- Email: [email protected]
Nonoperative Treatment of Enteric Fistulas: Results in 53 Patients
Hepatic, Biliary, and
Gastrointestinal Intervention
Nonoperative Treatment of Enteric Fistulas: Results in 53 Patients l Jeanne M. laBerge, MD Robert K. KerJan, Jr, MD Roy L. Gordon, MD Ernest J. Ring, MD
Index terms: Abscess, percutaneous drainage • Fistula, gastrointestinal • Gastrointestinal tract, interventional procedure JVIR 1992; 3:353-357 Abbreviation: sure rate
SCR = spontaneous clo-
Fifty-three patients with enterocutaneous fistulas were treated nonoperatively. While most of the fistulas resulted as a complication of surgery, other causes included inflammatory disease, trauma, and malignancy. Fistulas originated from the esophagus (n = 4), stomach (n = 4), duodenum (n = 5), small bowel (n = 27), colonic-small bowel anastomosis (n = 2), colon (n = 10), and Hartmann pouch (n = 1), and included both high-output (n = 29) and low-output (n = 24) types. Nonoperative management resulted in spontaneous fistula closure in 57%. Most of the spontaneous closures (83%) occurred within 2 months of treatment. Spontaneous closure rates were similar for low-output and for high-output fistulas (52% vs 63%), but the treatment duration was longer for high-output than for low-output fistulas (mean treatment duration, 17 days vs 41 days). Colonic fistulas were particularly difficult to manage and closed spontaneously in only 10% of patients (one of 10).
T HE treatment of patients with sig-
1 From the Department of Radiology, Box 0628, University of California, San Francisco, 505 Parnassus, San Francisco, CA 94143-0628. Received June 3, 1991; revision requested August 24; revision received December 6; accepted December 14. Address reprint requests to J.M.L.
" SCVIR, 1992
nificant gastrointestinal fistulas is generally a two-staged process. Initial management efforts center on controlling bowel effluent, clearing associated abscesses, and improving the patient's nutritional status, while the ultimate therapeutic goal is closure of the enteric fistula. Recent advances in nonoperative therapy have improved the success of initial management. Aggressive nutritional support and fluid and electrolyte management have lowered associated mortality (1), and interventional radiologic techniques have aided efforts to control sepsis and eradicate associated abscesses (2,3). The value of nonoperative measures in accomplishing the second task, fistula closure, is less clearly demonstrated in the literature. Reported rates of spontaneous closure have varied substantially (29%-100%) (4,5) and seem dependent on the type of nonoperative management employed as well as the location, underlying cause, and output volume of the fistulas. We report our experience in the nonoperative treatment of gastro-
intestinal fistulas with use of interventional radiologic techniques. The focus of our study is the rate of spontaneous closure. We analyze the success of nonoperative management with respect to the underlying cause, location, and output volume of the fistulas.
PATIENTS AND METHODS A retrospective review was performed in all patients referred to Interventional Radiology for management of enterocutaneous fistulas at the University of California, San Francisco and the Scripps' Memorial Hospital of La Jolla between June 1985 and January 1990. Patients who had undergone percutaneous abscess drainage and were subsequently proved to have occult enteric fistulas were also included. Patients with incomplete medical records or insufficient follow-up were excluded. Fifty-three patients, including 28 men and 25 women, whose ages ranged from 29 to 84 years, qualified for the study. The locations of the
353
354 • Journal of Vascular and Interventional Radiology May 1992
enteric communications are listed in Table 1. The causes of these leaks are summarized in Table 2. Interventional radiologic management was performed in accordance with previously established principles (5,6). These principles included control of the bowel effluent as well as evacuation of all associated abscesses. Abdominal fluid collections were drained with use of commercially available sump tubes (RMS; Cook, Bloomington, Ind) and gravity drainage tubes (Rob-Nel, Sherwood Medical, St Louis, Mo; Malecot, DowCorning, Upland, Mich). The diameter of the drainage catheters varied from 12 F to 30 F. Catheters were either placed percutaneously under computed tomographic or ultrasound guidance, or by catheterization of an existing percutaneous tract. An immediate sinogram was obtained to confirm proper catheter position. Routine postdrainage sinograms were obtained in all patients 3-6 days following the initial drainage. When a fistula was discovered, a sump drainage catheter was positioned adjacent to the bowel to control the fistula effluent and a catheter was placed in the dependent portion of any associated abscesses. The clinical status of the patient and the output from the drainage catheters were recorded by the nursing staff and were monitored daily by the interventional radiology staff. Sump catheters were irrigated on daily rounds; malfunctioning catheters were repositioned or replaced in the fluoroscopy suite. Nutritional support included enteral and parenteral nutrition consistent with current practice (1). However, the form of nutritional supplementation was not considered as an independent factor in this study. Low-output fistulas were defined as having an output of less than 200 mL per day following the initial drainage procedure (6,7). The remainder were considered high-output fistulas. A successful outcome of the interventional radiologic procedure was defined as healing of the fistula and all associated abscesses without any type of open surgical interven-
Table 1 Distribution of Enteric Fistulas in 53 Patients
Output Fistula Location (n = 53) Esophagus (n = 4) Stomach (n = 4) Duodenum (n = 5) Small intestine (n = 27) Small bowel-colon anastomosis (n = 2) Colon (n = 10) Hartmann anastomotic pouch (n = 1)
tion. All other outcomes were considered failures.
RESULTS The results of nonoperative therapy in 53 patients are summarized in Table 3. Conservative management resulted in spontaneous fistula closure in 57%. Results were similar for low-output (spontaneous closure rate [SCR] = 63%) and high-output (SCR = 52%) fistulas. When sorted by underlying cause, results were best for posttraumatic fistulas (SCR = 100%) and poorest for fistulas secondary to malignancy (SCR = 0%). Results with postsurgical and postinflammatory fistulas were similar (SCR = 57% vs 60%). When sorted by location of fistula, the poorest results occurred with gastric and colonic fistulas (SCR = 25% and 10%, respectively). Of the 10 colonic fistulas, eight were low-output and two were highoutput; three were secondary to inflammatory disease, one was secondary to malignancy, and six developed postoperatively. Four of 10 patients with colonic fistulas underwent surgery within 2 weeks. Colonic fistulas were generally treated with largebore sump tubes (16-30 F). Catheter obstruction from large particulate debris was treated by means of daily catheter irrigation and fluoroscopic catheter manipulation when necessary. Of the four gastric fistulas, three were postoperative and one was sec-
Low
High
(n = 24)
(n = 29)
3
2 4 2
10
17
1
1 2
2
o 8 1
o
ondary to malignancy; all four were high-output fistulas. Two of the four patients with gastric fistulas died during nonoperative treatment, and two patients underwent surgery within 1-2 months. Information regarding the exact duration of therapy was available for 49 of 53 patients. The data are summarized in Table 4. Treatment duration varied from 5 to 500 days. In those patients in whom treatment was ultimately successful (29 of 49), the duration of therapy varied from 7 to 63 days (mean = 28 days). Therapy was successful within 1 month in 18, between 1 and 2 months in six, and after a period longer than 2 months in five of 29. The mean duration of successful therapy in patients with low-output fistulas was 17 days, while the mean duration of successful therapy in patients with high-output fistulas was 41 days. In those patients in whom conservative therapy failed (20 of 49 patients), the duration of treatment ranged from 5 to 500 days (mean = 31 days, excluding one patient treated for 500 days). Among those patients in whom conservative therapy failed and in whom the treatment duration was known, four died and 16 underwent a surgical procedure. Surgery was performed within less than 1 month in six patients, at 1-2 months in six, and after 2 months in four. Six patients died during the period of nonoperative management (mortality = 11%); two died in the first 2 weeks of treatment and four died after more than 1 month of conserva-
LaBerge et al • 355 Volume 3 Number 2
Table 2 Underlying Cause of Enteric Fistulas in 53 Patients Output Low
High
Cause (n = 53)
(n = 24)
(n = 29)
Postsurgical complication (n = 44) Inflammatory disease* (n = 5) Trauma (n = 2) Malignancy (n = 2)
18
26 1
4 2 0
o 2
* Includes peptic ulcer disease, Crohn disease, and diverticulitis.
Table 3 Results of Nonoperative Management in 53 Patients Breakdown of Results Success Failure Output 9 (37) Low(n = 24) 15 (63) 14 (48) 15 (52) High (n = 29) Etiology 25 (57) Postsurgical (n = 44) 19 (43) Inflammatory (n = 5) 3 (60) 2 (40) 0(0) 2 (100) Trauma (n = 2) 0(0) Malignancy (n = 2) 2 (100) Location Esophagus (n = 24) 4 (l00) 0 (0) Stomach (n = 4) 1 (25) 3 (75) Duodenum (n = 5) 3 (60) 2 (40) Small bowel (n = 27) 18 (67) 9 (33) Colonic anastomosis (n = 2) 2 (100) 0 (0) Colon (n = 10) 1 (10) 9 (90) Hartmann pouch (n = 1) 1 (l00) 0 (0) Note.-Numbers in parentheses are percentages. Overall, nonoperative management was considered a success in 30 patients (57%) and a failure in 23 patients (43%).
Table 4 Duration of Nonoperative Therapy in 49 Patients No. of Patients with Successful Duration of Treatment Outcomes (n = 29) < 1 month (n = 26) 18 1-2 months (n = 14) 6 >2 months (n = 9) 5
No. of Patients with Unsuccessful Outcomes (n = 20)
8 8 4
Note.-Data available for 49 of the 53 patients treated.
tive management. All of the patients who died had an underlying malignancy; five of six developed fistulas as a postoperative complication. These fistulas were located in the stomach (n = 2), duodenum (n = 1) and small bowel (n = 3).
DISCUSSION Nonoperative therapy is the preferred initial approach to the treatment of patients with enteric fistulas. Surgery, because of its attendant
risks and technical difficulty, is usually reserved for patients who fail to heal with conservative management. The initial goals of conservative management are control of bowel effluent, drainage of associated abscesses, and improvement in the patient's fluid, electrolyte, and nutritional status. The ultimate goal of conservative management is closure of the enteric fistula. In some cases, conservative therapy alone will result in spontaneous closure of enteric fistulas. Before the advent of total parenteral nutrition and percutaneous interventional radiologic techniques, the SCR with conservative management was reported to be 10%-26% (7,8). While there is still debate about its impact, parenteral hyperalimentation seems to have improved the SCR, with reported closure rates of 35%-71% (7,9-12). A further improvement in the SCR was reported after the introduction of percutaneous interventional techniques, with success rates ranging from 53% to 100% (5,6,1315). Enteric fistulas can be categorized by the output volume, the underlying cause, and the segmental location of the fistula. Many authors have suggested that these factors influence the likelihood of spontaneous fistula closure and the time required for accomplishing closure. We reviewed the outcome of conservative treatment in our 53 patients with respect to each of these factors. A comparison of our data with those reported in the literature follows.
• Fistula Output The SCR for enteric fistulas in our study was 63% for low-output fistulas and 52% for high-output fistulas. This difference is not statistically significant (P = .43 with use of the x2 method). However, the time required for spontaneous fistula closure was shorter for low-output fistulas (17 days) than for high-output fistulas (41 days). Also of note is the mortality in the two groups. The overall mortality rate in our study was 11% (six of 53 patients). Of the six pa-
356 • Journal of Vascular and Interventional Radiology May 1992
tients who died during conservative treatment, five had high-output fistulas and one had a low-output fistula. The influence of output volume on prognosis has diminished with recent advances in medical management. Before the advent of total parenteral nutrition, aggressive fluid and electrolyte replacement, and interventional radiology techniques for managing effluent, high-output volumes clearly affected outcome adversely. High-output fistulas were associated with a higher mortality and a lower spontaneous closure rate (16). More effective medical management has decreased the mortality and increased the SCR for both high- and low-output fistulas and diminished the significance of output volume on outcome. A review of the radiologic literature does not demonstrate a clear-cut difference in prognosis between the two groups. In a study of 17 postoperative enteric fistulas, Lambiase et al (15) noted a higher SCR with high-output fistulas than with low-output fistulas (71% overall, 83% with high-output vs 64% with low-output fistulas). McLean et al (5) reported a 100% SCR in 12 patients with high-output fistulas, while KerIan et al (6) recorded 84% of patients with low-output fistulas "improved" with medical therapy. A review of the surgical literature also suggests that output volume is no longer a significant prognostic factor. Many authors believe that output volume can usually be reduced and controlled using a regimen of bowel rest and adjunctive total parenteral nutrition, cimetadine, or somatostatin (7,17-19). A notable exception was reported by Sitges-Serra et al (12), who found a marked decrease in SCR of high-output upper gastrointestinal fistulas compared with low-output upper gastrointestinal fistulas (SCR = 54% vs 97%).
• Underlying Cause Most of the fistulas in our series (83%) were postoperative complications. Since the number of patients with other underlying causes was
small, no definitive conclusions regarding the influence of etiology on outcome can be drawn from our data. Of note, however, is the fact that the two fistulas which arose spontaneously as a result of malignancy did not close with conservative therapy, while the two fistulas that were secondary to trauma both healed spontaneously with nonoperative management. In the two patients with Crohn disease, one fistula healed with conservative measures and one did not. A review of the literature suggests that enteric fistulas resulting from intrinsic disease in the bowel wall frequently do not heal as well as fistulas that occur in the presence of normal bowel. Numerous reports have suggested, for example, that enteric fistulas which arise spontaneously in patients with Crohn disease are less likely to heal with conservative measures (1,7). Reported closure rates with use ofinterventional radiologic techniques in patients with enteric fistulas secondary to Crohn disease range from 0% to 40% (2022). Enteric fistulas that result from other forms of inflammatory disease seem to have better results with conservative therapy. Closure rates of up to 100% have been reported when diverticulitis or appendicitis is the underlying cause of the enteric fistula (23-25).
• Segmental Location We noted a high success rate for anastomotic and primary esophageal fistulas (SCR = 100%), intermediate success for duodenal and small bowel fistulas, (SCR = 60% and 67%, respectively), and low success with gastric and colonic fistulas (SCR = 25% and 10%, respectively). The number of patients with gastric fistula was small (n = 4), and the success of conservative treatment may have been adversely affected by the high mortality in this group (two offour patients died). The number of patients with colonic fistulas was greater (n = 10), and the statistics are more compelling: only one fistula in 10 healed with conservative therapy. Our poor success in closing colonic fistulas is in
keeping with our general belief that colonic fistulas can be difficult to manage. The large particulate size of the colonic effluent generally requires large-bore drainage catheters and frequent tube manipulations to maintain patency. Additionally, total parenteral nutrition is often necessary when nutritional supplementation is required. Our findings with regard to colonic fistulas are at odds with other published reports. Many surgical reviews have suggested that colonic fistulas are more successfully treated than are upper gastrointestinal tract or small bowel fistulas (1,8,11,12). A wide differential in success with colonic versus upper gastrointestinal fistulas was reported by AlexanderWilliams and Irving (1) (SCR = 75% vs 25%, respectively), while a smaller discrepancy was noted by SitgesSerra et al (12) (SCR = 92% vs 85%, respectively). In the only study in the radiologic literature with a significant number of colonic fistulas, Lambiase et al (15) did not note a significant difference in outcome based on segmental fistula location. We found only one study in which a low success with colonic fistulas was reported. Papanicolaou et al (13) recorded a lower rate of spontaneous closure for colonic fistulas than for small bowel fistulas (SCR = 33% vs 82%, respectively); however, their study included only three colonic fistulas.
• Duration of Therapy In our study, spontaneous closure occurred within the first month of therapy in 62%, within the second month in 21%, and within the third month in 17%. This temporal distribution of successful management is consistent with other radiologic and surgical reports which suggest that most spontaneous fistula closures occur in the first month and almost all within 3 months. The temporal distribution of surgical intervention in our study is also of note. Six patients underwent surgery in the first 2 weeks for resolution of acute management problems. These data are consistent with most surgical reviews
LaBerge et al • 357 Volume 3 Number 2
which suggest that a small percentage of patients will require early operations for resolution of acute problems, such as persistent sepsis, distal bowel obstruction, or total bowel transection. It can be difficult to decide when to operate on patients whose fistulas have been initially controlled with conservative techniques. Many authors recommend a 6-week to 2-month trial of conservative therapy, since most spontaneous closures will occur during this time (1,7). The decision whether to operate on patients in whom 2 months of conservative therapy has failed must be based on individual factors, such as operative risk and projected operative recurrence rate, balanced against the inconvenience and cost of further conservative therapy. One can expect that a small but significant percentage of fistulas will heal spontaneously after an initial 2-month trial period. Interestingly, in our study, more than half of the patients treated for longer than 2 months (five of nine patients) were cured without an operation. In summary, conservative management employing interventional radiologic techniques resulted in spontaneous fistula closure in 57% of patients. The output volume of the fistula was not a significant determinant of the success of conservative therapy. The duration of therapy required for spontaneous closure (less than 2 months in 83%) is consistent with that in other published series. Our poorer results with colonic fistulas (SCR = 10%) is at odds with other reports but is supported by our general impression that colonic fistulas can be difficult to manage. References 1. Alexander-Williams J, Irving M. Intestinal fistulas. John Wright & Sons, Bristol, England, 1982. 2. Flancbaum L, Nosher JL, Brolin RG. Percutaneous catheter drainage of abdominal abscesses associated with perforated viscus. Am Surg 1990; 56:52-56. 3. Gerzof SG, Robbins AH, Johnson WC, Birkett DH, Nabseth DC. Per-
4.
5.
6.
7.
8.
9.
10. 11.
12.
13.
14.
15.
16. 17.
cutaneous catheter drainage of abdominal abscesses. N Engl J Med 1981; 305:653-657. Aguirre A, Fischer JE, Welch CEo The role of surgery and hyperalimentation in therapy of gastrointestinal-cutaneous fistulae. Ann Surg 1974; 180:393-401. McLean GK, Mackie JA, Freiman DB, Ring EJ. Enterocutaneous fistulae: interventional radiologic management. AJR 1982; 138:615-619. Kerlan RK, Jeffrey B, Pogany AC, Ring EJ. Abdominal abscess with low-output fistula: successful percutaneous drainage. Radiology 1985; 155:73-75. Reber HA, Roberts C, Way LW, Dunphy JE. Management of external gastrointestinal fistulas. Ann Surg 1978; 188:460-467. Soeter PB, Ebeid AM, Fisher JE. Review of 404 patients with gastrointestinal fistulas. Ann Surg 1979; 190:189-201. Thomas RJ. The response ofpatients with fistulas of the gastrointestinal tract to parenteral nutrition. Surg Gynecol Obstet 1981; 153:77-80. Allardyce DB. Management of small bowel fistulas. Am J Surg 1983; 145:593-595. MacFadyen BV, Dudrick SJ, Ruberg RL. Management of gastrointestinal fistulas with parenteral hyperalimentation. Surgery 1973; 74:100105. Sitges-Serra A, Jaurrieta E, SitgesCreus A. Management of postoperative enterocutaneous fistulas: the roles of parenteral nutrition and surgery. Br J Surg 1982; 69:147150. Papanicolaou N, Mueller PR, Ferrucci JT, et al. Abscess-fistula association: radiologic recognition and percutaneous management. AJR 1984; 143:811-815. Lang EK, Springer RM, Glorioso LW, Cammarata CA. Abdominal abscess drainage under radiologic guidance: causes of failure. Radiology 1986; 159:329-336. Lambiase RE, Cronan JJ, Dorfman GS, Paolella LP, Haas RA. Postoperative abscesses with enteric communication: percutaneous treatment. Radiology 1989; 171:497-500. Fischer JE. The management of high-output intestinal fistulas. Adv Surg 1975; 9:139-176. Hollender LF, Meyer C, Avet D,
18.
19.
20.
21.
22.
23.
24.
25.
Zeyer B. Postoperative fistulas of the small intestine: therapeutic principles. World J Surg 1983; 7:474-480. Geerdsen JP, Pedersen VM, Kjaergard HK. Small bowel fistulas treated with somatostatin: preliminary results. Surgery 1986; 100: 811-814. Mulvihill S, Pappas TN, Passaro E, Debas HT. The use of somatostatin and its analogs in the treatment of surgical disorders. Surgery 1986; 100:467-475. Lambiase RE, Cronan JJ, Dorfman GS, Paolella LP, Haas RA. Percutaneous drainage of abscesses in patients with Crohn's disease. AJR 1988; 150:1043-1045. Safrit HD, Mauro MA, Jaques PF. Percutaneous abscess drainage in Crohn disease. AJR 1987; 148:859862. Casola G, vanSonnenberg E, Neff CC, Saba RM, Withers C, Emarine CWo Abscesses in Crohn disease: percutaneous drainage. Radiology 1987; 163:19-22. Neff CC, vanSonnenberg E, Casola G, et al. Diverticular abscesses: percutaneous drainage. Radiology 1987; 163:15-18. Nunez D, Huber JS, Yrizarry JM, Mendez G, Russell E. Nonsurgical drainage of appendiceal abscesses. AJR 1986; 146:587-589. Jeffrey RB, Federle MP, Tolentino CS. Periappendiceal inflammatory masses: CT-directed management and clinical outcome in 70 patients. Radiology 1988; 167:13-16.
Gastrointestinal Intervention
Nonoperative Treatment of Enteric Fistulas: Results in 53 Patients l Jeanne M. laBerge, MD Robert K. KerJan, Jr, MD Roy L. Gordon, MD Ernest J. Ring, MD
Index terms: Abscess, percutaneous drainage • Fistula, gastrointestinal • Gastrointestinal tract, interventional procedure JVIR 1992; 3:353-357 Abbreviation: sure rate
SCR = spontaneous clo-
Fifty-three patients with enterocutaneous fistulas were treated nonoperatively. While most of the fistulas resulted as a complication of surgery, other causes included inflammatory disease, trauma, and malignancy. Fistulas originated from the esophagus (n = 4), stomach (n = 4), duodenum (n = 5), small bowel (n = 27), colonic-small bowel anastomosis (n = 2), colon (n = 10), and Hartmann pouch (n = 1), and included both high-output (n = 29) and low-output (n = 24) types. Nonoperative management resulted in spontaneous fistula closure in 57%. Most of the spontaneous closures (83%) occurred within 2 months of treatment. Spontaneous closure rates were similar for low-output and for high-output fistulas (52% vs 63%), but the treatment duration was longer for high-output than for low-output fistulas (mean treatment duration, 17 days vs 41 days). Colonic fistulas were particularly difficult to manage and closed spontaneously in only 10% of patients (one of 10).
T HE treatment of patients with sig-
1 From the Department of Radiology, Box 0628, University of California, San Francisco, 505 Parnassus, San Francisco, CA 94143-0628. Received June 3, 1991; revision requested August 24; revision received December 6; accepted December 14. Address reprint requests to J.M.L.
" SCVIR, 1992
nificant gastrointestinal fistulas is generally a two-staged process. Initial management efforts center on controlling bowel effluent, clearing associated abscesses, and improving the patient's nutritional status, while the ultimate therapeutic goal is closure of the enteric fistula. Recent advances in nonoperative therapy have improved the success of initial management. Aggressive nutritional support and fluid and electrolyte management have lowered associated mortality (1), and interventional radiologic techniques have aided efforts to control sepsis and eradicate associated abscesses (2,3). The value of nonoperative measures in accomplishing the second task, fistula closure, is less clearly demonstrated in the literature. Reported rates of spontaneous closure have varied substantially (29%-100%) (4,5) and seem dependent on the type of nonoperative management employed as well as the location, underlying cause, and output volume of the fistulas. We report our experience in the nonoperative treatment of gastro-
intestinal fistulas with use of interventional radiologic techniques. The focus of our study is the rate of spontaneous closure. We analyze the success of nonoperative management with respect to the underlying cause, location, and output volume of the fistulas.
PATIENTS AND METHODS A retrospective review was performed in all patients referred to Interventional Radiology for management of enterocutaneous fistulas at the University of California, San Francisco and the Scripps' Memorial Hospital of La Jolla between June 1985 and January 1990. Patients who had undergone percutaneous abscess drainage and were subsequently proved to have occult enteric fistulas were also included. Patients with incomplete medical records or insufficient follow-up were excluded. Fifty-three patients, including 28 men and 25 women, whose ages ranged from 29 to 84 years, qualified for the study. The locations of the
353
354 • Journal of Vascular and Interventional Radiology May 1992
enteric communications are listed in Table 1. The causes of these leaks are summarized in Table 2. Interventional radiologic management was performed in accordance with previously established principles (5,6). These principles included control of the bowel effluent as well as evacuation of all associated abscesses. Abdominal fluid collections were drained with use of commercially available sump tubes (RMS; Cook, Bloomington, Ind) and gravity drainage tubes (Rob-Nel, Sherwood Medical, St Louis, Mo; Malecot, DowCorning, Upland, Mich). The diameter of the drainage catheters varied from 12 F to 30 F. Catheters were either placed percutaneously under computed tomographic or ultrasound guidance, or by catheterization of an existing percutaneous tract. An immediate sinogram was obtained to confirm proper catheter position. Routine postdrainage sinograms were obtained in all patients 3-6 days following the initial drainage. When a fistula was discovered, a sump drainage catheter was positioned adjacent to the bowel to control the fistula effluent and a catheter was placed in the dependent portion of any associated abscesses. The clinical status of the patient and the output from the drainage catheters were recorded by the nursing staff and were monitored daily by the interventional radiology staff. Sump catheters were irrigated on daily rounds; malfunctioning catheters were repositioned or replaced in the fluoroscopy suite. Nutritional support included enteral and parenteral nutrition consistent with current practice (1). However, the form of nutritional supplementation was not considered as an independent factor in this study. Low-output fistulas were defined as having an output of less than 200 mL per day following the initial drainage procedure (6,7). The remainder were considered high-output fistulas. A successful outcome of the interventional radiologic procedure was defined as healing of the fistula and all associated abscesses without any type of open surgical interven-
Table 1 Distribution of Enteric Fistulas in 53 Patients
Output Fistula Location (n = 53) Esophagus (n = 4) Stomach (n = 4) Duodenum (n = 5) Small intestine (n = 27) Small bowel-colon anastomosis (n = 2) Colon (n = 10) Hartmann anastomotic pouch (n = 1)
tion. All other outcomes were considered failures.
RESULTS The results of nonoperative therapy in 53 patients are summarized in Table 3. Conservative management resulted in spontaneous fistula closure in 57%. Results were similar for low-output (spontaneous closure rate [SCR] = 63%) and high-output (SCR = 52%) fistulas. When sorted by underlying cause, results were best for posttraumatic fistulas (SCR = 100%) and poorest for fistulas secondary to malignancy (SCR = 0%). Results with postsurgical and postinflammatory fistulas were similar (SCR = 57% vs 60%). When sorted by location of fistula, the poorest results occurred with gastric and colonic fistulas (SCR = 25% and 10%, respectively). Of the 10 colonic fistulas, eight were low-output and two were highoutput; three were secondary to inflammatory disease, one was secondary to malignancy, and six developed postoperatively. Four of 10 patients with colonic fistulas underwent surgery within 2 weeks. Colonic fistulas were generally treated with largebore sump tubes (16-30 F). Catheter obstruction from large particulate debris was treated by means of daily catheter irrigation and fluoroscopic catheter manipulation when necessary. Of the four gastric fistulas, three were postoperative and one was sec-
Low
High
(n = 24)
(n = 29)
3
2 4 2
10
17
1
1 2
2
o 8 1
o
ondary to malignancy; all four were high-output fistulas. Two of the four patients with gastric fistulas died during nonoperative treatment, and two patients underwent surgery within 1-2 months. Information regarding the exact duration of therapy was available for 49 of 53 patients. The data are summarized in Table 4. Treatment duration varied from 5 to 500 days. In those patients in whom treatment was ultimately successful (29 of 49), the duration of therapy varied from 7 to 63 days (mean = 28 days). Therapy was successful within 1 month in 18, between 1 and 2 months in six, and after a period longer than 2 months in five of 29. The mean duration of successful therapy in patients with low-output fistulas was 17 days, while the mean duration of successful therapy in patients with high-output fistulas was 41 days. In those patients in whom conservative therapy failed (20 of 49 patients), the duration of treatment ranged from 5 to 500 days (mean = 31 days, excluding one patient treated for 500 days). Among those patients in whom conservative therapy failed and in whom the treatment duration was known, four died and 16 underwent a surgical procedure. Surgery was performed within less than 1 month in six patients, at 1-2 months in six, and after 2 months in four. Six patients died during the period of nonoperative management (mortality = 11%); two died in the first 2 weeks of treatment and four died after more than 1 month of conserva-
LaBerge et al • 355 Volume 3 Number 2
Table 2 Underlying Cause of Enteric Fistulas in 53 Patients Output Low
High
Cause (n = 53)
(n = 24)
(n = 29)
Postsurgical complication (n = 44) Inflammatory disease* (n = 5) Trauma (n = 2) Malignancy (n = 2)
18
26 1
4 2 0
o 2
* Includes peptic ulcer disease, Crohn disease, and diverticulitis.
Table 3 Results of Nonoperative Management in 53 Patients Breakdown of Results Success Failure Output 9 (37) Low(n = 24) 15 (63) 14 (48) 15 (52) High (n = 29) Etiology 25 (57) Postsurgical (n = 44) 19 (43) Inflammatory (n = 5) 3 (60) 2 (40) 0(0) 2 (100) Trauma (n = 2) 0(0) Malignancy (n = 2) 2 (100) Location Esophagus (n = 24) 4 (l00) 0 (0) Stomach (n = 4) 1 (25) 3 (75) Duodenum (n = 5) 3 (60) 2 (40) Small bowel (n = 27) 18 (67) 9 (33) Colonic anastomosis (n = 2) 2 (100) 0 (0) Colon (n = 10) 1 (10) 9 (90) Hartmann pouch (n = 1) 1 (l00) 0 (0) Note.-Numbers in parentheses are percentages. Overall, nonoperative management was considered a success in 30 patients (57%) and a failure in 23 patients (43%).
Table 4 Duration of Nonoperative Therapy in 49 Patients No. of Patients with Successful Duration of Treatment Outcomes (n = 29) < 1 month (n = 26) 18 1-2 months (n = 14) 6 >2 months (n = 9) 5
No. of Patients with Unsuccessful Outcomes (n = 20)
8 8 4
Note.-Data available for 49 of the 53 patients treated.
tive management. All of the patients who died had an underlying malignancy; five of six developed fistulas as a postoperative complication. These fistulas were located in the stomach (n = 2), duodenum (n = 1) and small bowel (n = 3).
DISCUSSION Nonoperative therapy is the preferred initial approach to the treatment of patients with enteric fistulas. Surgery, because of its attendant
risks and technical difficulty, is usually reserved for patients who fail to heal with conservative management. The initial goals of conservative management are control of bowel effluent, drainage of associated abscesses, and improvement in the patient's fluid, electrolyte, and nutritional status. The ultimate goal of conservative management is closure of the enteric fistula. In some cases, conservative therapy alone will result in spontaneous closure of enteric fistulas. Before the advent of total parenteral nutrition and percutaneous interventional radiologic techniques, the SCR with conservative management was reported to be 10%-26% (7,8). While there is still debate about its impact, parenteral hyperalimentation seems to have improved the SCR, with reported closure rates of 35%-71% (7,9-12). A further improvement in the SCR was reported after the introduction of percutaneous interventional techniques, with success rates ranging from 53% to 100% (5,6,1315). Enteric fistulas can be categorized by the output volume, the underlying cause, and the segmental location of the fistula. Many authors have suggested that these factors influence the likelihood of spontaneous fistula closure and the time required for accomplishing closure. We reviewed the outcome of conservative treatment in our 53 patients with respect to each of these factors. A comparison of our data with those reported in the literature follows.
• Fistula Output The SCR for enteric fistulas in our study was 63% for low-output fistulas and 52% for high-output fistulas. This difference is not statistically significant (P = .43 with use of the x2 method). However, the time required for spontaneous fistula closure was shorter for low-output fistulas (17 days) than for high-output fistulas (41 days). Also of note is the mortality in the two groups. The overall mortality rate in our study was 11% (six of 53 patients). Of the six pa-
356 • Journal of Vascular and Interventional Radiology May 1992
tients who died during conservative treatment, five had high-output fistulas and one had a low-output fistula. The influence of output volume on prognosis has diminished with recent advances in medical management. Before the advent of total parenteral nutrition, aggressive fluid and electrolyte replacement, and interventional radiology techniques for managing effluent, high-output volumes clearly affected outcome adversely. High-output fistulas were associated with a higher mortality and a lower spontaneous closure rate (16). More effective medical management has decreased the mortality and increased the SCR for both high- and low-output fistulas and diminished the significance of output volume on outcome. A review of the radiologic literature does not demonstrate a clear-cut difference in prognosis between the two groups. In a study of 17 postoperative enteric fistulas, Lambiase et al (15) noted a higher SCR with high-output fistulas than with low-output fistulas (71% overall, 83% with high-output vs 64% with low-output fistulas). McLean et al (5) reported a 100% SCR in 12 patients with high-output fistulas, while KerIan et al (6) recorded 84% of patients with low-output fistulas "improved" with medical therapy. A review of the surgical literature also suggests that output volume is no longer a significant prognostic factor. Many authors believe that output volume can usually be reduced and controlled using a regimen of bowel rest and adjunctive total parenteral nutrition, cimetadine, or somatostatin (7,17-19). A notable exception was reported by Sitges-Serra et al (12), who found a marked decrease in SCR of high-output upper gastrointestinal fistulas compared with low-output upper gastrointestinal fistulas (SCR = 54% vs 97%).
• Underlying Cause Most of the fistulas in our series (83%) were postoperative complications. Since the number of patients with other underlying causes was
small, no definitive conclusions regarding the influence of etiology on outcome can be drawn from our data. Of note, however, is the fact that the two fistulas which arose spontaneously as a result of malignancy did not close with conservative therapy, while the two fistulas that were secondary to trauma both healed spontaneously with nonoperative management. In the two patients with Crohn disease, one fistula healed with conservative measures and one did not. A review of the literature suggests that enteric fistulas resulting from intrinsic disease in the bowel wall frequently do not heal as well as fistulas that occur in the presence of normal bowel. Numerous reports have suggested, for example, that enteric fistulas which arise spontaneously in patients with Crohn disease are less likely to heal with conservative measures (1,7). Reported closure rates with use ofinterventional radiologic techniques in patients with enteric fistulas secondary to Crohn disease range from 0% to 40% (2022). Enteric fistulas that result from other forms of inflammatory disease seem to have better results with conservative therapy. Closure rates of up to 100% have been reported when diverticulitis or appendicitis is the underlying cause of the enteric fistula (23-25).
• Segmental Location We noted a high success rate for anastomotic and primary esophageal fistulas (SCR = 100%), intermediate success for duodenal and small bowel fistulas, (SCR = 60% and 67%, respectively), and low success with gastric and colonic fistulas (SCR = 25% and 10%, respectively). The number of patients with gastric fistula was small (n = 4), and the success of conservative treatment may have been adversely affected by the high mortality in this group (two offour patients died). The number of patients with colonic fistulas was greater (n = 10), and the statistics are more compelling: only one fistula in 10 healed with conservative therapy. Our poor success in closing colonic fistulas is in
keeping with our general belief that colonic fistulas can be difficult to manage. The large particulate size of the colonic effluent generally requires large-bore drainage catheters and frequent tube manipulations to maintain patency. Additionally, total parenteral nutrition is often necessary when nutritional supplementation is required. Our findings with regard to colonic fistulas are at odds with other published reports. Many surgical reviews have suggested that colonic fistulas are more successfully treated than are upper gastrointestinal tract or small bowel fistulas (1,8,11,12). A wide differential in success with colonic versus upper gastrointestinal fistulas was reported by AlexanderWilliams and Irving (1) (SCR = 75% vs 25%, respectively), while a smaller discrepancy was noted by SitgesSerra et al (12) (SCR = 92% vs 85%, respectively). In the only study in the radiologic literature with a significant number of colonic fistulas, Lambiase et al (15) did not note a significant difference in outcome based on segmental fistula location. We found only one study in which a low success with colonic fistulas was reported. Papanicolaou et al (13) recorded a lower rate of spontaneous closure for colonic fistulas than for small bowel fistulas (SCR = 33% vs 82%, respectively); however, their study included only three colonic fistulas.
• Duration of Therapy In our study, spontaneous closure occurred within the first month of therapy in 62%, within the second month in 21%, and within the third month in 17%. This temporal distribution of successful management is consistent with other radiologic and surgical reports which suggest that most spontaneous fistula closures occur in the first month and almost all within 3 months. The temporal distribution of surgical intervention in our study is also of note. Six patients underwent surgery in the first 2 weeks for resolution of acute management problems. These data are consistent with most surgical reviews
LaBerge et al • 357 Volume 3 Number 2
which suggest that a small percentage of patients will require early operations for resolution of acute problems, such as persistent sepsis, distal bowel obstruction, or total bowel transection. It can be difficult to decide when to operate on patients whose fistulas have been initially controlled with conservative techniques. Many authors recommend a 6-week to 2-month trial of conservative therapy, since most spontaneous closures will occur during this time (1,7). The decision whether to operate on patients in whom 2 months of conservative therapy has failed must be based on individual factors, such as operative risk and projected operative recurrence rate, balanced against the inconvenience and cost of further conservative therapy. One can expect that a small but significant percentage of fistulas will heal spontaneously after an initial 2-month trial period. Interestingly, in our study, more than half of the patients treated for longer than 2 months (five of nine patients) were cured without an operation. In summary, conservative management employing interventional radiologic techniques resulted in spontaneous fistula closure in 57% of patients. The output volume of the fistula was not a significant determinant of the success of conservative therapy. The duration of therapy required for spontaneous closure (less than 2 months in 83%) is consistent with that in other published series. Our poorer results with colonic fistulas (SCR = 10%) is at odds with other reports but is supported by our general impression that colonic fistulas can be difficult to manage. References 1. Alexander-Williams J, Irving M. Intestinal fistulas. John Wright & Sons, Bristol, England, 1982. 2. Flancbaum L, Nosher JL, Brolin RG. Percutaneous catheter drainage of abdominal abscesses associated with perforated viscus. Am Surg 1990; 56:52-56. 3. Gerzof SG, Robbins AH, Johnson WC, Birkett DH, Nabseth DC. Per-
4.
5.
6.
7.
8.
9.
10. 11.
12.
13.
14.
15.
16. 17.
cutaneous catheter drainage of abdominal abscesses. N Engl J Med 1981; 305:653-657. Aguirre A, Fischer JE, Welch CEo The role of surgery and hyperalimentation in therapy of gastrointestinal-cutaneous fistulae. Ann Surg 1974; 180:393-401. McLean GK, Mackie JA, Freiman DB, Ring EJ. Enterocutaneous fistulae: interventional radiologic management. AJR 1982; 138:615-619. Kerlan RK, Jeffrey B, Pogany AC, Ring EJ. Abdominal abscess with low-output fistula: successful percutaneous drainage. Radiology 1985; 155:73-75. Reber HA, Roberts C, Way LW, Dunphy JE. Management of external gastrointestinal fistulas. Ann Surg 1978; 188:460-467. Soeter PB, Ebeid AM, Fisher JE. Review of 404 patients with gastrointestinal fistulas. Ann Surg 1979; 190:189-201. Thomas RJ. The response ofpatients with fistulas of the gastrointestinal tract to parenteral nutrition. Surg Gynecol Obstet 1981; 153:77-80. Allardyce DB. Management of small bowel fistulas. Am J Surg 1983; 145:593-595. MacFadyen BV, Dudrick SJ, Ruberg RL. Management of gastrointestinal fistulas with parenteral hyperalimentation. Surgery 1973; 74:100105. Sitges-Serra A, Jaurrieta E, SitgesCreus A. Management of postoperative enterocutaneous fistulas: the roles of parenteral nutrition and surgery. Br J Surg 1982; 69:147150. Papanicolaou N, Mueller PR, Ferrucci JT, et al. Abscess-fistula association: radiologic recognition and percutaneous management. AJR 1984; 143:811-815. Lang EK, Springer RM, Glorioso LW, Cammarata CA. Abdominal abscess drainage under radiologic guidance: causes of failure. Radiology 1986; 159:329-336. Lambiase RE, Cronan JJ, Dorfman GS, Paolella LP, Haas RA. Postoperative abscesses with enteric communication: percutaneous treatment. Radiology 1989; 171:497-500. Fischer JE. The management of high-output intestinal fistulas. Adv Surg 1975; 9:139-176. Hollender LF, Meyer C, Avet D,
18.
19.
20.
21.
22.
23.
24.
25.
Zeyer B. Postoperative fistulas of the small intestine: therapeutic principles. World J Surg 1983; 7:474-480. Geerdsen JP, Pedersen VM, Kjaergard HK. Small bowel fistulas treated with somatostatin: preliminary results. Surgery 1986; 100: 811-814. Mulvihill S, Pappas TN, Passaro E, Debas HT. The use of somatostatin and its analogs in the treatment of surgical disorders. Surgery 1986; 100:467-475. Lambiase RE, Cronan JJ, Dorfman GS, Paolella LP, Haas RA. Percutaneous drainage of abscesses in patients with Crohn's disease. AJR 1988; 150:1043-1045. Safrit HD, Mauro MA, Jaques PF. Percutaneous abscess drainage in Crohn disease. AJR 1987; 148:859862. Casola G, vanSonnenberg E, Neff CC, Saba RM, Withers C, Emarine CWo Abscesses in Crohn disease: percutaneous drainage. Radiology 1987; 163:19-22. Neff CC, vanSonnenberg E, Casola G, et al. Diverticular abscesses: percutaneous drainage. Radiology 1987; 163:15-18. Nunez D, Huber JS, Yrizarry JM, Mendez G, Russell E. Nonsurgical drainage of appendiceal abscesses. AJR 1986; 146:587-589. Jeffrey RB, Federle MP, Tolentino CS. Periappendiceal inflammatory masses: CT-directed management and clinical outcome in 70 patients. Radiology 1988; 167:13-16.