Article Text
PDF
Abstract
Objective Severe endoscopic lesions (SEL) in patients with colonic Crohn's disease (CD) have been linked to higher risk of colectomy. The aims of this study were to reassess the predictive value of colonoscopy compared against MRI for requirement of resection surgery in patients with CD and determine the influence of current therapeutic options.
Design In this single-centre, observational, prospective, longitudinal study, patients with an established diagnosis of CD and suspected activity were included. After baseline assessment, including colonoscopy and MRI, patients were followed until resection surgery or the end of study.
Results 112 patients were eligible for analysis. Ulcers were present in 94/112 (84%) of patients at colonoscopy (SELs in 51/112 (46%)) and stenosis in 38/112 (34%). MRI identified ulcers in 79/112 (71%) of patients, stenosis in 36/112 (32%) and intra-abdominal fistulae in 20/112 (18%). Surgical resection requirements (29/112 (26%)) were not associated with the presence of SELs at colonoscopy. The presence of stenosis (p<0.001) or intra-abdominal fistulae (p<0.001) at MRI correlated with a higher risk of surgery. In the multivariate analysis, perianal disease (OR 9 (2 to 39), p=0.003), stenosis (OR 3.4 (1 to 11), p=0.04) and fistulae at MRI (OR 10.6 (2 to 46), p=0.002) increased the risk of abdominal resection surgery, while months under immunomodulators (OR 0.94 (0.90 to 0.98), p=0.002) and/or antitumor necrosis factor (anti-TNF) therapy (OR 0.97 (0.94 to 1), p=0.04) during follow-up decreased this risk.
Conclusions Perianal disease, stenosis and/or intra-abdominal fistulae at MRI independently predict an increased risk of resection surgery in patients with CD, whereas immunosuppressants and/or anti-TNF therapy reduce such risk. Under current therapeutic strategies, the presence of SELs is not a predictor of resection surgery in patients with CD.
- INFLAMMATORY BOWEL DISEASE
- SURGERY FOR IBD
- ABDOMINAL MRI
- ABDOMINAL SURGERY
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Severe endoscopic colonic lesions in patients with Crohn's disease (CD) have been associated with higher requirements for surgical resection.
Optimal use of therapeutic options may reduce the risk of surgical resection in patients with CD.
What are the new findings?
The future need for surgery in patients with CD is best established by a combined clinical and MRI assessment, taking into account the presence of perianal disease, and the presence of stricturing and fistulising lesions.
Patients receiving immunomodulators and/or antitumor necrosis factor therapy have a lower risk of surgery, independently of the presence of other prior factors.
Under current therapeutic algorithms, the presence of endoscopic ulcerations is not a predictor of the need for surgery.
How might it impact on clinical practice in the foreseeable future?
These findings should be taken into account when making therapeutic decisions for patients with CD.
Introduction
Various studies evaluating patients with Crohn's disease (CD) have shown that steroid use, age below 40 years, ileal or ileocolonic location and perianal disease at the time of diagnosis are predictors of a more severe disease course, which in turn is associated with greater surgical requirements.1–3 Although the predictive value of these clinical factors has been confirmed in an independent cohort of patients, the prevalence of at least two of the factors is very high, and these criteria have not been incorporated into therapeutic management decisions in clinical practice.4
In 2002, Allez et al undertook another approach for establishing prognosis in CD based on the severity of endoscopic lesions. This study established relationships between severe endoscopic colonic lesions (SEL) in patients with CD and a more aggressive clinical course, higher risk of penetrating complications and higher rates of abdominal surgery.5 However, these findings have not been validated in an independent cohort of patients.
The aim of this study was to assess the predictive value of identification of SELs at the time of endoscopic examinations, compared with predictive capacity of MRI findings for requirement of abdominal resection surgery in patients with CD. As a secondary objective, the influence of current therapeutic options including immunomodulators (IMM) and antitumour necrosis factor (anti-TNF) agents on disease course was also evaluated.
Methods
Inclusion criteria and baseline examinations
In this observational prospective longitudinal study, patients with an established diagnosis of CD of at least 6 months duration and suspected activity based on presence of clinical symptoms or altered biomarkers were included between June 2007 and May 2011 at the Gastroenterology Department of the Hospital Clínic de Barcelona. Patients who agreed to participate, had a complete initial assessment and at least 4 months of follow-up (one evaluation visit) were included in the study. Assessment of disease phenotype, activity and presence of complications was carried out in the context of three prospective studies evaluating a clinical activity index, biomarkers, endoscopy and MRI.6–8 Clinical disease activity was determined using the Harvey–Bradshaw index. Biomarkers analysed included C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), haematocrit, haemoglobin and albumin.
Endoscopic examinations were carried out by four experienced endoscopists in inflammatory bowel diseases (BG, JL, IO and ER). All patients underwent MRI within 1 week of ileocolonoscopy examination. Patients followed a bowel cleansing protocol as previously described.6–8 The severity and extent of inflammatory lesions was evaluated using the Crohn's Disease Endoscopic Index of Severity (CDEIS).9 For CDEIS calculation, the endoscopic variables were as originally defined: deep ulcers and superficial ulcers (presence or absence), ulcerated surface and affected surface (evaluated on a 10 cm linear analogue scale), and ulcerated and non-ulcerated stenosis. Stenosis was defined, according to the CDEIS, as luminal narrowing, impossible or difficult to pass with an adult endoscope. These variables were evaluated in the terminal ileum, ascending, transverse, descending and sigmoid colon, and in the rectum. Additionally, inflammatory lesions were categorised as SELs or moderate endoscopic lesions (MEL) according to the definitions used by Allez et al.5 SELs were defined as deep ulcerations covering more than 10% of the mucosal area of at least one intestinal segment, and MELs included deep ulcerations affecting <10% of the mucosa, superficial ulcerations, aphthoid lesions, stenosis and pseudopolyps.
All MRI examinations were performed using a 3.0 T MRI unit (TrioTim; Siemens Medical Solutions, Erlangen, Germany), and both small intestine and colon were distended. Methodological procedures and MRI sequences have been previously described.6–8 Predictors of disease severity at MRI were determined by magnetic resonance index of activity (MaRIA) score and its individual components including wall thickness, relative contrast enhancement, mural oedema and mucosal ulceration.6 The presence of internal intestinal fistula (defined as hyperenhancing tracts with or without internal air or fluid that originate from the bowel and connect bowel loops, adjacent organs or the skin), and abscesses were also recorded. Perianal fistulae were not analysed by MRI, as the technical requirements for an MRI examination of perianal lesions are different from the settings used for abdominal examinations. Stenosis at MRI was defined as a narrowing of 50% or more of the bowel lumen, according to Ochsenkühn et al.10
Follow-up
After the initial assessment with clinical, biological, endoscopic and MRI evaluations, medical visits were scheduled every 4 months. Unscheduled visits were made if the patient developed symptoms. Clinical activity assessed by the Harvey–Bradshaw index and biomarkers (CRP, ESR, haematocrit, haemoglobin and albumin) were recorded at each visit. Endoscopy and/or MRI examinations were performed whenever active disease or development of a complication was suspected during follow-up. Medical treatment and resection surgery during follow-up were recorded in an ad hoc case report form. Patients were followed-up until abdominal resection surgery was performed or until the end of the study in October 2013.
Outcomes
The primary outcome was requirement of abdominal resection surgery during follow-up as a binary variable (yes/no). The secondary outcome was time free of abdominal resection surgery. Surgery for perianal disease and abdominal abscess drainage without resection were not considered as surgical events, although the latter did not occur in any case during follow-up.
Statistics
A formal sample-size calculation was not performed for this study. We sought to include a number of patients similar to the study by Allez et al (n=102), in which a prognostic value for the presence of SELs was demonstrated.
Statistical analysis was performed using the statistical package SPSS V.18. Numerical data with normal distribution were analysed by Student t test (or analysis of variance, for more than two groups) and results are presented as the mean plus SD. Non-parametric tests (Mann Whitney or Kruskal–Wallis) were applied in variables lacking normal distribution, and results obtained are expressed as median and percentiles 25 and 75. Categorical data were compared using a χ2 test. Probability of resection surgery was calculated via the Kaplan–Meier method. Logistic regression was used to identify variables with predictive values for resection surgery. Cox regression was applied to examine the influence of variables on the length of time that elapsed until surgery. The significance level α was set at 0.05 two-tailed.
Results
Patients
Initially, 119 patients were enrolled in the study. Seven patients had <4 months of follow-up and were not included in the final analysis (figure 1). Baseline demographic characteristics are displayed in table 1.
Flow diagram of enrolment and patient outcomes.
Demographic characteristics
Colonoscopy was complete (including ileoscopy) in 69 (62%) of the cases, and in 90 cases (80%) reached the caecum. At the time of the baseline colonoscopy, ulcers (deep or superficial) were present in 94/112 (84%) patients (46% were SELs), and stenosis in 38/112 (34%) of cases. The median (P25, P75) CDEIS was 10 (4, 16). At baseline MRI, ulcers were detected in 79/112 (71%) of patients, stenosis in 36/112 (31%) and fistulae in 20/112 (18%) of cases. The median (P25, P75) MaRIA score was 63 (50, 82).
With regard to stenosis, discrepancies were observed in five cases between endoscopy and MRI. In all cases, lesions identified as stenosis at colonoscopy were considered inflammatory lesions at MRI due to marked hyperenhancement without proximal dilatation. One of these cases required resection surgery during follow-up, and both gross pathology examination and histological examination determined severe inflammatory lesions without stenosis. However, three stricturing lesions identified at MRI were not confirmed at colonoscopy, with the endoscopic examination being incomplete, not reaching the level of the stenosis in one case. These three patients underwent resection surgery, which confirmed the presence of stenosis in two cases in the postsurgical examination.
Abdominal surgical resection requirements
The median follow-up until abdominal resection surgery or until end of follow-up was 49 months (22, 65). Over this period, 87 patients (78%) received IMMs and 80 (71%) received anti-TNF agents.
Twenty-nine patients (26%) underwent surgical resection over the follow-up period, 23 out of 29 within the first 2 years after inclusion figure 2). Among the 29 operated patients, 12 (41%) had previously undergone abdominal surgical resection. Indications for abdominal surgical resections were stenosis in 14 cases (48%), penetrating complications in 7 (24%), severe inflammatory activity not responding to medical therapy in 4 (14%), ileal perforation in 1 (1%), dysplasia in 1 (1%), colon perforation after colonoscopy in 1 (1%) and intestinal adhesions in 1 (1%). Resection surgery involved 13 ileal or ileocaecal resections (one associated with upper small bowel resection and another with sigmoid resection), three anastomosis resections, four ascending colon resections, four descending colon or sigmoid resections and five total colectomies. Among the 83 non-operated patients, there were two (2%) who died during follow-up (causes non-related to CD). Balloon dilatation of stricturing lesions was performed in eight patients during follow-up. Surgical resection was required in three of these cases due to primary or secondary failure of endoscopic treatment.
Survival rate free of resection surgery. In this Kaplan–Meier analysis, the intestinal rate free of resection surgery-free is shown. During the first year of follow-up, 15 patients underwent resection surgery (52% of the total number of surgeries) and during the first 2 years 23 (79%).
Predictors of resection surgery
An analysis of the predictors of resection surgery is summarised in table 2.
Factors associated with requirement of abdominal surgical resection
The univariate analysis showed that patients requiring abdominal surgical resection during follow-up had a significantly younger age at diagnosis (23 years (18, 34)) compared with those not needing surgery (28 years (22, 37), p=0.049). Disease duration was longer in patients requiring abdominal surgical resection (9 years (4, 12) vs 5 years (1, 9), p=0.008). Perianal disease was present at the time of inclusion in 32 (29%) patients, and the need for abdominal surgical resection was higher in these patients compared with those without perianal disease (13/32 (41%) vs 16/80 (20%), p=0.02). In this subgroup of patients, the most common indication for abdominal surgical resection was also stenosis (7/13 patients). Biomarkers of disease severity including CRP, ESR, haemoglobin and albumin were similar in operated and non-operated patients. Importantly, disease severity as measured by endoscopy (CDEIS) or MRI (MaRIA), did not differ between patients requiring and those not requiring surgical resection during follow-up.
Surgical resection requirements were not associated with the presence of SELs at baseline colonoscopy or with the identification of ulcerations at MRI. SELs at colonoscopy were identified in 11/29 (38%) patients who underwent surgery and in 40/83 (48%) of non-operated patients (p=0.3). Ulcers were detected by MRI in 22/29 (76%) patients who underwent surgery and in 57/83 (69%) of those who did not (p=0.5). Stenosis at colonoscopy was identified in a numerically higher proportion of patients needing surgical resection 14/29 (48%) vs 24/83 (29%) of those who did not, although this difference was not statistically significant (p=0.06). By contrast, the detection of stenosis and intra-abdominal fistulae at MRI was associated with a significantly increased requirement of surgical resection: stenosis was observed in 17/29 (59%) of operated versus 19/83 (23%) non-operated patients (p<0.001) and fistulae in 12/29 (41%) versus 8/83 (10%) (p<0.001). Over the follow-up period, 87 (78%) patients received IMM and 80 (71%) anti-TNF therapy. Treatment with IMMs was associated with a significant reduction of surgery requirements per month of treatment (OR 0.95 (0.92 to 0.98), p<0.001), whereas this reduction was not significant in the univariate analysis for anti-TNF agents (OR 0.9 (0.9 to 1), p=0.06).
Multivariate logistic regression analysis identified the following independent predictors of abdominal resection surgery: presence of perianal disease at baseline (OR 9 (2 to 39), p=0.003), stenosis at MRI (OR 3.4 (1 to 11), p=0.04) and intra-abdominal fistulae at MRI (OR 10.6 (2 to 46), p=0.002). All were indicators of an increased risk of surgery, whereas duration (in months) of IMM (OR 0.94 (0.90 to 0.98), p=0.002) and anti-TNF therapy during follow-up (OR 0.97 (0.94 to 1), p=0.04) decreased this risk (table 3). The power of the model's predicted values was quantified by the area under the curve (AUC) performed in a receiver operating characteristic curve analysis, with an AUC 0.89 (0.82–0.96), p<0.001.
Predictors of abdominal resection surgery: logistic regression analysis
Multivariate survival Cox's regression analysis was used to identify factors associated with time to surgical resection. It was observed that disease duration at inclusion (HR 1.1 (1.03 to 1.14), p<0.001), presence of perianal disease (HR 3.2 (1.5 to 7.1), p=0.003) and presence of intra-abdominal fistulae at MRI (HR 6.9 (3 to 17), p<0.001) significantly reduced the surgical resection-free survival, whereas months under IMMs (HR 0.95 (0.93 to 0.98), p=0.001) and anti-TNF treatment (HR 0.97 (0.95 to 0.99), p=0.006) during follow-up increased survival free of surgical resection. These results are summarised in table 4.
Factors associated with abdominal resection surgery-free survival: Cox regression model
As in the study of Allez et al, the risk of surgery was evaluated in patients with colonic CD, a subanalysis of the 79 patients with colonic lesions in our cohort was performed, and the results are collated in the online supplementary table S1. In this subgroup of patients, 21 (27%) required abdominal surgical resection during follow-up. The proportion of patients requiring abdominal surgical resection was similar in those with SELs (9/44, 20%), compared with those without SELs at baseline colonoscopy (12/35, 34%), p=0.17. As in the whole cohort, in the subgroup of patients with colonic lesions, both stenosis (p=0.01) and fistulae (p=0.002) at MRI were associated with a higher risk of abdominal resection surgery.
Discussion
In this observational prospective longitudinal study, the presence of perianal disease and identification of stenosis or fistulae at baseline MRI were independent predictors of an increased risk of abdominal resection surgery in patients with CD, whereas duration of treatment with IMMs and/or anti-TNF agents reduced this risk.
The first remarkable finding of the current study is the inability to replicate the observation by Allez et al5 showing that the presence of SELs at endoscopy carries an increased risk of surgical resection, despite the fact that the proportion of patients who underwent surgical resection in both studies was similar. Even when we analysed the subpopulation of patients with exclusive colonic involvement, we could not confirm this finding. Various factors may contribute to this discrepant result. The most plausible, based on our own current observations, is the potential effect of cotreatments. The population included in the study by Allez et al was not treated with anti-TNF agents, whereas 70% of the patients included in the current study received anti-TNF therapy during follow-up. Even if anti-TNF therapy heals ulcerations in 24%–50% of patients, such treatment may halt lesion progression and the appearance of more severe (fistulae) or irreversible (fibrotic) lesions.11 ,12 In our cohort, we show that time under anti-TNF therapy was one of the independent predictors associated with lower requirements of resection surgery, which is in keeping with previous observations.13 ,14 Additionally, the proportion of patients receiving IMMs, mostly azathioprine, in the current study, was 78%, whereas this proportion was lower (58%) in the study by Allez et al. We also demonstrate that the length of time under IMM therapy helps protect patients against requiring resection surgery, which is also in keeping with recent population-based studies and meta-analysis.15 ,16 Therefore, it is conceivable that time under medical therapy (anti-TNF agents and/or IMMs) in patients with CD, even if it does not achieve complete healing, may halt lesion progression.
Another important negative finding of the current study is that the severity of a particular flare, measured by clinical indices, biomarkers, a validated endoscopic index of severity or a validated MRI index of severity, is not a predictor of the need for abdominal surgical resection. Presence of perianal disease, younger age at diagnosis and longer disease duration are the only phenotypical characteristics that predict the requirement of intestinal resection surgery. This observation is also consistent with previous data, and we show that it has an independent predictive value even when using the best available assessment tools at hand, including endoscopy and MRI.1 ,17
In recent years, MRI has gained widespread acceptance as a routine tool for assessing CD.18 It has the same diagnostic accuracy as CT and does not expose the patient to ionising radiation. In this study, the presence of stenosis at MRI doubled the risk of subsequent resection surgery, while the presence of intra-abdominal fistulae increased this risk fourfold. Thus, MRI provided the most valuable information for establishing the prognosis for surgical resection in patients with CD. In the current study, we used both intravenous and luminal contrast for MRI examinations, including both oral and rectal administration of luminal contrast. Although administration of rectal contrast generally improves the assessment of inflammatory lesions, it is not essential for identifying stenosis or fistula lesions, which are the relevant findings for establishing prognosis.18 ,19
An intriguing observation of the current study is the apparent discrepancy between MRI and endoscopy in the detection of some stricturing lesions. Individual analyses of these cases suggest that MRI may be superior to endoscopy in categorising the nature of a stenosis based on the characteristics of the lesion and its functional consequence (proximal dilatation). Even with an optimal use of MRI, endoscopy is still indispensable in the initial assessment of suspected CD whenever biopsies are needed during follow-up, and for screening of dysplasia.
This study has some weaknesses. The physicians who recommended surgery were not blinded to any of the assessments carried out in the current study, including clinical evaluations, biomarkers, endoscopies and MRI, and some of the indications for surgery may have been directly related to the observation of stenosis or intra-abdominal fistula at MRI. However, the proportion of cases in which this sequence occurred was low, since only 5/29 (17%) of operated patients underwent resection surgery within 3 months of an endoscopy and MRI assessment. We would also like to emphasise the main strengths of the study, including the prospective design, which resulted in a more robust capture of data, particularly for endoscopy and clinical variables, the fact that colonoscopy and MRI were performed in very close sequence and that both examinations were performed and interpreted by experienced gastroenterologists and radiologists.
In conclusion, the future need for resection surgery in patients with CD is best established by a combined clinical and MRI assessment, taking into account the presence of perianal disease, and the presence of stricturing and fistulising lesions at MRI. Patients receiving IMMs and/or anti-TNF therapy have a lower risk of abdominal resection surgery independently of the presence of clinical and MRI findings. These results need to be validated in an independent cohort before they are integrated into clinical decision algorithms. This could help physicians to select those patients with higher risk of surgery in order to intensify medical therapy or propose an early surgery.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Footnotes
Contributors JP designed the study; JP, ER, IO and AJ-A did patient recruitment; ER, IO, BG-S and JL performed endoscopic examinations; JR and SR did MRI examinations; AR-M, MG, MCM and AJ-A worked on acquisition of data and follow-up; JP and AJ-A were on charge of the analysis and interpretation of data and wrote the manuscript. All the authors helped on the critical revision of the manuscript. JP is guarantor of the article.
Funding The study was supported in part by an unrestricted research grant from Abbvie. JP is supported by grant SAF 2012-33560 from Ministerio de Economía y competitividad.
Competing interests None.
Ethics approval Ethics Committee—Hospital Clínic de Barcelona.
Provenance and peer review Not commissioned; externally peer reviewed.