Serum TGF- β 1 and CD14 Predicts Response to Anti-TNF- α Therapy in IBD

Abstract

Background: Tumor necrosis factor-alpha (TNF-α) agonists revolutionized therapeutic algorithms in inflammatory bowel disease (IBD) management. However, approximately every third IBD patient does not respond to this therapy in the long term, which delays efficient control of the intestinal inflammation.

Methods: We analyzed the power of serum biomarkers to predict the failure of anti-TNF-α. We collected serum of 38 IBD patients at therapy prescription and 38 weeks later and analyzed them with relation to therapy response (no-, partial-, and full response). We used enzyme-linked immunosorbent assay to quantify 16 biomarkers related to gut barrier (intestinal fatty acid-binding protein, liver fatty acid-binding protein, trefoil factor 3, and interleukin (IL)-33), microbial translocation, immune system regulation (TNF-α, CD14, lipopolysaccharide-binding protein, mannan-binding lectin, IL-18, transforming growth factor-β1 (TGF-β1), osteoprotegerin (OPG), insulin-like growth factor 2 (IGF-2), endocrine-gland-derived vascular endothelial growth factor), and matrix metalloproteinase system (MMP-9, MMP-14, and tissue inhibitors of metalloproteinase-1).

Results: We found that future full-responders have different biomarker profiles than non-responders, while partial-responders cannot be distinguished from either group. When future non-responders were compared to responders, their baseline contained significantly more TGF-β1, less CD14, and increased level of MMP-9, and concentration of these factors could predict non-responders with high accuracy (AUC = 0.938). Interestingly, during the 38 weeks, levels of MMP-9 decreased in all patients, irrespective of the outcome, while OPG, IGF-2, and TGF-β1 were higher in non-responders compared to full-responders both at the beginning and the end of the treatment.

Conclusions: The TGF-β1 and CD14 can distinguish non-responders from responders. The changes in biomarker dynamics during the therapy suggest that growth factors (such as OPG, IGF-2, and TGF-β) are not markedly influenced by the treatment and that anti-TNF-α therapy decreases MMP-9 without influencing the treatment outcome.

Figure 1

Patients who do not respond to biological therapy have a distinct serum biomarker profile at baseline, as measured by (a) hierarchical clustering heatmap, (b) nearest shrunken centroids classification, and (c) principal component analysis (PCA). PCA biplots show 1st and 2nd and 1st and 3rd PCA axis. Ordination space occupied by each sample group is highlighted by standard 95% ellipses.

Figure 2

Changes in biomarker level during anti-TNF-α therapy: (a) molecules associated with gut barrier damage; (b) molecules involved in immune response associated with microbial translocation; (c) molecules regulating the immune response; (d) molecules of the matrix metalloproteinase system. The level of biomarkers is depicted as a mean of values; statistically significant differences between groups or between time points are marked with  ^∗ ( ^∗p < 0.05).

Figure 3

Diagnostic performance to predict therapy response: (a) relative importance of individual biomarkers in distinguishing NR from FR and PR group within the best model found by regression analysis; (b) quantitative plot of the two most efficient discriminating factors analyzed by Mann–Whitney test, the biomarker levels are depicted as mean ± SD of the values ( ^∗p < 0.05;  ^∗∗∗p < 0.001).

Figure 4

The network depicting the protein–protein interaction between selected biomarkers delineating the gut epithelium damage and inflammatory response. The interconnection biomarkers shown by protein–protein network functional enrichment analysis based on coexpression (black line), co-occurrence (dark-blue line), experimentally determined (pink line), and both text mining (yellow line) and curated databases (blue line). The statistically significant differences between NR and FR groups in biomarker level at the baseline are marked by  ^∗. Network was produced through the STRING database and STRING Consortium 2022 web source. The dark-red dotted circle depicting the molecules associated with the damage of gut epithelium, the blue dotted circle depicting molecules associated with the regulation of immune response, and the red dotted circle showing the TNF-α and MMP-9 in the center of this protein–protein interaction network of molecules analyzed in IBD pathogenesis and therapy response.