Article Text

Download PDFPDF

Surgery for adult Crohn's disease: what is the actual risk?
  1. Guillaume Bouguen1,
  2. Laurent Peyrin-Biroulet2
  1. 1Inserm UMR991, Department of Hepato-Gastroenterology, University Hospital of Rennes, Pontchaillou, Rennes, France
  2. 2Inserm U954, Department of Hepato-Gastroenterology, University Henri Poincaré 1, University Hospital of Nancy, Vandoeuvre-les-Nancy, France
  1. Correspondence to Professor Laurent Peyrin-Biroulet, Inserm U954, and Department of Hepato-Gastroenterology, University Hospital of Nancy, Allée du Morvan, 54511 Vandœuvre-lès-Nancy, France; peyrin-biroulet{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


When Crohn and his colleagues first described regional ileitis in 1932,1 intestinal resection was the only effective treatment available for these patients. Until the introduction of steroids during the 1950s, medical treatments for Crohn's disease (CD) were limited to digestive rest or ‘High vitamin, high protein, high carbohydrate, low residue diet with liver, iron and calcium supplements and the judicious use of sedatives and antispasmodics’.2 3

In the mid-1990s, the advent of antitumour necrosis factor α (TNFα) agents changed the treatment of CD refractory to standard medications.4 5 In 2011, the ultimate therapeutic goal in CD should be to reduce the long-term risk for intestinal resection. Despite an increasing use of immunosuppressants and anti-TNFα treatments, surgery is still required for many patients with CD. Whether these drugs impact on reducing the long-term requirement for surgery remains debated.

The aim of this article is to review the risk of surgery before and in the era of biologics (based on infliximab approval in 1998) and to discuss the impact of medications on this risk, with a focus on adult luminal CD. Data were analysed using three types of studies—randomised controlled trials, referral centre studies and population-based studies (table 1).

Table 1

Summary of data available from population-based studies, randomised controlled trials and referral centre studies reporting the overall rate of major abdominal surgery according to the period of enrolment

Randomised controlled trials

Before the era of anti-TNFα

To our knowledge, only one randomised controlled trial has reported the rate of surgery in adult CD before the advent of anti-TNFα therapy. In this randomised crossover study of 11 patients who were treated with azathioprine at a maintenance dose of 50 mg, one patient in each group flared and required surgery (figure 1A and table 1).6

Figure 1

(A) Overall rate of surgery with its range before and in the era of anti-tumour necrosis factor α (TNFα) adapted from data coming from the three types of studies: randomised controlled trials, referral centre studies and population-based studies. (B) Rates of surgery and use of immunosuppressants and anti-TNFα agents from the description of Crohn's disease (CD) in the 1930s until today, and projection of their outcome in forthcoming decades. As depicted in (B) future directions in treatment strategies, with a wider and earlier use of disease-modifying agents, might modify the course of CD and the requirements for surgery.

In the era of anti-TNFα

The landmark ACCENT I trial evaluated the efficacy of infliximab in 573 patients with moderately to severely active CD. At enrolment, patients had median disease duration of between 7.5 and 9.3 years, and between 44% and 60% of patients had previous segmental resections.7 Infliximab scheduled treatment was associated with a twofold lower rate of surgery at week 54 as compared with on-demand treatment (11/385 patients (3%) vs 14/188 patients (7.5%), respectively; p=0.01).8

Another randomised controlled trial compared episodic versus scheduled maintenance treatment with infliximab.9 Baseline characteristics of patients with CD showed a median duration of disease of 7.5 and 8.7 years and a previous segmental resection rate of 48% and 51% in the placebo and control groups, respectively. At week 54, significantly more patients in the episodic strategy group—that is, 14 of 188 (7.4%)—required surgical resections as compared with 22 of 770 (2.8%) patients in the scheduled therapy group (p<0.05).9

The pivotal CHARM trial evaluated the efficacy of adalimumab for the risk of surgery in 778 patients with CD. The rate of surgery at 1 year was nine times greater in the placebo group than in the adalimumab group, 3.8% versus 0.6%, respectively (p=0.001).10

In summary, the requirement for surgery from randomised controlled trials evaluating the efficacy of anti-TNF agents in CD ranged from 0.6% to 7.4% at 1 year of treatment (figure 1A and table 1).

Referral centre data

As outlined above, randomised controlled trials provide robust short-term efficacy data; however, the risk of requiring surgery beyond 12 months is not widely reported. To obtain these medium to longer term rates for surgical requirements, the use of referral centre study data is required.

Before the era of anti-TNFα

Between 1938 and 1970, of 189 patients with CD followed at the Radcliffe Infirmary, Oxford, 74% of patients required surgical resections.11 Similarly, among 592 patients diagnosed at the Cleveland Clinic between 1966 and 1969, 74% (n=438) of patients with CD underwent surgery after a mean follow-up time of 13 years.12

More recently, among 2573 French patients with CD diagnosed between 1978 and 2002, 1070 patients underwent 1426 intestinal resections from 1978 to 2003.13 The cumulative risk for requiring surgery at 5 years was 35% and this rate appeared to be stable over time.13 From an American multicentre study of patients newly diagnosed with CD between 1991 and 1997, 69 patients (20%) required surgery within 3 years.14 In a study from Korea in 96 patients with CD, who never received immunosuppressants or anti-TNFα agents, the cumulative surgery rate increased from 9% after 6 months to 17% after 24–36 months.15

In summary, before the era of biologics, the requirement for surgery in referral centre data ranged from 17% to 35% within 5 years after initial diagnosis (figure 1A and table 1).

In the era of anti-TNFα

The Leuven group recently reported on 614 patients with CD who were treated with infliximab. Over a median time of 4.6 years a total of 24% (144/614) of patients required major abdominal surgery.16

An Austrian study has assessed the outcomes of 153 patients treated with infliximab between 1999 and 2001. At short-term follow-up (12±7 months), surgery had been performed in 8 (14%) of 58 patients with luminal CD. At long-term follow-up (29±8 months), 10 (20%) of 50 patients had been operated on for luminal CD.17

In an Italian study of 40 patients with CD, eight patients (20%) underwent surgery for intestinal strictures and/or abdominal abscesses after a 27 month median duration of infliximab treatment.18

From a well-defined referral centre cohort in France, the ‘Nancy IBD cohort’, among 296 patients newly diagnosed with CD from 2000 to 2008, 26% underwent at least one major abdominal surgical procedure after a median follow-up of 57 months.19 The reported cumulative probabilities for the first major abdominal surgery at 1, 5 and 9 years from the time of diagnosis were 7, 26 and 44%, respectively.19 From the same referral centre, 53 adult patients were treated with adalimumab as a second-line therapy; the cumulative probability of major abdominal surgery was 18% at week 130.20

In summary, in the era of biologics, the need for surgery in referral centres ranged from 18% to 33% within 5 years after diagnosis (figure 1A and table 1). Sensitivity analysis excluding the cohort study in which only a subset of patients had received anti-TNFα therapy19 did not significantly modify these results.

Population-based cohorts

Population-based studies are the optimal approach to reflect the natural history of CD.21 22

Before the era of anti-TNFα

In a Europe-wide study, 40% of 706 patients with CD diagnosed between 1991 and 1993 underwent surgical intervention during the first 10 years after diagnosis.23 In the IBSEN study from Norway of 237 patients newly diagnosed with CD from 1990 to 1993, the probability of surgery was 14, 27 and 38% at 1, 5, and 10 years, respectively.24 In a study in Copenhagen County, surgery was performed in 131 (35%) of 373 patients with CD in the year of diagnosis between 1962 and 1987,25 and the cumulative probability of surgery was greater, reaching 55% after 10 years.26 In a study in Stockholm County, 1936 patients were diagnosed with CD between 1955 and 1989; the cumulative rate of intestinal resection was 44, 61 and 71% at 1, 5, and 10 years, respectively.27

From 1986 to 2003, 341 patients were diagnosed with CD in a study carried out in Cardiff and 148 (43%) patients underwent surgery. The population was divided into three separate time periods (1986–1991, 1992–1998 and 1998–2003). During the two first periods, the cumulative probability for surgery was 25–32% and 37–59%, at 1 and 5 years, respectively.28 The latter period will be discussed in the following section.

In the Olmsted County study, of the patients with CD diagnosed between 1935 and 1975, 41% underwent at least one surgical procedure after a median follow-up of 8.5 years from initial diagnosis.29 A recent update demonstrated that of all patients with CD diagnosed between 1974 and 2004, 152 patients (49%) underwent surgery and the cumulative risk for first intestinal resection was 38, 48 and 61% at 5, 10 and 30 years, respectively.30 In the Statistics Canada's Health Person Oriented Information Database evaluating patients with CD from 1994 to 2001, surgical resections per year remained stable, involving 43% of all patients with CD.31

In summary, before the era of anti-TNFα therapies, the rate of surgical requirements ranged from 27% to 61% at 5 years (figure 1A and table 1).

In the era of anti-TNFα

In the Copenhagen County study, of the 209 patients with CD diagnosed between 2003 and 2005, 12% (n=25) had a surgical resections performed within 1 year after diagnosis, with a median time to resection after diagnosis of 1 month (range 0–8 months).32 In the Stockholm County study, between 1999 and 2001, 191 patients with CD received a mean of 2.6 infliximab infusions (range 1–11). A total of 33% of patients required major surgical interventions.33

From the Cardiff cohort, the cumulative probability for surgery was 19% and 25% at 1 and 5 years, respectively, for the latter chronological period (1998–2003).28

In summary, in the era of anti-TNFα therapy the rate of surgery ranged from 25% to 33% within 5 years (figure 1A and table 1).

Impact of medications on the requirement for surgery

In a French referral centre, despite an increasing use of immunosuppressant treatments, the rate of surgery was stable over time at 32, 37, 42, 41, 50 and 45% in the time periods 1978–1982, 1983–1987, 1988–1992, 1993–1997 and 1998–2003, respectively.13 However, these results should be interpreted with caution as <10% of patients during this period received azathioprine before surgery.13

In another referral centre study from the University of Pittsburgh, the medical records of 227 patients with CD from 1995 to 2008 were reviewed and divided into four separate time periods.34 While anti-TNF use increased from 0% (in 1995–1998) to 35% of patients (in 2005–2008) (p=0.0002), the annual rate of small bowel resections did not change over time (1.6% to 1.9%, p=0.93).34

In a recent report, two patient cohorts with newly diagnosed CD followed in a Spanish referral centre were reviewed.35 The first cohort had included 146 patients from 1994 to 1997 and the second, 182 patients from 2000 to 2003. A total of 14% of patients received infliximab in the second cohort. The cumulative probability of intestinal resection did not change significantly between groups; 11, 15 and 21% at 1, 2 and 3 years after diagnosis, respectively.35

From the ‘Nancy IBD cohort’, long-term anti-TNFα treatments are associated with a lower risk for surgery whereas azathioprine only modestly lowers this risk.19

Two American population-based cohorts assembled from nationwide inpatient samples of hospital admission with CD did not observe a decrease in overall rate of surgery between 1993 and 2004 or from 1998 to 2005, but neither assessed the impact of medications.36 37

In a population-based cohort from Manitoba of 1737 patients with CD, 126 patients received infliximab, and 33 (26%) underwent surgery, with no overall difference in surgerical procedures between infliximab and steroids after initial drug prescription.38

In a Danish population-based cohort, Vind et al compared their cohort assembled between 2003 and 2005 with a previous cohort formed between 1962 and 1987, and observed a significant reduction in early operation rates (12% vs 35%, p=0.0001).25 32

The Cardiff group assembled data between 1986 and 2003 and observed a significant decrease in the cumulative probability for surgery at 5 years from 59% to 25% for each chronological group that was associated with early use of thiopurines (HR 0.47, 95% CI 0.27 to 0.79; p=0.005). The impact of infliximab could not be evaluated as it was prescribed in only 22 (16%) patients.28


Despite a significant decrease from the 1930s up to the 1990s, the requirement for surgery remains high in the biologic era (figure 1A,B). Up to one-third of patients with CD require major abdominal surgery at 5 years in both referral centre studies and population-based cohorts (figure 1A). Data from both referral centre studies and population-based cohorts suggested only a modest or no decrease in the need for surgery in the era of biologics. The reduction of surgery in randomised controlled trials should be interpreted with caution as nearly half of the patients had been previously operated on after a median disease duration of 8 years and because the randomised controlled trials were not designed to assess the impact of anti-TNFα therapy on surgery rates.

Unfortunately data on the therapeutic impact of disease-modifying agents, such as immunosuppressants and anti-TNFα therapies, to reduce the requirement for surgery beyond 1 year are scarce. Only two retrospective studies suggested that azathioprine and anti-TNF agents may reduce the need for surgery in CD.19 28

One might argue that early introduction of anti-TNFα therapies may further reduce the need for surgery (figure 1B). In the step-up/top-down trial, the rate of surgery was similar in both groups,39 but no definite conclusions can be drawn due to small sample size. In addition, the authors used an empirical definition of early CD.39 A formal definition of early CD is eagerly awaited.40 In the SONIC trial, the rates of surgery have not been published.41 A disease modification trial comparing a top-down approach (anti-TNFα with concomitant immunosuppressant) with a step-up approach according to current recommendations42 and enrolling several hundred patients with a well-defined definition of early CD and with long-term follow-up could address this issue further. The development of large population-based cohorts to assess the impact of current therapeutic strategies on the long-term need for surgery appears more realistic.

In a similar approach to rheumatoid arthritis, the definitions of new outcome measures and new end points, such as mucosal healing, bowel damage scoring systems and disability,43 44 are a prerequisite to the development of disease-modifying anti-CD drugs capable of reducing the long-term requirements for surgery.



  • Competing interests GB received lecture fees from Abbott and Sherring Plough. LPB received consulting and lecture fees from Abbott and Merck.

  • Provenance and peer review Commissioned; externally peer reviewed.