Society: AGA

Background:
The pathogenesis of inflammatory bowel disease (IBD) involves crosstalk between genetic and microbial factors that influence immune reactions. Tofacitinib, a pan JAK-STAT inhibitor, is a small molecule indicated for the treatment of moderate-to-severe ulcerative colitis (UC). However, approximately 30-40% of patients will not response to tofacitinib, and underscores the importance of identifying individuals who may respond to drug prior to therapy initiation. We utilized patient-derived intestinal organoids to predict responsiveness to tofacitinib.

Methods:
We cultured colonic organoids derived from endoscopic mucosal pinch biopsies from 4 healthy donors (HD) and 24 patients with UC initiating tofacitinib at weeks 0 and 24. Mechanically disrupted organoids were dissociated into 96-well tissue culture plate. The organoids were cultured in differentiation media (IF media) supplemented with Y-27632 for the first 2 days, and then stimulated with 20 ng/ml of human(h) TNFα and 1,000 IU/ml of hIFNβ or 20 ng/ml of hTNFα and hIFNγ in the presence of DMSO, 1, or 10 μM of tofacitinib for 5 days. Percent viable organoids was determined by daily quantification of the number of intact organoids. Change in total Mayo scores at weeks 0 and 24 were correlated with organoid viability.

Results:
Tofacitinib at 1 and 10 μM significantly rescued the decreased viability of organoids-derived from both HD and UC patients under hTNFα+hIFNβ or hTNFα+hIFNγ stimulation, consistent with tofacitinib sensitivity (Figure 1A). UC patient-derived organoids showed heterogenous viability when treated with 1 μM tofacitinib (Figure 1A). 10 UC organoid lines were sensitive to tofacitinib (UC-S) and phenocopied HD-derived organoids. 14 UC organoid lines displayed comparable viability under hTNFα+hIFNγ stimulation despite 1μM of tofacitinib, consistent with tofacitinib tolerance (UC-T1 and 2; Figure 1B). 7 organoid lines (UC-T2) demonstrated impaired viability rescue under both cytokine combinations with tofacitinib (Figure 1B). Total Mayo scores of patients corresponding to UC-S were significantly decreased after tofacitinib treatment (Figure 1C). Total Mayo scores were unchanged in the patients corresponding to UC-T1 and T2 (Figure 1C).

Conclusion:
Our findings indicate that patient-derived intestinal organoids display heterogeneous sensitivity to tofacitinib. The ability of tofacitinib to rescue organoid viability directly correlated with change in the total Mayo score, suggesting that organoids are a useful platform to predict drug responsiveness of individual patients.

Background:
While biological treatment has proven effective in treating Crohn’s disease (CD), we still face an important therapeutic ceiling, as a large proportion of patients fail to demonstrate primary response or lose response over time. Therefore, pre-treatment selection of patients with increased potential to objectively respond is of paramount importance, yet biomarkers facilitating such selection are lacking. Previous epigenome-wide association studies associated differential DNA methylation with CD-specific phenotypes, of which the majority suggests a potential use in classification and prediction of response to treatment.

Methods:
We therefore prospectively measured peripheral blood DNA methylation signatures of 184 adult CD patients before (T1) and after a median of 28 weeks (T2) of treatment with adalimumab (ADA), vedolizumab (VEDO) or ustekinumab (USTE) in a discovery (n=88) and independently collected internal validation cohort (n=96) using the Illumina EPIC BeadChip array. Response (R) was defined as the combination of endoscopic response (≥50% reduction in SES-CD score) and steroid-free clinical response (≥3 point drop in HBI or HBI ≤4 AND no systemic steroids) and/or biochemical response (≥50% reduction in C-reactive protein (CRP) and fecal calprotectin or a CRP ≤5 g/mL and fecal calprotectin ≤250 µg/g). Samples taken at T1 were used for biomarker identification and classification through stability selected gradient boosting, while samples taken at T2 as well as publicly available intraclass correlation (ICC) data were used to evaluate the long-term stability of our identified CpG markers.

Results:
A total of 58 ADA-patients (N_R=29, N_NR=29), 64 VEDO-patients (N_R=36, N_NR=28) and 62 USTE-patients (N_R=30, N_NR=32) were included (table 1). Classification analyses at T1 selected distinct CpG-signatures that, in combination, predict clinical- and endoscopic response to ADA (100 loci), VEDO (22 loci) and USTE (68 loci) with high performance (AUC ADA=0.73, VEDO=0.89 and USTE=0.94) upon validation. In addition, these signatures did not significantly differ between T1 and T2 and the majority demonstrated ICC-values ≥0.90, indicating long-term hyper stability irrespective of inflammatory status or external exposure. Furthermore, genes annotated to the CpGs of interest suggest drug specific involvement in TNF-signaling, endothelial cell-cell adhesion, integrin dependent T-cell homing, the innate immune system and Th17/Treg differentiation, corroborating to the mode of action of each drug.

Conclusion: Here, we report on 3 novel validated signatures of highly stable CpG-loci that predict endoscopic- and clinical response in CD patients treated with ADA, VEDO or USTE. Additional external- and clinical validation as part of EPIC-CD and the OMICROHN clinical trial are currently ongoing.

Figure 1. Machine learning methodology, model performance and radar plots.

Figure 1. Machine learning methodology, model performance and radar plots.

Table 1. Baseline characteristics

Background: Mirikizumab (miri), a p19-directed IL-23 antibody, showed efficacy in ulcerative colitis (UC)¹. Miri modulates transcripts associated with disease activity and anti-TNF resistance in UC and increases expression of transcripts associated with healthy mucosa². Here, we evaluate the transcription profile of patients with UC who respond to placebo (PBO, 20.6%) or miri 200mg treatment (59.7%) versus non-responders and build a model that can predict response in patients with UC.

Methods: In the phase 2 AMAC trial (NCT02589665), patients were randomized 1:1:1:1 to receive intravenous PBO (N=63), 50mg miri (N=63), or 200mg miri (n=62) with possible exposure-based dose increases or fixed 600mg miri (n=61) at Week (W)0, 4 and 8. Colonic biopsies were collected at baseline, W12, and W52. An Affymetrix HTA2.0 microarray measured gene expression. Limma comparing values at baseline and W12 determined differentially expressed genes (DEGs). DEGs maintaining their W12 expression level through W52 among responders in either the PBO and miri 200mg arms were designated as similarly expressed genes (DEGSEGs).
XGBoost models³ were trained to predict clinical response (decrease in 9-point Mayo subscore [rectal bleeding, stool frequency, endoscopy] of ≥2 points and ≥35% from baseline [BL] with either a decrease of RB subscore of ≥1 or an RB subscore of 0 or 1) using baseline levels of DEGSEGs and TNF biomarkers. Leave-one-out cross validation of the models using AMAC data and validation experiments by training on the PROTECT data⁴ then predicting in AMAC were used to evaluate our models.

Results: Miri 200mg responders were associated with 84 DEGSEGs, PBO responders were associated with 26 DEGSEGs, and 21 DEGSEGs were common to both. When using the PROTECT data as the training data and AMAC data as the validation cohort (Figure A), we were able to achieve high ROC-AUC and PR-AUCs using the top 36 (FC > 1.5) miri DEGSEGs (PR-AUC=0.78), TNF biomarkers (PR-AUC=0.91), or combined (PR-AUC=0.93) gene sets as the features (Figure B-E) and found that the response prediction was significantly higher in AMAC responders vs. non-responders (Figure F).

Conclusions: Baseline transcript features were predictive of clinical response in patients with UC treated with miri 200mg and our model distinguished responders from non-responders. Future studies may further validate this model.

References:
¹Sandborn WJ, et al. Gastroenterology. 2020;158(3):537-549.e10.
²Steere B, et al. J Crohns Colitis. 2020;S103-S104.
³Chen T, et al. Proceedings of the 22^nd ACM SIGKDD International Conference. 2016;7285-794.
⁴Hyams JS, et al. Lancet Gastroenterol Hepatol. 2017;2(12):855-868.

<b>Figure 1.</b> Predictive accuracy of models predicting treatment response in AMAC patients after being trained on the PROTECT cohort (A), Miri stringent DEGSEGs (B), and TNF-associated genes (C). The union of both gene sets (D-E) are predictive of response. Responders had higher predicted values for response than non-responders (F).

Figure 1. Predictive accuracy of models predicting treatment response in AMAC patients after being trained on the PROTECT cohort (A), Miri stringent DEGSEGs (B), and TNF-associated genes (C). The union of both gene sets (D-E) are predictive of response. Responders had higher predicted values for response than non-responders (F).

Venkatesh Krishnan

Appearances