Study Identifies Novel Plasma Proteomic Biomarkers for Crohn Disease Complications

Investigators used a machine learning strategy to seek out plasma protein biomarkers for stricturing and penetrating complications.

Applying a machine learning approach to an inception cohort of pediatric patients with Crohn disease led to the identification of novel blood proteomic biomarkers for stricturing (B2) and penetrating (B3) complications, according to findings from a study published in Alimentary Pharmacology & Therapeutics. When assayed in the blood at the time of diagnosis, the biomarkers “have predictive capacity for [the] development of complications” in pediatric patients with the inflammatory bowel disease, the investigators stated.

Currently, biomarkers for Crohn-specific complications are “limited,” translating to a need for improved risk stratification in this disease setting. Thus, the investigators of the case-cohort study used a pediatric inception patient group derived from the Risk Stratification and Identification of Immunogenetic and Microbial Markers of Rapid Disease Progression in Children with Crohn Disease (RISK) cohort to identify plasma protein biomarkers and assess their potential for correlation with subsequent complications. The outcomes were development of B2 (stricturing) or B3 (penetrating) complications.

All 265 of the pediatric patients in the inception cohort had inflammatory disease (B1) at baseline. The mean patient age was 11.6 years.

The study authors assayed 92 inflammation-related proteins in baseline plasma using a proximity extension assay and ultimately included 85 of these proteins in the analysis. A random survival forests (RSF) machine learning strategy was used to select variables predictive of B2 and B3 complications.

“RSF modeling selected 9 proteins (IL12B, CXCL9, IL7, CCL3, CD6, IL15RA, MMP10, CCL11, IL10) and 3 serologic markers (LnCbir, LnASCA IgA, LnOMPC) that were most predictive for any new complication,” the investigators said.

Specifically, IL7, MMP10, IL12B, and CCL11 and LnASCA IgA and LnCbir were the 4 proteins and 2 serologic markers deemed the “most predictive” for B2 complications, respectively. TNFSF14, CCL4, IL15RA, TNFB, and CD40 emerged as the 5 proteins “most predictive” for B3 complications. The 3 serologic markers with the highest predictive capacity for B3 complications included LnASCA IgA, LnANCA, and LnCbir.

The investigators found that 73 patients developed B2 complications within a mean of 1123 days (SD, 477 days). Thirty-four patients presented with B3 complications within a median of 1251 days.

A model with 5 protein markers predicted B3 complications with an area under the curve (AUC) of 0.79 (95% CI, 0.76-0.82) compared with 0.69 for serologies (95% CI, 0.66-0.72) and 0.74 (95% CI, 0.71-0.77) for clinical variables. A model with 4 protein markers predicted B2 complications with an AUC of 0.68 (95% CI, 0.65-0.71) vs 0.62 (95% CI, 0.59-0.65) for serologies and 0.52 (95% CI, 0.50-0.55) for clinical variables.

B2 analytes were found to be highly expressed in ileal stromal cells while B3 analytes were prominent in peripheral blood and ileal T cells. Of note, the protein-based models implemented during the study period were found to “outperform clinical and serologic-based models” for both B2 and B3.

The study authors noted several limitations, including the observational nature of the RISK cohort. Further, existing uncertainty about the “best definition” of a stricture in Crohn disease could have led to imprecise assessment of this B2 complication, they said.

“Although further studies with external cohorts are needed before these markers could be implemented clinically, our findings support the use of blood biomarkers in assisting with risk stratification of [patients with Crohn] disease at the time of diagnosis,” the investigators concluded.

Disclosure: Several study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of the authors’ disclosures.

Follow @Gastro_Advisor


Ungaro RC, Hu L, Ji J, et al. Machine learning identifies novel blood protein predictors of penetrating and stricturing complications in newly diagnosed paediatric Crohn’s disease. Aliment Pharmacol Ther. Published online November 1, 2020. doi:10.1111/apt.16136