GLP-1 receptor agonists have moved beyond diabetes and obesity. With 34 active pipeline candidates spanning 14 therapeutic indications — from Alzheimer’s to addiction to autoimmune disease — protocol designers are facing a set of challenges that didn’t exist 18 months ago.
We analyzed the full GLP-1 trial landscape using data from ClinicalTrials.gov, openFDA, and PubMed to identify the protocol design patterns that distinguish successful trials from those heading for amendments. This article breaks down what the data tells us and how to apply it to your next GLP-1 protocol.
If you haven’t seen it yet, our GLP-1 Trial Landscape Intelligence Brief tracks the full pipeline in real time — 10 companies, 5 pending FDA decisions, and every active safety signal.
The GLP-1 Protocol Design Problem Has Changed
Two years ago, designing a GLP-1 trial meant working within well-established parameters: Type 2 diabetes or obesity indication, HbA1c or body weight primary endpoint, 24–68 week duration, standard GI tolerability monitoring. The protocol design playbook was mature.
That playbook no longer covers the landscape.
Novo Nordisk is running a Phase 3 semaglutide trial for Alzheimer’s disease. Harvard’s Brigham and Women’s Hospital has active trials for opioid and alcohol use disorder. Eli Lilly is testing tirzepatide in combination with ixekizumab for psoriasis and psoriatic arthritis, and with mirikizumab for ulcerative colitis and Crohn’s disease. Pfizer is advancing a monthly injectable with 20+ planned studies across its obesity pipeline.
Each of these indications requires fundamentally different endpoint strategies, patient populations, safety monitoring protocols, and study durations. A protocol designer working on a GLP-1 MASH trial has more in common with a hepatology trialist than with the team running the next obesity study.
This is the new challenge: GLP-1s are becoming a multi-system therapeutic platform, and protocol design must adapt accordingly.
What the Pipeline Data Reveals About Endpoint Strategy
Looking at the current distribution of GLP-1 trials across indications, three endpoint patterns emerge that protocol designers should be building around.
Metabolic Indications: The Established Framework
For obesity and Type 2 diabetes — which still represent the majority of GLP-1 trial volume — the endpoint framework is well validated. Primary endpoints cluster around percent change in body weight from baseline, HbA1c reduction, and achievement of clinically meaningful weight loss thresholds (typically ≥5%, ≥10%, ≥15%, and increasingly ≥20% as newer agents show greater efficacy).
CagriSema (cagrilintide + semaglutide) demonstrated approximately 20% mean body weight loss at 68 weeks in clinical trials, and retatrutide (Eli Lilly’s triple agonist) is pushing these thresholds even higher. Protocol designers working on next-generation metabolic GLP-1 trials should calibrate their power calculations against these benchmarks — a new entrant showing 10% weight loss at 52 weeks will struggle to differentiate in a market where 20%+ is becoming the bar.
The Protocol Design tools we covered in our Protocol Design AI Stack guide are well suited for this category. TriNetX can pressure-test eligibility criteria against real-world patient populations, and Medidata’s protocol optimization suite is trained on thousands of comparable metabolic trials.
Cardiovascular and Cardiometabolic: Imaging Endpoints Enter the Picture
GLP-1 trials in cardiovascular indications — heart failure with preserved ejection fraction (HFpEF), peripheral artery disease (PAD), and MACE risk reduction — introduce imaging-based and functional endpoints that add protocol complexity.
For HFpEF trials, primary endpoints typically involve the Kansas City Cardiomyopathy Questionnaire (KCCQ) combined with hierarchical composite endpoints including cardiovascular death, heart failure hospitalization, and urgent heart failure visits. Echocardiography-based secondary endpoints (left atrial volume index, E/e’ ratio) require standardized imaging protocols and core lab adjudication — a significant operational consideration that pure metabolic trials don’t face.
Protocol designers in this space should be planning for centralized imaging review from the outset. Our Medical Imaging AI Stack guide covers the AI tools available for imaging endpoint standardization and quality control.
Non-Metabolic Indications: Uncharted Protocol Territory
This is where the most significant protocol design challenges lie. GLP-1 trials in Alzheimer’s, addiction, autoimmune conditions, and PCOS are operating without established endpoint precedents for this drug class.
Alzheimer’s trials are using cognitive scales (Montreal Cognitive Assessment), tau-PET imaging, TSPO-PET for neuroinflammation, GFAP protein levels, and neurofilament light chain as biomarkers. The semaglutide Alzheimer’s program includes MRI-based hippocampal volume measurement as a secondary endpoint. These are multi-modal assessment batteries that require specialized sites, trained raters, and imaging infrastructure that most GLP-1-experienced CROs are not set up to deliver.
Addiction trials present unique endpoint challenges around substance use biomarkers, self-reported use diaries, and relapse rates. There are currently more than 15 GLP-1 trials globally for substance use disorders, but the measurement frameworks are still being established. Validated patient-reported outcomes for addiction don’t transfer cleanly from the metabolic PRO instruments that GLP-1 trial teams are accustomed to.
Autoimmune indications (psoriasis, psoriatic arthritis, ulcerative colitis, Crohn’s disease) use dermatology-specific endpoints like PASI-90, EASI-90, and IGA 0/1, or gastroenterology endpoints like clinical remission and endoscopic response. These are well established within their respective therapeutic areas but new territory for GLP-1 trial teams.
The protocol design implication is clear: teams designing GLP-1 trials in non-metabolic indications need to bring in therapeutic area expertise for endpoint selection and measurement strategy, rather than relying on the metabolic trial playbook. AI tools like Medidata Protocol Optimization can benchmark your proposed design against comparable trials in the target indication — not just comparable GLP-1 trials.
Safety Monitoring: What openFDA Data Tells Protocol Designers
The FDA safety signal landscape for GLP-1s should directly inform your safety monitoring plan. Our GLP-1 Intelligence dashboard tracks nine active safety signals; here’s how they translate to protocol design decisions.
Standard GI Monitoring Is Table Stakes
Gastrointestinal adverse events (nausea, vomiting, diarrhea) remain the most common class effect across all GLP-1 receptor agonists. Every GLP-1 protocol needs a structured dose-titration scheme to manage GI tolerability, with pre-specified criteria for dose modification and discontinuation. This is well understood but still the most common cause of treatment discontinuation in clinical trials.
Signals That Should Change Your Protocol Design
Pancreatitis monitoring. Acute and necrotizing pancreatitis are listed as post-market adverse reactions for semaglutide and tirzepatide. Protocols should include baseline and periodic amylase/lipase monitoring, with clear stopping rules for suspected pancreatitis. Exclusion criteria should address history of pancreatitis — and the specificity of that criterion (acute vs. chronic, timeframe) directly affects your eligible patient pool.
Diabetic retinopathy. Semaglutide trials have shown higher rates of diabetic retinopathy complications compared to placebo, particularly in patients with existing retinopathy. For T2D trials, protocols should include baseline retinal examination and periodic monitoring, with stratification or exclusion criteria for patients with unstable retinopathy. This is a criterion that AI feasibility tools like TriNetX can pressure-test: how much does excluding “unstable retinopathy” shrink your eligible pool?
Pulmonary aspiration risk. This is a newer safety signal — GLP-1 receptor agonists delay gastric emptying, and pulmonary aspiration has been reported in patients undergoing procedures requiring general anesthesia or deep sedation. Any GLP-1 trial protocol that includes procedures under sedation (imaging studies, biopsies, endoscopies) should address pre-procedural fasting requirements that account for delayed gastric emptying.
Suicidal ideation. In March 2026, the FDA issued a warning letter to Novo Nordisk for failing to adequately report serious adverse events related to semaglutide, including unreported cases of suicidal ideation. Protocols should incorporate validated mental health screening instruments (PHQ-9 or C-SSRS) at baseline and periodically throughout the study, particularly for non-metabolic indications where the patient population may have higher baseline psychiatric comorbidity.
Eligibility Criteria: What the Competitive Landscape Tells You
One of the highest-value uses of pipeline intelligence for protocol design is benchmarking your eligibility criteria against active competitors.
The Enrollment Competition Problem
With 34 active GLP-1 candidates recruiting simultaneously, many targeting the same patient populations, site-level competition for eligible patients is real. In obesity, where trial volume is highest, overly restrictive eligibility criteria don’t just reduce your theoretical pool — they put you in direct competition with less restrictive trials recruiting at the same sites.
Using the GLP-1 Intelligence dashboard, you can identify exactly which companies are recruiting for which indications and at which phases. If you’re designing a Phase 3 obesity trial, knowing that Novo Nordisk, Eli Lilly, Pfizer, Amgen, Viking, and Roche all have active or upcoming Phase 2/3 obesity programs tells you that your criteria need to be competitive — not just scientifically justified, but operationally feasible in a crowded recruitment landscape.
BMI Thresholds Are Shifting
The traditional BMI ≥30 kg/m² (or ≥27 with comorbidities) inclusion criterion for obesity trials is being re-evaluated as the therapeutic landscape expands. Some newer trials are exploring broader metabolic phenotyping beyond BMI alone, incorporating waist circumference, visceral fat measurements, and metabolic biomarkers. Protocol designers should evaluate whether a BMI-only criterion is still the optimal approach for their specific program, or whether composite metabolic criteria could improve both the scientific rationale and the recruitment funnel.
Cross-Indication Exclusions
For GLP-1 trials in non-metabolic indications, a key criteria decision is whether to exclude patients already taking a GLP-1 for diabetes or obesity. Given the large and growing prescribed population, this exclusion could significantly narrow enrollment — but including them introduces confounding. AI feasibility platforms like TriNetX can quantify this trade-off using real-world patient data.
Protocol Duration and Retention: Lessons from the Data
GLP-1 trials face a specific retention challenge: the drugs work, patients feel better, and the temptation to discontinue study participation (especially in placebo-controlled trials) is high. The pipeline data suggests several design considerations.
Longer Trials Are Becoming Standard
CagriSema’s pivotal trials run 68+ weeks. Retatrutide’s Phase 3 program is designed for extended follow-up. Cardiovascular outcome trials (CVOTs) require years of follow-up for MACE endpoints. The days of 24-week GLP-1 efficacy trials are largely over for pivotal programs.
Protocols need to plan for retention over these longer horizons. Structured patient engagement strategies, flexible visit schedules, telehealth options, and clear communication about the importance of trial completion are operational considerations that should be designed into the protocol — not bolted on after enrollment begins.
Oral vs. Injectable: Protocol Implications
The approval of oral Wegovy 25mg in December 2025 and oral Ozempic tablets in February 2026, plus Eli Lilly’s pending oral orforglipron, means protocol designers now have oral comparator options for the first time. Oral GLP-1 trials have different adherence profiles, different PK considerations (fasting requirements for oral semaglutide), and potentially different patient populations willing to participate. Protocols should address whether oral vs. injectable formulation is a design variable or a fixed choice, and how it affects blinding strategy.
Building Your GLP-1 Protocol Design Stack
For teams designing GLP-1 clinical trials in 2026, we recommend a four-layer approach using the tools from our Protocol Design AI Stack guide:
Layer 1 — Competitive intelligence. Start with the GLP-1 Intelligence dashboard to understand the competitive landscape: who’s running what, at which phase, in which indications. This informs your differentiation strategy before you write a single protocol section.
Layer 2 — Feasibility and criteria optimization. Use TriNetX to pressure-test your proposed eligibility criteria against real-world patient populations. For GLP-1 trials specifically, pay attention to how criteria like BMI thresholds, prior GLP-1 use, retinopathy history, and psychiatric comorbidity exclusions affect your eligible pool in the context of 34 competing trials recruiting from the same population.
Layer 3 — Protocol simulation and authoring. Use Medidata Protocol Optimization to benchmark your proposed design against comparable trials — but benchmark against the target indication (Alzheimer’s trials, MASH trials, addiction trials), not just against other GLP-1 trials. Then use Clinion eProtocol to generate the structured protocol document, which your medical and regulatory team refines.
Layer 4 — Safety monitoring design. Map the known GLP-1 safety signals from our dashboard’s Safety tab into your monitoring plan. Use the signal severity and trend data to calibrate the intensity of your safety assessments: stable signals get standard monitoring, increasing or new signals get enhanced surveillance.
This layered approach ensures your protocol is designed against competitive reality, validated against real-world feasibility, and informed by the latest safety intelligence — not just based on the team’s best guess at what a GLP-1 trial should look like.
What’s Next
The GLP-1 pipeline will continue to evolve rapidly in 2026. Five FDA decisions are pending, CagriSema and retatrutide Phase 3 data readouts are expected, and new entrants from Pfizer, Amgen, Viking, and others will reshape the competitive landscape.
We’ll be updating the GLP-1 Intelligence dashboard as new data emerges. For teams that need a deeper analysis — custom competitive intelligence for a specific indication, protocol benchmarking against the full trial landscape, or safety signal analysis for a specific compound — reach out to discuss a custom briefing.
This article is part of the EmergingAIHub Clinical Research AI series. For the complete AI tool recommendations across all 8 stages of the clinical trial workflow, see The Complete AI Stack for Clinical Research.
Affiliate Disclosure: Some links in this article are affiliate links. EmergingAIHub may earn a commission at no extra cost to you when you use these links. We only recommend tools we’ve evaluated and believe add genuine value to clinical research workflows.