Quantitative Visceral Fat Findings Using Analytic Morphomics are Associated with Future Surgical Bowel Resection in Stricturing Crohn’s Disease.

Stidham RK, Ruma JA, Enchakalody B, Waljee AK, Govani SM, Higgins P, Su GL. Quantitative Visceral Fat Findings Using Analytic Morphomics are Associated with Future Surgical Bowel Resection in Stricturing Crohn’s Disease. Poster session presented at: Digestive Disease Week Annual Conference; 2015 May 16; Washington, DC.

Related HSR&D Project(s)

• CDA 11-217 – Risk Stratification and Targeted Therapy for HELP Diseases in Veterans

Quantitative Visceral Fat Findings Using Analytic Morphomics are Associated with Future Surgical Bowel Resection in Stricturing Crohn's Disease. View Session Detail Presentation Number: Sa1104 Author Block: Ryan W. Stidham1 , Julie A. Ruma3 , Binu Enchakalody4 , Akbar K. Waljee5,1 , Shail M. Govani1,2 , Peter Higgins1 , Grace L. Su2 1 Internal Medicine, Division of Gastroenterology and Hepatology, University of Michigan, Ann Arbor, Michigan, United States; 2 Internal Medicine, Division of Gastroenterology and Hematology, Veterans Administration Ann Arbor Healthcare Systems, Ann Arbor, Michigan, United States; 3 Radiology, Veterans Affairs Healthcare Systems, Ann Arbor, Michigan, United States; 4 Surgery, University of Michigan, Ann Arbor, Michigan, United States; 5 Health Services Research and Development Center for Clinical Management Research, Veterans Affairs Healthcare Systems, Ann Arbor, Michigan, United States Abstract: Background: Changes in abdominal fat characteristics in proximity to active Crohn's disease (CD) are a common finding on computed tomography (CT) scans. Qualitative assessments of fat density and stranding have been correlated to endoscopic disease activity. We hypothesize that quantitative fat assessment using analytic morphomic image analysis can capture this disease activity finding and may be a predictor of therapeutic response in stricturing CD. Methods: We conducted a retrospective review of a large tertiary center's electronic medical records (2000-2010) identifying CD patients (ICD 555.x) hospitalized with obstructive symptoms, ileal disease, small bowel dilation > 3.5cm and CT scan within 30 days of admission. CT scans underwent analytic morphomic image analysis to extract fat data. Subcutaneous (SCF) and visceral (VF) fat areas were standardized to total body area (L3 spinal level). Localized VF density was calculated by isolating VF within a 5cm radius around the point of maximal disease activity (identified by radiologist) and measuring the mean Hounsfield unit (HU) value. Patient demographics, radiographic report data, and outcomes of bowel resection within 2 years of index imaging were collected by chart review. The Student's t-test was used with logistic regression in SAS 9.3. Results: Of 281 eligible admissions, complete data was present in 204 admissions. Surgical bowel resection occurred 24% of the cohort. On univariate analysis, raw and standardized VF and SCF areas and the VF-SCF ratio were not associated with surgical outcomes. In addition, VF density (mean HU) did not differ between those avoiding and requiring bowel resection (-90.16 10.70 vs. -89.53 10.61, p = 0.756). Localized VF density demonstrated a significant difference between those avoiding and requiring surgery (-49.43 20.23 vs. -59.68 19.47, p = 0.032). Standardization of localized VF density to mean whole abdominal VF density demonstrated a value of 0.8633 0.2489 in those avoiding surgery, compared to 1.0049 0.2359 in those requiring bowel resection, p = 0.0062. When adjusting for age, gender, medications, and radiologist report of active CD inflammation, standardized localized VF density measurements closer to 1 (no difference from surrounding fat) demonstrated a small association with future bowel resection (OR = 1.20,95%CI 1.09-1.45, p = 0.0279). Conclusions: Differences in visceral fat density near the point of maximum disease activity are associated with the likelihood of surgery in CD. While differences in visceral fat density were small in this pilot study using a restricted range of CD phenotype, we demonstrate proof of principle that analytic morphomics can provide a quantitative method for reporting presently subjectively imaging findings. Further studies in larger cohorts may better define the role of fat quantitation in image based-CD activity scoring tools.