Results of the Seventh Scientific Workshop of ECCO: Precision Medicine in IBD—Prediction and Prevention of Inflammatory Bowel Disease

Abstract

Inflammatory bowel disease [IBD] is a complex chronic disorder with no clear aetiology and no known cure. Despite recent advances in overall disease management and improved therapeutics, patients with IBD still experience a substantial burden. Furthermore, as the incidence continues to increase in developing areas of the world, it is expected that the burden of IBD to society will increase and exert tremendous pressure on health care systems worldwide. Therefore, new strategies to prevent the global increase of IBD are urgently required. Data are being progressively acquired on the period preceding disease diagnosis, which support the concept that IBD has a preclinical period that may reveal the triggers of disease and may be amenable to early intervention. Having a better knowledge of this preclinical period will increase the potential not only for improved understanding of disease pathogenesis and improved therapeutics, but also for disease prediction and prevention.

1. Introduction

Inflammatory bowel disease [IBD] is a chronic intestinal disorder. More than 2 million North Americans and 2 million Europeans currently suffer from IBD.^1,2 IBD incidence is steadily increasing in recently industrialiszed areas, such as Asia, Africa, and South America.¹ Patients experience a lifelong disease course and are exposed to a variety of interventions, including medications, hospitalisation, and surgery. This long-lasting disease course may affect the mental and general health of the individual with IBD. At the societal level, IBD results in loss of productivity due to sick leave and premature retirement.^1–3 Although the incidence of IBD appears to have stabilised in most parts of Europe and North America, disease prevalence continues to increase in these regions due to the young age at diagnosis, an ageing population, and decreasing mortality from IBD.³ Hence, the burden of IBD has become a worldwide challenge to health care systems and societies.

Once the diagnosis of IBD is made, bowel damage has already occurred in a significant number of patients and disease is irreversible.^4,5 All currently available therapeutic interventions are directed towards established disease. Drug-free remission is seldom possible, as relapse or disease progression upon drug discontinuation frequently occurs.⁶ To truly change the natural history and long-term consequences of IBD, an effective intervention should ideally occur as early as possible in the disease course and should target the primary processes that trigger the disease or drive disease from a preclinical to clinical stage.⁷ Whereas prediction and prevention of IBD may be an ambitious goal, it is in fact already being implemented in other complex diseases, such as type 1 diabetes and rheumatoid arthritis.⁷ The recruitment of ‘at-risk’ family cohorts, with prospective biospecimen collection, close follow-up, and biobank repositories with access to preclinical samples, has led to improved knowledge of pathogenesis, more refined biomarkers, and reformulation of diagnostic criteria.^8–10 As a first but important step in understanding the preclinical phases of IBD, consensus in the terminology that will be applied to preclinical studies is needed. This is necessary to help define specific research questions and to design the framework for future efforts in this direction.^9,11,12

2. Developing the conceptual network of the preclinical phases of IBD

IBD is diagnosed after a variable period of non-specific clinical symptoms and laboratory abnormalities, leading to endoscopic and histological diagnosis. Many efforts have been undertaken to shorten diagnostic delay, reduce the disabling course of disease, and alter its natural history. However, there is limited research on the period preceding clinical symptoms, and the main triggers of disease initiation and expansion have not yet been identified. Based on the currently available evidence, we propose four biologically plausible stages that describe the preclinical stage of disease [Figure 1]. The preclinical phase of disease may be initiated when interactions between the host and environmental risk factors occur in a genetically susceptible individual [Stage 1—At risk]. Whether environmental exposure(s) occur as single or multiple events, in sequence or cumulatively, or whether there are windows of susceptibility when triggers act more deleteriously, is unknown. Nonetheless, these currently unidentified steps appear to initiate a sequence of events that may occur in tandem, sequentially, or in more than one set order. These events are characterised by altered gut microbiota [dysbiosis], altered intestinal permeability, and dysregulated immune responses [Stage 2—Disease initiation]. During this stage, biomarkers of altered barrier integrity and abnormal innate or adaptive immunity are likely to emerge, which reflect immune activation at the mucosal level initiating chronic inflammation and causing mucosal injury [Stage 3—Disease expansion]. Even if symptoms are still absent, it is likely that a progressive rise in serum and faecal biomarkers of inflammation is occurring from stage 2 to stage 3. However, these biomarkers may or may not be detectable with current methodologies. Mucosal tissue damage is a progressive process that is expected to begin with microscopic changes that evolve towards macroscopic lesions in the gut. Nonetheless, these endoscopically visible lesions may not trigger the initial symptoms until they reach a certain degree of severity that prompts functional and sensory derangements leading to symptoms [Stage 4—Diagnosis].

Whether all patients go through every phase, in a random or sequential order before developing full-blown inflammation and IBD symptoms, or whether some patients rapidly progress from an early phase of subclinical IBD to clinically obvious disease, is not yet known. We should acknowledge that the whole process probably occurs as a continuum and over a period of time that varies from patient to patient. We also hypothesise that some stages may be reversible, whereas established disease is not. Unfortunately, we cannot yet determine the exact sequence of events or the point from where pathological processes become irreversible. Finally, it is also important to highlight that no clear evidence supports the linearity and stepwise occurrence of these four stages, and it remains to be established whether one stage is required before entering the next stage and how these stages overlap.

2.1. Stage 1: at-risk population: genetic, familial, and environmental risk factors

2.1.1. Genetic risk factors

NOD2 was the first susceptibility gene identified in CD,^13,14 and nearly two decades later genome-wide association studies [GWAS] have identified over 240 single-nucleotide polymorphisms [SNP] in IBD. The majority of these are driven by common variants (minor allele frequency [MAF] >5%) of small effect size [increasing risk by 1.1–1.3-fold], which together only explain 13% and 8% of the variance in disease susceptibility for Crohn’s disease [CD] and ulcerative colitis [UC], respectively.^15–17

Considering the polygenic nature of IBD, it is possible that IBD will be better predicted by polygenic risk scores, rather than by a single allelic odds ratio [OR] for a rare monogenic mutation.¹⁸ Initial attempts at calculating genetic risk scores entailed analysing genotype data using the ImmunoChip [Illumina, Inc., CA].¹⁹ The weighted genetic risk score [GRS] was calculated by extracting 163 distinct IBD risk loci. For each risk allele, the genetic contribution was calculated by multiplying the log OR of its association with disease and allele dose [wild type, heterozygous, or homozygous]. UC patients with GRS in the highest quartile were twice as likely to have a first-degree relative [FDR] with IBD [p_trend = 0.03]. Similarly, CD patients with the highest quartile GRS were over two times more likely to present with ileal disease location [p_trend <0.0001] and had an almost 5-fold higher risk of younger age of onset [p_trend = 0.008]. The same group of investigators then calculated a weighted GRS based on data extracted from 201 distinct IBD-risk SNPs²⁰ and reported that patients with familial IBD had a stronger genetic predisposition to develop IBD than those with sporadic IBD [p = 0.006].

Khera et al.¹⁸ used large datasets and improved algorithms to overcome the previous limitations to the creation of a genome-wide polygenic risk score [PRS]. PRS considers not only the IBD-risk variants [identified at a genome-wide significance level], but also is calculated by using genetic variants that have been identified at various levels of p-values. PRSs that include variants with lower p-values seem to improve disease prediction, since they tend to explain more of the phenotypic variance. These novel PRSs may help identify individuals who are at high risk of developing IBD.²¹ Borren et al. demonstrated that individuals with the top 1% of risk scores had a 3.87-fold increased risk for IBD.²⁰ Khera et al. in their large dataset revealed that the polygenic predictor identified 3.2%, 0.8%, and 0.2% of the population with a 3-, 4-, and 5-fold increased risk of IBD, respectively.¹⁸ However, the accuracy of PRS is highly dependent upon matching IBD association data to the population in which the risk is being predicted. Use of association data from populations that differ genetically reduces the performance of PRS, particularly when calculating PRS for individuals of African ancestry.²² Currently, most large-scale IBD association datasets are based on populations of European ancestry, making calculation of accurate PRS in other populations contingent upon additional GWAS studies in non-European populations.

The ‘Genetic, Environmental, Microbial [GEM] Project’ and the ‘Multiplex Orthodox Jewish Family’ study on behalf of the Road To Prevention Study Group have presented data on PRS within unaffected family members. The GEM study, based on long-term follow-up of unaffected FDR of probands with CD, demonstrated that the CD polygenic risk score was significantly associated with the level of faecal calprotectin [FCal]. These findings suggest that there is an influence of genetic factors on preclinical inflammation in unaffected FDR.²³ Spencer et al. reported that a substantive fraction of familial risk in an Ashkenazi Jewish multiplex cohort is driven by the composite of common variation from the over 200 other IBD loci that are also present in non-Ashkena`i Jewish IBD. There were no unaffected individuals above the age of 25 with a PRS in the top 10%, suggesting that children and young adults carrying a PRS in the top 10% in this high-risk cohort should be carefully monitored for disease inception.²⁴

2.1.2. Family history

Population-based twin studies have demonstrated proband concordance rates of 12–18% for UC and 38–60% for CD among monozygotic twins.^25–28 In a nationwide Danish study, up to 12% of all IBD cases were familial cases, with a higher occurrence of familial CD than familial UC. In FDR of patients with CD, the risk of CD was almost 8-fold increased, whereas this risk was increased 2.4-fold and 1.9-fold in second- and third-degree relatives of patients with CD, respectively.²⁹ Although less pronounced, the risk of UC was significantly increased in both first-degree [4-fold], second-degree [1.9-fold], and third-degree relatives [1.5-fold] of patients with UC. The risk of both CD and UC increased with two or more IBD-affected relatives.³⁰ The relatively common familial occurrence of IBD leads to the occurrence of multiplex IBD families, ie, families with several IBD cases. In such families, a strong concordance for disease type and location has been observed.³¹ Importantly, the familial occurrence of IBD is not fully explained by genetics, pointing towards shared environmental factors between family members as a contributor to disease risk.

Co-occurrence of IBD in individuals with other immune-mediated diseases [IMIDs] has also been described. In a nationwide population-based cohort study, 22.5% of IBD patients were diagnosed with another IMID.³² In 80% of cases, the other IMID occurred before the onset of IBD. The most common IMIDs in IBD patients were psoriasis, asthma, type 1 diabetes, and iridocyclitis.³² These findings suggest the existence of shared pathogenic pathways, and may explain the overlapping efficacy of specific medications for different IMIDs.^33,34

2.1.3. Environmental risk factors

2.1.3.1. Prenatal and perinatal risk factors for IBD

Early life exposures have a crucial role in modulating the gut microbiome and maturation of the immune system.^35,36 Therefore, exposures during this crucial stage could contribute to subsequent IBD susceptibility, as supported by mounting epidemiological evidence.^37–40

Mode of delivery is a major modulator of the infant’s gut microbiota.^41–43 Existing data on the possible association between mode of delivery and IBD risk are conflicting.^44,45 However, in the most recent meta-analysis, delivery by caesarean section was not associated with increased risk of IBD (OR: 1.01; 95% confidence interval [CI]: 0.81–1.27), CD [OR: 1.15; 95% CI: 0.94–1.42], or UC [OR: 0.94; 95% CI: 0.61–1.45].⁴⁵

Breastfeeding is of importance for shaping the infant’s gut microbiota, as approximately 30% of the bacteria in the infant gut comes from breast milk.⁴⁶ Human breast milk is enriched with bioactive substances and other immune or non-immune factors that both play crucial roles in shaping immunity.⁴⁷ Studies addressing the relationship between breastfeeding and IBD are associated with pronounced heterogeneity, but a protective effect against the development of CD [OR: 0.71; 95% CI: 0.59–0.85] and UC [OR: 0.78; 95% CI: 0.67–0.91] has been reported.⁴⁸

Heavy metal exposure during intrauterine and early postnatal life was recently identified by Nair et al. as a plausible risk factor for IBD, based on analyses of deciduous teeth.⁴⁹ Increased uptake of lead, copper, zinc, and chromium was observed in the baby teeth matrix of individuals who subsequently developed IBD, when compared with controls who did not have IBD during follow-up. The observed difference was time dependent and occurred at various developmental time periods for each metal.⁴⁹ A previous study did not identify pollutants as a risk factor for IBD, except for the environmental toxicant [p,p’-DDT].⁵⁰

2.1.3.2. Drug exposure

Antibiotic exposure has been observed in several studies to show a positive association with subsequent CD, but not UC.^51–57 A positive dose-response relationship between antibiotic exposure and later CD has also been found in some of these studies. Based on a pooled data meta-analysis, the relative risk estimate seems to be higher for paediatric-onset CD (hazard ratio [HR]: 2.75; 95% CI: 1.72–4.38) than for adult-onset disease.⁵⁷ Intriguingly, exposure to antibiotics during pregnancy increased the risk for very early-onset CD, but not UC, in a nationwide population-based birth cohort study of more than 800 000 Swedish children.⁵⁸

Oestrogen is known to enhance cellular proliferation, influence the immune system, and modify gut barrier function.⁵⁹ Use of oral contraceptives has been associated with increased risk of IBD, primarily CD.^60,61

2.1.3.3. Lifestyle and hygiene

Smoking affects several key mechanisms related to IBD pathogenesis, including gut barrier integrity, immune responses, and the diversity and composition of the gut microbiota. Smoking may also cause epigenetic modifications that influence gene expression.^62,63 Current smokers have an approximately 2–4 fold increased risk of CD and a decreased risk of UC compared with never smokers; smoking cessation is a risk factor for UC.^44,62,64

Physical exercise may downregulate inflammatory cytokine and apoptotic protein expression.^65,66 A prospective study of the women in the Nurses’ Health Study [NHS] I and II cohorts reported a protective effect of physical activity, with a 44% reduction in the risk of developing CD compared with inactive women, but no relationship with UC.⁶⁷ Additionally, analyses of Swedish nationwide registries demonstrated that low fitness among males at military conscription assessments in late adolescence was associated with an increased risk of both CD and UC.⁶⁵

Obesity has not been associated with future development of IBD in a prospective cohort.⁶⁸ In the prospective EPIC cohort, exposure to ambient air pollution was not associated with IBD development⁶⁹; however, some data have suggested that air pollution may correlate with hospitalisations for IBD.⁷⁰ Vitamin D status, a surrogate for sunlight exposure, was not associated with the development of CD or UC.⁷¹

Diet is thought to play an important role in IBD pathogenesis.⁷² The most compelling evidence comes from the epidemiological transition in IBD, with a rapid rise in disease incidence as developing societies adopt a westernised diet.⁷³ Previous cohort studies have shown that higher intake of fibre could protect from CD development.^74,75 A recent large prospective study showed that adherence to a Mediterranean diet may lower the risk for developing CD, but not UC.⁷⁶ Dietary arachidonic acid and oleic acid intake have been positively and inversely associated with UC development, respectively.⁷⁷ No associations have been found between alcohol use and incidental diagnosis of IBD.⁷⁸ In a recently published study using data from the NHS, NHS II, and Health Professionals Follow-up Study, empirical dietary inflammatory pattern [EDIP] scores were calculated. Diets associated with a high inflammatory potential were associated with an increased risk of developing CD but not UC.⁷⁹

2.2. Stage 2: disease initiation

2.2.1. Alterations in gut microbiota [‘dysbiosis’]

The central role of the microbiota in the pathogenesis of UC and CD has long been proposed.^80–82 Whereas variation exists, case-control studies of IBD patients frequently suggest reduced microbial diversity, increased instability, reduction in phyla with anti-inflammatory properties, such as butyrate producers [including Firmicutes], and increased levels of E. coli to be characteristic of IBD.⁸³ However, studies of individuals with established disease do not preclude the possibility that the observed abnormalities are merely a consequence of inflammation rather than having a role in disease pathogenesis. Studies addressing microbial biomarkers as predictors of IBD onset in the general population are difficult to perform, due to the challenges of executing large prospective cohort studies and the uncertainty around the length of the preclinical period. Moreover, defining truly unaffected individuals requires confirmation of lack of disease or intestinal damage; however, it would be difficult if not impossible to perform endoscopic and histological assessments, or even use videocapsule assessments, in this setting. Another study design that may allow the discovery of microbiological markers of IBD risk is to compare unaffected siblings of IBD patients [who themselves have a higher risk of developing IBD than the background population^28,84] with the affected sibling, and more interestingly, with unrelated healthy controls. Several studies have confirmed that healthy but at-risk siblings of IBD patients have altered gut microbiota, confirming an at-risk dysbiosis.^85–89 The majority of studies have revealed that at-risk dysbiosis is a ‘milder’ version of IBD dysbiosis.⁹⁰ This evidence strongly suggests that dysbiosis exists in at-risk individuals and gives hope for the use of microbiological markers to predict IBD onset. Studies of twins discordant for IBD phenotype offer great opportunities for insights, as genetic predisposition and early life and childhood exposures are similar. Small studies on the microbiome of twins discordant for IBD phenotype have been published, highlighting different microbiome composition and altered microbial functional pathways, carbohydrate metabolism, and human host-secreted enzymes in the affected twin when compared with the healthy co-twin.^91–95

Further studies have examined potential mechanisms by which dysbiosis may confer an increased risk of IBD. An example is the presence of the β-glucuronidase locus, possessed by some microbiota including E. coli, which may allow the generation of toxic compounds and is increased in both CD patients and their unaffected relatives.⁹⁶ Recent data suggest that the inheritance of IBD risk may be mediated not only genetically but also through vertical transmission of pro-inflammatory microbiota from IBD mothers to their offspring.⁹⁷ Lower bacterial diversity and an enrichment of Gammaproteobacteria along with a depletion in Bifidobacteria were observed in stool samples collected at the end of the first week of life in infants of IBD mothers when compared with controls. Intriguingly, inoculation of germ-free mice with 90-day stools of infants of IBD mothers led to fewer class-switched memory B cells in the colonic lamina propria and in the mesenteric lymph nodes, and fewer homeostatic IgA+ switched memory B cells in the lamina propria.⁹⁷

Several studies are ongoing, including the GEM project, where FDRs of CD patients are characterised in terms of genotype, environmental exposures, and microbiota and then followed for up to 10 years.⁹⁸ FDRs who develop CD [92 so far] can then be compared in a nested case-control design with matched individuals from the approximately 5000 FDRs enrolled in the study who remain IBD free, to determine if pre-disease gut microbial composition is a marker of future risk of developing CD. Assessing the association of these signals with genetic makeup, markers of the physiological state of gut barrier function, and markers of preclinical inflammation as measured by FCal may provide insights into the relationship of these alterations with the future development of CD. The Exploring MEChanisms Of disease traNsmission In Utero through the Microbiome [MECONIUM] study is enrolling pregnant women with and without IBD to focus in detail on how the process of microbial acquisition in early life may confer risk of IBD.⁹⁷ The ‘Multiplex Orthodox Jewish Family’ prospective cohort study plans to compare the microbiome of the affected versus unaffected siblings within multiplex [more than two FDRs] Ashkenazi Jewish families, and at the same time to prospectively follow and biosample the unaffected FDRs in the hope of identifying enterotype and metabotype markers that identify disease initiation and expansion.⁹⁹

2.2.2. Altered barrier integrity

Altered barrier function can modify the interplay between luminal microbes and the mucosal immune system. However, the exact mechanisms and their sequence are still not clearly understood. Conversely, a primary inflammatory process in the gut could secondarily alter the normal barrier function, leading to an increased intestinal permeability and further bacterial translocation.¹⁰⁰ It has long been suggested that increased permeability is a preceding event in the development of CD. Data from IBD animal models suggest that the gut barrier represents a key component of disease pathogenesis, given that restricted epithelial defects have been observed to cause intestinal inflammation even in the presence of normal gut microbiota and immunity.^101,102

Increased mucosal permeability in FDRs of patients with CD was observed in the 1990s using sugar-probe based methods,^103,104 even though some subsequent studies reported conflicting results.^105–107 Keita et al. reported increased⁵¹ Cr-EDTA permeability and decreased levels of the tight-junction proteins claudin-5 and tricellulin, when ileal biopsies from healthy twin siblings of patients with CD were mounted in Ussing chambers and compared with biopsies from healthy controls.¹⁰⁸ The increased paracellular permeability was primarily seen in monozygotic pairs, pointing to an influence of genetic risk factors. However, a GWAS of 1075 healthy FDRs of patients with CD failed to identify risk loci with respect to intestinal permeability, as defined by the lactulose mannitol fractional excretion ratio.¹⁰⁹

Approximately 10–30% of FDRs of patients with IBD have an altered intestinal permeability,^110–114 but this has not been consistently observed across all populations.^115,116 The GEM study recently showed that abnormal intestinal permeability in FDRs predicted future development of CD. This finding was not confounded by measures of early inflammation, such as FCal.¹¹⁷ Interestingly, this observation was time dependent, as intestinal permeability increased towards disease diagnosis. This was the first study to formally demonstrate that altered intestinal permeability is an early biomarker of risk of CD development, and suggests that an abnormal barrier function may be a primary event that sets the stage for CD onset. How the relationship between altered barrier function and the host genetic and intestinal microbiome leads to disease remains to be fully elucidated. It is tempting to speculate that this increased intestinal permeability is present in the early stages of the pre-disease state in the absence of macroscopic lesions, and this may be expected to promote an increased exposure of the local immune response to microbial antigens and possibly to inappropriate pro-inflammatory responses. This hypothesis is further supported by the presence of altered gut barrier function in FDR, which also seems to be a primary defect in twin studies.¹⁰⁸ Teshima et al. examined small-bowel permeability in a cohort of 223 FDRs of Canadian patients with CD. Relatives with an increased permeability and randomly selected relatives with normal permeability were then examined by video capsule endoscopy. In total, 28% of the relatives with increased intestinal permeability and 20% of relatives with normal intestinal permeability were found to have three or more small-bowel ulcers.¹⁰⁶ This again suggests that altered barrier function precedes mucosal damage and its relationship with genetic or environmental factors,¹¹¹ further exploration is required.¹¹⁸ Ideally, further studies should combine determination of intestinal permeability with endoscopic examination of the intestinal mucosa and a prospective evaluation of the risk of developing IBD.

2.2.3. Innate and adaptive immunity

It has been suggested that dysbiosis, plausibly in combination with impaired barrier integrity, may trigger an abnormal mucosal innate immune response. Activation and recruitment of neutrophils, defined as increased concentrations of FCal, has been reported in several studies of healthy FDRs of patients with IBD, including healthy twin siblings.¹¹⁹ On the other hand, eosinophil activation seems to be a consequence of inflammation, as eosinophil-derived neurotoxin and eosinophil cationic protein levels measured in faecal samples from discordant twin pairs and healthy controls did not differ between the healthy twins and controls.¹²⁰ Increased expression of myeloperoxidase, a marker of activated neutrophils,^121,122 has also been observed in colonic biopsies from healthy twin siblings.¹¹⁹ Increased FCal in ‘at-risk individuals’ has also been identified as a predictive marker of subsequent CD. The activation of innate immunity is followed by activation of the adaptive immune system, via antigen presentation, as demonstrated by the presence of serological markers primarily targeted against microbial antigens. The first study examining the prevalence of serological markers in pre-diagnostic serum samples identified patients with IBD in the Israeli army and cross-linked their records with those from the army’s serum repository.¹²³ Overall, 31% of patients with CD versus 0% of the matched controls were positive for anti-Saccharomyces cerevisiae antibody [ASCA]. Consistently, 25% of patients with UC and 0% of the controls were anti-neutrophil cytoplasmic antibody [ANCA] positive. These findings were confirmed and expanded in the large European Prospective Investigation into Cancer and Nutrition cohort. Pre-diagnosis samples were tested for ASCA IgG, ASCA IgA, pANCA, E. coli outer membrane porin C [anti-OmpC], and CBir1 flagellin [anti-CBir1].¹²⁴ The predictive accuracy of combining markers was higher than the accuracy of each marker alone and increased towards diagnosis.

Zhulina et al. reported increased NFκB activity in mucosal colonic biopsies from healthy twins of patients with CD and UC when compared with controls.¹¹⁹ Data from NHS cohorts showed increased levels of serum interleukin [IL]-6 and high-sensitivity C-reactive protein [CRP] in pre-diagnostic samples from individuals who later developed CD and UC, when compared with matched controls.¹²⁵ Furthermore, there was a positive correlation between the concentration of these markers and IBD risk, but no significant correlation between any of these markers and the period from blood collection to the diagnosis of IBD. However, the median period to diagnosis of CD or UC was 6.6–6.8 years, which may have limited the possibility to detect a correlation. Consistently, elevated erythrocyte sedimentation rate [≥15 mm] in males at military conscription assessments in late adolescence was associated with a diagnosis of CD later in life [HR: 5.95; 95% CI: 4.47–7.92], based on data from Swedish nationwide registries.⁶⁵

A large study using longitudinal pre-diagnostic samples from the US Department of Defense Serum repository, the Proteomic Evaluation and Discovery in an IBD Cohort of Tri-service Subjects [PREDICTS] study, assessed the predictive value of a panel of antimicrobial markers and a proteomic panel [1129 proteins] in the years preceding IBD diagnosis. The study identified 51 proteins that could predict CD with 76% accuracy up to 5 years preceding diagnosis and 87% accuracy at 1 year before diagnosis. The most consistent proteomic markers preceding disease diagnosis were markers of inflammation [CRP, serum amyloid P, trypsin 2], markers involved in cytokine signalling, innate immunity, and response to bacteria [complement factors, TNF-receptor subunits, lipopolysaccharide binding protein, proteinase-3, and several interleukins].¹²⁶ Antimicrobial markers were also significantly higher in preclinical samples from CD patients compared with controls. The combination of selected markers across several years before diagnosis yielded an excellent discriminative probability in developing CD as compared with healthy controls, especially when combined with ASCA IgA.¹²⁶ By mapping the markers to annotated databases, several dysregulated pathways in preclinical CD were identified, such as the lysosome pathway, pathways involved in glycosaminoglycan metabolism, and the complement pathway.¹²⁶ Using samples from the same cohort, antibodies against granulocyte macrophage-colony stimulating factor [GM-CSF] were shown to be upregulated in the years preceding CD diagnosis and to predict complicated disease at diagnosis.¹²⁷ This is interesting, given the role of GM-CSF in the generation and maintenance of immunosuppressive regulatory T cells.^128,129 Together, these findings support the hypothesis of altered host-microbiome interactions as one of the earliest events in CD pathogenesis, and offers the possibility to explore preventive interventions.¹²⁶

2.3. Stage 3: disease expansion

2.3.1. Subclinical inflammation and tissue injury

Activation of the innate and the adaptive immune system may lead to subclinical intestinal inflammation. Mucosal tissue damage is a progressive process that is expected to begin with microscopic changes that evolve towards macroscopic lesions in the gut. This process would be present in the months to years before disease onset, but the duration of this period remains unclear. Recent findings from the MECONIUM study revealed that indirect markers of intestinal inflammation may be detected early in life in babies at risk for IBD based on maternal IBD status.¹³⁰

One assessment method is FCal measurement, which reflects infiltration of activated neutrophils in the mucosa. Indeed, many studies have revealed elevated FCal in unaffected FDRs compared with control subjects. In the GEM study, FCal >100 µg/g was associated with an increased risk of developing CD [HR: 7.76; 95% CI: 3.99–15.11]. FCal was weakly correlated with intestinal permeability at recruitment [R = 0.069; p = 0.012, Spearman correlation].¹¹⁷

Following activation of the immune system, asymptomatic endoscopic lesions may develop. A recent study recruited 480 healthy FDRs [siblings, offspring, or parents] and collected samples for genotyping. A combined risk score was calculated based on a PRS [derived from 72 risk variants] and smoking history. Those individuals falling into the highest or lowest risk-score quartiles were asked to donate blood [for ASCA and inflammatory/immune proteins] and stool [FCal] and undergo a video capsule endoscopy.^127,131 From these, 35% had elevated FCal [≥50 µg/g] and 21% [mostly from the high-quartile risk score, n = 22] presented a Lewis score [LS] ≥135 [abnormal inflammation]. Notably, in 11 participants again from the high-risk group, LS was ≥790 [moderate-to-severe small bowel inflammation] and one individual was diagnosed with CD during the 3-year follow up.^127,131 Although not without limitations, this study represents a first step in stratifying an at-risk population using a multidimensional tool. Those with a higher polygenic risk score, positive family history, and elevated FCal are more likely to present asymptomatic small-intestinal lesions and therefore have, hypothetically, a higher risk of future IBD.^127,131 Additional follow-up of high-risk individuals will clarify the significance of small-bowel abnormalities found by video capsule endoscopy in the risk of developing CD.

2.4. Stage 4: diagnosis

The presence of subclinical endoscopic lesions found incidentally in colorectal cancer screening programmes has been described in different cohorts,^132–139 and can be expected in 0.35% of patients undergoing a colorectal cancer screening endoscopy.¹³³ A positive result in the faecal occult blood test may be due to the ability of the test to detect intestinal inflammation at least in established UC.^140–143 Incidental endoscopic findings of IBD lesions support the hypothesis of a subclinical period where the macroscopic damage is already present, but patients still remain asymptomatic.^132,133,144 It has been described that some of these cases had a negative faecal test in the previous 2 years, suggesting that the endoscopic lesions may have appeared during this period of time. A more detailed description of the histological characteristics of incidental CD has shown some specific features, including a low rate of granuloma [4%].¹⁴⁵ Moreover specific microscopic findings, such as crypt distortion, are associated with progression to symptomatic disease which occurs in approximately one-third of patients.^133,145 However, there remains to be determined the exact risk of progression from these early non-specific lesions to CD or UC.

3. Development of symptomatic disease

It is expected that the presence of mucosal lesions will induce symptomatic disease at some point. However, it is necessary to determine if all of these subclinical endoscopic lesions lead to overt disease and also to establish the influence of different triggers during this process. Interestingly, during the period preceding diagnosis, patients with IBD have an excess rate of gastrointestinal symptoms and an increased societal cost in the years before the diagnosis compared with the general population.^146–149 These recent findings suggest that subclinical disease or mild symptoms due to CD may be present well before the final diagnosis. Additionally, data from preclinical cohorts has reported that increased risk of stenosing or penetrating complications can be detected years before diagnosis.^129,150 Considering this evidence together, it can be argued that the initial mechanisms leading to CD may be present at least 5 years before,¹⁵⁰ and macroscopic lesions may appear approximately 2 years before, the development of symptoms.¹³³ Importantly, complicated disease [defined by Montreal classification B2 and B3] can be observed in approximately 30–40% of patients already at diagnosis,^151,152 suggesting that advanced lesions could appear in subjects without long-standing gastrointestinal symptoms. As has been demonstrated in other IMIDs like rheumatoid arthritis, disease modification strategies are possible if we target early disease before advanced tissue damage. Despite this critical importance, results from trials focused on the potential modification of natural history based on this definition are still scarce in the IBD field.^153,154 We hypothesise that interventions during an even earlier stage [during the initiation or expansion stage] may more efficiently ameliorate or halt the operation of dysregulated pathways, thus intercepting, delaying, or even preventing disease onset.

3. Challenges and priorities for research

We have identified several challenges and research gaps in the field [Figure 2]. Although different biomarkers have been found to predict disease development, validation in independent cohorts is necessary. Expansion of ongoing and development of new cohorts with careful phenotyping and follow-up, and with standardised operating procedures for sample collection, are required. IBD is a heterogeneous condition and it is plausible that the disease has multiple triggers, genetic predispositions, and immune dysregulation pathways that lead to an overt disease outcome.⁷ Therefore, we anticipate that a predictive tool(s) must also integrate information from multiple domains, and that a high-dimension model will be necessary for accurate prediction.¹⁰ Additionally, different risk prediction could be possible at each stage of preclinical disease, with risk augmenting progressively. Ideally such a predictive tool, which is currently not available, should have sufficiently high sensitivity and predictive value to justify potential adverse effects from screening, including emotional stress, unnecessary health care costs due to false positivity, potentially toxic interventions, or combinations thereof.⁷ Likewise, this predictive tool should be adapted to different age ranges, geographical locations, and different populations [familial versus non-familial IBD], among other considerations, as genetic and environmental triggers may vary. Importantly, as data are now accumulating on the preclinical stages of CD, very few clues into the preclinical stages of UC exist, perhaps implying a shorter preclinical period; this is consistent with the clinical observation that UC usually has a more abrupt clinical presentation.¹²⁶

As multiomics data will be generated from different cohorts, data integration will be a challenge and collaboration with bioinformaticians and development of appropriate analytical tools will be needed.¹⁰ Likewise, to develop appropriate interventions [both pharmaceutical and non-pharmaceutical], mechanistic studies will be required to understand the fundamental processes that the biomarkers reflect. However, it currently remains unclear exactly which interventions or which modifiable environmental exposures could reverse the preclinical period, and how they should be applied differently throughout the different stages of disease. Importantly, little is known on which therapies or preventive measures high-risk individuals would be willing to accept when faced with the prospect of developing IBD. It is also unknown for how long a preventive strategy should be continued or, importantly, what the outcome should be in disease prevention trial development.

Acknowledging these challenges, we must also reflect and learn from other IMIDs where multiple disease-prevention trials have provided breakthrough insights into disease pathogenesis, but have so far failed to prevent disease or have only transiently met this goal.^155,156 It is crucial that investigators come together to increase collaborative research and build the appropriate structures and framework to conduct timely and relevant research programmes in prediction and prevention, supported by IBD scientific societies and funding agencies.¹⁵⁷

4. Conclusions

We propose a common terminology for the different stages of disease initiation and progression, in parallel with known risk factors, known preclinical biomarkers, and data obtained from preclinical cohorts.

Ongoing cohorts, like the GEM study,^{23,98,117,118,158} the Multiplex Orthodox Jewish Family at Mount Sinai,^24,31 the PREDICTS study, the UK Twin study,¹⁵⁹ the EPIC cohort,^{68,74,77,78,160} the Dutch TWIN-IBD study,¹⁶¹ the Swedish IBD Twin Study,^{25,28,91,92,94,95,107,108,119,162} the EARLY cohort from GETECCU [Grupo Español de Trabajo en Enfermedad de Crohn y Colitis Ulcerosa], the MECONIUM Study,^97,163 and others offer the potential to gain important insights into the preclinical period and on possible prevention strategies that may be applied throughout the different stages. These cohorts hold great promise to help build predictive models of IBD in those at highest risk, with the eventual goal of intercepting disease and perhaps even capturing the events that surround initiation of the disease itself.

Acknowledgements

We would like to acknowledge academic medical illustrator Jill Gregory, CMI, FAMI, for help with figure design. We would like to thank Professor Paula Borralho, MD, PhD, for providing a histology picture for Figure 1. We would like to thank Catarina Gomes, MD, for providing a capsule endoscopy for Figure 1. We would like to thank Joseph Murray and Mark Riddle [from the PREDICTS Study] for their careful review and critical revisions of the manuscript.

Scientific Workshop Steering Committee

Bram Verstockt^a,b, Claudio Fiocchi^l, Joana Torres^k, Michael Scharl^maUniversity Hospitals Leuven Department of Gastroenterology and Hepatology, KU Leuven, Leuven, Belgium ^bKU Leuven Department of Chronic Diseases, Metabolism and Ageing, Translational Research Center for Gastrointestinal Disorders [TARGID], Leuven, Belgium ^kDivision of Gastroenterology, Hospital Beatriz Ângelo, Loures, Portugal ^lDepartment of Inflammation & Immunity, Lerner Research Institute, and Department of Gastroenterology, Hepatology & Nutrition, Digestive Disease Institute, Cleveland Clinic, Cleveland, OH, USA ^mDepartment of Gastroenterology and Hepatology, University Hospital Zürich, Switzerland.

Disclaimer

ECCO Scientific Workshop Papers are targeted at health care professionals only and are based on a standardised drafting procedure. Any treatment decisions are a matter for the individual clinician and may not be based exclusively on the content of the ECCO Scientific Workshop Papers. The European Crohn’s and Colitis Organisation and/or any of its staff members and/or any paper contributor may not be held liable for any information published in good faith in the ECCO Scientific Workshop Papers.

Conflict of Interest

ECCO has diligently maintained a disclosure policy of potential conflicts of interests [CoI]. The conflict of interest declaration is based on a form used by the International Committee of Medical Journal Editors [ICMJE]. The CoI disclosures are not only stored at the ECCO Office and the editorial office of JCC, but are also open to public scrutiny on the ECCO website [https://www.ecco-ibd.eu/about-ecco/ecco-disclosures.html], providing a comprehensive overview of potential conflicts of interest of the authors.

References