- Email: [email protected]
Acute lower gastrointestinal bleeding in Crohn’s disease: characteristics of a unique series of 34 patients
THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 1999 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.
Vol. 94, No. 8, 1999 ISSN 0002-9270/99/$20.00 PII S0002-9270(99)00347-0
Acute Lower Gastrointestinal Bleeding in Crohn’s Disease: Characteristics of a Unique Series of 34 Patients Jacques Belaiche, M.D., Edouard Louis, M.D., Geert D’Haens, M.D., Marc Cabooter, M.D., Serge Naegels, M.D., Martine De Vos, M.D., Fernand Fontaine, M.D., Piet Schurmans, M.D., Filip Baert, M.D., Marc De Reuck, M.D., Rene´ Fiasse, M.D., Jan Holvoet, M.D., Alain Schmit, M.D., Marc Van Outryve, M.D., and the Belgian IBD Research Group Department of Gastroenterology CHU Sart Tilman, Lie`ge; University Hospital Gasthuisberg, Leuven, Brugge; Academic Hospital Free University of Brussels, Brussels; University Hospital, Gent; Clinique Saint Joseph, Lie`ge, Hasselt, Roselare; Hopital Universitaire Brugmann, Bruxelles; Clinique Universitaire St. Luc, Bruxelles, Antwerpen; Universite´ Libre de Bruxelles, Bruxelles; Universitair Ziekenhuis, Antwerpen, Belgium
OBJECTIVE: Acute lower gastrointestinal bleeding is a rare complication of Crohn’s disease, which represents a diagnostic and therapeutic challenge. The aim of this study was to define epidemiological characteristics and therapeutic options of hemorrhagic forms of Crohn’s disease. METHODS: Thirty-four cases of hemorrhagic forms of Crohn’s disease were studied retrospectively. Acute lower gastrointestinal hemorrhage was defined as acute rectal bleeding originating in diseased bowel and requiring a transfusion of at least 2 units of red blood cells within 24 h. Upper gastrointestinal tract hemorrhage or anal lesions and postoperative bleeding were excluded. RESULTS: Mean age at time of hemorrhage was 34.2 ⫾ 14 yr. Mean duration of disease before the hemorrhage was 5.6 ⫾ 6 yr. The hemorrhage occurred during a flare up of the disease in 35% of cases. The hemorrhage revealed Crohn’s disease in 23.5% of cases. The hemorrhage was more frequent in colonic disease (85%) than in isolated small bowel disease (15%) (p ⬍ 0.0001). The origin of bleeding was identified in 65% of cases, by colonoscopy (60%), by angiography (3 patients), or at surgery (1 patient). The bleeding lesion was an ulcer in 95% of cases, most often in the left colon. The treatment was surgical in 20.5% (colectomy in 36%), endoscopical (7 patients, including 5 successes), or medical. Hemorrhage recurred in 12 patients (35%) within a mean time of 3 yr (4 days– 8 yr), requiring surgery in 3 cases. No death was observed. CONCLUSIONS: This study performed in a series characterized by a nonsurgical recruitment, the largest to date, shows that hemorrhagic forms of Crohn’s disease may reveal disease in 23.5%, occurs in quiescent Crohn’s disease in twothirds of cases. Given the potential efficacy of endoscopical or medical treatment, as well as the absence of mortality, a conservative approach may be suggested as first-line ther-
apy in the majority of patients. (Am J Gastroenterol 1999; 94:2177–2181. © 1999 by Am. Coll. of Gastroenterology)
INTRODUCTION Gastrointestinal (GI) hemorrhage in a setting of inflammatory bowel disease has traditionally been associated with ulcerative colitis. Crohn’s disease (CD) is mentioned in most textbooks as an unusual cause of massive lower bleeding occurring in 0.9 – 6% of patients (1–3). However, more recent studies cite an equal frequency of severe gastrointestinal hemorrhage in both CD and ulcerative colitis (1, 3). Hemorrhagic forms of CD are not well known and mainly reported in surgical series (1– 4). They represent a diagnostic and therapeutic challenge. The aim of this study was to define clinical, epidemiological characteristics, and therapeutic options for hemorrhagic forms of CD through a retrospective nonsurgical series of 34 cases.
MATERIALS AND METHODS Patients Thirty-four patients with CD presenting with acute low GI bleeding at several medical centers affiliated with the Belgian Inflammatory Bowel Disease Research Group, were included in this study. CD was diagnosed according to Lennard-Jones criteria (5). Acute lower GI bleeding was defined as an acute rectal bleeding originating in diseased bowel and requiring at least 2 units of blood transfusion within 24 h. Exclusion criteria were as follows: bleeding originating from upper GI tract or anal lesions, and massive hemorrhage occurring within 30 days after surgery. Clinical Data In this retrospective study, the following data were evaluated in each patient: sex, age at diagnosis, age at time of first
2178
Belaiche et al.
AJG – Vol. 94, No. 8, 1999
Table 1. Characteristics of the 34 Patients at Inclusion Characteristics
Male
Female
All Patients
No. of patients Age at onset of disease (yr) Age at time of first bleeding (yr) Duration of disease from onset to bleeding (yr) Familial inflammatory bowel disease (%) Smoker (%) Extraintestinal manifestations Previous intestinal resection (%) Disease location (%) Small bowel only Colon only Small bowel and colon Anal lesions Medical treatment at time of bleeding (%) Steroid Sulfasalazine or mesalamine Azathioprine NSAID Warfarin No treatment
17 29.4 ⫾ 14.6 34 ⫾ 15.5 4.9 ⫾ 6.2 5 7 3 8
17 27.8 ⫾ 12.2 33.3 ⫾ 12.5 6.3 ⫾ 5.9 4 4 4 5
34 28.5 ⫾ 13.1 34.2 ⫾ 14 5.6 ⫾ 6 9 (27) 11 (34) 7 (22.5)* 13 (38)
3 3 11 3
2 7 8 5
5 (15) 10 (29) 19 (56) 8 (23)
4 4 4
6 2 2 1 1 7
10 (29) 6 (17) 6 (17) 1 1 10 (29)
3
Values are expressed as mean ⫾ SD. There was no significant difference between the two groups for any of these parameter distributions. * n ⫽ 31.
bleeding, duration of disease from onset to bleeding, smoking habit, familial history of inflammatory bowel disease, extraintestinal manifestations, previous surgery, disease location (small bowel only, colon only, small bowel and colon, anoperineal disease), medical treatment at time of bleeding, amount of blood transfused, bleeding site, treatment, and follow-up. Activity of CD was judged by each physician on pure clinical basis as, in this retrospective study; Crohn’s disease activity index (CDAI) was not always available. Statistical Methods Comparative statistics were carried out using the one-way analysis of variance (ANOVA), Student’s t test or MannWhitney U test as appropriate. Qualitative variables were analyzed by Fisher’s exact test. Correlations between severity of the hemorrhage and the time to rebleeding was assessed by Pearson test. Follow-up for nonoperated patients was modeled using Kaplan-Meier survival estimates.
RESULTS Patient Characteristics The clinical details of the patients are summarized in Table 1. The mean age at the time of the first bleeding episode was 34.2 yr ⫾ 14 yr (range: 18 –75 yr; median 31 yr). The mean duration of disease from onset to time of hemorrhage was 5.6 yr ⫾ 6 yr (range: 0 –23 yr; median: 5 yr). Twelve patients (35%) had active symptomatic CD and 22 patients (65%) had a quiescent CD at the time of bleeding. Severe bleeding was the first manifestation of disease in 8 patients (23.5%). The hemorrhage was more frequent in colonic disease (85%) than in isolated small bowel disease (15%) (p ⬍ 0.0001).
Blood Transfusion Requirements The mean serum hemoglobin at presentation was 7.8 ⫾ 1.8 g/dl (range: 3.4 –10.3 g/dl; median: 7.9 g/dl). Blood requirements averaged 5.2 units (1 unit ⫽ 400 ml of packed red blood cells) ⫾ 3 units (range: 2–20 units; median: 4 units) per patient. No significant differences were found between male and female patients and between operated and nonoperated patients. Bleeding Site The site of bleeding was identified in 22 patients (65%) (Table 2). Colonoscopy performed in 30 patients revealed a bleeding site in 18 of them (60%). The lesion found was considered to be the origin of the hemorrhage when it was actively bleeding (n ⫽ 16), or when an adherent clot was present (n ⫽ 2). In 95% of cases with identified bleeding site the hemorrhage was attributable to an ulcer or an ulcerated area, most often located in the sigmoid or the left colon. The ulcers involved had various shapes: deep round or serpiginous ulcers (n ⫽ 13), and more rarely superfical (n ⫽ 3) or solitary as a skip lesion (n ⫽ 1). In 1 patient the bleeding Table 2. Localization of the Bleeding Site Methods Colonoscopy (n ⫽ 30) Ulcers
Pseudopolyp Angiography (n ⫽ 4) Surgery
Bleeding Site
No. of Patients
Sigmoid or left colon Right colon Ileocaecal valve Ileum Ileocolic anastomosis Sigmoid Ileum Pseudopolyp of ileum
8 2 3 2 2 1 3 1
AJG – August, 1999
Acute Lower GI Bleeding in CD
2179
Table 3. Type of Surgery in 10 Operated Patients (7 for First Bleeding and 3 for Recurrent Bleeding) Type of Surgery
No. of Patients
Subtotal colectomy Subtotal colectomy with ileostomy Ileocolic resection Partial colectomy Partial colectomy with colostomy
3 1 4 1 1
originated from a large pseudopolyp in the sigmoid. Mesenteric angiography was performed on four occasions and revealed a brisk bleeding site in the distal small bowel in 3 cases. These 3 patients required surgery. In 1 patient a pseudopolyp of the terminal ileum was identified at surgery. The only two isotopic bleeding scan performed were noncontributive. Treatment Initial therapy at the moment of hemorrhage consisted of blood transfusions and supportive measures. A specific medical treatment of CD was started in 24 patients (70%), consisting of corticosteroids in 21 patients and cyclosporine in 3 patients. Endoscopic treatment of the bleeding lesion was attempted in 7 cases: laser coagulation in 3 cases (including 1 case with associated adrenalin injection), bipolar coagulation in 3 cases, and polypectomy of a pseudopolyp in the sigmoid colon in 1 case. These endoscopic treatments were successful in 5 cases. Seven patients (20.5%) were operated on during their first bleeding episode. Five of them had a clinically active CD and 2, a quiescent CD at time of surgery. In all cases, an intervention was required on an urgent basis due to persistent hemorrhage. The operative procedures performed are mentioned in Table 3. In one specific case, an ileocolic resection was performed due to small bowel ischemia after radiological embolization of the ileocecal artery. Follow-Up No death was observed in this series. The mean hospital stay was 14.7 ⫾ 11 days (range: 2– 40 days; median: 13.5 days). In nonoperated patients the treatment at the end of hospital stay comprised sulfasalazine or mesalamine and/or steroids (13 cases), azathioprine (4 cases), azathioprine and steroids (1 case). Long-term follow-up was available in 25 patients (mean: 3 yr, range: 0.5–11 yr; median 2 yr) (Fig. 1). Twelve nonoperated patients (35%) had recurrent bleeding: 9 bled twice, 2 bled three times, and 1 bled four times. Bleedingfree survival curve at 6 mo for these patients is shown in Figure 2. There was no bleeding recurrence in operated patients. Recurrence of bleeding occurred on average 3 yr after the first hemorrhage (mean: 3 mo, range: 4 days– 8 yr; median 1.5 yr). There was no significant difference between patients without recurrence and patients who rebled according to the severity of the hemorrhage assessed by both the hemoglobin level and the amount of blood transfusion required. In patients who rebled, no correlation was observed
Figure 1. Outcome of medical and surgical treatment of acute lower bleeding in 34 patients with Crohn’s disease. There was no bleeding recurrence in patient with surgical resection.
between the delay to rebleeding and the severity of the hemorrhage assessed by the same criteria. Three patients were operated after one (1 case), three (1 case), and four (1 case) recurrent bleeding episodes.
Figure 2. Bleeding-free survival curve at 6 months (Kaplan-Meier) in nonoperated patients.
2180
Belaiche et al.
DISCUSSION Acute lower GI hemorrhage is an unusual complication of CD, occurring in 0.9 – 6% of the cases (1–3). In our retrospective and multicenter series an estimation based on the frequency in the three centers having included ⬎50% of patients gave an incidence of 1.08% (data not shown). The definition of massive GI hemorrhage, which ranges from a very blurred one (need of transfusion for maintening normal vital signs) (4) to the transfusion of at least 5 units of blood (2), can explain this variation. In our study a bleeding necessitating at least 2 units of blood over a period of 24 h was chosen as inclusion criterion. The mean serum hemoglobin concentration at presentation (7.8 ⫾ 1.8 g/dl) and the number of blood units transfused (5.2 ⫾ 3 units) reflect the importance of the hemorrhage. In contrast to previously published surgical studies, the current one is characterized by a nonsurgical recruitment. Furthermore, it represents the largest published study to date. The data about the clinical and epidemiological characteristics of hemorrhagic forms of CD are not well defined in the literature. Our results support the statement that bleeding may occur at any age or stage of CD (2– 4, 6 –9). The oldest patient was 75 yr old, and the youngest was 18 yr, with disease duration before onset of GI hemorrhage of 0 to 23 yr. A recent review reported a nearly 2:1 ratio of men to women (2). The present series as well as that of Robert et al. (1) do not confirm this finding. Gastrointestinal hemorrhage was the initial symptom of CD in 23% of our patients, which is close to the frequency previously reported by Cirocco et al. (2) and confirmed by several case reports (3, 7, 9, 10 –12). Other series suggested a higher frequency up to 50% (3); in contrast, in the Mount Sinai Hospital series, the hemorrhage never revealed the disease (1). These differences may be explained by the different types of patient recruitment, nonsurgical or surgical. We have identified a significantly higher rate of hemorrhage among the patients with colonic involvement than among those with only small bowel disease. This observation is similar to other reports (1, 3, 7, 9, 11) but is different from the review by Cirocco et al. (2) where the site of bleeding corresponded to the general distribution of CD, with ileal disease predominating in 66% of cases. Localization of the site of hemorrhage remains a major challenge in this disease process, which may be along the bowel. The precise identification of the bleeding site may be difficult even at pathological examination of the resected specimens; the precise bleeding site in our study remained unknown in 35% of the cases. In our study, colonoscopy was useful as it identified the bleeding site in 18 patients (60%). Most of bleeding episodes originated from colonic ulcers, either isolated, or in an area of more widespread ulceration. In 2 cases, the ulcer was located at the ileocolonic anastomosis. Very rarely, the bleeding originated from a pseudopolyp, as exceptionally reported (11). However, colonoscopy can be difficult, due to the amount of blood in
AJG – Vol. 94, No. 8, 1999
the colonic lumen. In patients where colonoscopy is not diagnostic, enteroscopy should be evaluated to identify small bowel bleeding lesions. Angiography was also useful in our study, revealing the site of hemorrhage in 3 of 4 patients in which it was performed. Several studies confirmed the diagnostic yield of this exploration, which can demonstrate not only diffuse irregular hypervascularity of the bowel wall and mesenterium but also, as in our patients, an extravasation of contrast material from the ileal branch of the ileocolic artery (1, 2, 4, 6, 8 –11). The management of lower GI hemorrhage due to CD includes both medical and surgical modalities. Conservative therapy has been advocated in view of the likelihood of spontaneous cessation of the bleeding (7). In the present series, primary bleeding episodes subsided without surgery in 22 of 34 patients. In these patients, a specific medical treatment of CD was initiated consisting mostly of corticosteroids and cyclosporine (in 3 patients). In a recent case report, thalidomide was effective in stopping bleeding. This could be related to its anti-TNF (tumor necrosis factor) effect (13). The other nonsurgical treatments included endoscopic treatment and interventional angiography. Colonoscopy not only allows a precise localization of the bleeding site but also permits to perform hemostatic techniques that, in our series, controlled the hemorrhage in 5 patients by using laser or bipolar coagulation and an adrenalin injection in 1 case. A variable response to intraarterial infusion of vasopressin has been reported, with better results in bleeding originating for hyperemic lesions (6, 8, 11, 14, 15). Vasopressin infusion seems more appropriate than embolization, which was responsible for a bowel necrosis requiring surgical resection in one of our patients. In our series, an uncontrolled hemorrhage required surgery in 7 of 34 patients (20%) with a first bleeding episode. This frequency is lower than reported in previous studies where it ranged from 60% to 91% of the cases (2). This discrepancy may again be linked both to the definition of significant hemorrhage and to the surgical or nonsurgical recruitment of the series. In our series no significant differences of blood requirement were found between male and female patients, as well as between operated or nonoperated patients. The optimal extent of an intestinal resection is a difficult problem in CD. In contrast to general agreement concerning surgery for CD where conservative resection is the goal, it often becomes necessary, if not desirable, to remove all the involved bowel if possible. In our series, 36% of operated patients had a subtotal colectomy with associated ileostomy in 1 case. The frequency of subtotal colectomy is even higher in surgical series ranging from 45% to 70% (1, 3). During follow-up, we observed 12 (35%) recurrent bleeding episodes among nonoperated patients, which is close to the frequency reported in the literature (1). The rebleeding may sometimes occur very late, up to 8 years after the first episode in 1 case of our series. In contrast, there was no bleeding recurrence in our patients with primary excisional surgery. However, according to the literature, primary sur-
AJG – August, 1999
gical treatment does not completely protect against a recurrence. Even here, the frequency of recurrence still ranges from 3.5% to 10% (1, 2). No death was observed in our series, in contrast with reported mortality after either medical or surgical treatment reaching 20% in some series (1, 2). The etiology of massive GI bleeding in CD has not been fully elucidated. One plausible explanation is a transmural inflammation leading to the erosion of large vessels. As noted previously, concomitant administration of steroids did not appear to be involved in the development of severe bleeding because 21 of our 34 patients were not on this form of therapy at the time of hemorrhage. On the other hand, in our series, the bleeding may have been favored in 1 case by concomitant warfarin treatment and in another patient by nonsteroidal antiinflammatory drug intake (16). Finally some reports suggest that an acquired factor XIII deficiency, secondary to plasma protein depletion, may contribute to GI hemorrhage (17).
CONCLUSIONS In summary, massive intestinal hemorrhage remains a rare but worrying complication of CD. The main problems are the identification of the bleeding site, the therapeutic strategy, and the risk of recurrence. The results of this nonsurgical series, the largest to date, demonstrate that a nonsurgical treatment of CD either with specific drugs or with endoscopic hemostasis may be effective in 80% of patients with a first hemorrhage. Furthermore, no death was observed in our series with this conservative approach. Therefore, we recommend this first-line therapy in most hemorrhagic CD. Overall surgical treatment should be recommended in patients with massive bleeding not stabilized by transfusion, and in patients with life-threatening bleeding or recurrent massive hemorrhage. Furthermore, it is possible that hemorrhagic forms of CD represents a particular phenotype of the disease, as many patients tend to rebleed (one-third), whereas overall bleeding rate for CD patients is very low.
ACKNOWLEDGMENT E. Louis is supported by the National Fund for Scientific Research (FNRS). We thank ASTRA (Belgium) for their kind support to the Belgian IBD Research Group. Reprint requests and correspondence: Prof. Jacques Belaiche, M.D., Department of Gastroenterology, CHU Sart Tilman, 4000 Lie`ge, Belgium. Received Dec. 18, 1998; accepted Mar. 30, 1999.
Acute Lower GI Bleeding in CD
2181
REFERENCES 1. Robert JR, Sachar DB, Greenstein AJ. Severe gastrointestinal hemorrhage in Crohn’s disease. Ann Surg 1991;213:207–11. 2. Cirocco WC, Reilly JC, Rusin LC. Life-threatening hemorrhage and exsanguination from Crohn’s disease. Report of four cases. Dis Colon Rectum 1995;38:85–95. 3. Driver CP, Anderson DN, Keenan RA. Massive intestinal bleeding in association with Crohn’s disease. J R Coll Surg Edinb 1996;41:152– 4. 4. Homan WP, Tang CK, Thorbjarnarson B. Acute massive hemorrhage from intestinal Crohn’s disease. Arch Surg 1976; 111:901–5. 5. Lennard-Jones JE. Classification of inflammatory bowel disease. Scand J Gastroenterol 1989(suppl 170):2– 6. 6. Asakura H, Takagi T, Kobayashi K, et al. Microangiographic findings of massive intestinal bleeding in a patient with Crohn’s disease: A case report. Angiology 1985;36:802–7. 7. Ciccarelli O, Coley GM. Massive rectal bleeding in Crohn’s colitis. Conn Med 1985;50:301–3. 8. McGarrity TJ, Manasse JS, Koch KL, et al. Crohn’s disease and massive lower gastrointestinal bleeding: Angiographic appearance and two case reports. Am J Gastroenterol 1987; 82:1096 –9. 9. Smith-Behn J, Banez A, Brown T, et al. Acute massive rectal bleeding as a presenting sign of Crohn’s disease. NY State J Med 1988;88:545– 6. 10. Mesleh G, Lemons J, Lawn O. Massive lower gastrointestinal hemorrhage in Crohn’s disease. Illinois Med J 1983;164: 182–5. 11. Renison DM, Forouhar FA, Levine JB, et al. Filiform polyposis of the colon presenting as massive hemorrhage: An uncommon complication of Crohn’s disease. Am J Gastroenterol 1983;78:413– 6. 12. Dent MT, Freeman AH, Dickinson RJ. Massive gastrointestinal bleeding in Crohn’s disease. JR Soc Med 1985;78:628 –9. 13. Wettstein AR, Meagher AP. Thalidomide in Crohn’s disease. Lancet 1997;350:1445– 6. 14. Podolny GA. Crohn’s disease presenting with massive lower gastrointestinal hemorrhage. AJR 1978;130:368 –70. 15. Mellor JA, Chandler GM, Chapman AH, et al. Massive lower gastrointestinal bleeding in Crohn’s disease: Successful control by intraarterial vasopressin infusion. Gut 1982;23:872– 4. 16. Aalykhe C, Lauritsen JM, Hallas J, et al. Drugs as risk factors for relapse of inflammatory bowel disease: Results of a casecontrol study. Gastroenterology 1998;114:G3757 (abstract). 17. Wisen O, Gardlund B. Haemostasis in Crohn’s disease: Low factor XIII levels in active disease. Scand J Gastroenterol 1988;23:961– 6.
ADDENDUM Since this manuscript was accepted, a new series of 28 cases of major gastrointestinal hemorrhage in CD from the Mayo Clinic has been published (Pardi DS et al. Gastrointest Endosc 1999;49:153–7).
Vol. 94, No. 8, 1999 ISSN 0002-9270/99/$20.00 PII S0002-9270(99)00347-0
Acute Lower Gastrointestinal Bleeding in Crohn’s Disease: Characteristics of a Unique Series of 34 Patients Jacques Belaiche, M.D., Edouard Louis, M.D., Geert D’Haens, M.D., Marc Cabooter, M.D., Serge Naegels, M.D., Martine De Vos, M.D., Fernand Fontaine, M.D., Piet Schurmans, M.D., Filip Baert, M.D., Marc De Reuck, M.D., Rene´ Fiasse, M.D., Jan Holvoet, M.D., Alain Schmit, M.D., Marc Van Outryve, M.D., and the Belgian IBD Research Group Department of Gastroenterology CHU Sart Tilman, Lie`ge; University Hospital Gasthuisberg, Leuven, Brugge; Academic Hospital Free University of Brussels, Brussels; University Hospital, Gent; Clinique Saint Joseph, Lie`ge, Hasselt, Roselare; Hopital Universitaire Brugmann, Bruxelles; Clinique Universitaire St. Luc, Bruxelles, Antwerpen; Universite´ Libre de Bruxelles, Bruxelles; Universitair Ziekenhuis, Antwerpen, Belgium
OBJECTIVE: Acute lower gastrointestinal bleeding is a rare complication of Crohn’s disease, which represents a diagnostic and therapeutic challenge. The aim of this study was to define epidemiological characteristics and therapeutic options of hemorrhagic forms of Crohn’s disease. METHODS: Thirty-four cases of hemorrhagic forms of Crohn’s disease were studied retrospectively. Acute lower gastrointestinal hemorrhage was defined as acute rectal bleeding originating in diseased bowel and requiring a transfusion of at least 2 units of red blood cells within 24 h. Upper gastrointestinal tract hemorrhage or anal lesions and postoperative bleeding were excluded. RESULTS: Mean age at time of hemorrhage was 34.2 ⫾ 14 yr. Mean duration of disease before the hemorrhage was 5.6 ⫾ 6 yr. The hemorrhage occurred during a flare up of the disease in 35% of cases. The hemorrhage revealed Crohn’s disease in 23.5% of cases. The hemorrhage was more frequent in colonic disease (85%) than in isolated small bowel disease (15%) (p ⬍ 0.0001). The origin of bleeding was identified in 65% of cases, by colonoscopy (60%), by angiography (3 patients), or at surgery (1 patient). The bleeding lesion was an ulcer in 95% of cases, most often in the left colon. The treatment was surgical in 20.5% (colectomy in 36%), endoscopical (7 patients, including 5 successes), or medical. Hemorrhage recurred in 12 patients (35%) within a mean time of 3 yr (4 days– 8 yr), requiring surgery in 3 cases. No death was observed. CONCLUSIONS: This study performed in a series characterized by a nonsurgical recruitment, the largest to date, shows that hemorrhagic forms of Crohn’s disease may reveal disease in 23.5%, occurs in quiescent Crohn’s disease in twothirds of cases. Given the potential efficacy of endoscopical or medical treatment, as well as the absence of mortality, a conservative approach may be suggested as first-line ther-
apy in the majority of patients. (Am J Gastroenterol 1999; 94:2177–2181. © 1999 by Am. Coll. of Gastroenterology)
INTRODUCTION Gastrointestinal (GI) hemorrhage in a setting of inflammatory bowel disease has traditionally been associated with ulcerative colitis. Crohn’s disease (CD) is mentioned in most textbooks as an unusual cause of massive lower bleeding occurring in 0.9 – 6% of patients (1–3). However, more recent studies cite an equal frequency of severe gastrointestinal hemorrhage in both CD and ulcerative colitis (1, 3). Hemorrhagic forms of CD are not well known and mainly reported in surgical series (1– 4). They represent a diagnostic and therapeutic challenge. The aim of this study was to define clinical, epidemiological characteristics, and therapeutic options for hemorrhagic forms of CD through a retrospective nonsurgical series of 34 cases.
MATERIALS AND METHODS Patients Thirty-four patients with CD presenting with acute low GI bleeding at several medical centers affiliated with the Belgian Inflammatory Bowel Disease Research Group, were included in this study. CD was diagnosed according to Lennard-Jones criteria (5). Acute lower GI bleeding was defined as an acute rectal bleeding originating in diseased bowel and requiring at least 2 units of blood transfusion within 24 h. Exclusion criteria were as follows: bleeding originating from upper GI tract or anal lesions, and massive hemorrhage occurring within 30 days after surgery. Clinical Data In this retrospective study, the following data were evaluated in each patient: sex, age at diagnosis, age at time of first
2178
Belaiche et al.
AJG – Vol. 94, No. 8, 1999
Table 1. Characteristics of the 34 Patients at Inclusion Characteristics
Male
Female
All Patients
No. of patients Age at onset of disease (yr) Age at time of first bleeding (yr) Duration of disease from onset to bleeding (yr) Familial inflammatory bowel disease (%) Smoker (%) Extraintestinal manifestations Previous intestinal resection (%) Disease location (%) Small bowel only Colon only Small bowel and colon Anal lesions Medical treatment at time of bleeding (%) Steroid Sulfasalazine or mesalamine Azathioprine NSAID Warfarin No treatment
17 29.4 ⫾ 14.6 34 ⫾ 15.5 4.9 ⫾ 6.2 5 7 3 8
17 27.8 ⫾ 12.2 33.3 ⫾ 12.5 6.3 ⫾ 5.9 4 4 4 5
34 28.5 ⫾ 13.1 34.2 ⫾ 14 5.6 ⫾ 6 9 (27) 11 (34) 7 (22.5)* 13 (38)
3 3 11 3
2 7 8 5
5 (15) 10 (29) 19 (56) 8 (23)
4 4 4
6 2 2 1 1 7
10 (29) 6 (17) 6 (17) 1 1 10 (29)
3
Values are expressed as mean ⫾ SD. There was no significant difference between the two groups for any of these parameter distributions. * n ⫽ 31.
bleeding, duration of disease from onset to bleeding, smoking habit, familial history of inflammatory bowel disease, extraintestinal manifestations, previous surgery, disease location (small bowel only, colon only, small bowel and colon, anoperineal disease), medical treatment at time of bleeding, amount of blood transfused, bleeding site, treatment, and follow-up. Activity of CD was judged by each physician on pure clinical basis as, in this retrospective study; Crohn’s disease activity index (CDAI) was not always available. Statistical Methods Comparative statistics were carried out using the one-way analysis of variance (ANOVA), Student’s t test or MannWhitney U test as appropriate. Qualitative variables were analyzed by Fisher’s exact test. Correlations between severity of the hemorrhage and the time to rebleeding was assessed by Pearson test. Follow-up for nonoperated patients was modeled using Kaplan-Meier survival estimates.
RESULTS Patient Characteristics The clinical details of the patients are summarized in Table 1. The mean age at the time of the first bleeding episode was 34.2 yr ⫾ 14 yr (range: 18 –75 yr; median 31 yr). The mean duration of disease from onset to time of hemorrhage was 5.6 yr ⫾ 6 yr (range: 0 –23 yr; median: 5 yr). Twelve patients (35%) had active symptomatic CD and 22 patients (65%) had a quiescent CD at the time of bleeding. Severe bleeding was the first manifestation of disease in 8 patients (23.5%). The hemorrhage was more frequent in colonic disease (85%) than in isolated small bowel disease (15%) (p ⬍ 0.0001).
Blood Transfusion Requirements The mean serum hemoglobin at presentation was 7.8 ⫾ 1.8 g/dl (range: 3.4 –10.3 g/dl; median: 7.9 g/dl). Blood requirements averaged 5.2 units (1 unit ⫽ 400 ml of packed red blood cells) ⫾ 3 units (range: 2–20 units; median: 4 units) per patient. No significant differences were found between male and female patients and between operated and nonoperated patients. Bleeding Site The site of bleeding was identified in 22 patients (65%) (Table 2). Colonoscopy performed in 30 patients revealed a bleeding site in 18 of them (60%). The lesion found was considered to be the origin of the hemorrhage when it was actively bleeding (n ⫽ 16), or when an adherent clot was present (n ⫽ 2). In 95% of cases with identified bleeding site the hemorrhage was attributable to an ulcer or an ulcerated area, most often located in the sigmoid or the left colon. The ulcers involved had various shapes: deep round or serpiginous ulcers (n ⫽ 13), and more rarely superfical (n ⫽ 3) or solitary as a skip lesion (n ⫽ 1). In 1 patient the bleeding Table 2. Localization of the Bleeding Site Methods Colonoscopy (n ⫽ 30) Ulcers
Pseudopolyp Angiography (n ⫽ 4) Surgery
Bleeding Site
No. of Patients
Sigmoid or left colon Right colon Ileocaecal valve Ileum Ileocolic anastomosis Sigmoid Ileum Pseudopolyp of ileum
8 2 3 2 2 1 3 1
AJG – August, 1999
Acute Lower GI Bleeding in CD
2179
Table 3. Type of Surgery in 10 Operated Patients (7 for First Bleeding and 3 for Recurrent Bleeding) Type of Surgery
No. of Patients
Subtotal colectomy Subtotal colectomy with ileostomy Ileocolic resection Partial colectomy Partial colectomy with colostomy
3 1 4 1 1
originated from a large pseudopolyp in the sigmoid. Mesenteric angiography was performed on four occasions and revealed a brisk bleeding site in the distal small bowel in 3 cases. These 3 patients required surgery. In 1 patient a pseudopolyp of the terminal ileum was identified at surgery. The only two isotopic bleeding scan performed were noncontributive. Treatment Initial therapy at the moment of hemorrhage consisted of blood transfusions and supportive measures. A specific medical treatment of CD was started in 24 patients (70%), consisting of corticosteroids in 21 patients and cyclosporine in 3 patients. Endoscopic treatment of the bleeding lesion was attempted in 7 cases: laser coagulation in 3 cases (including 1 case with associated adrenalin injection), bipolar coagulation in 3 cases, and polypectomy of a pseudopolyp in the sigmoid colon in 1 case. These endoscopic treatments were successful in 5 cases. Seven patients (20.5%) were operated on during their first bleeding episode. Five of them had a clinically active CD and 2, a quiescent CD at time of surgery. In all cases, an intervention was required on an urgent basis due to persistent hemorrhage. The operative procedures performed are mentioned in Table 3. In one specific case, an ileocolic resection was performed due to small bowel ischemia after radiological embolization of the ileocecal artery. Follow-Up No death was observed in this series. The mean hospital stay was 14.7 ⫾ 11 days (range: 2– 40 days; median: 13.5 days). In nonoperated patients the treatment at the end of hospital stay comprised sulfasalazine or mesalamine and/or steroids (13 cases), azathioprine (4 cases), azathioprine and steroids (1 case). Long-term follow-up was available in 25 patients (mean: 3 yr, range: 0.5–11 yr; median 2 yr) (Fig. 1). Twelve nonoperated patients (35%) had recurrent bleeding: 9 bled twice, 2 bled three times, and 1 bled four times. Bleedingfree survival curve at 6 mo for these patients is shown in Figure 2. There was no bleeding recurrence in operated patients. Recurrence of bleeding occurred on average 3 yr after the first hemorrhage (mean: 3 mo, range: 4 days– 8 yr; median 1.5 yr). There was no significant difference between patients without recurrence and patients who rebled according to the severity of the hemorrhage assessed by both the hemoglobin level and the amount of blood transfusion required. In patients who rebled, no correlation was observed
Figure 1. Outcome of medical and surgical treatment of acute lower bleeding in 34 patients with Crohn’s disease. There was no bleeding recurrence in patient with surgical resection.
between the delay to rebleeding and the severity of the hemorrhage assessed by the same criteria. Three patients were operated after one (1 case), three (1 case), and four (1 case) recurrent bleeding episodes.
Figure 2. Bleeding-free survival curve at 6 months (Kaplan-Meier) in nonoperated patients.
2180
Belaiche et al.
DISCUSSION Acute lower GI hemorrhage is an unusual complication of CD, occurring in 0.9 – 6% of the cases (1–3). In our retrospective and multicenter series an estimation based on the frequency in the three centers having included ⬎50% of patients gave an incidence of 1.08% (data not shown). The definition of massive GI hemorrhage, which ranges from a very blurred one (need of transfusion for maintening normal vital signs) (4) to the transfusion of at least 5 units of blood (2), can explain this variation. In our study a bleeding necessitating at least 2 units of blood over a period of 24 h was chosen as inclusion criterion. The mean serum hemoglobin concentration at presentation (7.8 ⫾ 1.8 g/dl) and the number of blood units transfused (5.2 ⫾ 3 units) reflect the importance of the hemorrhage. In contrast to previously published surgical studies, the current one is characterized by a nonsurgical recruitment. Furthermore, it represents the largest published study to date. The data about the clinical and epidemiological characteristics of hemorrhagic forms of CD are not well defined in the literature. Our results support the statement that bleeding may occur at any age or stage of CD (2– 4, 6 –9). The oldest patient was 75 yr old, and the youngest was 18 yr, with disease duration before onset of GI hemorrhage of 0 to 23 yr. A recent review reported a nearly 2:1 ratio of men to women (2). The present series as well as that of Robert et al. (1) do not confirm this finding. Gastrointestinal hemorrhage was the initial symptom of CD in 23% of our patients, which is close to the frequency previously reported by Cirocco et al. (2) and confirmed by several case reports (3, 7, 9, 10 –12). Other series suggested a higher frequency up to 50% (3); in contrast, in the Mount Sinai Hospital series, the hemorrhage never revealed the disease (1). These differences may be explained by the different types of patient recruitment, nonsurgical or surgical. We have identified a significantly higher rate of hemorrhage among the patients with colonic involvement than among those with only small bowel disease. This observation is similar to other reports (1, 3, 7, 9, 11) but is different from the review by Cirocco et al. (2) where the site of bleeding corresponded to the general distribution of CD, with ileal disease predominating in 66% of cases. Localization of the site of hemorrhage remains a major challenge in this disease process, which may be along the bowel. The precise identification of the bleeding site may be difficult even at pathological examination of the resected specimens; the precise bleeding site in our study remained unknown in 35% of the cases. In our study, colonoscopy was useful as it identified the bleeding site in 18 patients (60%). Most of bleeding episodes originated from colonic ulcers, either isolated, or in an area of more widespread ulceration. In 2 cases, the ulcer was located at the ileocolonic anastomosis. Very rarely, the bleeding originated from a pseudopolyp, as exceptionally reported (11). However, colonoscopy can be difficult, due to the amount of blood in
AJG – Vol. 94, No. 8, 1999
the colonic lumen. In patients where colonoscopy is not diagnostic, enteroscopy should be evaluated to identify small bowel bleeding lesions. Angiography was also useful in our study, revealing the site of hemorrhage in 3 of 4 patients in which it was performed. Several studies confirmed the diagnostic yield of this exploration, which can demonstrate not only diffuse irregular hypervascularity of the bowel wall and mesenterium but also, as in our patients, an extravasation of contrast material from the ileal branch of the ileocolic artery (1, 2, 4, 6, 8 –11). The management of lower GI hemorrhage due to CD includes both medical and surgical modalities. Conservative therapy has been advocated in view of the likelihood of spontaneous cessation of the bleeding (7). In the present series, primary bleeding episodes subsided without surgery in 22 of 34 patients. In these patients, a specific medical treatment of CD was initiated consisting mostly of corticosteroids and cyclosporine (in 3 patients). In a recent case report, thalidomide was effective in stopping bleeding. This could be related to its anti-TNF (tumor necrosis factor) effect (13). The other nonsurgical treatments included endoscopic treatment and interventional angiography. Colonoscopy not only allows a precise localization of the bleeding site but also permits to perform hemostatic techniques that, in our series, controlled the hemorrhage in 5 patients by using laser or bipolar coagulation and an adrenalin injection in 1 case. A variable response to intraarterial infusion of vasopressin has been reported, with better results in bleeding originating for hyperemic lesions (6, 8, 11, 14, 15). Vasopressin infusion seems more appropriate than embolization, which was responsible for a bowel necrosis requiring surgical resection in one of our patients. In our series, an uncontrolled hemorrhage required surgery in 7 of 34 patients (20%) with a first bleeding episode. This frequency is lower than reported in previous studies where it ranged from 60% to 91% of the cases (2). This discrepancy may again be linked both to the definition of significant hemorrhage and to the surgical or nonsurgical recruitment of the series. In our series no significant differences of blood requirement were found between male and female patients, as well as between operated or nonoperated patients. The optimal extent of an intestinal resection is a difficult problem in CD. In contrast to general agreement concerning surgery for CD where conservative resection is the goal, it often becomes necessary, if not desirable, to remove all the involved bowel if possible. In our series, 36% of operated patients had a subtotal colectomy with associated ileostomy in 1 case. The frequency of subtotal colectomy is even higher in surgical series ranging from 45% to 70% (1, 3). During follow-up, we observed 12 (35%) recurrent bleeding episodes among nonoperated patients, which is close to the frequency reported in the literature (1). The rebleeding may sometimes occur very late, up to 8 years after the first episode in 1 case of our series. In contrast, there was no bleeding recurrence in our patients with primary excisional surgery. However, according to the literature, primary sur-
AJG – August, 1999
gical treatment does not completely protect against a recurrence. Even here, the frequency of recurrence still ranges from 3.5% to 10% (1, 2). No death was observed in our series, in contrast with reported mortality after either medical or surgical treatment reaching 20% in some series (1, 2). The etiology of massive GI bleeding in CD has not been fully elucidated. One plausible explanation is a transmural inflammation leading to the erosion of large vessels. As noted previously, concomitant administration of steroids did not appear to be involved in the development of severe bleeding because 21 of our 34 patients were not on this form of therapy at the time of hemorrhage. On the other hand, in our series, the bleeding may have been favored in 1 case by concomitant warfarin treatment and in another patient by nonsteroidal antiinflammatory drug intake (16). Finally some reports suggest that an acquired factor XIII deficiency, secondary to plasma protein depletion, may contribute to GI hemorrhage (17).
CONCLUSIONS In summary, massive intestinal hemorrhage remains a rare but worrying complication of CD. The main problems are the identification of the bleeding site, the therapeutic strategy, and the risk of recurrence. The results of this nonsurgical series, the largest to date, demonstrate that a nonsurgical treatment of CD either with specific drugs or with endoscopic hemostasis may be effective in 80% of patients with a first hemorrhage. Furthermore, no death was observed in our series with this conservative approach. Therefore, we recommend this first-line therapy in most hemorrhagic CD. Overall surgical treatment should be recommended in patients with massive bleeding not stabilized by transfusion, and in patients with life-threatening bleeding or recurrent massive hemorrhage. Furthermore, it is possible that hemorrhagic forms of CD represents a particular phenotype of the disease, as many patients tend to rebleed (one-third), whereas overall bleeding rate for CD patients is very low.
ACKNOWLEDGMENT E. Louis is supported by the National Fund for Scientific Research (FNRS). We thank ASTRA (Belgium) for their kind support to the Belgian IBD Research Group. Reprint requests and correspondence: Prof. Jacques Belaiche, M.D., Department of Gastroenterology, CHU Sart Tilman, 4000 Lie`ge, Belgium. Received Dec. 18, 1998; accepted Mar. 30, 1999.
Acute Lower GI Bleeding in CD
2181
REFERENCES 1. Robert JR, Sachar DB, Greenstein AJ. Severe gastrointestinal hemorrhage in Crohn’s disease. Ann Surg 1991;213:207–11. 2. Cirocco WC, Reilly JC, Rusin LC. Life-threatening hemorrhage and exsanguination from Crohn’s disease. Report of four cases. Dis Colon Rectum 1995;38:85–95. 3. Driver CP, Anderson DN, Keenan RA. Massive intestinal bleeding in association with Crohn’s disease. J R Coll Surg Edinb 1996;41:152– 4. 4. Homan WP, Tang CK, Thorbjarnarson B. Acute massive hemorrhage from intestinal Crohn’s disease. Arch Surg 1976; 111:901–5. 5. Lennard-Jones JE. Classification of inflammatory bowel disease. Scand J Gastroenterol 1989(suppl 170):2– 6. 6. Asakura H, Takagi T, Kobayashi K, et al. Microangiographic findings of massive intestinal bleeding in a patient with Crohn’s disease: A case report. Angiology 1985;36:802–7. 7. Ciccarelli O, Coley GM. Massive rectal bleeding in Crohn’s colitis. Conn Med 1985;50:301–3. 8. McGarrity TJ, Manasse JS, Koch KL, et al. Crohn’s disease and massive lower gastrointestinal bleeding: Angiographic appearance and two case reports. Am J Gastroenterol 1987; 82:1096 –9. 9. Smith-Behn J, Banez A, Brown T, et al. Acute massive rectal bleeding as a presenting sign of Crohn’s disease. NY State J Med 1988;88:545– 6. 10. Mesleh G, Lemons J, Lawn O. Massive lower gastrointestinal hemorrhage in Crohn’s disease. Illinois Med J 1983;164: 182–5. 11. Renison DM, Forouhar FA, Levine JB, et al. Filiform polyposis of the colon presenting as massive hemorrhage: An uncommon complication of Crohn’s disease. Am J Gastroenterol 1983;78:413– 6. 12. Dent MT, Freeman AH, Dickinson RJ. Massive gastrointestinal bleeding in Crohn’s disease. JR Soc Med 1985;78:628 –9. 13. Wettstein AR, Meagher AP. Thalidomide in Crohn’s disease. Lancet 1997;350:1445– 6. 14. Podolny GA. Crohn’s disease presenting with massive lower gastrointestinal hemorrhage. AJR 1978;130:368 –70. 15. Mellor JA, Chandler GM, Chapman AH, et al. Massive lower gastrointestinal bleeding in Crohn’s disease: Successful control by intraarterial vasopressin infusion. Gut 1982;23:872– 4. 16. Aalykhe C, Lauritsen JM, Hallas J, et al. Drugs as risk factors for relapse of inflammatory bowel disease: Results of a casecontrol study. Gastroenterology 1998;114:G3757 (abstract). 17. Wisen O, Gardlund B. Haemostasis in Crohn’s disease: Low factor XIII levels in active disease. Scand J Gastroenterol 1988;23:961– 6.
ADDENDUM Since this manuscript was accepted, a new series of 28 cases of major gastrointestinal hemorrhage in CD from the Mayo Clinic has been published (Pardi DS et al. Gastrointest Endosc 1999;49:153–7).