Current thinking on recurrence

McLeod RS, Wolff BG, Ross S, et al. Recurrence of Crohn's disease after ileocolic resection is not affected by anastomotic type: results of a multicenter, randomized, controlled trial. Dis Colon Rectum. 2009;52:919–927. An ambitious randomized controlled study, involving 17 participating sites in the United States, Britain, and Canada, compared rates of endoscopic and symptomatic recurrence in 66 patients who had a stapled side-to-side anastomosis following ileocecal resection for Crohn's disease (CD), versus 73 patients who had a handsewn end-to-end anastomosis. For purposes of the study, endoscopic recurrence was defined as any recurrence with more than 5 aphthous ulcers at the time of a mandated 12-month colonoscopy (i.e., a Rutgeert's score of 2 or higher). Symptomatic recurrence was defined as the presence of endoscopic disease plus symptoms that “were severe enough to require medical or surgical treatment” based on the Crohn's Disease Activity Index (CDAI) score, C-reactive protein, and the subject's unspecified “symptom status.” At 1 year of follow-up, the 2 groups had similar rates of endoscopic recurrence (37.9% versus 42.5%, P = 0.55) and symptomatic recurrence (22.7% versus 21.9%, P = 0.92). The authors then performed a secondary analysis, which considered other factors that might be associated with recurrence: length of time since diagnosis, number of previous resections, smoking, the presence of fistula or abscess at surgery, length of small bowel affected by CD, type of surgery (open or laparoscopic), postoperative azathioprine therapy, compliance with postoperative azathioprine maintenance, and CDAI at 6 weeks postoperatively. On multivariate logistic analysis, endoscopic and symptomatic recurrence were each higher in patients who had a previous resection and lower in patients who were compliant in azathioprine maintenance therapy. (Anastomotic technique was not included as a variable in the secondary analysis, perhaps due to the assumption that randomization obviated the need to control for confounding.)1 Surgeons have long hypothesized that stapled anastomoses—with their longer length and side-to-side orientation—were associated with a lower rate of recurrence following ileocecal resections for CD. The hypothesis was based on several assumptions. Staples might have lower immunogenicity than sutures. They might well reduce the effects of colonic reflux. In addition, a wider, stapled anastomosis would logically lead to a decrease in fecal stasis. Consistent with these assumptions, some early data tended to support the hypothesis that stapled anastomoses were associated with better long-term recurrence rates. For example, 1 retrospective study compared 69 patients with stapled anastomoses to an equal number of patients with handsewn anastomoses. The study reported that patients who had a stapled anastomosis had a 12% symptomatic recurrence rate at 1 year, as compared to 28% in those with end-to-end handsewn anastomoses.2 At 5 years, 20% of patients treated with a handsewn anastomosis needed repeat surgery, whereas only 11% of the patients in the stapled anastomosis group did. Another retrospective study of 141 consecutive patients also reported a decreased rate of surgical recurrence in patients with wider side-to-side anastomotic lumens. The effect did not depend on whether the anastomosis was handsewn or stapled. As a result, these authors suggested that the configuration of the anastomosis matters, not necessarily by altering rates of endoscopic recurrence, but by delaying the onset of obstructing symptoms.3 Still other studies, however, did not seem to support this hypothesis.4,5 Moreover, 1 meta-analysis of 661 patients with CD (who collectively underwent 712 anastomoses) concluded that the 2 anastomotic techniques did not have a statistically significant difference in recurrence.6 This analysis included and weighed very heavily a randomized controlled study by Cameron et al.7 However, the Cameron et al study randomized patients to a handsewn end-to-end versus a handsewn end-to-side, i.e., none of the anastomoses that Cameron et al analyzed generated the wide lumen usually seen with a functional end-to-end. (If, as some have postulated, the functional end-to-end anastomosis leads to a decrease in symptomatic recurrence due to its mere length, then the Cameron et al study, which did not analyze functional end-to-end anastomoses, may be of less than optimal utility.) Against this somewhat confusing backdrop, one can only applaud the extraordinary effort that has gone into the present, multicenter endeavor. Unfortunately, the results of the endeavor, so far at least, may still fall short of settling the issues definitively. For example, study accrual was stopped at 171 patients because, the authors say, recruitment was difficult and slow. As a result, the authors tell us, the study was only powered to detect an absolute difference of 23% between the cohorts. None of the prior retrospective studies reported such an enormous difference in recurrence rates between anastomotic techniques at just 1 year. In other words, the study was not adequately powered to discern the 1-year differences that prior studies had suggested one might see. (To rule out a 10% absolute risk reduction, which is typical of most retrospective studies on the topic, with a 5% level of significance and a 90% power, the study authors would have needed to continue accruing patients until they had 309 in each study arm.) One may also raise concerns about the randomization of patients in the study. On close examination, the patients treated with a stapled anastomosis seem to have had a 9% higher rate of previous resections and a 5% lower rate of postoperative azathioprine maintenance, both of which the study identified as predictors of recurrence on multivariate analysis. We are told that the differences between the 2 groups favored the handsewn arm but were not statistically significant; i.e., the absolute difference did not reach 23%. Still, one may wonder about the equivalency of the 2 groups, and wish that the study design had better controlled for such an important predictor of recurrence. Had the authors included stapled anastomoses in their multivariate model, the question might have been addressed. (On the other hand, one can also appreciate the study authors' decision not to do this, since a truly stable model should have included no more than 3 predictors, for 31 symptomatic recurrences, and their model had 9 already.) Finally, the study reported that every fifth patient had a symptomatic recurrence at 1 year, a rate that seems rather high. One part of the explanation may be that the authors' “symptomatic recurrences” are not very well defined. Were these symptomatic recurrences obstructive in nature (emesis, crampy discomfort, need for nasogastric tube placement for decompression)? Or were they inflammatory (abdominal pain, malaise, and diarrhea)? Both presentations might prompt initiation of medical therapy and an increase in the CDAI and the C-reactive protein, criteria used to define symptomatic recurrence in this study. However, the first scenario is far more likely to lead to a surgical reintervention, and the second is not. By not differentiating between the 2 categories of symptoms, the study authors may have missed an important distinction that, at least in theory, could be seen as an advantage of a wider anastomotic lumen. In the end, it may well be true, as the study authors suggest, that anastomotic technique does not alter the natural progression of CD, at least on a pathological level. In analyzing the topic, this study represents a promising beginning. However, it falls short of providing a definite answer either to the question of whether technique has an impact on disease progression or to the further question of whether, when recurrences do occur, a particular anastomotic technique can forestall or prevent reoperation and reintervention.