Bleeding from Peristomal Varices: Perspectives on Prevention and Treatment Claudio Fucini, M.D., Bruce G. Wolff, M.D., Roger R. Dozois, M.D. From the Section of Colon a n d Rectal Surgery, Mayo Clinic a n d Mayo Foundation, Rochester, Minnesota
Peristomal variceal bleeding is a serious complication in patients with chronic liver disease undergoing colon surgery with a stoma. Our aim was to examine the morbidity of bleeding for peristomal, perianastomotic, and esophageal varices in a group of patients with chronic liver disease who underwent colectomy at the Mayo Clinic between 1970 and 1988. Morbidity was evaluated in terms of the number of major bleeding episodes and the number of units of blood transfused. The treatment of bleeding was also evaluated. One hundred seventeen patients (74 males and 43 females) aged 11-78 years were studied. Sixty-two patients (53 percent) had a permanent stoma, while 55 patients (47 percent) had a colonic resection and anastomosis. Sixty-seven patients (62 percent) had chronic ulcerative colitis and primary sclerosing cholangitis. In the stoma group, bleeding appeared from stomal and/or esophageal varices in 19 patients (31 percent), while, in the non-stoma group, bleeding exclusively from the esophageal varices occurred in eight patients (15 percent). Perianastomotic variceal bleeding was never observed. The 5-year cumulative probabilities of one major bleed occurring from gastrointestinal varices appeared to be similar between the two groups. Patients who bled from peristomal varices with or without esophageal bleeding (n --- 17) rebled more frequently (6.5 --. 5.5 vs. 3 --. 1.6; P < 0.05) and were transfused more often (14.9 --- 12.3 vs. 7.5 4- 4.1; P < 0.05) than patients who bled exclusively from esophageal varices (n = 10). No difference was found in the incidence of recurrent bleeding and the number of units of blood transfused between patients who bled exclusively from peristomal varices (n - 10) and those who bled from both peristomal and esophageal varices (n 7). Medical and local measures were more effective in controlling esophageal bleeding than in controlling peristomal bleeding. Therefore, patients with chronic liver disease who must undergo colectomy should have a distal anastomosis rather than a terminal stoma. [Key words: Intestinal stomas; Portal hypertension; Colectomy; Cirrhosis] Fucini C, Wolff BG, Dozois RR. Bleeding from peristomal varices: perspectives on prevention and treatment. Dis Colon Rectum 1991 ;34:1073-1078. Read in part at the meeting of The American Society of Colon and Rectal Surgeons, St. Louis, Missouri, April 27 to May 4, 1990. Address reprint requests to Dr. Dozois: Mayo Clinic, 200 First Street Southwest, Rochester, Minnesota 55905. 1073
l e e d i n g from varices in the abdominal wall s u r r o u n d i n g an intestinal stoma has b e e n rep o r t e d in at least 62 patients with portal hypertension. 1-27 Patients with chronic liver disease u n d e r g o i n g abdominal surgery (e.g., a b d o m i n o perineal resection, p r o c t o c o l e c t o m y , and ileal conduit) and requiring an abdominal stoma are exp o s e d to the risk of adhesions b e t w e e n the mesentery and the abdominal wall with portosystemic shunting if portal h y p e r t e n s i o n develops. 28 Varices are primarily located at the peristomal mucocutan e o u s junction, w h e r e b l e e d i n g may originate. Muc o c u t a n e o u s varices are occasionally a continuation of submucosal varices of the e n d portion of the intestine.14, ,7 T h e y have also b e e n r e p o r t e d as prestomal, with b l e e d i n g occurring within the ileal lumen. 2~ The bleeding, w h i c h is often heavy, may be difficult to control, and the frequently observed episodes of h e m o r r h a g e are responsible for their high morbidity. 7 In a recent study from our institutior129 on primary sclerosing cholangitis (PSC), a cholestatic and cirrhogenic liver disease often associated with chronic ulcerative colitis (CUC), peristomal variceal b l e e d i n g was observed in 53 p e r c e n t of patients u n d e r g o i n g total p r o c t o c o l e c t o m y and fieostomy, whereas it did not occur in patients u n d e r g o i n g ileoanal anastomosis. Based on this observation, the authors r e c o m m e n d e d ileoanal or at least ileorectal anastomoses in PSC patients requiring c o l e c t o m y for CUC. Such an indication might be e x t e n d e d to patients with other types of cirrhosis and c o l o n disease in w h i c h the fashioning of a distal anastomosis (rectal or anal) is an alternative to a terminal stoma (i.e., low rectal cancers and polyposis of the colon). Because the occurrence of perianastomotic and esophageal variceal
B
1074
FUCINI E T A L
bleeding, which were also responsible for a high morbidity, has occasionally been reported by some authors, 7 we examined the frequency of bleeding from stomal or other gastrointestinal varices in advanced chronic liver disease and/or PSC patients who underwent a colonic resection with the fashioning of a distal anastomosis or the creation of a terminal stoma. Our aim was to verify whether the actual fashioning of a stoma would increase the risk of bleeding and morbidity. At the same time, we examined some clinical aspects of stomal variceal bleeding in an attempt to assess the value of therapeutic procedures used to control bleeding. PATIENTS A N D M E T H O D S A survey was conducted of portal hypertension, PSC, and/or other biopsy-proven advanced chronic liver disease patients seen at the Mayo Clinic between January 1970 and June 1988 who had had a distal anastomosis (ileorectal, ileoanal, colorectal, or coloanal) or a permanent intestinal stoma at the time of colonic resection. Patients who died in the postoperative period and those followed for less than 4 months were excluded. The records of 117 patients (74 men and 43 women) with an age range of 11 to 78 years were reviewed. The following information was recorded: diagnosis of intestinal and liver disease at the time of operation; type of intestinal operation; occurrence, site, and number of bleeding episodes from gastrointestinal varices which prompted blood transfusions; number of units of blood transfused; values of liver function tests at the time of the first episode of bleeding; types of therapy used to control the hemorrhage; and outcome of the most recent follow-up. Analysis of the risk of variceal bleeding in patients with a stoma vs. those without a stoma was carried out, and the morbidity of peristomal vs. other types of intestinal variceal bleeding was evaluated. A separate analysis was made of all 23 registered patients who had had one or more episodes of stomal bleeding requiring blood transfusions in the same period. Twenty had a terminal ileostomy (two with a Kock pouch), and three had an end colostomy. Liver disease appeared in one of these patients after the colonic operation. In five patients, the initial colonic operation had been performed either before 1970 or outside the Mayo Clinic. Statistical analysis was performed with the unpaired Student's t-test and the •2 test.
Dis Colon Rectum, December 1991 RESULTS
Permanent abdominal stomas (40 Brooke ileostomies, 8 Kock pouches, and 14 colostomies) had been constructed in 62 patients (53 percent), while, in the other 55 patients (47 Percent), colonic resection with anastomosis was established (20 J-ileal pouch-anal anastomoses, 23 ileorectostomies, 1 ileoanal anastomosis without reservoir, and 11 colorectostomies). Twenty-five patients of the latter group received a temporary loop ileostomy or colostomy. In 24 patients, the loop ostomy was closed over a period ranging from 4 to 16 weeks. Because no stomal bleeding appeared in any of these patients, they were included in the non-stoma group. One patient whose loop ileostomy was left in place because of an ileoanal anastomosis failure was included in the stoma patient group. Seventy patients (60 percent) were diagnosed as having PSC, 25 of them after the introduction of endoscopic retrograde cholangiopancreatography (ERCP). Forty-seven patients (40 percent) had other forms of cirrhosis (e.g., La~nnec's, biliary, and postnecrotic). At the time of operation, 33 patients had features of portal hypertension such as splenomegaly and/or esophageal varices. Eighty-one patients (69 percent) had CUC, associated with colon cancer in 17 patients. The other 36 patients (31 percent) had other colonic diseases such as rectal cancer, Crohn's disease, diverticulitis, or polyposis. Sixty-seven patients (62 percent) had both CUC and PSC. The occurrence of bleeding and the general mortality observed in patients with or without a stoma are reported in Table 1. In the stoma group, the bleeding sites included the esophagus (two patients), the stomal area (10 patients), and both sites (seven patients). In the non-stoma group, bleeding originated from the esophagus (eight patients). No perianastomotic variceal bleeding was observed from either ileorectostomy or ileoanal anastomosis sites. PSC patients showed a significantly greater risk of bleeding than patients with other forms of cirrhosis (Table 2), even though they were equally represented in both groups (56 percent vs. 63 percent). The two groups were also similar in terms of patient age and number and stages of cancers (Table 1). The 5-year cumulative probabilities of one major bleed occurring from intestinal varices appeared
Vol. 34, No. 12
BLEEDING PERISTOMALVARICES
1075
Table 1.
Clinical Features and Frequency of Bleeding in Patients With or Without a Stoma Stoma
Clinical Data No. of patients Mean age -+-SD (and range) (yr)
Yes
No
62
55 42 + 15.2
44 _+ 15.1 (17-78) 20 (32) 61 _+ 39.2 (5-155) 19 (31) 19 (31)
No. of patients with cancer (and %) Mean + SD follow-up (and range) (months) No. of patients with variceal bleeding (and %) No. of deaths (and %)
(11-68)
14 (25) 32 ___28.8 (5-156)
8 (15) 9 (16)
Table 2.
Frequency of Gastrointestinal Variceal Bleeding in PSC vs. Non-PSC Patients Patients
Patient Population Total no. of patients Mean + SD follow-up (and range) (months) No. of patients who bled (and %)
With PSC
Without PSC
70 47 + 37.3 (5-155) 22 (31)*
47 47 _+ 38.2 (5-156) 5 (11)*
* Fisher's exact test (P = 0.007).
to be similar between the two groups (Fig. 1). The cumulative probabilities of survival were also similar (Fig. 2). However, when we evaluated the morbidity from peristomal v s . other variceal bleeding sites in terms of the number of rebleeding episodes and the number of units of blood transfused, we found that patients who bled from peristomal varices, regardless of esophageal bleeding, had a significantly higher number of recurrent hemorrhagic episodes
and units of blood transfused than did patients who bled exclusively from esophageal varices (Table 3). The values of liver function tests, such as serum albumin, bilirubin, prothrombin time, transaminase, alkaline phosphatase, and platelet count, were not significantly different between the two groups at the time of the first episode of bleeding. When the patients who bled exclusively from peristomal varices (n = 10) were compared with those who bled from both peristomal and esopha-
'1
'1
0,9
-1
O,II /
b,I
1
O,I t O.,Y
0,6 O,S
=
9
9,11
o,s o,+
0,3
0,3 0,2
0,1 0
i gD
i
, 4m
i
0
J
~0
I0
(aom=) flTOVJk
+
X~)Y
Figure 1. Five-year cumulative probability of a major variceal bleed in stoma vs. non-stoma patients.
i
i
,
+
i
I)
frFolLt.
+
1t'0)1' g T O M I
Figure 2. Five-year cumulative probability of survival in
stoma vs. non-stoma patients.
FUCINI E T A L
1076
Dis Colon Rectum, December 1991
Table 3.
Sites and Number of Major Bleeds and Number of Units of Blood Transfused Before a Major Operation Bleeding Site Patient Population
No. of patients Mean __.SD no. of bleeds (and range)
Mean _+ SD no. of units transfused (and range) Mean _+ SD follow-up (and range) (months)?
Esophagus 10 3 + 1.6" (1-5) 7.5 _ 4.1" (3-15) 25.7 _+ 30.2 (2-101)
Stoma With/Without Esophageal Bleeding 17 6.5 + 5.5* (2-21) 14.9 +_ 12.3"?
(5-49) 25.5 ___17.9 (0-58)
* P < 0.05. t From the first episode of bleeding.
geal varices (n = 7), the number of recurrent bleeds did not differ; nor did the number of units of blood transfused. It would appear that the primary medical or local measures used to control esophageal bleeding (esophageal balloon tamponage with or without vasopressin infusion or sclerotherapy with ethanolamine oleate with or without propanolol) were more effective in controlling the hemorrhage than were the medical or local methods (vasopressin infusion to oral propanolol, compression of varices, sclerotherapy, fulguration, transstomal ligation, laser coagulation, mucocutaneous disconnection, and stoma refashioning) employed to control peristomal bleeding. In a majority of these patients, several means were tried. Repeated episodes of major bleeding were the main indication for surgery in 2 of the 10 patients (20 percent) who bled from esophageal varices, i.e., the Sugiura procedure (one patient) and liver transplantation (one patient). Seven of the 17 patients (41 percent) who bled from peristomal varices underwent a major operation for the same reason (five liver transplantations, one splenorenal shunt, and one mesocaval shunt). In a comparable mean follow-up from the first episode of bleeding (29 vs. 31 months), mortality was similar between the patients who bled exclusively from esophageal varices and those who bled from peristomal varices (40 percent vs. 53 percent; P > 0.05). Based on the parameters assessed in 23 patients seen in our institution, we made the following observations. First, CUC was the most common colonic disease (60 percent of peristomal bleeders), and PSC was the most frequent liver disease (65 percent). The CUC-PSC association accounted for 13 (57 percent) of the patients who bled from stomal varices. Second, 18 of the patients (78 per-
cent) with stomal varices also had esophageal varices. Ten patients (43 percent) bled from esophageal varices after a peristomal bleed. Third, in the 17 stomal bleeders who were operated on after 1970 and who had liver disease at the time of the operation, the time that elapsed between the creation of the stoma and the bleeding from stomal varices varied from 6 to 133 months (mean, 47 months). Fourth, the therapeutic procedures used to stem peristomal bleeding fell into three main categories: medical or local surgery, portosystemic shunts, and liver transplantation. Thirteen patients had only medical and/or local surgery. Five patients had a portosystemic shunt (three side-to-side splenorenal shunts, one central splenorenal shunt, and one mesocaval shunt); in all of them, the shunt was performed after repeated attempts to stem the bleeding with local measures. Five patients with PSC had liver transplantation. Four of them had previously been treated with a trial of local procedures. One patient who underwent emergency liver transplantation at the time of the first episode of peristomal bleeding died postoperatively. A second patient who had a rejection after 1 year underwent retransplantation. Table 4 summarizes the results of the different treatments. Despite the failure of the shunt in one patient (obstruction owing to thrombosis), the operation provided better control of the bleeding (P < 0.05) and better survival (P < 0.05) than local treatment. Similar results were obtained in patients who underwent transplantation. In patients who had colon cancer, all deaths were attributable to liver disease. One of these patients with concomitant primary biliary cirrhosis and metastatic liver disease had undergone abdominoperineal excision for a rectal cancer.
BLEEDING PERISTOMALVARICES
Vol. 34, No. 12
1077
Table 4. Results of Different Therapies Types of Surgeries Patient Population No. of patients Mean _+ SD follow-up (and range) (months) Bleeding control (and %) Mortality (and %)
Local
Shunting
Transplantation
13 27 _+ 18.1 (2-63) 3 (23)1 10 (77)~:
5 24.8 ___19.9" (8-58) 4 (80):1: 1 (20)1:
5 22 + 16.6" (0-38) 4 (80) 1 (20)
* Data refer to follow-up from shunt or from transplantation. The mean _+ SD periods from first bleeding to shunting and to transplantation were 13.2 _+ 19.2 and 13.8 _+ 14 months, respectively. 1 The number refers to patients who were free from bleeding after a local procedure for at least 12 months; this corresponds to the shortest follow-up for a successful shunt. :l= Fisher's exact test (P < 0.05).
DISCUSSION Bleeding from peristomal varices is a serious event in the natural history of cirrhotic patients. Twenty-nine (37 percent) of the 78 patients who could be evaluated (55 from the medical literature and 23 from our own institution) died from liver disease within 3 years of the first episode of bleeding. Death from exsanguination occurred rarely. 13 This complication can be considered frequent, as it occurred in 17 (27 percent) of the 62 Mayo Clinic patients operated on between 1970 and 1988 who had advanced chronic liver disease as demonstrated at the time the stoma was fashioned and who were followed for a median period of 5 years. The complication is more frequent in CUC-PSC patients. In the same group of patients, Peck and Boyden 23 reported a bleeding incidence of 24 percent, as compared with 53 percent reported by Wiesner e t al. 29 To eliminate the morbidity derived from this type of bleeding, some authors, 29 have proposed that ileostomy be avoided in CUC-PSC patients requiting proctocolectomy and that ileoanal and/or ileorectal anastomosis, which are less susceptible to variceal formation and bleeding, be performed. Our retrospective study, which examined all patients with cirrhosis and colon disease with an abdominal stoma or with a distal anastomosis, confirmed that perianastomotic variceal bleeding is a very rare complication. However, it showed that the probability of major bleeding from esophageal varices occurring in cirrhotic patients who have had an anastomosis does not differ from that of bleeding from peristomal and/or esophageal varices in patients who have an abdominal stoma. Thus, peristomal bleeding did not act as a "safety valve" for esophageal bleeding, as suggested by some authors. 2
However, when we examined the morbidity derived from peristomal bleeding in comparison with esophageal bleeding as assessed in terms of the number of episodes of major rebleeding, the number of units of blood transfused, and the efficacy of medical or local methods to control bleeding, we found that esophageal bleeders had a lower morbidity than peristomal bleeders. This was also evident in patients who bled from both sites. Trivial trauma from the stomal appliance or from usual stoma care may have played a role in triggering bleeding. Thus, cirrhotic patients who undergo a colon resection, in whom the decision between fashioning a distal anastomosis and fashioning a stoma is not governed by the type or extent of the colonic disease, should preferably have an anastomosis. CUC-PSC patients represent the largest group of these patients. When major stomal bleeding occurred, the medical or local management was merely palliative, and recurrent hemorrhage was common. However, it has not yet been established whether repeated episodes of bleeding precipitate liver failure. 18' 21 For this reason, some authors 21 challenge the indication of shunting procedures which could control bleeding but do not increase the chances of survival and at the same time expose the patients to the risk of encephalopathy. Our experience does show that patients having shunts had a better chance of survival than those having local procedures alone. The only instance of encephalopathy observed was in the patient whose shunt failed. The complete absence of the colon in the majority of these patients may be relevant. Therefore, it seems reasonable to consider a definitive procedure and avoid a stoma unless the extent of the disease precludes such a course of
1078
FUCINI E T A L
action. H o w e v e r , the success of liver transplantation, which is b e c o m i n g an established therapy in PSC, 3~ raises other considerations. If a stoma is avoided, only local control of esophageal b l e e d i n g may be necessary until the patient n e e d s a transplantation. The p r e s e n c e of a previously constructed ileal pouch-anal anastomosis should not p r e c l u d e orthotopic liver transplantation, but special attention to m e s e n t e r i c vessels will be necessary.
16.
REFERENCES
18.
1. Resnick RH, Ishihara A, Chalmers TC, Schimmel EM, The Boston Inter-Hospital Liver Group. Controlled trial of colon bypass in chronic hepatic encephalopathy. Gastroenterology 1968;54:1057-69. 2. Eade MN, Williams JA, Cooke WT. Bleeding from an ileostomy caput medusa. Lancet 1969;2:1166-8. 3. Cameron AD, Fone DJ. Portal hypertension and bleeding ileal varices after colectomy and ileostomy for chronic ulcerative colitis. Gut 1970;11:755-9. 4. Hamyln AN, Lunzer MR, Morris JS, Puritz H, Dick R. Portal hypertension with varices in unusual sites. Lancet 1974;2:1531-4. 5. Foulkes J, Wallace DM. Haemorrhage from stomal varices in an ileal conduit. BrJ Urol 1975;47:630. 6. Graeber GM, Ratner MH, Ackerman NB. Massive hemorrhage from ileostomy and colostomy stomas due to mucocutaneous varices in patients with coexisting cirrhosis. Surgery 1976;79:107-10. 7. Adson MA, Fulton RE. The ileal stoma and portal hypertension: an uncommon site of variceal bleeding. Arch Surg 1977;112:501-4. 8. Crooks KK, Hensle TW, Heney NM, Waltman A, Irwin RJ Jr. Ileal conduit hemorrhage secondary to portal hypertension. Urology 1978;12:689-93. 9. Firlit RS, Firlit CF, Canning J. Exsanguinating hemorrhage from urinary ileal conduit in patient with portal hypertension. Urology 1978;12:710-1. 10. Finemore RG. Repeated haemorrhage from a terminal colostomy due to mucocutaneous varices with coexisting hepatic metastatic rectal adenocarcinoma: a case report. BrJ Surg 1979;66:806. 11. Federle M, Clark RA. Mesenteric varices: a source of mesosystemic shunts and gastrointestinal hemorrhage. Gastrointest Radiol 1979;4:331-7. 12. Goldstein WZ, Edoga J, Crystal R. Management of colostomal hemorrhage resulting from portal hypertension. Dis Colon Rectum 1980;23:86-90. 13. Ackerman NB, Graeber GM, Fey J. Enterostomal varices secondary to portal hypertension: progression of disease in conservatively managed cases. Arch Surg 1980;115:1454-5. 14. Ricci RL, Lee KR, Greenberger NJ. Chronic gastrointestinal bleeding from ileal varices after total proc-
15.
17.
19.
20.
21.
22. 23.
24.
25.
26.
27. 28.
29.
30.
Dis Colon Rectum, December 1991 tocolectomy for ulcerative colitis: correction by mesocaval shunt. Gastroenterology 1980;78:1053-8. Eckhauser FE, Sonda LP, Strodel WE, Edgcomb LP, Turcott JG. Parastomal ileal conduit hemorrhage and portal hypertension. Ann Surg 1980;192:620-4. Watkins RM. Variceal haemorrhage from a colostomy due to portal hypertension secondary to intrahepatic metastases from rectal carcinoma. BMJ 1981;282: 189-90. Cooper MJ, Mackie CR, Dhorajiwala J, Baker AL, Moossa AR. Hemorrhage from ileal varices after total proctocolectomy. AmJ Surg 1981;141:178-9. Hollands MJ. Parastomal haemorrhage from an ileal conduit secondary to portal hypertension. Br J Surg 1982;69:675. Perreault JP, Laroze M, Faucher R, Paquin JM, Mauffette F. H~morragie d'un conduit ileal secondaire une hypertension portale. J Urol (Paris) 1982;88: 47-9. Goldstein MB, Brandt LJ, Bernstein LH, Sprayragen S. Hemorrhage from ileal varices: a delayed complication after total proctocolectomy in a patient with ulcerative colitis and cirrhosis. Am J Gastroenterol 1983;78:351-4. Grundfest-Broniatowski S, Fazio V. Conservative treatment of bleeding stomal varices. Arch Surg 1983;118:981-5. Johnson AG, Simms JM. Correspondecne. Br J Surg 1983;70:187-90. PeckJJ, BoydenAM. Exigent ileostomyhemorrhage: a complication of proctocolectomy in patients with chronic ulcerative colitis and primary sclerosing cholangitis. Am J Surg 1985;150:153-8. Wang MM, McGrew W, Dunn GD. Variceal bleeding from an ileostomy stoma. South Med J 1985;78: 733-7. Morgan TR, Feldshon SD, Tripp MR. Recurrent stomal variceal bleeding: successful treatment using injection sclerotherapy. Dis Colon Rectum 1986;29: 269-7O. Hesterberg R, Stahlknecht CD, R6her HD. Sclerotherapy for massive enterostomy bleeding resulting from portal hypertension. Dis Colon Rectum 1986; 29:275-7. HendersonJM, Lumsden A. Variceal bleeding. South Med J 1986;79:921. Lebrec D, Benhamou JP. Ectopic varices in portal hypertension. Baillieres Clin Gastroenterol 1985;14: 105-21. Wiesner RH, LaRusso NF, Dozois RR, Beaver SJ. Peristomal varices after proctocolectomy in patients with primary sclerosing cholangitis. Gastroenterology 1986;90:316-22. MarshJWJr, Iwatsuki S, Makowka L, e t al. Orthotopic liver transplantation for primary sclerosing cholangitis. Ann Surg 1988;207:21-5.
Peristomal variceal bleeding is a serious complication in patients with chronic liver disease undergoing colon surgery with a stoma. Our aim was to examine the morbidity of bleeding for peristomal, perianastomotic, and esophageal varices in a group of patients with chronic liver disease who underwent colectomy at the Mayo Clinic between 1970 and 1988. Morbidity was evaluated in terms of the number of major bleeding episodes and the number of units of blood transfused. The treatment of bleeding was also evaluated. One hundred seventeen patients (74 males and 43 females) aged 11-78 years were studied. Sixty-two patients (53 percent) had a permanent stoma, while 55 patients (47 percent) had a colonic resection and anastomosis. Sixty-seven patients (62 percent) had chronic ulcerative colitis and primary sclerosing cholangitis. In the stoma group, bleeding appeared from stomal and/or esophageal varices in 19 patients (31 percent), while, in the non-stoma group, bleeding exclusively from the esophageal varices occurred in eight patients (15 percent). Perianastomotic variceal bleeding was never observed. The 5-year cumulative probabilities of one major bleed occurring from gastrointestinal varices appeared to be similar between the two groups. Patients who bled from peristomal varices with or without esophageal bleeding (n --- 17) rebled more frequently (6.5 --. 5.5 vs. 3 --. 1.6; P < 0.05) and were transfused more often (14.9 --- 12.3 vs. 7.5 4- 4.1; P < 0.05) than patients who bled exclusively from esophageal varices (n = 10). No difference was found in the incidence of recurrent bleeding and the number of units of blood transfused between patients who bled exclusively from peristomal varices (n - 10) and those who bled from both peristomal and esophageal varices (n 7). Medical and local measures were more effective in controlling esophageal bleeding than in controlling peristomal bleeding. Therefore, patients with chronic liver disease who must undergo colectomy should have a distal anastomosis rather than a terminal stoma. [Key words: Intestinal stomas; Portal hypertension; Colectomy; Cirrhosis] Fucini C, Wolff BG, Dozois RR. Bleeding from peristomal varices: perspectives on prevention and treatment. Dis Colon Rectum 1991 ;34:1073-1078. Read in part at the meeting of The American Society of Colon and Rectal Surgeons, St. Louis, Missouri, April 27 to May 4, 1990. Address reprint requests to Dr. Dozois: Mayo Clinic, 200 First Street Southwest, Rochester, Minnesota 55905. 1073
l e e d i n g from varices in the abdominal wall s u r r o u n d i n g an intestinal stoma has b e e n rep o r t e d in at least 62 patients with portal hypertension. 1-27 Patients with chronic liver disease u n d e r g o i n g abdominal surgery (e.g., a b d o m i n o perineal resection, p r o c t o c o l e c t o m y , and ileal conduit) and requiring an abdominal stoma are exp o s e d to the risk of adhesions b e t w e e n the mesentery and the abdominal wall with portosystemic shunting if portal h y p e r t e n s i o n develops. 28 Varices are primarily located at the peristomal mucocutan e o u s junction, w h e r e b l e e d i n g may originate. Muc o c u t a n e o u s varices are occasionally a continuation of submucosal varices of the e n d portion of the intestine.14, ,7 T h e y have also b e e n r e p o r t e d as prestomal, with b l e e d i n g occurring within the ileal lumen. 2~ The bleeding, w h i c h is often heavy, may be difficult to control, and the frequently observed episodes of h e m o r r h a g e are responsible for their high morbidity. 7 In a recent study from our institutior129 on primary sclerosing cholangitis (PSC), a cholestatic and cirrhogenic liver disease often associated with chronic ulcerative colitis (CUC), peristomal variceal b l e e d i n g was observed in 53 p e r c e n t of patients u n d e r g o i n g total p r o c t o c o l e c t o m y and fieostomy, whereas it did not occur in patients u n d e r g o i n g ileoanal anastomosis. Based on this observation, the authors r e c o m m e n d e d ileoanal or at least ileorectal anastomoses in PSC patients requiring c o l e c t o m y for CUC. Such an indication might be e x t e n d e d to patients with other types of cirrhosis and c o l o n disease in w h i c h the fashioning of a distal anastomosis (rectal or anal) is an alternative to a terminal stoma (i.e., low rectal cancers and polyposis of the colon). Because the occurrence of perianastomotic and esophageal variceal
B
1074
FUCINI E T A L
bleeding, which were also responsible for a high morbidity, has occasionally been reported by some authors, 7 we examined the frequency of bleeding from stomal or other gastrointestinal varices in advanced chronic liver disease and/or PSC patients who underwent a colonic resection with the fashioning of a distal anastomosis or the creation of a terminal stoma. Our aim was to verify whether the actual fashioning of a stoma would increase the risk of bleeding and morbidity. At the same time, we examined some clinical aspects of stomal variceal bleeding in an attempt to assess the value of therapeutic procedures used to control bleeding. PATIENTS A N D M E T H O D S A survey was conducted of portal hypertension, PSC, and/or other biopsy-proven advanced chronic liver disease patients seen at the Mayo Clinic between January 1970 and June 1988 who had had a distal anastomosis (ileorectal, ileoanal, colorectal, or coloanal) or a permanent intestinal stoma at the time of colonic resection. Patients who died in the postoperative period and those followed for less than 4 months were excluded. The records of 117 patients (74 men and 43 women) with an age range of 11 to 78 years were reviewed. The following information was recorded: diagnosis of intestinal and liver disease at the time of operation; type of intestinal operation; occurrence, site, and number of bleeding episodes from gastrointestinal varices which prompted blood transfusions; number of units of blood transfused; values of liver function tests at the time of the first episode of bleeding; types of therapy used to control the hemorrhage; and outcome of the most recent follow-up. Analysis of the risk of variceal bleeding in patients with a stoma vs. those without a stoma was carried out, and the morbidity of peristomal vs. other types of intestinal variceal bleeding was evaluated. A separate analysis was made of all 23 registered patients who had had one or more episodes of stomal bleeding requiring blood transfusions in the same period. Twenty had a terminal ileostomy (two with a Kock pouch), and three had an end colostomy. Liver disease appeared in one of these patients after the colonic operation. In five patients, the initial colonic operation had been performed either before 1970 or outside the Mayo Clinic. Statistical analysis was performed with the unpaired Student's t-test and the •2 test.
Dis Colon Rectum, December 1991 RESULTS
Permanent abdominal stomas (40 Brooke ileostomies, 8 Kock pouches, and 14 colostomies) had been constructed in 62 patients (53 percent), while, in the other 55 patients (47 Percent), colonic resection with anastomosis was established (20 J-ileal pouch-anal anastomoses, 23 ileorectostomies, 1 ileoanal anastomosis without reservoir, and 11 colorectostomies). Twenty-five patients of the latter group received a temporary loop ileostomy or colostomy. In 24 patients, the loop ostomy was closed over a period ranging from 4 to 16 weeks. Because no stomal bleeding appeared in any of these patients, they were included in the non-stoma group. One patient whose loop ileostomy was left in place because of an ileoanal anastomosis failure was included in the stoma patient group. Seventy patients (60 percent) were diagnosed as having PSC, 25 of them after the introduction of endoscopic retrograde cholangiopancreatography (ERCP). Forty-seven patients (40 percent) had other forms of cirrhosis (e.g., La~nnec's, biliary, and postnecrotic). At the time of operation, 33 patients had features of portal hypertension such as splenomegaly and/or esophageal varices. Eighty-one patients (69 percent) had CUC, associated with colon cancer in 17 patients. The other 36 patients (31 percent) had other colonic diseases such as rectal cancer, Crohn's disease, diverticulitis, or polyposis. Sixty-seven patients (62 percent) had both CUC and PSC. The occurrence of bleeding and the general mortality observed in patients with or without a stoma are reported in Table 1. In the stoma group, the bleeding sites included the esophagus (two patients), the stomal area (10 patients), and both sites (seven patients). In the non-stoma group, bleeding originated from the esophagus (eight patients). No perianastomotic variceal bleeding was observed from either ileorectostomy or ileoanal anastomosis sites. PSC patients showed a significantly greater risk of bleeding than patients with other forms of cirrhosis (Table 2), even though they were equally represented in both groups (56 percent vs. 63 percent). The two groups were also similar in terms of patient age and number and stages of cancers (Table 1). The 5-year cumulative probabilities of one major bleed occurring from intestinal varices appeared
Vol. 34, No. 12
BLEEDING PERISTOMALVARICES
1075
Table 1.
Clinical Features and Frequency of Bleeding in Patients With or Without a Stoma Stoma
Clinical Data No. of patients Mean age -+-SD (and range) (yr)
Yes
No
62
55 42 + 15.2
44 _+ 15.1 (17-78) 20 (32) 61 _+ 39.2 (5-155) 19 (31) 19 (31)
No. of patients with cancer (and %) Mean + SD follow-up (and range) (months) No. of patients with variceal bleeding (and %) No. of deaths (and %)
(11-68)
14 (25) 32 ___28.8 (5-156)
8 (15) 9 (16)
Table 2.
Frequency of Gastrointestinal Variceal Bleeding in PSC vs. Non-PSC Patients Patients
Patient Population Total no. of patients Mean + SD follow-up (and range) (months) No. of patients who bled (and %)
With PSC
Without PSC
70 47 + 37.3 (5-155) 22 (31)*
47 47 _+ 38.2 (5-156) 5 (11)*
* Fisher's exact test (P = 0.007).
to be similar between the two groups (Fig. 1). The cumulative probabilities of survival were also similar (Fig. 2). However, when we evaluated the morbidity from peristomal v s . other variceal bleeding sites in terms of the number of rebleeding episodes and the number of units of blood transfused, we found that patients who bled from peristomal varices, regardless of esophageal bleeding, had a significantly higher number of recurrent hemorrhagic episodes
and units of blood transfused than did patients who bled exclusively from esophageal varices (Table 3). The values of liver function tests, such as serum albumin, bilirubin, prothrombin time, transaminase, alkaline phosphatase, and platelet count, were not significantly different between the two groups at the time of the first episode of bleeding. When the patients who bled exclusively from peristomal varices (n = 10) were compared with those who bled from both peristomal and esopha-
'1
'1
0,9
-1
O,II /
b,I
1
O,I t O.,Y
0,6 O,S
=
9
9,11
o,s o,+
0,3
0,3 0,2
0,1 0
i gD
i
, 4m
i
0
J
~0
I0
(aom=) flTOVJk
+
X~)Y
Figure 1. Five-year cumulative probability of a major variceal bleed in stoma vs. non-stoma patients.
i
i
,
+
i
I)
frFolLt.
+
1t'0)1' g T O M I
Figure 2. Five-year cumulative probability of survival in
stoma vs. non-stoma patients.
FUCINI E T A L
1076
Dis Colon Rectum, December 1991
Table 3.
Sites and Number of Major Bleeds and Number of Units of Blood Transfused Before a Major Operation Bleeding Site Patient Population
No. of patients Mean __.SD no. of bleeds (and range)
Mean _+ SD no. of units transfused (and range) Mean _+ SD follow-up (and range) (months)?
Esophagus 10 3 + 1.6" (1-5) 7.5 _ 4.1" (3-15) 25.7 _+ 30.2 (2-101)
Stoma With/Without Esophageal Bleeding 17 6.5 + 5.5* (2-21) 14.9 +_ 12.3"?
(5-49) 25.5 ___17.9 (0-58)
* P < 0.05. t From the first episode of bleeding.
geal varices (n = 7), the number of recurrent bleeds did not differ; nor did the number of units of blood transfused. It would appear that the primary medical or local measures used to control esophageal bleeding (esophageal balloon tamponage with or without vasopressin infusion or sclerotherapy with ethanolamine oleate with or without propanolol) were more effective in controlling the hemorrhage than were the medical or local methods (vasopressin infusion to oral propanolol, compression of varices, sclerotherapy, fulguration, transstomal ligation, laser coagulation, mucocutaneous disconnection, and stoma refashioning) employed to control peristomal bleeding. In a majority of these patients, several means were tried. Repeated episodes of major bleeding were the main indication for surgery in 2 of the 10 patients (20 percent) who bled from esophageal varices, i.e., the Sugiura procedure (one patient) and liver transplantation (one patient). Seven of the 17 patients (41 percent) who bled from peristomal varices underwent a major operation for the same reason (five liver transplantations, one splenorenal shunt, and one mesocaval shunt). In a comparable mean follow-up from the first episode of bleeding (29 vs. 31 months), mortality was similar between the patients who bled exclusively from esophageal varices and those who bled from peristomal varices (40 percent vs. 53 percent; P > 0.05). Based on the parameters assessed in 23 patients seen in our institution, we made the following observations. First, CUC was the most common colonic disease (60 percent of peristomal bleeders), and PSC was the most frequent liver disease (65 percent). The CUC-PSC association accounted for 13 (57 percent) of the patients who bled from stomal varices. Second, 18 of the patients (78 per-
cent) with stomal varices also had esophageal varices. Ten patients (43 percent) bled from esophageal varices after a peristomal bleed. Third, in the 17 stomal bleeders who were operated on after 1970 and who had liver disease at the time of the operation, the time that elapsed between the creation of the stoma and the bleeding from stomal varices varied from 6 to 133 months (mean, 47 months). Fourth, the therapeutic procedures used to stem peristomal bleeding fell into three main categories: medical or local surgery, portosystemic shunts, and liver transplantation. Thirteen patients had only medical and/or local surgery. Five patients had a portosystemic shunt (three side-to-side splenorenal shunts, one central splenorenal shunt, and one mesocaval shunt); in all of them, the shunt was performed after repeated attempts to stem the bleeding with local measures. Five patients with PSC had liver transplantation. Four of them had previously been treated with a trial of local procedures. One patient who underwent emergency liver transplantation at the time of the first episode of peristomal bleeding died postoperatively. A second patient who had a rejection after 1 year underwent retransplantation. Table 4 summarizes the results of the different treatments. Despite the failure of the shunt in one patient (obstruction owing to thrombosis), the operation provided better control of the bleeding (P < 0.05) and better survival (P < 0.05) than local treatment. Similar results were obtained in patients who underwent transplantation. In patients who had colon cancer, all deaths were attributable to liver disease. One of these patients with concomitant primary biliary cirrhosis and metastatic liver disease had undergone abdominoperineal excision for a rectal cancer.
BLEEDING PERISTOMALVARICES
Vol. 34, No. 12
1077
Table 4. Results of Different Therapies Types of Surgeries Patient Population No. of patients Mean _+ SD follow-up (and range) (months) Bleeding control (and %) Mortality (and %)
Local
Shunting
Transplantation
13 27 _+ 18.1 (2-63) 3 (23)1 10 (77)~:
5 24.8 ___19.9" (8-58) 4 (80):1: 1 (20)1:
5 22 + 16.6" (0-38) 4 (80) 1 (20)
* Data refer to follow-up from shunt or from transplantation. The mean _+ SD periods from first bleeding to shunting and to transplantation were 13.2 _+ 19.2 and 13.8 _+ 14 months, respectively. 1 The number refers to patients who were free from bleeding after a local procedure for at least 12 months; this corresponds to the shortest follow-up for a successful shunt. :l= Fisher's exact test (P < 0.05).
DISCUSSION Bleeding from peristomal varices is a serious event in the natural history of cirrhotic patients. Twenty-nine (37 percent) of the 78 patients who could be evaluated (55 from the medical literature and 23 from our own institution) died from liver disease within 3 years of the first episode of bleeding. Death from exsanguination occurred rarely. 13 This complication can be considered frequent, as it occurred in 17 (27 percent) of the 62 Mayo Clinic patients operated on between 1970 and 1988 who had advanced chronic liver disease as demonstrated at the time the stoma was fashioned and who were followed for a median period of 5 years. The complication is more frequent in CUC-PSC patients. In the same group of patients, Peck and Boyden 23 reported a bleeding incidence of 24 percent, as compared with 53 percent reported by Wiesner e t al. 29 To eliminate the morbidity derived from this type of bleeding, some authors, 29 have proposed that ileostomy be avoided in CUC-PSC patients requiting proctocolectomy and that ileoanal and/or ileorectal anastomosis, which are less susceptible to variceal formation and bleeding, be performed. Our retrospective study, which examined all patients with cirrhosis and colon disease with an abdominal stoma or with a distal anastomosis, confirmed that perianastomotic variceal bleeding is a very rare complication. However, it showed that the probability of major bleeding from esophageal varices occurring in cirrhotic patients who have had an anastomosis does not differ from that of bleeding from peristomal and/or esophageal varices in patients who have an abdominal stoma. Thus, peristomal bleeding did not act as a "safety valve" for esophageal bleeding, as suggested by some authors. 2
However, when we examined the morbidity derived from peristomal bleeding in comparison with esophageal bleeding as assessed in terms of the number of episodes of major rebleeding, the number of units of blood transfused, and the efficacy of medical or local methods to control bleeding, we found that esophageal bleeders had a lower morbidity than peristomal bleeders. This was also evident in patients who bled from both sites. Trivial trauma from the stomal appliance or from usual stoma care may have played a role in triggering bleeding. Thus, cirrhotic patients who undergo a colon resection, in whom the decision between fashioning a distal anastomosis and fashioning a stoma is not governed by the type or extent of the colonic disease, should preferably have an anastomosis. CUC-PSC patients represent the largest group of these patients. When major stomal bleeding occurred, the medical or local management was merely palliative, and recurrent hemorrhage was common. However, it has not yet been established whether repeated episodes of bleeding precipitate liver failure. 18' 21 For this reason, some authors 21 challenge the indication of shunting procedures which could control bleeding but do not increase the chances of survival and at the same time expose the patients to the risk of encephalopathy. Our experience does show that patients having shunts had a better chance of survival than those having local procedures alone. The only instance of encephalopathy observed was in the patient whose shunt failed. The complete absence of the colon in the majority of these patients may be relevant. Therefore, it seems reasonable to consider a definitive procedure and avoid a stoma unless the extent of the disease precludes such a course of
1078
FUCINI E T A L
action. H o w e v e r , the success of liver transplantation, which is b e c o m i n g an established therapy in PSC, 3~ raises other considerations. If a stoma is avoided, only local control of esophageal b l e e d i n g may be necessary until the patient n e e d s a transplantation. The p r e s e n c e of a previously constructed ileal pouch-anal anastomosis should not p r e c l u d e orthotopic liver transplantation, but special attention to m e s e n t e r i c vessels will be necessary.
16.
REFERENCES
18.
1. Resnick RH, Ishihara A, Chalmers TC, Schimmel EM, The Boston Inter-Hospital Liver Group. Controlled trial of colon bypass in chronic hepatic encephalopathy. Gastroenterology 1968;54:1057-69. 2. Eade MN, Williams JA, Cooke WT. Bleeding from an ileostomy caput medusa. Lancet 1969;2:1166-8. 3. Cameron AD, Fone DJ. Portal hypertension and bleeding ileal varices after colectomy and ileostomy for chronic ulcerative colitis. Gut 1970;11:755-9. 4. Hamyln AN, Lunzer MR, Morris JS, Puritz H, Dick R. Portal hypertension with varices in unusual sites. Lancet 1974;2:1531-4. 5. Foulkes J, Wallace DM. Haemorrhage from stomal varices in an ileal conduit. BrJ Urol 1975;47:630. 6. Graeber GM, Ratner MH, Ackerman NB. Massive hemorrhage from ileostomy and colostomy stomas due to mucocutaneous varices in patients with coexisting cirrhosis. Surgery 1976;79:107-10. 7. Adson MA, Fulton RE. The ileal stoma and portal hypertension: an uncommon site of variceal bleeding. Arch Surg 1977;112:501-4. 8. Crooks KK, Hensle TW, Heney NM, Waltman A, Irwin RJ Jr. Ileal conduit hemorrhage secondary to portal hypertension. Urology 1978;12:689-93. 9. Firlit RS, Firlit CF, Canning J. Exsanguinating hemorrhage from urinary ileal conduit in patient with portal hypertension. Urology 1978;12:710-1. 10. Finemore RG. Repeated haemorrhage from a terminal colostomy due to mucocutaneous varices with coexisting hepatic metastatic rectal adenocarcinoma: a case report. BrJ Surg 1979;66:806. 11. Federle M, Clark RA. Mesenteric varices: a source of mesosystemic shunts and gastrointestinal hemorrhage. Gastrointest Radiol 1979;4:331-7. 12. Goldstein WZ, Edoga J, Crystal R. Management of colostomal hemorrhage resulting from portal hypertension. Dis Colon Rectum 1980;23:86-90. 13. Ackerman NB, Graeber GM, Fey J. Enterostomal varices secondary to portal hypertension: progression of disease in conservatively managed cases. Arch Surg 1980;115:1454-5. 14. Ricci RL, Lee KR, Greenberger NJ. Chronic gastrointestinal bleeding from ileal varices after total proc-
15.
17.
19.
20.
21.
22. 23.
24.
25.
26.
27. 28.
29.
30.
Dis Colon Rectum, December 1991 tocolectomy for ulcerative colitis: correction by mesocaval shunt. Gastroenterology 1980;78:1053-8. Eckhauser FE, Sonda LP, Strodel WE, Edgcomb LP, Turcott JG. Parastomal ileal conduit hemorrhage and portal hypertension. Ann Surg 1980;192:620-4. Watkins RM. Variceal haemorrhage from a colostomy due to portal hypertension secondary to intrahepatic metastases from rectal carcinoma. BMJ 1981;282: 189-90. Cooper MJ, Mackie CR, Dhorajiwala J, Baker AL, Moossa AR. Hemorrhage from ileal varices after total proctocolectomy. AmJ Surg 1981;141:178-9. Hollands MJ. Parastomal haemorrhage from an ileal conduit secondary to portal hypertension. Br J Surg 1982;69:675. Perreault JP, Laroze M, Faucher R, Paquin JM, Mauffette F. H~morragie d'un conduit ileal secondaire une hypertension portale. J Urol (Paris) 1982;88: 47-9. Goldstein MB, Brandt LJ, Bernstein LH, Sprayragen S. Hemorrhage from ileal varices: a delayed complication after total proctocolectomy in a patient with ulcerative colitis and cirrhosis. Am J Gastroenterol 1983;78:351-4. Grundfest-Broniatowski S, Fazio V. Conservative treatment of bleeding stomal varices. Arch Surg 1983;118:981-5. Johnson AG, Simms JM. Correspondecne. Br J Surg 1983;70:187-90. PeckJJ, BoydenAM. Exigent ileostomyhemorrhage: a complication of proctocolectomy in patients with chronic ulcerative colitis and primary sclerosing cholangitis. Am J Surg 1985;150:153-8. Wang MM, McGrew W, Dunn GD. Variceal bleeding from an ileostomy stoma. South Med J 1985;78: 733-7. Morgan TR, Feldshon SD, Tripp MR. Recurrent stomal variceal bleeding: successful treatment using injection sclerotherapy. Dis Colon Rectum 1986;29: 269-7O. Hesterberg R, Stahlknecht CD, R6her HD. Sclerotherapy for massive enterostomy bleeding resulting from portal hypertension. Dis Colon Rectum 1986; 29:275-7. HendersonJM, Lumsden A. Variceal bleeding. South Med J 1986;79:921. Lebrec D, Benhamou JP. Ectopic varices in portal hypertension. Baillieres Clin Gastroenterol 1985;14: 105-21. Wiesner RH, LaRusso NF, Dozois RR, Beaver SJ. Peristomal varices after proctocolectomy in patients with primary sclerosing cholangitis. Gastroenterology 1986;90:316-22. MarshJWJr, Iwatsuki S, Makowka L, e t al. Orthotopic liver transplantation for primary sclerosing cholangitis. Ann Surg 1988;207:21-5.
