P0759 - Optimizing Treatment Selection in Crohn’s Disease Using Patient-Specific Features: An Individual Participant Data Meta-Analysis of Fifteen Randomized Controlled Trials

Sunday, October 22, 2023

3:30 PM – 7:00 PM PT

Location: Exhibit Hall

Has Audio

Presenting Author(s)

VR

Vivek Rudrapatna, MD, PhD

UCSF
San Francisco, CA

Vivek Rudrapatna, MD, PhD¹, Vignesh Ravindranath, MS¹, Douglas Arneson, PhD¹, Arman Mosenia, MD¹, Atul Butte, MD, PhD¹, Shan Wang, PhD²
¹UCSF, San Francisco, CA; ²University of San Francisco, San Francisco, CA

Introduction: Crohn’s disease is characterized by diverse clinical manifestations that likely reflect differences in disease biology and thus treatment susceptibility. Prior studies have found anti-tumor necrosis factor (anti-TNF) drugs to be more efficacious than other drugs based on their cohort-averaged effects. We performed a subgroup analysis to determine if it is possible to achieve even more efficacious outcomes by personalizing treatment selection.

Methods: See Fig. 1A-D for an overview. We obtained participant-level data from 15 trials of FDA-approved treatments (N=5703). We used sequential regression and simulation to model the week six Crohn’s Disease Activity Index as a function of drug class, demographics, and disease-related features. We performed post-hoc statistical hypothesis testing to classify participants into treatment subgroups, defined by distinct rank orderings of treatment efficacy for three drug classes (anti-TNF, anti-Integrin, anti-IL12/23). We queried health records data from University of California Health (UCH; 2012-2023) to estimate the potential impacts these could have on practice. We computed the sample size needed to prospectively test a key prediction of our models. Lastly, we prototyped a treatment recommendation tool that uses patient features as inputs.

Results: 55% of all participants did not show superior efficacy with any one drug class. We classified the remainder into six subgroups (Table 1), two of which showed greatest efficacy with anti-TNFs (36%, N=2061). We also identified a subgroup of women over 50 with superior responses to anti-IL12/23s (Fig. 1E). Although they represent 2% of the trial-based cohort, 25% of Crohn’s patients at UCH are women over 50 (n=5,647). During the period when all three drug classes were FDA-approved, 75% of biologic exposed women over 50 did not receive an anti-IL12/23 first line. We calculate that a trial with 250 subjects per arm will have 97% power to show anti-IL12/23s as being superior to anti-TNFs in this subgroup.

Discussion: Patients with Crohn’s disease likely harbor different underlying tendencies to benefit from different treatments. Prior results suggesting the superiority of anti-TNFs may reflect majority-vote methods, where many would not benefit from that drug class. Crohn’s trials appear to under-enroll many patient subgroups, including women over 50 who may benefit from anti-IL12/23s. We are releasing a decision support tool (crohnsrx.org) to facilitate early user testing in advance of confirmatory studies.

Figure: Figure 1: Study overview (A-D) and summary of subgroup analyses (E).  
A. Clinical trials were found using clinicaltrials.gov and sought for retrieval on the YODA and Vivli platforms. Individual participant data (IPD) from trials that collected CDAI scores at week 6 visits were then aggregated and harmonized.  
B. Using sequential regression and simulation, a method for normalizing clinical trial data against a common placebo rate, a placebo-attributable model and three drug-attributable models - anti-integrin, anti-interleukin-12/23, and anti-TNF - were developed. Disease activity reduction was partitioned into placebo attributable (square), and drug-attributable (circle) effects based on baseline covariates (age, sex, BMI, etc.). IPD (solid lines) were used to predict or simulate data (dashed lines).  
C. The drug-attributable models were utilized to simulate patient-level outcomes post-treatment (counterfactuals). Pairwise t-tests (p < 0·05) were conducted to compare and rank the mean responses for all drug classes - anti-integrin vs anti-interleukin-12/23, anti-integrin vs anti-TNF, and anti-interleukin-12/23 vs anti-TNF - and assign patients into one of seven subgroup memberships (see Table 3).  
D. Lastly, the models were re-packaged into a prototype decision support tool that uses manual inputs and optionally, OMOP-formatted data, to recommend treatments for individual patients.  
E. Detailed comparison of three major subgroup cohorts found in the trial-based cohort (N=5703): prefer anti-TNF only (N = 2061, red), prefer anti-TNF or anti-IL-12/23 (N = 355, blue), and prefer anti-IL-12/23 only (N = 139, green). The first row of bar plots show the average placebo (P) and drug-class (D) attributable effects for each subgroup. Superior drug classes (left of bolded vertical line) reduce disease activity (CDAI reduction) by 30-40 points more on average compared to non-superior drug-classes (right of bolded vertical line). The second and third rows of plots compare the proportions and distributions of covariates for each subgroup.

Drug Class Preference	Subgroup	N (%)
Anti-TNF	TNF > IL > INT	44 (0.8)
Anti-TNF	TNF > (IL = INT)	2,017 (35)
Anti-TNF, Anti-Interleukin-12/23	(IL = TNF) > INT	355 (6)
Anti-Interleukin-12/23	IL > (TNF = INT)	138 (2.5)
Anti-Interleukin-12/23	IL > TNF > INT	1 (0.02)
Other	(TNF = INT) > IL	4 (0.07)
No Preference	(TNF = IL = INT)	3,144

Table: Table 1: Treatment subgroups. We calculated the potential outcomes (mean response, residual uncertainty) for every patient under scenarios where they were assigned to different drug classes, and pairwise compared treatment scenarios to define preference orderings (treatment subgroups). TNF = anti-tumor necrosis factor, IL = anti-interleukin-12/23, INT = anti-integrin.

Disclosures:

Vivek Rudrapatna: Alnylam – Grant/Research Support. Blueprint Medicines – Grant/Research Support. Genentech – Grant/Research Support. Janssen – Grant/Research Support. Merck – Grant/Research Support.

Vignesh Ravindranath indicated no relevant financial relationships.

Douglas Arneson: Bristol Meyers Squibb – Employee. Bristol Myers Squibb – Employee, Stock-publicly held company(excluding mutual/index funds).

Arman Mosenia indicated no relevant financial relationships.

Atul Butte: Personalis and NuMedii,Mango Tree Corporation, Samsung, 10x Genomics, Helix, Pathway Genomics, and Verinata,Geisinger Health, Regenstrief Institute, – Advisor or Review Panel Member, Advisory Committee/Board Member, Consultant, Employee, Grant/Research Support, Independent Contractor, Intellectual Property/Patents, Owner/Ownership Interest, Royalties, Stock Options, Stock-privately held company, Stock-publicly held company(excluding mutual/index funds). Regenstrief Institute, Gerson Lehman Group, AlphaSights, Covance, Novartis, Genentech, and Merck, and Roche; Personalis and NuMedii;Apple, Meta, – Advisor or Review Panel Member, Advisory Committee/Board Member, Consultant, Employee, Grant/Research Support, Independent Contractor, Intellectual Property/Patents, Owner/Ownership Interest, Royalties, Speakers Bureau, Stock Options, Stock-privately held company, Stock-publicly held company(excluding mutual/index funds).

Shan Wang indicated no relevant financial relationships.

Vivek Rudrapatna, MD, PhD¹, Vignesh Ravindranath, MS¹, Douglas Arneson, PhD¹, Arman Mosenia, MD¹, Atul Butte, MD, PhD¹, Shan Wang, PhD². P0759 - Optimizing Treatment Selection in Crohn’s Disease Using Patient-Specific Features: An Individual Participant Data Meta-Analysis of Fifteen Randomized Controlled Trials, ACG 2023 Annual Scientific Meeting Abstracts. Vancouver, BC, Canada: American College of Gastroenterology.