Abstract

Abstract Background Several transcript signatures have been published that predict anti-TNF resistance in ulcerative colitis (UC). However, clinical translation is hampered by suboptimal validation performance, partly because of insufficient emphasis on potential pathophysiology. Moreover, predictors of ustekinumab response are lacking. We interrogated transcriptomic datasets and anti-TNF predictive signatures to investigate resistance mechanisms and applicability to ustekinumab. Methods Transcript profiles of colonic biopsies taken before treatment for moderate-to-severe UC were analysed from Gene Expression Omnibus. Patients were then treated with infliximab (n = 24 and n = 22), golimumab (n = 84) or ustekinumab (n = 358). Molecular pathways from the Reactome database that were differentially enriched between mucosal healing responders and non-responders post-induction were found, along with shared pathways between cohorts. The largest cohort was used to calculate enrichment of published resistance signatures in actively inflamed mucosa using Spearman correlation. Hierarchical clustering was undertaken for non-responders. Analyses were performed using R 4.2.2 (Vienna, Austria). Results Many biological pathways were associated with resistance across the cohorts (1079 pathways, adjusted p value<0.01). Sixty pathways were shared between 3 or 4 datasets. Those relating to extracellular matrix (ECM) activity, neutrophil degranulation and interleukin signalling were most prominent. Network analyses revealed that proteins forming these pathways interact in vivo, forming a discrete functional module. The enrichment of published transcript signatures correlated highly. For instance, correlation between enrichment of UC1 (Czarnewski et al, 2019) and M5 (Friedrich et al 2021) was 0.96. Over-representation analyses showed that these signatures represent processes such as interleukin signalling and neutrophil activity. Hierarchical clustering of non-responders revealed two distinct groups, including those with high (NRhi) and low (NRlo) signature enrichment (Figure 1). Cellular deconvolution analyses showed that neutrophils are more prevalent in NRhi compared to NRlo and that eosinophils may be more frequent in NRlo than NRhi. Conclusion ECM dysregulation and neutrophil activity may underlie a core ‘resistome’ that heralds poor treatment outcomes to anti-TNFs and ustekinumab. Supporting this, the enrichment scores of published transcript signatures in inflamed tissue correlate highly, suggesting that many of these represent shared, or similar, molecular mechanisms of resistance. However, there is a separate resistant cohort with different molecular drivers. These findings have implications for precision medicine approaches in UC.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call