Clinical and biological predictors of response to standardised paediatric colitis therapy (PROTECT): a multicentre inception cohort study

Jeffrey S. Hyams, Sonia Davis Thomas, Nathan Gotman, Yael Haberman, Rebekah Karns, Melanie Schirmer, Angela Mo, David R. Mack, Brendan Boyle, Anne M. Griffiths, Neal S. LeLeiko, Cary G. Sauer, David J. Keljo, James Markowitz, Susan S. Baker, Joel Rosh, Robert N. Baldassano, Ashish Patel, Marian Pfefferkorn, Anthony OtleyMelvin Heyman, Joshua Noe, Maria Oliva-Hemker, Paul A. Rufo, Jennifer Strople, David Ziring, Stephen L. Guthery, Boris Sudel, Keith Benkov, Prateek Wali, Dedrick Moulton, Jonathan Evans, Michael D. Kappelman, M. Alison Marquis, Francisco A. Sylvester, Margaret H. Collins, Suresh Venkateswaran, Marla Dubinsky, Vin Tangpricha, Krista L. Spada, Bradley Saul, Jessie Wang, Jose Serrano, Kevin Hommel, Urko M. Marigorta, Greg Gibson, Ramnik J. Xavier, Subra Kugathasan, Thomas Walters, Lee A. Denson

Research output: Contribution to journalArticlepeer-review

114 Scopus citations


Background: Lack of evidence-based outcomes data leads to uncertainty in developing treatment regimens in children who are newly diagnosed with ulcerative colitis. We hypothesised that pretreatment clinical, transcriptomic, and microbial factors predict disease course. Methods: In this inception cohort study, we recruited paediatric patients aged 4–17 years with newly diagnosed ulcerative colitis from 29 centres in the USA and Canada. Patients initially received standardised mesalazine or corticosteroids, with pre-established criteria for escalation to immunomodulators (ie, thiopurines) or anti-tumor necrosis factor-α (TNFα) therapy. We used RNA sequencing to define rectal gene expression before treatment, and 16S sequencing to characterise rectal and faecal microbiota. The primary outcome was week 52 corticosteroid-free remission with no therapy beyond mesalazine. We assessed factors associated with the primary outcome using logistic regression models of the per-protocol population. This study is registered with, number NCT01536535. Findings: Between July 10, 2012, and April 21, 2015, of 467 patients recruited, 428 started medical therapy, of whom 400 (93%) were evaluable at 52 weeks and 386 (90%) completed the study period with no protocol violations. 150 (38%) of 400 participants achieved week 52 corticosteroid-free remission, of whom 147 (98%) were taking mesalazine and three (2%) were taking no medication. 74 (19%) of 400 were escalated to immunomodulators alone, 123 (31%) anti-TNFα therapy, and 25 (6%) colectomy. Low baseline clinical severity, high baseline haemoglobin, and week 4 clinical remission were associated with achieving week 52 corticosteroid-free remission (n=386, logistic model area under the curve [AUC] 0·70, 95% CI 0·65–0·75; specificity 77%, 95% CI 71–82). Baseline severity and remission by week 4 were validated in an independent cohort of 274 paediatric patients with newly diagnosed ulcerative colitis. After adjusting for clinical predictors, an antimicrobial peptide gene signature (odds ratio [OR] 0·57, 95% CI 0·39–0·81; p=0·002) and abundance of Ruminococcaceae (OR 1·43, 1·02–2·00; p=0·04), and Sutterella (OR 0·81, 0·65–1·00; p=0·05) were independently associated with week 52 corticosteroid-free remission. Interpretation: Our findings support the utility of initial clinical activity and treatment response by 4 weeks to predict week 52 corticosteroid-free remission with mesalazine alone in children who are newly diagnosed with ulcerative colitis. The development of personalised clinical and biological signatures holds the promise of informing ulcerative colitis therapeutic decisions. Funding: US National Institutes of Health.

Original languageEnglish
Pages (from-to)1708-1720
Number of pages13
JournalThe Lancet
Issue number10182
StatePublished - 27 Apr 2019


Dive into the research topics of 'Clinical and biological predictors of response to standardised paediatric colitis therapy (PROTECT): a multicentre inception cohort study'. Together they form a unique fingerprint.

Cite this