinary tract obstruction developed severe pyelonephritis requiring nephrectomy and ultimately had end-stage renal failure. Eight patients with end-stage malignancy (seven patients) or prohibitive surgical risk (one) were managed without operation. None of them had obstruction of the urinary tract. Antibiotics aimed at suppression of urinary sepsis were used in each. All patients died of causes unrelated to complications of the fistula 1 to 14 months after presentation. No cases of systemic sepsis from the urinary tract resulted. The intestinal site of the fistula was the sigmoid colon in 27 patients (53%), ileum in 10 (20%), rectum in 8 (16%), nonsigmoid colon in 3 (5%), and unknown in 3 (5%). The dome was the most common site of fistulization on the bladder, occurring in 28 patients (62%). The fistula was to the posterior bladder wall in 10 patients (22%) and the trigone area in 7 (16%). Two fistulas directly involved the ureterovesical junction. Resection of the bladder was performed in seven patients (16%), only two for malignancy. The bladder was closed in 25 patients, utilizing a wide variety of sutures and techniques. The incidence of bladder resection and the method of closure were not related to outcome. COMMENTS The presenting symptoms and complications of enterovesical fistulas occur primarily in the urinary tract. However, the pathologic process is almost always intestinal, and the treatment is usually surgical resection of the involved bowel. Effective management depends both on 516
THE A M E R I C A N J O U R N A L OF SURGERY
recognizing the presence of a fistula and determining its cause preoperatively. Proper management should lead to a successful outcome in almost every case. The causes of enterovesieal fistulas in our series were similar to the findings of previous investigators [1-4]. Diverticulitis continues to be the most common cause. It has been estimated that 2% of patients with diverticulitis will develop this complication during their lifetime [5]. An increased awareness of this disorder is desirable, as prolonged delay in diagnosis frequently occurs. Hafner and co-workers [5] reported in 1962 that the patient's history is at least as important as any diagnostic test in the diagnosis of enterovesical fistula. This was confirmed by our finding that in all but one patient, the diagnosis was made on clinical grounds. Nevertheless, almost every recent review of this problem emphasizes the difficulty of making the diagnosis endoscopically or radiographically. Certainly, confirmatory studies may be helpful. However, we would argue that the primary purpose of preoperative studies should not be to see the fistula, which is clinically apparent, but rather to identify the patient with malignancy and determine the anatomic site and extent of the bowel disease. Utilized in this way, the preoperative evaluation will be most helpful in developing an operative plan. Cystoscopy was the most frequently used test in our series and most often the first procedure in a given patient. We believe this was appropriate for several reasons. It confirmed the clinical impression of a fistula in about two thirds of patients. Furthermore, it can be useful in ruling out the other causes in the differential diagnosis of recurrent urinary tract infection, i.e., bladder calculi or interstitial cystitis. Finally, biopsy of the fistulous tract may reveal malignancy. Although cystoscopy is useful, preoperative evaluation of the nature and extent of diseased bowel is more important. Barium enema and flexible sigmoidoscopy or eolonoscopy rarely reveal the fistula, yet are important studies. Both are useful in delineating diverticular disease from colorectal carcinoma. Intravenous pyelography was done infrequently in our series and in others; we found that it never affected patient management. Perhaps it is only of use when upper urinary tract disease is suspected. Sarr et al [4] reported that CT scanning of the abdomen and pelvis revealed air in the bladder, providing the presumptive diagnosis of enterovesical fistula in 20 of 23 patients. On this basis, they suggested that CT should be the first test of choice in these patients. Though our experience with CT was confined to six patients, we found it to be abnormal only a third of the time. More importantly, the CT scan does not distinguish benign from malignant disease and is not useful in planning the extent of necessary resection. We believe that its use in enterovesical fistula is limited to those cases in which abscess is suspected or in confirmed malignant cases that require further staging. Malignancy was an important cause of enterovesical fistula in our series and in others. We found that our effectiveness in making the diagnosis was poor, particularly with respect to colorectal cancer. Colon cancer has
VOLUME 159
MAY 1990
ENTEROVESICALFISTULAS
been reported to be adherent to the bladder as much as 15% of the time [6]. A recent report from our institution demonstrated a significantly decreased survival when colon cancer adherent to other organs was managed by separation of organs rather than by en bloc resection [7]. Consistent use of lower gastrointestinal endoscopy may improve our ability to diagnose colorectal cancer preoperatively. All cases of colovesical fistula should have endoscopic evaluation of the involved segment of colon, with biopsy if indicated. Carcinoembryonic antigen levels may help guide the evaluation and therapy. In cases where a definitive diagnosis is not made preoperatively, frozensection examination of the specimen should be used to determine whether bladder resection is necessary. One-stage resection of the bowel with primary anastomosis is the operative procedure of choice whenever possible. The merits of this approach in appropriately selected patients have been discussed previously [1]. Investigators have reported an incidence of effective one-stage resection from 18% to 47% [2,3,4,8-10]. Sixty-six percent of our patients treated for cure of the fistula were successfully treated in one stage with no major complications. We had a higher complication rate for the multiple-stage procedures, although all patients who underwent resection ultimately did well. This undoubtedly reflects the fact that only patients with abscess or bowel obstructions were managed with multi-stage procedures. The increased risk of surgery in these patients has been discussed by McConneU et al [9]. Bladder resection is only indicated for fistulas of malignant cause. Simple debridement and closure with or without drainage were used in the majority of benign cases in our series with success. In both our series and another recent report, a significant portion of patients were treated by fecal diversion alone [4]. Goligher [I 1] pointed out that resection of the diseased bowel is necessary for cure of the fistula. This was again discussed by Amin and co-workers [12]. All of our patients who did not undergo resection had persistence of the fistula. We would argue that if operative treatment of the fistula is planned, it must include resection of the involved bowel. In a canine study of enterovesical fistulas, Heiskell et al [13] noted that the fistula was well tolerated as long as there was no distal obstruction to urinary flow. Amin and
co-workers [i2] were the first to report a number of patients with this disorder treated nonoperatively for several years. They found that patients did quite well with chronic antibiotic suppression. Our eight patients treated in this manner were similarly free of serious sequelae of the fistula. The importance of an unobstructed urinary tract is emphasized by a patient described by Amin and co-workers [12]. Despite the presence of a diverting colostomy, this patient with urethral stricture developed severe complications. It should be emphasized that the experience with nonoperative management of these patients is small and confined to a highly selected group. The only definitive treatment for enterovesical fistula is resection of the involved bowel. This should be done whenever possible. When resection is not feasible, medical management is preferable to colostomy.
REFERENCES 1. King RM, Beart RW, Mcllrath DC. Colovesicaland rectovesical fistulas. Arch Surg 1982; 117: 680-3. 2. Krco M J, Jacobs SC, Malongoni MA, Lawson RK. Colovesical fistulas. Urology 1984; 23: 340-2. 3. Karamchandani MC, West CF. Vesicoenteric fistulas. Am J Surg 1984; 147: 631-3. 4. Sarr MG, Fishman EK, Goldman SM, et al. Enterovesical fistula. Surg Gynecol Obstet 1987; 164: 41-8. 5. Hafner CD, Ponka JL, Brush BE. Genitourinary manifestations of diverticulitis of the colon: a study of 500 cases. JAMA 1962; 179: 174-6. 6. Aldrete JS, ReMine WH. Vesicocolicfistula: a complication of colonic cancer. Long term results of its surgical treatment. Arch Surg 1976; 94: 627-37. 7. Hunter JA, Ryan JA. En bloc resectionof colon cancer adherent to other organs. Am J Surg 1987; 154: 67-71. 8. Rao PN, Knox R, Barnard RJ, et al. Management of colovesical fistula. Br J Surg 1987; 74: 362-3. 9. McConnell DB, Sasaki TM, Vetto RM. Experience with colovesical fistula. Am J Surg 1980; 140: 80-4. 10. Suits GS, Knoepp LF. A community experience with enterovesical fistula. Am Surg 1985; 51: 523-8. 11. Goligher J. Surgery of the anus, rectum, and colon. London: Balliere Tindall, 1984: 1109. 12. Amin M, Nallinger R, Polk HC. Conservative treatment of selected patients with colovesicalfistula due to diverticulitis. Surg Gynecol Obstet 1984; 159: 442-4. 13. Heiskell LA, Vjiki GT, Beal JM. A study of experimental colovesical fistula. Am J Surg 1975; 129: 316-8.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990 517
The records of 51 patients diagnosed with enterovesical fistulas at Virginia Mason Medical Center from 1974 to 1988 were reviewed. Diverticulitis ( 4 1 % ) , Crohn's disease ( 1 7 % ) , and colorectal cancer ( 1 6 % ) were the m a j o r causes. In 50 of 51 patients, the diagnosis was made on the basis of the clinical history and the urine culture. Radiologic and endoscopic studies failed to identify the fistula in 20%. though all were confirmed at operation or autopsy. In four of eight patients with fistulas secondary to colorecta] cancer, malignancy was not diagnosed preoperatively. Operation was performed in 84% of the patients. One-stage resection of the bowel was performed in 66% of patients with the intent of removing the fistula. The complication rate was 8% with no deaths. Al] multi-stage procedures were performed for fistulas complicated by abscess or bowel obstruction. There were two postoperative deaths in patients with metastatic cancer undergoing palliative diversion. All eight patients treated by diverting colostomy had persistent fistulas and urinary sepsis. All eight patients treated with antibiotics but without operation were free of complications of the fistula until death from other causes. Enterovesical fistula is a clinical diagnosis. Preoperative studies should be used to delineate the bowel disease and search for malignancy rather than to see the fistula, which is clinically apparent. One-stage resection of the involved bowel is the procedure of choice in the absence of abscess or bowel obstruction. When resection is not feasible, medical management with antibiotics is preferable to colostomy.
1 ~ nterovesical fistula is an uncommon but significant F__~complication of inflammatory and neoplastic disease. Although the existence of this disorder has been known for centuries, its diagnosis and management continue to be controversial. Frequently, patients with enterovesical fistulas are followed months or years before the condition is recognized and treated effectively. This study Fromthe Virginia Mason MedicalCenter. Seattle,Washington. Requestsfor reprintsshouldbe addressedto JohnA. Ryan,Jr, MD, Section of General Surgery,Virginia Mason Clinic, 1100 Ninth Avenue, Seattle, Washington98112. Presentedat the 76th Annual Meetingof the North PacificSurgical Association.Victoria.BritishColumbia,Canada,November10-11. 1989. 514
of enterovesical fistulas was undertaken to answer the following questions: (1) How is the diagnosis best made? (2) Can malignancy be diagnosed preoperatively? (3) What is the best operative plan? (4) What is the role of palliative diverting colostomy? and (5) Is nonoperative treatment effective? PATIENTS AND M E T H O D S A review of the records of Virginia Mason Medical Center, Seattle, Washington, from 1974 to 1988 revealed 51 patients with enterovesical fistulas. The progress records, operative notes, laboratory data, radiographs, and pathology reports were reviewed. Long-term patient status was obtained from the medical record for 33 patients and telephone interview for 18. The mean duration of information was 47 months. RESULTS The 51 patients identified with enterovesical fistulas included 31 men (61%) and 20 women (39%). The causes of fistula are summarized in Table I. Diverticulitis was the most frequent cause, with Crohn's disease and colorectal cancer as the two other major causes. Causes not listed included adenocarcinoma of the bladder, lymphoma, and ovarian carcinoma in one patient each. In one patient, the cause was unknown at the time of the patient's death of congestive heart failure. Eight patients (16%) had received pelvic radiation prior to the development of the fistula. This includes patients with prostatic cancer (four patients), cervical cancer (two), rectal cancer (one), and ovarian carcinoma (one). The mean age at diagnosis was 59 years, with a range of 16 to 85 years. Age was related to cause, with fistulas secondary to Crohn's disease or cervical cancer occurring in younger patients (Table I). Irritative voiding symptoms and recurrent cystitis were the most frequent presenting symptoms. Dysuria was the most common complaint, present in 37 patients (73%). Pneumaturia occurred in 35 patients (69%). Urine cultures were performed in all patients and were positive for enteric organisms in 88%. Escherichia coli was the most common pathogen. Twelve patients (24%) had cultures revealing the presence of multiple organisms in the urine. Significant complaints referable to the gastrointestinal tract (diarrhea, abdominal pain, nausea) were present in half of the patients. Physical examination was generally of little value in identifying the fistula or determining the underlying disease process. An abdominal mass was present in 25%, including 41% of patients with malignant disease and 18% with benign disease. Abdominal tenderness was present in 29% of patients. Overall, the clinical history and the urine culture provided a presumptive diagnosis of the fistula in 50 of 51 patients. In only one patient (2%) was an enterovesical fistula found incidentally at operation. Table II summarizes the 149 diagnostic studies per-
THE AMERICAN JOURNAL OF SURGERY VOLUME159 MAY 1990
ENTEROVESICAL FISTULAS
formed in these patients. A diagnostic study may be helpful in three ways: It may reveal the fistula directly, allow the presumptive diagnosis of the fistula by indirect evidence, or identify the etiologic disease with or without revealing the fistula. Direct confirmation of a fistula may be provided by visualization or catheterization of the fistulous tract at endoscopy or by radiographic demonstration of the tract with contrast. Indirect evidence indudes findings such as an inflammatory papule at cystoscopy or air in the bladder seen radiographically. Important related abnormalities include findings such as diverticuli or strictures, cancer, or inflammatory bowel disease revealed by contrast radiographs or endoscopy. Cystoscopy was the most frequently used study and also the most likely to confirm diagnosis. However, even this most reliable study revealed evidence of fistulization only 67% of the time. Barium enema, though it infrequently revealed the fistula (17%), was abnormal in a great majority of cases (800) and very helpful in delineating a colonic cause. Flexible sigmoidoscopy was frequently abnormal (58%), but proctoscopy was least likely to reveal any abnormality. Computed tomography (CT) was used in only six patients and was abnormal (air seen in the bladder) in two. Intravenous pyelography rarely revealed the fistula; though 39% were abnormal, only two revealed upper tract abnormalities and none changed the operative plan. In 10 patients (20%), radiologic and endoscopic studies failed to provide either direct or indirect evidence of the fistulas, all of which were confirmed at operation or autopsy. The average time from the onset of symptoms to diagnosis was 14 weeks. The range was from immediate diagnosis in some to a 2-year delay in one patient, who was treated for recurrent urinary tract infections ultimately found to be from a sigmoid carcinoma perforating into the bladder. The delay in diagnosis was the same for all causes of enterovesical fistula, benign or malignant. In four of eight patients (50%) with fistulas secondary to colorectal cancer, the diagnosis of malignancy was not made preoperatively. In two of these patients, the cancer was an unexpected finding at stage 2 of a multistage procedure 7 and 16 weeks after the initial operation. In one patient, the colon containing the malignancy was separated from the bladder and then found to contain malignancy after excision. In three of these four patients with unsuspected cancer, diagnostic endoscopy was not performed preoperatively (Table IH). Operation was performed in 43 of 51 patients, while 8 patients were managed medically. The surgical treatment consisted of 35 resections for cure of the fistula and 8 diverting colostomies without resection. Twenty-three resections (66%) were done in 1 stage, 7 (20%) in 2, and 5 (14%) in 3 stages. The choice of operation was made on the basis of the presence or absence of intraperitoneal sepsis, the extent of involved bowel, the degree of intestinal obstruction, and the clinical condition of the patient. Three minor complications and two major complications occurred. Both major complications consisted of adhesive small bowel obstruction and pelvic abscesses requiring reoperation with successful outcomes. Two patients died
TABLE I Causes of Enteroveslcal Fistula in 51 Patients
Cause
Number*
Mean Age at Diagnosis (y)
Diverticulitis Crohn's disease Colorectal cancer Prostate cancer Complication ot previous operation Cervical cancer Other
21 (41) 9 (17) 8 (16) 4 (8) 3 46) 2 (4) 4 (8)
66 36 65 68 56 46 64
* Values in parentheses are percentages.
T A B L E II
Diagnostic Studies in 51 Patients with Enteroveslcal Fistula
Test
Cystoscopy Cystography IVP Barium enema Flexible sigmoldoscopy Proctoscopy CT scan
No. Done
Direct Evidence of Fistula (%)
Any Evidence of Fistula (%)
Abnormal (%)
36 13 31 35 12 16 6
56 46 3 11 0 0 O
67 53 10 17 0 0 33
94 61 39 80 58 6 33
IVP = intravenous pyelography.
T A B L E III
Fistulas Secondary t o C o l o r e c t a l C a n c e r Age Sex 64 82 67 65 43 55 79 68
M M M M M M M F
Cancer Dx Preop
DukesKlrklln Stage
Barium Enema
Endoscopy
Biopsy
Operation
No No No No Yes Yes Yes Yes
B-2 D B-2 C-2 --* D D D
Yes Yes Yes Yes Yes No No No
No No No Yes Yes No Yes Yes
No No No No Yes No Yes No
Yes Yes Yes Yes Yes Yes No No
9 Unknown Dx = diagnosis.
postoperatively; both had undergone palliative procedures for widely metastatic cancer (Table IV). Among 33 patients with benign disease, all but 1 were operated on. One patient with a fistula from diverticular disease had widely metastatic lymphoma and was managed medically. Eleven of 17 patients (65%) with malignant disease had surgery. For diverticulitis, 11 of 17 (65%) resections were done in one stage. For Crohn's disease, eight of nine (89%) resections were done in one stage. Single-stage procedures were performed with only 2 complications in 23 resections (8%). Both complications were minor; one was a wound infection and the other, a
THE AMERICAN JOURNAL OF SURGERY
VOLUME 159
MAY 1990
515
MOSS AND RYAN
T A B L E IV
Results in 43 Surgically Treated Patients with Enterovesical Fislula
Procedure
n
Complications
Death
One-stage Two-stage Three-stage Colostomy
23 7 5 8
2 2 1 2
0
Persistence of Fistula
0 0 2
0 1~ 0 8
* One patient underwent colostomy and reanastomosis without resection.
deep venous thrombosis. No fistulas recurred in this group. All multiple-stage procedures were done for fistulas complicated by intraperitoneal sepsis or bowel obstruction. The complication rate was higher in this group, as was expected, but there were no postoperative deaths. A fistula persisted in a patient who underwent a two-stage procedure without resection of the involved bowel. Diverting colostomy was done without an attempt to resect the involved bowel in eight patients: two with diverticulitis, three with colon cancer, and one each with prostate, cervical, and ovarian carcinoma. As noted above, both postoperative deaths in our series occurred in this group. All patients in this group had persistent fistulas and urinary tract infections. One patient with distal urinary tract obstruction developed severe pyelonephritis requiring nephrectomy and ultimately had end-stage renal failure. Eight patients with end-stage malignancy (seven patients) or prohibitive surgical risk (one) were managed without operation. None of them had obstruction of the urinary tract. Antibiotics aimed at suppression of urinary sepsis were used in each. All patients died of causes unrelated to complications of the fistula 1 to 14 months after presentation. No cases of systemic sepsis from the urinary tract resulted. The intestinal site of the fistula was the sigmoid colon in 27 patients (53%), ileum in 10 (20%), rectum in 8 (16%), nonsigmoid colon in 3 (5%), and unknown in 3 (5%). The dome was the most common site of fistulization on the bladder, occurring in 28 patients (62%). The fistula was to the posterior bladder wall in 10 patients (22%) and the trigone area in 7 (16%). Two fistulas directly involved the ureterovesical junction. Resection of the bladder was performed in seven patients (16%), only two for malignancy. The bladder was closed in 25 patients, utilizing a wide variety of sutures and techniques. The incidence of bladder resection and the method of closure were not related to outcome. COMMENTS The presenting symptoms and complications of enterovesical fistulas occur primarily in the urinary tract. However, the pathologic process is almost always intestinal, and the treatment is usually surgical resection of the involved bowel. Effective management depends both on 516
THE A M E R I C A N J O U R N A L OF SURGERY
recognizing the presence of a fistula and determining its cause preoperatively. Proper management should lead to a successful outcome in almost every case. The causes of enterovesieal fistulas in our series were similar to the findings of previous investigators [1-4]. Diverticulitis continues to be the most common cause. It has been estimated that 2% of patients with diverticulitis will develop this complication during their lifetime [5]. An increased awareness of this disorder is desirable, as prolonged delay in diagnosis frequently occurs. Hafner and co-workers [5] reported in 1962 that the patient's history is at least as important as any diagnostic test in the diagnosis of enterovesical fistula. This was confirmed by our finding that in all but one patient, the diagnosis was made on clinical grounds. Nevertheless, almost every recent review of this problem emphasizes the difficulty of making the diagnosis endoscopically or radiographically. Certainly, confirmatory studies may be helpful. However, we would argue that the primary purpose of preoperative studies should not be to see the fistula, which is clinically apparent, but rather to identify the patient with malignancy and determine the anatomic site and extent of the bowel disease. Utilized in this way, the preoperative evaluation will be most helpful in developing an operative plan. Cystoscopy was the most frequently used test in our series and most often the first procedure in a given patient. We believe this was appropriate for several reasons. It confirmed the clinical impression of a fistula in about two thirds of patients. Furthermore, it can be useful in ruling out the other causes in the differential diagnosis of recurrent urinary tract infection, i.e., bladder calculi or interstitial cystitis. Finally, biopsy of the fistulous tract may reveal malignancy. Although cystoscopy is useful, preoperative evaluation of the nature and extent of diseased bowel is more important. Barium enema and flexible sigmoidoscopy or eolonoscopy rarely reveal the fistula, yet are important studies. Both are useful in delineating diverticular disease from colorectal carcinoma. Intravenous pyelography was done infrequently in our series and in others; we found that it never affected patient management. Perhaps it is only of use when upper urinary tract disease is suspected. Sarr et al [4] reported that CT scanning of the abdomen and pelvis revealed air in the bladder, providing the presumptive diagnosis of enterovesical fistula in 20 of 23 patients. On this basis, they suggested that CT should be the first test of choice in these patients. Though our experience with CT was confined to six patients, we found it to be abnormal only a third of the time. More importantly, the CT scan does not distinguish benign from malignant disease and is not useful in planning the extent of necessary resection. We believe that its use in enterovesical fistula is limited to those cases in which abscess is suspected or in confirmed malignant cases that require further staging. Malignancy was an important cause of enterovesical fistula in our series and in others. We found that our effectiveness in making the diagnosis was poor, particularly with respect to colorectal cancer. Colon cancer has
VOLUME 159
MAY 1990
ENTEROVESICALFISTULAS
been reported to be adherent to the bladder as much as 15% of the time [6]. A recent report from our institution demonstrated a significantly decreased survival when colon cancer adherent to other organs was managed by separation of organs rather than by en bloc resection [7]. Consistent use of lower gastrointestinal endoscopy may improve our ability to diagnose colorectal cancer preoperatively. All cases of colovesical fistula should have endoscopic evaluation of the involved segment of colon, with biopsy if indicated. Carcinoembryonic antigen levels may help guide the evaluation and therapy. In cases where a definitive diagnosis is not made preoperatively, frozensection examination of the specimen should be used to determine whether bladder resection is necessary. One-stage resection of the bowel with primary anastomosis is the operative procedure of choice whenever possible. The merits of this approach in appropriately selected patients have been discussed previously [1]. Investigators have reported an incidence of effective one-stage resection from 18% to 47% [2,3,4,8-10]. Sixty-six percent of our patients treated for cure of the fistula were successfully treated in one stage with no major complications. We had a higher complication rate for the multiple-stage procedures, although all patients who underwent resection ultimately did well. This undoubtedly reflects the fact that only patients with abscess or bowel obstructions were managed with multi-stage procedures. The increased risk of surgery in these patients has been discussed by McConneU et al [9]. Bladder resection is only indicated for fistulas of malignant cause. Simple debridement and closure with or without drainage were used in the majority of benign cases in our series with success. In both our series and another recent report, a significant portion of patients were treated by fecal diversion alone [4]. Goligher [I 1] pointed out that resection of the diseased bowel is necessary for cure of the fistula. This was again discussed by Amin and co-workers [12]. All of our patients who did not undergo resection had persistence of the fistula. We would argue that if operative treatment of the fistula is planned, it must include resection of the involved bowel. In a canine study of enterovesical fistulas, Heiskell et al [13] noted that the fistula was well tolerated as long as there was no distal obstruction to urinary flow. Amin and
co-workers [i2] were the first to report a number of patients with this disorder treated nonoperatively for several years. They found that patients did quite well with chronic antibiotic suppression. Our eight patients treated in this manner were similarly free of serious sequelae of the fistula. The importance of an unobstructed urinary tract is emphasized by a patient described by Amin and co-workers [12]. Despite the presence of a diverting colostomy, this patient with urethral stricture developed severe complications. It should be emphasized that the experience with nonoperative management of these patients is small and confined to a highly selected group. The only definitive treatment for enterovesical fistula is resection of the involved bowel. This should be done whenever possible. When resection is not feasible, medical management is preferable to colostomy.
REFERENCES 1. King RM, Beart RW, Mcllrath DC. Colovesicaland rectovesical fistulas. Arch Surg 1982; 117: 680-3. 2. Krco M J, Jacobs SC, Malongoni MA, Lawson RK. Colovesical fistulas. Urology 1984; 23: 340-2. 3. Karamchandani MC, West CF. Vesicoenteric fistulas. Am J Surg 1984; 147: 631-3. 4. Sarr MG, Fishman EK, Goldman SM, et al. Enterovesical fistula. Surg Gynecol Obstet 1987; 164: 41-8. 5. Hafner CD, Ponka JL, Brush BE. Genitourinary manifestations of diverticulitis of the colon: a study of 500 cases. JAMA 1962; 179: 174-6. 6. Aldrete JS, ReMine WH. Vesicocolicfistula: a complication of colonic cancer. Long term results of its surgical treatment. Arch Surg 1976; 94: 627-37. 7. Hunter JA, Ryan JA. En bloc resectionof colon cancer adherent to other organs. Am J Surg 1987; 154: 67-71. 8. Rao PN, Knox R, Barnard RJ, et al. Management of colovesical fistula. Br J Surg 1987; 74: 362-3. 9. McConnell DB, Sasaki TM, Vetto RM. Experience with colovesical fistula. Am J Surg 1980; 140: 80-4. 10. Suits GS, Knoepp LF. A community experience with enterovesical fistula. Am Surg 1985; 51: 523-8. 11. Goligher J. Surgery of the anus, rectum, and colon. London: Balliere Tindall, 1984: 1109. 12. Amin M, Nallinger R, Polk HC. Conservative treatment of selected patients with colovesicalfistula due to diverticulitis. Surg Gynecol Obstet 1984; 159: 442-4. 13. Heiskell LA, Vjiki GT, Beal JM. A study of experimental colovesical fistula. Am J Surg 1975; 129: 316-8.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990 517
