CRP Predicts Biochemical but Not Clinical Remission in Ustekinumab-Treated Crohn's Disease: A Remission-Specific Whole-Blood Transcriptomic Signature
Transcriptomic biomarkers for ustekinumab response in Crohn's disease
DOI:
https://doi.org/10.64949/81rdwc86Keywords:
Crohn's disease, ustekinumab, transcriptomics, machine learning, biomarkers, C-reactive protein, remission;, Random Forest, SHAPAbstract
This article is a preprint and has not yet been peer-reviewed. Not for clinical use.
Background and Aims
Clinical outcome prediction in ustekinumab-treated Crohn's disease remains a significant unmet need. C-reactive protein (CRP) is widely used as a biochemical marker of inflammation, yet its utility as a predictor of clinical remission alongside blood transcriptomics has not been formally evaluated.
Methods
Predictive model training used GSE207465 (n=353, ustekinumab-treated, UNITI-2 trial; whole-blood RNA, HuGene 2.1 ST). Co-primary endpoints were CRP normalisation (serum CRP <5 mg/L at week 8) and clinical remission (CDAI score <150, week 8). Baseline CRP served as a reference comparator via logistic regression. A 60-gene baseline expression panel was selected from 1,165 remission-specific differentially expressed genes; a Random Forest classifier was trained with five-fold stratified cross-validation and SHAP feature attribution. External validation used GSE186963 (n=23, infliximab-treated; Clariom S) via a 25-gene cross-platform subset and performance was quantified by AUC with 2,000-iteration bootstrap 95% confidence intervals.
Results
Baseline CRP strongly predicted CRP normalisation (AUC 0.840, 95% CI 0.756-0.911) but was near chance for clinical remission (AUC 0.564, 95% CI 0.454-0.675). The 60-gene panel achieved a higher point-estimate AUC for remission than CRP alone (0.641, 95% CI 0.525-0.751), though confidence intervals overlapped substantially. The top SHAP feature, PDE2A-AS1 (phosphodiesterase 2A antisense lncRNA), was non-significant in the overall differential expression analysis (logFC 0.013, Padj 0.808), but it was a significant remission-specific transcriptional marker (logFC 0.288, -log10P 3.376). BANK1 showed opposing associations with the two endpoints. External validation was negative (AUC 0.517, 95% CI 0.265-0.768), consistent with biologic-specific transcriptomic prediction.
Conclusions
Baseline CRP and whole-blood gene expression predict discordant endpoints in ustekinumab-treated Crohn's disease. A remission-specific signature including PDE2A-AS1 adds incremental predictive value beyond CRP alone. Prospective validation in adequately powered ustekinumab cohorts is required before clinical utility can be established.
References
1. Feagan BG, Sandborn WJ, Gasink C, et al. Ustekinumab as Induction and Maintenance Therapy for Crohn’s Disease. New England Journal of Medicine. 2016;375(20):1946-1960. doi:10.1056/nejmoa1602773 DOI: https://doi.org/10.1056/NEJMoa1602773
2. Sandborn WJ, Gasink C, Gao LL, et al. Ustekinumab Induction and Maintenance Therapy in Refractory Crohn’s Disease. New England Journal of Medicine. 2012;367(16):1519-1528. doi:10.1056/NEJMoa1203572 DOI: https://doi.org/10.1056/NEJMoa1203572
3. Torres J, Bonovas S, Doherty G, et al. ECCO guidelines on therapeutics in Crohn’s disease: Medical treatment. J Crohns Colitis. 2020;14(1):4-22. doi:10.1093/ecco-jcc/jjz180 DOI: https://doi.org/10.1093/ecco-jcc/jjz180
4. Turner D, Ricciuto A, Lewis A, et al. STRIDE-II: An Update on the Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE) Initiative of the International Organization for the Study of IBD (IOIBD): Determining Therapeutic Goals for Treat-to-Target strategies in IBD. Gastroenterology. 2021;160(5):1570-1583. doi:10.1053/j.gastro.2020.12.031 DOI: https://doi.org/10.1053/j.gastro.2020.12.031
5. Vermeire S, Assche G Van, Rutgeerts P. C-Reactive Protein as a Marker for Inflammatory Bowel Disease. Inflamm Bowel Dis. 2004;10(5):661-665. doi:10.1097/00054725-200409000-00026. DOI: https://doi.org/10.1097/00054725-200409000-00026
6. Arijs I, Li K, Toedter G, et al. Mucosal gene signatures to predict response to infliximab in patients with ulcerative colitis. Gut. 2009;58(12):1612-1619. doi:10.1136/gut.2009.178665 DOI: https://doi.org/10.1136/gut.2009.178665
7. Gerassy-Vainberg S, Starosvetsky E, Gaujoux R, et al. A personalized network framework reveals predictive axis of anti-TNF response across diseases. Cell Rep Med. 2024;5(1). doi:10.1016/j.xcrm.2023.101300 DOI: https://doi.org/10.1016/j.xcrm.2023.101300
8. Lundberg SM, Allen PG, Lee SI. A Unified Approach to Interpreting Model Predictions. In: Guyon I, Luxburg UV, Bengio S, Wallach H, Fergus R, Vishwanathan S, Garnett R, editors. Advances in Neural Information Processing Systems 30. 2017. p. 4765-4774.
9. National Center for Biotechnology Information. GSE207465. Gene Expression Omnibus. November 8, 2022. Accessed June 5, 2026. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE207465
10. Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30(1):207-210. doi:10.1093/nar/30.1.207 DOI: https://doi.org/10.1093/nar/30.1.207
11. National Center for Biotechnology Information. GSE186963. Gene Expression Omnibus. November 1, 2024. Accessed June 5, 2026. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE186963
12. Harvey RF, Bradshaw JM. A simple index of Crohn’s-disease activity. The Lancet. 1980;315(8167):514. doi:10.1016/s0140-6736(80)92767-1 DOI: https://doi.org/10.1016/S0140-6736(80)92767-1
13. Ritchie ME, Phipson B, Wu D, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47. doi:10.1093/nar/gkv007 DOI: https://doi.org/10.1093/nar/gkv007
14. Breiman L. Random forests. Mach Learn. 2001;45(1):5-32. DOI: https://doi.org/10.1023/A:1010933404324
15. Pedregosa F, Michel V, Grisel O, Blondel M, Prettenhofer P, Weiss R, et al. Scikit-learn: Machine learning in Python. J Mach Learn Res. 2011;12:2825-30. Available from: http://scikit-learn.sourceforge.net
16. Chen T, Guestrin C. XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery; 2016. p. 785-794. doi:10.1145/2939672.2939785. DOI: https://doi.org/10.1145/2939672.2939785
17. Lundberg SM, Erion G, Chen H, et al. From local explanations to global understanding with explainable AI for trees. Nat Mach Intell. 2020;2(1):56-67. doi:10.1038/s42256-019-0138-9 DOI: https://doi.org/10.1038/s42256-019-0138-9
18. Kozyrev S V., Abelson AK, Wojcik J, et al. Functional variants in the B-cell gene BANK1 are associated with systemic lupus erythematosus. Nat Genet. 2008;40(2):211-216. doi:10.1038/ng.79 DOI: https://doi.org/10.1038/ng.79
19. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): The TRIPOD Statement. BMC Med. 2015;13(1). doi:10.1186/s12916-014-0241-z DOI: https://doi.org/10.1186/s12916-014-0241-z
Downloads
Published
Data Availability Statement
Data Availability Statement All gene expression data are publicly available through the NCBI Gene Expression Omnibus under accession numbers GSE207465 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE207465) and GSE186963 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE186963). Analysis code is available from the corresponding author upon reasonable request.Issue
Section
License
Copyright (c) 2026 Simbarashe G. Magwenzi
This work is licensed under a Creative Commons Attribution 4.0 International License.
