A97 DEVELOPMENT OF PREDICTION MODELS FOR THE TRIAGING OF REFERRALS OF INDIVIDUALS WITH SUSPECTED INFLAMMATORY BOWEL DISEASE TO IMPROVE PROMPT ACCESS TO CARE

Background

The negative impact of a delayed inflammatory bowel disease (IBD) diagnosis has been well established. We created a clinical pathway referred to as the “High-Risk IBD clinic” within a centralized referral program in a tertiary referral centre, in order to improve access to subspecialist care for individuals suspected but not yet diagnosed with IBD. Despite the creation of this specialized clinic, wait times continue to be above the recommended benchmarks established by the Canadian Association of Gastroenterology (CAG).

Aims

The purpose of our study was to create predictive models to identify factors associated with an IBD diagnosis in order to improve triage of referrals of individuals with features highly suggestive of IBD. We hypothesized that features suggestive of IBD could be used to create discriminating prediction models between IBD and IBS.

Methods

We conducted a retrospective cohort study of referrals to the High-Risk IBD clinic from February 2014 to December 2018. Referral information, investigations, endoscopic findings and final diagnosis were obtained from 316 consented individuals. Information required included symptoms (e.g. diarrhea, abdominal pain, rectal bleeding), risk factors (e.g. family history, rheumatological disease) and investigations (e.g. hemoglobin, CRP, abdominal imaging). Univariate logistic regression was performed to explore the association between factors included in the referral form, and a diagnosis of Crohn’s disease (CD) and ulcerative colitis (UC). For creation of predictive models, any variable with a p-value of <0.1 in univariate logistic regression was selected for entry into the multivariate model for CD and UC.

Results

For UC, the predictive model included weight loss, the presence of rectal bleeding and abdominal pain. Using these criteria, the sensitivity and specificity of the model were 62.5% and 74.1%, respectively. The negative predictive value (NPV) was high at 94.2%. For CD, the predictive model included male gender, elevated CRP, presence of anemia and presence of weight loss. The sensitivity and specificity of this model were 61.7% and 71.2%, respectively. As for UC, the NPV was also high (89.2%). For IBS, the most common diagnosis encountered in patients referred to the HR-IBD clinic, the model included absence of weight loss, presence of abdominal pain and female gender. The sensitivity and specificity were 71.6% and 64.0%, respectively. The positive predictive value was 60.6% and NPV was 74.5%.

Conclusions

We established predictive tools associated with a final diagnosis of IBD and IBS as a means to expedite the care of individuals with undiagnosed IBD.

Funding Agencies

CCC

A150 ACCESS TIMES TO GASTROENTEROLOGY FOR HIGH-RISK IBD REFERRALS IN THE GREATER CALGARY REGION

Background

Delay in the diagnosis of inflammatory bowel disease (IBD) can lead to adverse outcomes. In 2006, the CAG Wait Time Consensus Group recommended that wait times for patients with symptoms highly suggestive of IBD should be seen within two weeks. In 2007, the greater Calgary region established a central access and triage system to improve access to care as well as the “High-Risk IBD clinic” (HR-IBD) to further expedite the access of patients with IBD alarm symptoms. These included diarrhea, rectal bleeding, weight loss, abnormalities in laboratory and stool investigations.

Aims

The current study aimed to evaluate whether patient access to the HR-IBD clinic in the Calgary region was within recommended wait times.

Methods

We conducted a cross-sectional study of charts from consented patients pulled from the EMR of five Gastroenterologists in the Calgary region that received HR-IBD referrals from Feb 2014 to Jan 2018. Of the 206 patients included, the majority were female (139 vs 65) and the mean age was 34.4 y, with no statistical difference in age between genders (p=0.81). Data analysis was done with Stata (StataCorp 2019).

Results

The mean time to initial consult was 74.8 days (median 64), whereas time to endoscopy was 85.5 days (median 77). There was no statistical difference in the mean wait times between genders. Of the patient charts reviewed, 27% of referrals had a confirmed diagnosis of IBD (CD 17%, UC 11%). Patients with a diagnosis of UC waited a mean of 60.1 days (median 60) until initial consultation and patients with a diagnosis of CD waited 77 days (median 63.5), although this was not statistically different (p=0.27). The mean time to endoscopy for patients with UC was 77 days (median 67), and 85.4 days for patients with CD (median 78.5), again not statistically different. These wait times are below the reported wait times for all GI complaints, of 92 days from referral to consultation and 155 days from referral to procedure, as reported in the SAGE survey (2012). Although there were no differences in time to consult and endoscopy between groups, there were notable differences in alarm symptoms reported in the referral. For example, rectal bleeding was reported in 81.8% of referrals that culminated in a diagnosis of UC, as compared to 50% in CD and 47.6% of non-IBD patients. Further analysis in which alarm symptoms correlate with a final diagnosis of IBD may guide triaging of referrals to decrease the time to diagnosis.

Conclusions

Timely access for consultation and endoscopy for patients presenting with high-risk features for IBD by Gastroenterology in the Calgary region remains above the CAG recommended wait times. Further correlation of high-risk features with a final diagnosis of IBD will help risk-stratify referrals in order to decrease time to IBD diagnosis.

Funding Agencies

CIHRAlberta Innovates Health Solutions

Differential Effect of Genetic Burden on Disease Phenotypes in Crohn's Disease and Ulcerative Colitis in a Canadian Cohort

Background and Aims

Crohn’s disease (CD) and ulcerative colitis (UC) demonstrate considerable phenotypic heterogeneity and course. Accurate predictors of disease behaviour are lacking. The contribution of genetics and specific polymorphisms is widely appreciated; however, their cumulative effect(s) upon disease behaviour remains poorly understood. Here, we investigate the relationship between genetic burden and disease phenotype in a Canadian inflammatory bowel disease (IBD) Cohort.

Methods

We retrospectively examined a cohort of CD and UC patients recruited from a single tertiary referral center genotyped using a Goldengate Illumina platform. A genetic risk score (GRS) incorporating strength of association (log odds ratio) and allele dose for 151 IBD-risk loci was calculated and evaluated for phenotypic associations.

Results

Among CD patients, higher GRS was associated with earlier onset of disease (regression coefficient −2.19, 95% confidence interval [CI] −3.77 to −0.61, P = 0.007), ileal disease (odds ratio [OR] 1.45), stricturing/penetrating disease (OR 1.72), perianal disease (OR 1.57) and bowel resection (OR 1.66). Higher GRS was associated with use of anti-tumor necrosis factor (TNF) (P < 0.05) but not immunomodulators. Interestingly, we could not demonstrate an association between higher GRS and family history of IBD (OR 1.27, P = 0.07). Onset of disease remained statistically significant for never smokers (P = 0.03) but not ever smokers (P = 0.13). For UC, having a higher GRS did not predict the age of diagnosis nor was it predictive of UC disease extent (P = 0.18), the need for surgery (P = 0.74), nor medication use (immunomodulators P = 0.53, anti-TNF P = 0.49). We could not demonstrate an association between increased GRS and having a family history of IBD in the UC group.

Conclusions

Increasing genetic burden is associated with early age of diagnosis in CD and may be useful in predicting disease behaviour in CD but not UC.

N-Terminomics/TAILS Profiling of Proteases and Their Substrates in Ulcerative Colitis

Dysregulated protease activity is often implicated in the initiation of inflammation and immune cell recruitment in gastrointestinal inflammatory diseases. Using N-terminomics/TAILS (terminal amine isotopic labeling of substrates), we compared proteases, along with their substrates and inhibitors, between colonic mucosal biopsies of healthy patients and those with ulcerative colitis (UC). Among the 1642 N-termini enriched using TAILS, increased endogenous processing of proteins was identified in UC compared to healthy patients. Changes in the reactome pathways for proteins associated with metabolism, adherens junction proteins (E-cadherin, liver-intestinal cadherin, catenin alpha-1, and catenin delta-1), and neutrophil degranulation were identified between the two groups. Increased neutrophil infiltration and distinct proteases observed in ulcerative colitis may result in extensive break down, altered processing, or increased remodeling of adherens junctions and other cellular functions. Analysis of the preferred proteolytic cleavage sites indicated that the majority of proteolytic activity and processing comes from host proteases, but that key microbial proteases may also play a role in maintaining homeostasis. Thus, the identification of distinct proteases and processing of their substrates improves the understanding of dysregulated proteolysis in normal intestinal physiology and ulcerative colitis.

The effects of pregnancy on the pharmacokinetics of infliximab and adalimumab in inflammatory bowel disease

BACKGROUND

Transplacental transfer of infliximab and adalimumab results in detectable drug levels in the cord blood and infant.

AIM

To determine if pregnancy influenced the pharmacokinetics of anti-TNF agents in women with inflammatory bowel disease.

METHODS

Twenty-five women from the University of Calgary inflammatory bowel disease(IBD) pregnancy clinic on maintenance infliximab or adalimumab were recruited prospectively with serum bio-banking performed each trimester. Infliximab trough and adalimumab steady-state levels were the outcomes of interest and were analysed using the ANSER infliximab and adalimumab assays. Multivariate linear mixed-effects models were constructed to assess infliximab and adalimumab drug levels during pregnancy adjusting for the clinical covariates of albumin, BMI and CRP.

RESULTS

Fifteen women (eight Crohn's disease, seven ulcerative colitis) received infliximab and 10 women with 11 pregnancies were treated with adalimumab. Median age was 29.6 years (IQR: 27.6-31.2 years). Median disease duration was 9.2 years (IQR: 3.16-15.0 years). Median trough infliximab concentrations were 8.50 μg/mL (IQR: 7.23-10.07 μg/mL), 10.31 μg/mL (IQR: 7.66-15.63 μg/mL) and 21.02 μg/mL (IQR: 16.01-26.70 μg/mL) at trimesters 1, 2 and 3 respectively. Significant changes in albumin and BMI (P < 0.05) but not CRP (P > 0.05) were documented throughout pregnancy. After adjusting for albumin, BMI and CRP, infliximab trough levels increased during pregnancy, by 4.2 μg/mL per trimester (P = 0.02), while adalimumab drug levels remained stable (P > 0.05).

CONCLUSIONS

Infliximab levels rise during pregnancy, whereas adalimumab levels remain stable after accounting for changes in albumin, BMI and CRP. Therapeutic drug monitoring in the second trimester may be useful in guiding dosing in the third trimester.

Ciprofloxacin: drug of choice in the treatment of malignant external otitis (MEO)

Ciprofloxacin, a fluorinated quinolone with high efficacy against Pseudomonas aeruginosa, was used in the treatment of 10 consecutive patients with malignant external otitis. All patients had skull base osteomyelitis documented by nuclear and computed tomography (CT) scans. Dosages of 1.5 g of ciprofloxacin daily were used for a mean average of 10 weeks. All patients were considered cured with a minimum follow-up of 18 months after completion of therapy. A new classification of malignant external otitis (MEO) is presented.