Predicting Outcomes to Optimize Disease Management in Inflammatory Bowel Diseases

Background and Aims: Efforts to slow or prevent the progressive course of inflammatory bowel diseases [IBD] include early and intensive monitoring and treatment of patients at higher risk for complications. It is therefore essential to identify high-risk patients – both at diagnosis and throughout disease course. Methods: As a part of an IBD Ahead initiative, we conducted a comprehensive literature review to identify predictors of long-term IBD prognosis and generate draft expert summary statements. Statements were refined at national meetings of IBD experts in 32 countries and were finalized at an international meeting in November 2014. Results: Patients with Crohn’s disease presenting at a young age or with extensive anatomical involvement, deep ulcerations, ileal/ileocolonic involvement, perianal and/or severe rectal disease or penetrating/stenosing behaviour should be regarded as high risk for complications. Patients with ulcerative colitis presenting at young age, with extensive colitis and frequent flare-ups needing steroids or hospitalization present increased risk for colectomy or future hospitalization. Smoking status, concurrent primary sclerosing cholangitis and concurrent infections may impact the course of disease. Current genetic and serological markers lack accuracy for clinical use. Conclusions: Simple demographic and clinical features can guide the clinician in identifying patients at higher risk for disease complications at diagnosis and throughout disease course. However, many of these risk factors have been identified retrospectively and lack validation. Appropriately powered prospective studies are required to inform algorithms that can truly predict the risk for disease progression in the individual patient.


Introduction
Crohn's disease [CD] and ulcerative colitis [UC] are chronic gastrointestinal inflammatory diseases characterized by disabling bowel symptoms. Ongoing inflammation leads to progressive bowel damage and complications, often requiring surgery. These diseases are associated with significant morbidity, resource utilization and costs to society. Furthermore, as these inflammatory bowel diseases [IBD] often present early in life, they can compromise education, career development and family planning.
The recognition that chronic uncontrolled inflammation in IBD ultimately results in poor outcomes has led to a recent paradigm shift in treatment, with the belief that early intervention with immunosuppressant and biological therapy can prevent disease progression and avoid complications. 1,2,3,4,5 However, treatment of all patients with biologics and/or combination therapy is economically unsustainable and would risk exposing those with an indolent disease course to unnecessary risks or side effects of potent therapy. The challenge remains to select the patients who will benefit most from early intensive therapy, while sparing those who will derive minimal benefit from such treatment. 6 The ability to predict specific disease complications such as progression of phenotype from inflammatory to penetrating or fibrostenotic disease, need for surgery or development of dysplasia or cancer would be of particular value, as would being able to identify 'red flags' that could alert the clinician to an impending flare or relapse.
As part of the IBD Ahead 2014 educational programme, we conducted a comprehensive literature review to identify predictors of long-term IBD prognosis and generate draft expert summary statements relating to prognosis. Statements were refined at national meetings of IBD experts and were finalized at an international meeting in November 2014. Here we present the agreed statements, together with a summary of published evidence, to guide clinicians on the best use of these predictors in the individual patient.

Methods
In February 2014, the Global Steering Committee [GSC] of the IBD Ahead 2014 educational programme identified key topics of interest or uncertainty in understanding prognostic factors in IBD and developed clinical questions relating to these topics. Six bibliographical fellows [FC, KK, TL, DM, JT and MZ] were nominated by the GSC to identify and evaluate the published evidence on prognostic factors under the mentorship of JOL, GVA, DR, J-FC and JCL, respectively. PubMed, Embase and the Cochrane Library were searched using predefined search strings and limits, and additional searches were conducted by hand as required. Searches were restricted to manuscripts published in English after 1993. Abstracts from the following conferences were also searched: European Crohn's and Colitis Organisation and simplified the statements. National meetings were held in August-November 2014 in the 32 participating countries to allow participants to review, vote and provide their expert opinions and local perspective on the statements. Based on this feedback, a set of consolidated statements was generated by the GSC focusing on predictors of long-term IBD prognosis. An international meeting was held in November 2014 with 99 experts from each of the participating countries. Participants voted on their level of agreement with each statement using a scale of 1 to 9 [where 1=strong disagreement and 9=strong agreement]. If ≥75% of participants scored within the 7-9 range, then the statement was deemed to be agreed upon. If <75% of participants scored within this range, the statement was debated and revised, and a second vote was taken. Again, if ≥75% of participants scored within the 7-9 range, the statement was deemed to be agreed upon. If agreement was not reached at this stage, a lack of agreement was noted. The agreed statements and a summary of the supporting evidence are presented here. CD is an inflammatory disease that may affect the entire GI tract, with repeated flares resulting in bowel damage that leads to a complicated disease course and short-and long-term disability. Therapy may involve repeated courses of corticosteroids and ultimately surgery. 7 Our literature search identified a number of studies evaluating risk factors for complicated CD outcomes [ Table 1]; however, interpretation was hindered by the lack of consistent or standard definitions of disease progression. Interestingly, we did not find any studies documenting risk factors for bowel damage or participationrelated disability. Two large tertiary-centre studies provided the greatest insight into clinical predictors of disability and severe disease in CD. 8,9 The primary outcome in these studies was disabling disease, defined as a requirement for more than two steroid courses, steroid dependence, hospitalization for disease flare or complication, disabling chronic symptoms for a cumulative time of more than 12 months, or need for immunosuppressive therapy, intestinal resection or surgery for perianal disease. Variables assessed included gender, ethnicity, age at onset, disease location, previous appendectomy, smoking status, EIMs, perianal lesions at diagnosis and the need for steroids for treating the first flare-up. In the first study, an initial requirement for steroids ( 9 Another study found that older age at diagnosis and the absence of perianal lesions predicts a milder CD course. 10 One of the most important clinical risk factors for disease progression is disease location [ileal and ileocolonic vs colonic, rectal involvement, and upper GI involvement]. [11][12][13][14][15] In a population-based cohort from North America, disease in the terminal ileum, ileocolonic disease and upper GI involvement were strongly associated with change in disease behaviour in a multivariate model. 16 Other factors that may act as predictors of disease progression include smoking, endoscopic disease severity and serological status.

3.
In several studies in CD patients, active smoking was associated with stenosing/fistulizing behaviour, 17,18,19 the need for steroids or immunomodulators, 20 requirement for surgery 21,22 and disease recurrence following surgery. 23,24 In one study, passive smoking was associated with increased need for immunosuppressants and biologics in CD patients. 25 In the only study identified that specifically assessed the value of endoscopic findings in CD, severe endoscopic lesions [defined as extensive and deep ulcerations covering more than 10% of the mucosal area of at least one segment of the colon] at index colonoscopy were associated with penetrating complications in CD. 26 While the prognostic value of inflammatory markers on CD course is limited, 27,28,29 results from several independent adult and paediatric cohorts have indicated that circulating antibodies against bacterial antigens are associated with complicated CD and evolution towards stricturing or penetrating behaviour. 30,31 Interestingly, a direct correlation was observed between the magnitude of immune response to microbial antigens and frequency of penetrating/stenosing disease in children. Additionally, paediatric CD patients who were positive for one or more immune response progressed to penetrating or stricturing disease sooner after diagnosis compared with those negative for all immune responses. 31 An association between mutations in the NOD2 gene and complicated disease phenotype has been suggested in several independent cohorts 32,33 and a meta-analysis. 34 However, the low effect size of other CD-associated genetic polymorphisms means that a very large sample size is required to make true associations between genotype and phenotype. Current studies have lacked power to establish a true association and do not adequately take into account the effects of age, disease location, smoking and other variables on phenotype. Moreover, several studies have reported that NOD2 mutations are specifically associated with stricturing ileal disease, which almost certainly accounts for the reported association with increased rates of surgery. 35,36 Indeed, this association disappears if the data are corrected for disease location. 37

Question 2: What are the prognostic factors for surgery or multiple surgeries in CD?
The cumulative probability of surgery in CD has been evaluated in several population-based cohorts. 21,38,39,40 A number of clinical risk factors have been associated with an increased need for surgery [ Table 1]. In adult patients, younger age at diagnosis is a prognostic factor for surgery. 38,39,41 Within the paediatric population, available data suggest an opposite trend, however, with younger age at diagnosis associated with a decreased risk of surgery. 42,43,44 One potential explanation for this observation is that younger paediatric patients present more commonly with isolated colonic disease, 42 which has been repeatedly associated with lower surgical rates [see below].
Disease location is among the most important prognostic factors for surgery in CD, and should actively be incorporated into clinical decision-making. Disease located in the small bowel [ileal/ileocolonic disease] has been consistently identified as an independent risk factor for surgery in adult populations, 21,38,45,46,47,48 possibly because small bowel disease is more frequently associated with penetrating and stenosing behaviour. 22 Disease located in the jejunum and upper GI tract [L4] is also indicative of higher surgical risk for similar reasons, 11,12,13,14,15,49 while colonic disease [L2] is protective against major surgery. 22,42,49,50,51 Penetrating and stricturing disease at diagnosis is possibly the most important independent factor associated with the need for surgery. 21,38,42,48 Furthermore, patients who have surgery for penetrating complications have a higher likelihood of being re-operated on and a shorter time to second surgery. 15,52 Extensive and deep ulcerations on index colonoscopy have been shown in a retrospective study to represent an independent risk factor for surgery, in patients with colonic disease. 26 A large systematic review and meta-analysis found that the risk of any CD surgery was increased by 58% if any NOD2 mutation was present, with a pooled sensitivity of 41% and a specificity of 74%. 34 A meta-analysis of seven cohort studies and four case-control studies showed an association between ASCA-positive status and surgery risk [OR 1.64; 95% CI 1.37-1.95]. 53 In addition, greater immune responses to anti-Cbir1, anti-ompC, ASCA and perinuclear anti-neutrophil cytoplasmic antibody [p-ANCA] were shown to be predictive of surgery in a paediatric population. 43 The increased risk of surgery associated with these genetic and serological markers probably also reflects their association with ileal and ileocolonic disease and complicated disease behaviour.

Question 3: What are the prognostic factors for hospitalization in CD?
Hospitalization is generally regarded as a marker of high disease activity or severity in CD. However, our review of the literature revealed that very few studies have assessed predictive factors of hospitalization in patients with CD [ Table 1]. Generally, risk factors associated with disease progression and risk for surgery are also associated with higher likelihood of hospitalization because they are markers of disease aggressiveness and severity. In a population-based study from Olmsted County, MN, USA, factors associated with time to first hospitalization included ileocolonic disease (hazard ratio [HR] 3.3; 95% CI 1.8-5.8), small bowel disease [HR 3.4; 95% CI 1.9-6.1] and gastroduodenal disease [HR 4.0; 95% CI 1.2-13.8] as opposed to colitis. 21 Compared with non-penetrating and non-stricturing disease, penetrating disease also increased the risk of first major abdominal surgery [HR 2.7; 95% CI 1.1-6.7]. 21  Permanent stoma may be a therapeutic necessity in patients with refractory rectal or perianal CD and severe Cohn's proctocolitis. Identification of risk factors at an earlier stage of disease may guide more intensive medical intervention, preventing the development of strictures and fistulae that ultimately result in the need for irreversible surgery. In our review of the evidence [primarily in patients with perianal and/or colonic CD], the most frequent factors independently associated with a permanent stoma were: complex perianal fistulae, 55,56 anal-canal stricture, 57 perineum involvement and perineal granulomas, 58 perianal sepsis, 59 faecal incontinence, 55 colonic CD 59 and distal colonic involvement. 57,60 Additionally, patients undergoing rectal resection or temporary faecal diversion for perianal disease control have a higher rate of permanent faecal diversion. 55,61 Patients with permanent stoma have typically experienced a greater number of previous abdominal surgeries than those without a permanent stoma. 55,59 It is important to acknowledge that the likelihood of successful stoma reversal after temporary diversion for control of perianal sepsis or disease is low.
Intestinal failure is a rare complication of CD, with a lack of consistent risk factors reported in the literature. Nevertheless, bowel-preserving strategies, such as strictureplasty or stricture balloon dilation, should be used whenever possible to prevent intestinal failure.

Question 5: What are the prognostic factors for proximal disease extension in UC?
UC is a dynamic disease, with up to 50% of patients progressing from limited forms of disease [proctitis, left-sided colitis] to more extensive forms of colitis [extensive colitis, pancolitis]. 62,63 The extent of colitis is clinically relevant, as extensive colitis is associated with higher hospitalization rates, need for corticosteroids, greater likelihood of surgery, and increased risk of progression to dysplasia and colorectal cancer [CRC]. The ability to identify patients who are likely to experience disease extension would allow close monitoring and tight control, and perhaps more intensive treatment. It is important to note that most studies on this topic address association of potential risk factors with extensive colitis, rather than predicting proximal extension of disease, and do not allow assessment of whether the same factors that cause proximal disease extension are also predictive of extensive disease from the onset. This analysis was limited to studies that specifically evaluated risk factors for proximal disease extension [ Table 2]. In a paediatric cohort with UC, a delay in diagnosis of more than 6 months and a family history of IBD were associated with increased risk of proximal disease extension [OR 5.0; 95% CI 1.2-21.5 and OR 11.8; 95% CI 1.3-111.3, respectively]. 64 In an adult cohort, independent factors associated with disease proximal extension were younger age at diagnosis [HR 0. 98 65 More than three relapses in one year, a requirement for systemic steroid or immunosuppressive treatment and non-smoking were associated with risk of proximal extension in a retrospective cohort study of adults with UC. 66 Increased severity of disease and use of corticosteroids upon diagnosis were significantly associated with proximal disease extension in adults with ulcerative proctitis. 67 3.6. Question 6: What are the prognostic factors for acute severe UC? Acute severe UC is a medical emergency that requires intensive medical therapy or colectomy; however, robust prognostic factors for this event have not been established [ Table 2]. In a retrospective analysis of a cohort of 750 patients in the UK diagnosed with UC from 1996 to 2001, acute severe UC occurred more frequently in patients with more extensive disease, a younger age at diagnosis and a shorter duration of disease. 68 More recently, Cesarini et al. have shown that the likelihood of developing acute severe UC within 3 years of diagnosis was increased in patients with extensive colitis [E3 from Montreal classification] who presented with C-reactive protein [CRP] >10 mg/l and low haemoglobin [<13.5 g/dl for men or 12.1 g/dl for women]. 69 Data from this single-centre study require validation in external cohorts.
In the absence of studies evaluating acute severe UC as a definitive endpoint, hospitalization due to UC flares may act as a surrogate marker. A concurrent diagnosis of PSC may be a protective factor for hospitalization in UC patients, 70 which is in accordance with studies suggesting that PSC carries a milder course of colonic activity in UC patients. 71 Smoking has also been associated with a lower hospitalization rate in UC; 72 conversely, quitting smoking increases the risk of hospitalization. 73,74 Finally, extensive colitis, 75 cytomegalovirus infection 76 and C. difficile infection 77 are risk factors for hospitalization for UC flare.

Question 7: What are the prognostic factors for colectomy in UC?
In patients with UC, colectomy is undertaken in the emergency setting [typically for complications of fulminant colitis], in hospitalized patients who are non-responsive to maximal medical therapy and as elective treatment in patients who have persistent symptoms despite medical therapy, dysplasia or CRC. We recognize that it is important to distinguish prognostic factors according to colectomy context [  78 A review of the literature found that disease extent, particularly extensive disease 24,79 and pancolitis, 80 may predict subsequent colectomy. Longer disease duration was also identified as a risk factor, both in adults 24,81 and in children. 80 However, caution is needed when interpreting the evidence as the data are cumulative and are confounded by the differences in follow-up duration. Patients who are hospitalized at UC diagnosis 78,82 or require recurrent hospitalization for UC management 82 should be carefully monitored for colectomy indicators. While the presence of PSC reduces the risk of hospitalization for UC [discussed previously], it is also a risk factor for colectomy, primarily due to the associated risk of colorectal neoplasia. 83 There is limited evidence that active smoking is protective against colectomy 84 and smoking cessation in patients with established UC has not been shown conclusively to influence the rate of colectomy. 73

Question 8: What are the prognostic factors for dysplasia and CRC in IBD?
Early intervention with immunosuppressive or biological therapies may have contributed to an overall decline in the incidence of CRC in patients with IBD, although it is evident that some subgroups of patients with UC remain at increased risk of CRC. 85 In patients with IBD, independent risk factors for the development of CRC or highgrade dysplasia include longer duration of IBD 85,86,87,88 and greater extent of colonic involvement at diagnosis. 86,87,89 The excess risk becomes significant 7-10 years after diagnosis in the general IBD population 90 although it is immediately seen in patients with PSC. 85 In UC, disease extent is also a predictor for CRC development, with increased risk in patients with pancolitis compared with left-sided or distal colitis. 87,89,91,92 A number of studies have shown that the concurrent presence of PSC is a strong risk factor for dysplasia or CRC development, 85,93,94,95,96,97 warranting special surveillance protocols. Furthermore, when assessing for site of cancer development, PSC was a predictor of cancer proximal to the splenic flexure. 96 In a study in patients with colonic CD, the presence of PSC was weakly associated with the development of dysplasia or CRC. 98 Two referral centre studies have demonstrated an association between histological inflammation score and the development of dysplasia or CRC in patients with UC, 99,100 and one case-control study found that every 1-unit increase in histological score independently increased the odds of colorectal neoplasia by a factor of 4.69 [95% CI 2.10-10.48]. 101 The possible relationship between elevated CRP or erythrocyte sedimentation rate and CRC 102 may also support the association between poorly controlled inflammation and cancer risk.
Among IBD patients with a first-degree relative with CRC, the relative risk [ 109 One cohort study found that age at diagnosis and male gender were independently associated with mortality in patients with UC; in patients with CD, only age at diagnosis was statistically significant. 110 Another study found an independent association between PSC and mortality in IBD patients. 111

Implications for practice
Stratifying patients and individualizing therapy in IBD should be an ongoing process. It is important for treating physicians and patients to understand which factors are associated with different outcomes as this may influence important therapeutic decisions. Evaluating these factors should arguably be the first step in stratifying patients into low-risk or high-risk groups and should drive treatment target discussions. It is equally important to actively pursue and document resolution of inflammation and adjust therapy accordingly [treat to target], offering the clinician the opportunity to improve patient outcomes at every stage of the disease process. 112 Postponing adequate therapy in patients with aggressive disease may result in disease progression and complications; however, treating all patients intensively may lead to over-treatment and expose some patients to unnecessary risks of immunosuppression. 113 Therefore, it is important to be aware of the risk factors associated with specific complications and to use these to tailor therapy. Briefly, patients with CD presenting with a young age at diagnosis, extensive anatomical involvement, deep ulcerations at endoscopy, ileal or ileocolonic involvement, perianal disease and/or severe rectal disease and penetrating or stenosing behaviour should be regarded as those with the highest probability of developing complications and therefore merit intensive therapy and close follow-up. 114 Patients with UC presenting at young age at diagnosis, with extensive colitis and frequent flare-ups needing steroids or hospitalization bear a higher risk of colectomy. Concurrent presence of PSC, smoking status and superimposed infections may impact the course and activity of UC; therefore, these are clinical risk factors needing attention from the clinician.

Future directions
The use of prognostic factors to guide IBD management is an evolving field. We must be mindful that most of the currently available prognostic factors in IBD are clinical and lack precision. Furthermore, many of these risk factors have been identified retrospectively and have yet to be validated; appropriately powered prospective studies are still required to inform algorithms that can truly predict the risk for disease progression in the individual patient. While use of serological and genetic markers in prognostication has been hampered by lack of sensitivity and lack of wide availability, these are of increasing clinical interest. Indeed, composite scores incorporating clinical information and molecular profiling, and considering harder endpoints such as bowel damage 115,116 and disability, 117 will hopefully allow us to better personalize therapy in the future, both at diagnosis and throughout the disease course. 118

Funding
AbbVie provided funding to the Lucid Group, Loudwater, UK, to manage the IBD Ahead educational programme. AbbVie paid consultancy fees to the Global Steering Committee for their participation in the Global Steering Committee meeting in Copenhagen, Denmark, and speaker fees to the committee members who presented at the international IBD Ahead meeting in Frankfurt, Germany. Travel to and from these meetings was reimbursed. The bibliographical fellows were paid consultancy fees by AbbVie to conduct the literature reviews. No payments were made to the authors for the development or writing of this manuscript. Payment of fees for Open Access was made by AbbVie. Juliette Allport of the Lucid Group provided editorial support to the authors in the development of this manuscript; AbbVie paid the Lucid Group for this work. AbbVie selected the programme chairs and reviewed the final draft of the manuscript. The authors maintained complete control over the content of the paper and it reflects their opinions. All authors made substantial contributions to [1] the conception and design of the study, or acquisition of data, or analysis and interpretation of data, [2] drafting the article or revising it critically for important intellectual content and [3] final approval of the version to be submitted.