Given the lack of overt clinical symptoms in many patients with POR, multiple methods of POR detection have been investigated. Endoscopy is perhaps the most well studied. The importance in detecting endoscopic recurrence of CD lies in the downstream effects. Rutgeerts et al. demonstrated that endoscopic disease severity at 1 year directly correlated with progression to symptomatic recurrence and most strongly pre dicted clinical outcomes [3]. The authors then suggested an endoscopic grading system, the Rutgeerts score, identifying key endoscopic findings (Table 5.2) that correlated with outcomes. The Rutgeerts scoring system defines disease severity based on the extent of aphthous ulceration in the neoterminal ileum. Absence of lesions is classified as Rutgeerts i0, five or fewer aphthous ulcers; i1, more than five aphthous lesions with normal intervening mucosa or larger skip lesions; or i2, lesions confined to the ileocolonic anastomosis. Diffuse neoterminal ileitis defines i3, and the addition of large ulcers (≥5 mm), nodules, and/or luminal narrowing delineates the most severe classification, i4.

For patients with Rutgeerts score i0 or i1 at 1 year, only 8.6% had clinical symptoms at 8 years [3]. Conversely, patients with Rutgeerts score i4 had a 100% symptomatic recurrence rate at only 4 years. Those with severe endoscopic recurrence (i3–i4) were the most likely to progress to another Crohn’s disease-related surgery [3, 8, 9]. Consequently, postoperative clinical studies have designated endoscopic recurrence using the Rutgeerts scoring system as i2–i4, whereas endoscopic remission includes i0–i1. This designation of endoscopic recurrence or remission using the Rutgeerts score has not been validated as a measure of treatment response. Intra-observer reliability using the Rutgeerts scoring system has been shown to be fair to good with kappa between 0.43 and 0.67 [10, 11]. The point of most discrepancy likely results from the difference between i1 and i2 endoscopic appearance as the addition of a si ngle aphthous ulcer can upgrade an i1 lesion to i2. Despite the limitations, due to the correlation with clinical outcomes, the Rutgeerts scoring system has stood as the gold standard as detection of POR.

While sensitive for detecting recurrence, ileocolonoscopy is an invasive and somewhat costly procedure with associated risks. As such, there have been efforts to identify noninvasive detection methods of POR. One such method is fecal calprotectin . Fecal calprotectin (fCal) is a molecule produced by mucosal leukocytes and epithelial cells as sites of mucosal injury.

Initial studies evaluating the utility of fCal as a marker of POR were conflicting. Lasson et al. reported there was no difference in fCal levels in postoperative CD patients with endoscopic recurrence compared to patients with endoscopic remission at 1 year [12]. However, this study was limited by small size (n = 30). A subsequent, larger study of 86 asymptomatic postoperative CD patients demonstrated significantly higher levels of fCal in patient with endoscopic recurrence (i2–i4) than those in endoscopic remission (i0–i1) (mean ± s.e.m.: 473 ± 78 μg/g vs. 115 ± 18 μg/g; p < 0.0001) [13]. The same study suggested a cutoff value of 100 μg/g to detect endoscopic recurrence with a 95% sensitivity, 54% specificity, 69% positive predictive value (PPV), 93% negative predictive value (NPV), and 73% overall accuracy. In a meta-analysis of ten prospective studies totaling 613 postoperative CD patients, Qiu et al. estimated a pooled sensitivity of 82% (95% CI 73–89%) and pooled specificity of 61% (95% CI 51–71%) for detecting endosco pic recurrence with an overall PPV of 2.11 (95% CI 1.68–2.66) and NPV 0.29 (95% CI 0.197–0.44) [14]. Furthermore, these authors also analyzed fCal for detection of clinical recurrence and found a pooled sensitivity of 59% (95% CI 47–71%) and pooled specificity of 88% (95% CI 80–93%) with PPV of 5.10 and NPV 0.47. This study suggests that fecal calprotectin can be a useful, noninvasive screening tool for detecting of POR. In an analysis of a randomized controlled control of postoperative CD patients undergoing colonoscopy at 6 and 18 months with fCal measurements, Wright et al. found similar predictive capability of fCal and suggested a potential avoidance of colonoscopy in 47% postoperative patients using the testing characteristics of fCal [15].

Furthermore, Wright et al. also investigated fCal levels as a marker of response to treatment. In their study, patients randomized to receive step-up postoperative medical therapy or not (see POCER trial discussion under “Postoperative Prophylaxis” section). The authors found fCal concentrations significantly decrease in response to intensification of drug therapy in patients with evidence of endoscopic recurrence (from 324 to 180 μg/g at 12 months (p = 0.005) and to 109 μg/g at 18 months (p = 0.004)), whereas patients in endoscopic remission who did not step up medical therapy had increasing fCal concentrations (from 129 to 153 μg/g at 12 months (p = 0.194) and to 178 μg/g at 18 months (p = 0.245)) [15]. This suggests that fCal may also serve as a noninvasive, indirect measure of treatment response in treatment of POR.

The utility of serum inflammatory marker C-reactive protein (CRP) in predicting POR has been analyzed in several studies with discordant results. Boschetti et al. collected CRP data in their 86 asymptomatic postoperative CD patients and found a weak but significant difference in CRP concentrations between patients with endoscopic remission and endoscopic recurrence (3.0 ± 0.7 and 8.5 ± 1.4 mg/L, respectively; p = 0.001) [13]. Furthermore, a significant increase of CRP levels according to Rutgeerts score was also observed (p _trend = 0.02), but without significant differences between individual subscores. When compared to fCal, CRP was less accurate (53% vs. 77% for fCal) in predicting endoscopic recurrence, and the area under the curve for fCal was 0.86 compared to <0.70 with CRP suggesting fCal as the superior testing modality. Conversely, in the same randomized control trial for step-up medical therapy following surgical resection in CD patients, Wright et al. also collected CRP data and found that CRP was not significantly correlated with endoscopic recurrence (Rutgeerts i2–i4) or scored endoscopic severity (i0–i4) [15]. Given the conflicting results, further studies are needed on the utility of CRP in predicting endoscopic and clinical recurrence postoperatively.

Noninvasive radiographic studies including abdominal ultrasound have also been investigated in detecting POR. A study of traditional transabdominal ultrasound (TUS) in 32 CD patients who had undergone one or more intestinal resections revealed an accuracy of 93.7% in detecting POR confirmed by radiography and endoscopy and biopsy, 82% sensitive and 100% specificity when using a bowel wall thickness >5 mm as a positive detection [16]. This study was limited by small number of POR occurrences (n = 9). These findings were corroborated by Andreoli et al. in 41 postoperative CD patients with TUS and concurrent ileocolonoscopy using the same bowel wall thickness cutoff with 81% sensitivity, 86% specificity, 83% accuracy, 96% PPV, and 57% [17].

The addition of contrast improves the capability of US in a technique termed small intestine contrast ultrasonography (SICUS). Using SICUS and an oral contrast solution with a decreased bowel wall thickness cutoff of 3 mm for at least 4 cm at the perianastomotic area, bowel dilation (>25 mm), or stricture (<10 mm), Calabrese et al. analyzed 72 postoperative CD undergoing ileocolonoscopy and SICUS within 6 months and found an increased sensitivity of 93% [18]. Bowel wall thickness also strongly correlated with Rutgeerts score (p = 0.0001, r = 0.67). These findings were supported when using intravenous contrast-enhanced US as well. Paredes et al. using cutoffs of >5 mm bowel wall thickness or >46% contrast enhancement determined a 98% sensitivity, 100% sensitivity, 100% PPV, and 92% NPV for detecting endoscopic recurrence (i1–i4) [19]. While suggesting the utility of abdominal ultrasound in detecting POR, the clinical usefulness of these techniques in the United States remains limited due to the requirement of experienced radiologist with a dvanced training.

Many studies have evaluated factors influencing the development of POR. These are outlined in Table 5.3. These factors can be divided into patient-oriented, disease-related, and surgery-specific characteristics. The strongest and most consistent patient-specific factor is cigarette smoking after surgery. Sutherland et al. demonstrated both 5- and 10-year recurrence rates were significantly increased in smokers (36% and 70%, respectively) than in nonsmokers (20% and 41%, respectively) with an odds ratio (OR) of 2.1 (p = 0.007) [20]. Women smokers were also found to be at higher risk than men who smoked (OR 4.2; 95% CI 2.0–4.2 women; OR 1.5; 95% CI 0.8–6.0 men). The risk of recurrence with smoking is also dose dependent with patients smoking ≥15 cigarettes daily having higher rates or POR and other studies reporting a clear dose response [21, 22]. Patients who quit smoking postoperatively have a POR risk similar to nonsmokers. In a questionnaire study of 267 CD patients following ileocecal resection, Ryan et al. found that patients who quit smoking following surgical resection had significantly lower relative incidence rates (RIR) for one, two, and three reoperations for POR at any site (RIR 0.25, 95% CI 0.15–0.41; RIR 0.30, 95% CI 0.16–0.57; and RIR 0.25, 95% CI 0.10–0.71, respectively) as well as recurrent ileocecal CD (RIR 0.27, 95% CI 0.15–0.47) [23]. Thus postoperative smoking represents a significant modifiable risk factor for POR.

There are several patient-related factors that have had discordant associations resulting in inconclusive interpretation. Patient age at onset of disease has been evaluated in several studies with conflicting results. It is possible that positive association with disease recurrence could be related to increased duration of follow-up and thus likelihood of recurrence over time rather than a true causative relationship [24].

Similar to patient age at disease onset, shorter duration of disease prior to surgical resection may be a factor influencing POR, though this is still in question due to conflicting results. Varying definitions of “short duration” in individual studies have hampered pooling and comparative studies. One could imagine that a shorter duration of disease prior to requiring surgical resection may reflect a more aggressive disease phenotype, thus placing the patient at higher risk of POR.

A family history of inflammatory bowel disease was demonstrated by Unkart et al. to convey a 2.2-fold increased risk of repeat surgery in 176 postoperative CD patients though this finding has not been replicated [25].

There have also been studies evaluating genetic risk factors for POR. Fowler et al. examined 194 CD patients who underwent bowel resection with 69 patients requiring repeat resection. Patients who were homozygous for SMAD3 risk allele were independently associated with increased risk of repeat surgery (hazard ratio [HR] 4.04, p = 0.001) [26]. Similarly, Germain et al. in a study of 200 genetic variants demonstrated that patients with CARD8 risk allele homozygosity carried a sevenfold increased risk of surgical recurrence compared to non-risk allele carriers (OR 7.56, 95% CI 1.13–50.37) [27]. Several studies have examined the role of NOD2 (also known as CARD15), which has been previously associated with ileal and stricturing disease. These studies revealed conflicting results. A meta-analysis of six cohort studies comprising 1003 CD pat ients examining the risk of NOD2 polymorphisms suggested an increased risk of further surgical resection but failed to reach significance (OR 1.58, 95% CI 0.97–2.57, p = 0.06) which the authors felt was likely due to study heterogeneity (Cochran Q: 12.36, p = 0.03, I: 59.6%). Lastly, interleukin-10 has been studied by Meresse et al. in a group of 36 postoperative CD patients and did not detect any association with endoscopic recurrence [28]. Consequently, there likely exists various genetic signatures which may predispose patients to POR; however, the current strength of data is suboptimal, and larger cohort studies with defined and consistent protocols are needed.

Disease behavior is a frequently cited risk factor for surgical resection with stricturing and penetrating phenotypes at increased risk of surgery. However, relating disease behavior to postoperative recurrence is difficult given the fluctuating nature of CD and changes in the behavior pattern over time and in response to medical therapy. In a meta-analysis of 12 studies examining postoperative recurrence, Pascua et al. found that penetrating/fistulizing phenotype was a risk factor for endoscopic recurrence (OR 1.59, 95% CI 1.37–1.84 for every 10% placebo-treated patients with fistulizing disease) [29]. In the same study, patients who had prior surgery for CD indications were at significantly increased risk of POR (OR 1.14, 95% CI 1.04–1.26 for every 10% increase). This risk association has been replicated in other studies as well. Simillis et al. demonstrated that patients who have surgery with a particular disease behavior often have recurrence of that same behavior requiring reoperation [30]. It follows that any history of CD-related surgeries, regardless of disease behavior, is a strong predictor of postoperative recurrence. However, it should be noted that most studies did not differentiate between penetrating complications related to stricturing disease and de novo perforating disease without stricture.

The requirement of certain medications prior to surgery has also been shown to predict the risk of postoperative recurrence. The use of anti-TNF therapy presurgery has been associated in several studies to predict higher rates of POR [31, 32]. The medication themselves are not likely responsible for the disease recurrence, but they are more likely a reflection of disease activity, severity, or complicatio n(s) prior to resection.

Anastomotic technique has been suggested as influencing POR. A difference in outcomes has been postulated from the wider luminal capacity of a stapled anastomosis preventing fecal stasis and bacterial overgrowth compared to a hand-sewn end-to-end anastomosis. Yamamoto et al. followed 45 patients who underwent stapled side-to-side anastomosis (“functional end to end”), and 78 underwent conventional sutured end-to-end anastomosis and found that cumulative 1-, 2-, and 5-year ileocolonic recurrence rates requiring reoperation were significantly lower in the stapled anastomosis (0%, 0%, and 3%, respectively) compared to sutured end to end (5%, 11%, and 24%, respectively, p _log-rank = 0.007) [33]. These findings have been corroborated in several other, mostly retrospective, studies [24]. However, in two prospective randomized controlled trials of anastomosis type in 98 and 139 CD patients, both studies failed to show a significant difference in either clinical or endoscopic recurrence by an astomotic type.

Three studies have independently found myenteric plexitis to be a significant predictor of POR, both endoscopic and clinical [34, 35]. Furthermore, the severity of plexitis appears to correlate with severity of endoscopic recurrence at both early (3 months) and later (12 months) time points.

Characteristic findings in the surgical specimen have also been investigated as potentially related to POR. The degree of histologic inflammatory activity has been shown in several studies to correlate with increased rates of anastomotic recurrence in ileocolonic CD [24]. The presence of granulomas in surgical pathology has contradictory data with several large studies favoring a predisposition to POR if the surgical specimen contained granulomas [36–38]. However, the significance of this histologic finding in relation to POR remains uncertain.

Several early reports suggested an association between wide macroscopic margins and lower recurrence risk. Fazio et al. conducted a randomized controlled trial of 152 CD patients who underwent ileocolonic resection to limited (2 cm) or extended (12 cm) margin from macroscopic disease [39]. There were no significant differences in recurrence rates between the groups (25% limited, 18% extended). Of the group with microscopic activity at the margin, 31.7% had recurrence, whereas 17.8% of activity-free margin patients had POR though this difference failed to reach significance (p = 0.07). Thus, margin size or histologic activity does not seem to influence POR.

There is increasing evidenc e that biologic agents are the most effective therapy to prevent POR. The most well-studied agents in this class are the antitumor necrosis factor alpha (anti-TNFα) agents. The first report of successful use of prophylactic infliximab (IFX) in a CD colitis patient after a partial colonic resection occurred in 2006 by Sorrentino et al. [46]. Since this initial description, multiple studies have focused on the role of anti-TNFs in preventing POR. Regueiro et al. performed the first randomized, placebo-controlled trial examining the ability of IFX (initiated within 4 weeks of surgery) to prevent endoscopic recurrence 1 year after ileocolonic re section [47]. In this study of 24 CD patients at moderate to high risk for POR, patients randomized to IFX had significantly lower rates of endoscopic recurrence compared to placebo (1/11, 9.2% vs. 11/13, 84.6%, p = 0.0006). Following these patients out to 5 years postoperatively, patients assigned to IFX continued to have significantly lower rates of endoscopic recurrence (22.2% vs. 93.9%, p < 0.0001) and longer mean time to first endoscopic recurrence (1231 ± 747 days vs. 460 ± 121 days, p = 0.003) [48]. Patients who were initially assigned to IFX had significantly longer time to repeat surgery (1798 ± 359 days vs. 1058 ± 529 days, p = 0.04). Those who stayed on IFX for a longer period also had significantly lower rates of surgical recurrence (20.0% vs. 64.3%, p = 0.047) suggestive of a maintenance effect of prophylactic IFX. This effect was further shown by Sorrentino et al. when patients maintained on IFX (5 mg/kg) for 3 years postoperatively had IFX stopped [49]. Of 12 patients who had no evidence of endoscopic or clinical recurrence prior to cessation of IFX, 10/12 (83%) developed endoscopic recurrence after 4 months without IFX. Mucosal integrity was restored with retreatment with lower-dose IFX (3 mg/kg every 8 weeks). Yoshida and colleagues similarly demonstrated a durable effect of IFX when following 31 postoperative CD patients who were maintained on 5 mg/kg every 8 weeks IFX (n = 15) or placebo (n = 16). Both arms received oral mesalamine 1.5 g/day for trial duration. They found significantly higher rates of maintained clinical, serologic (CRP), and endoscopic remission in patients treated with IFX than placebo [50].