A CRO project manager running a 400-patient phase 2 triple-agonist obesity trial watched the screening log for eleven weeks before flagging the problem to the sponsor: 61% of self-referred candidates were failing screening, not because they lacked obesity, but because the protocol had stacked a BMI floor, an HbA1c ceiling, and an eight-week GLP-1 washout requirement on top of each other. Enrollment that was modeled to close in twelve months was tracking toward twenty. That scenario is not unusual in incretin-agonist trial design, and it is the reason protocol architecture — not just molecule potency — determines whether a GLP-3 class trial reads out on schedule.
What Investigators Mean by "GLP-3" in Current Protocols
There is no endogenous GLP-3 receptor in human pharmacology. The term functions as informal shorthand — used in sponsor decks, CRO proposals, and trade press — for a class of next-generation incretin agonists that act on more than one receptor simultaneously. Most commonly this means GIP/GLP-1/glucagon triple receptor agonists, of which retatrutide (LY3437943) is the most clinically advanced example, alongside dual-agonist and multi-agonist candidates in earlier development.
Protocol authors who use "GLP-3" informally in internal communication should not let the shorthand migrate into the background and rationale sections of the actual protocol document. IRBs, ethics committees, and FDA reviewers expect explicit receptor pharmacology — binding targets, agonist potency data, and mechanism-of-action rationale tied to the specific molecule under study, not a marketing-derived class label.
This distinction matters operationally. A protocol that conflates "GLP-3" with a single defined pharmacologic target risks inconsistent eligibility criteria across sites if investigators interpret the mechanism differently. Defining the molecule's actual receptor profile in section 1 of the protocol removes that ambiguity before the first site activates.
Endpoint Selection: Primary and Secondary Outcomes That Hold Up to Regulatory Scrutiny
Percent change in body weight from baseline, typically measured at week 48 or week 72, remains the standard primary endpoint for obesity-indicated incretin trials. The retatrutide phase 2 trial (n=338, NEJM 2023) reported mean weight loss up to 24.2% at 48 weeks in the highest-dose arm, while SURMOUNT-1 (n=2,539, NEJM 2022) reported up to 20.9% at 72 weeks with tirzepatide 15 mg. These figures function as the closest available benchmarks for power calculations in new triple-agonist protocols.
Secondary endpoints typically include waist circumference, HbA1c change in participants with prediabetes or type 2 diabetes, lipid panel shifts, systolic blood pressure, and patient-reported outcome instruments such as the Impact of Weight on Quality of Life-Lite (IWQOL-Lite). Composite endpoints combining weight loss thresholds (e.g., proportion achieving ≥15% or ≥20% weight reduction) are increasingly used as key secondary measures because they map more directly to clinically meaningful outcomes than mean percent change alone.
Estimand framework selection — treatment policy versus hypothetical (on-treatment) — should be locked before the statistical analysis plan is finalized, not retrofitted after unblinding. Sponsors that specify a treatment-policy estimand as primary and a hypothetical estimand as sensitivity analysis generally face fewer regulatory queries about how discontinuations and rescue therapy use are handled in the primary readout.
Inclusion and Exclusion Criteria: The Screen-Fail Problem
The scenario described at the start of this article is a design failure, not a recruitment failure. Each individual criterion — BMI ≥30 kg/m² or ≥27 kg/m² with a weight-related comorbidity, HbA1c below 9%, an eight-week washout from prior GLP-1 receptor agonist use — is defensible in isolation. Stacked together without modeling, they compound into a screen-fail rate that can exceed 50–60% in a self-referred population, as documented informally across multiple CRO feasibility reports on obesity-indication trials.
The corrective step that experienced protocol teams now build in as standard practice: run the proposed eligibility criteria against a retrospective electronic health record cohort of 500–1,000 patients matched to the target population before the protocol is finalized. This produces a realistic screen-fail estimate before a single site opens, rather than after three months of underperforming recruitment.
A tiered eligibility structure — a core cohort meeting strict criteria for the pivotal analysis, plus an expanded-access or open-label extension sub-study with relaxed criteria — preserves data integrity for the primary endpoint while widening the enrollable population and reducing site frustration with high screen-fail volume.
Dose-Escalation Schedules and Titration-Driven Attrition
Gastrointestinal adverse events — nausea, vomiting, diarrhea — are the dominant driver of early discontinuation in incretin-agonist trials, not lack of efficacy. In STEP 1, adverse-event-related discontinuation occurred in roughly 7% of the semaglutide arm versus approximately 3% of placebo, with GI events accounting for the largest share of that difference.
Standard titration schedules now step doses upward every four weeks in 2.5 mg to 5 mg increments (molecule-dependent) specifically to blunt this effect. Protocols that compress escalation intervals to two weeks in pursuit of faster dose-response data consistently show higher early dropout in comparative feasibility analyses, trading recruitment speed for retention risk later in the trial.
A practical mitigation used in newer protocols: a structured run-in period of 2–4 weeks on a sub-therapeutic starter dose before randomization, paired with scripted nurse-led counseling on anticipated GI symptoms. Sites that implement this consistently report fewer unplanned early-visit calls and lower rates of unscheduled discontinuation in the first 8 weeks of dosing.
Site Selection and Investigator Burden
Injectable metabolic trials typically enroll at 0.5 to 1.5 patients per site per month. For a 400-patient trial targeting a 12–18 month enrollment window, that arithmetic requires 40 to 60 actively enrolling sites, not the 15–20 sites a sponsor might budget for a simpler oral-drug study.
Site feasibility questionnaires should screen specifically for prior experience with injectable titration counseling and infrastructure for the 12–16 clinic visits typical of a 48–72 week metabolic trial protocol — this includes visit-day capacity for vitals, labs, and injection training, not just principal investigator credentials.
Investigator burden is frequently underestimated at the protocol design stage. A site enrolling eight patients simultaneously across staggered titration schedules is managing a materially higher coordination load than the per-patient visit count suggests, because escalation timing varies by randomization date. Sponsors that build in a coordinator-to-patient ratio requirement (commonly one full-time coordinator per 15–20 actively enrolled patients) during site selection see fewer protocol deviations related to missed titration windows.
Retention Strategy: What the Published Dropout Data Shows
STEP 1 reported an overall trial completion rate near 94.7%, achieved through structured injection training, scheduled phone check-ins between visits, and dedicated patient support materials. SURMOUNT-1 reported discontinuation of study drug in the range of 14% at the highest dose arm, concentrated in the first 24 weeks of dosing — the same window where GI adverse events peak during titration.
Retention tactics with the clearest track record in published protocols and CRO operational reports include reimbursed travel for in-person visits, telehealth options for lower-acuity check-in visits, and video-based injection training modules that patients can revisit between clinic visits. None of these substitute for adequate GI-symptom counseling during escalation, but they reduce the administrative friction that turns a manageable side effect into a dropout.
Protocols that track a leading indicator — missed-dose calls or unscheduled contact volume in weeks 1–8 — rather than waiting for the discontinuation rate itself give site staff and sponsors a window to intervene before a patient formally exits the trial.
Safety Monitoring and Adverse Event Reporting Requirements
Protocols for this class typically require serial amylase and lipase monitoring for pancreatitis surveillance, alongside the calcitonin-screening precaution carried over from GLP-1 receptor agonist labeling due to thyroid C-cell tumor findings in rodent models — a signal that has not been confirmed in human data but remains a standard monitoring inclusion pending longer-term human safety follow-up.
Cardiovascular risk assessment is built into obesity-drug development programs as a structural requirement, a pattern that followed prior market withdrawals of weight-loss drugs for cardiovascular signals in the 2000s. Sponsors typically address this either through a dedicated cardiovascular outcome trial run in parallel with the pivotal efficacy program or through a defined post-marketing commitment, and FDA guidance for the class should be consulted directly when finalizing this element of the protocol.
Adverse event reporting cadence follows 21 CFR 312 requirements for IND-stage trials, with serious adverse events reported to the FDA and IRB within regulatory timelines regardless of causality assessment. A Data Monitoring Committee reviewing unblinded safety data on a fixed schedule — commonly every six months — or triggered by a pre-specified adverse event count, is now standard rather than optional for trials of this size and duration.
Enrollment of Sub-populations and Diversity Targets
FDA's 2022 draft guidance on diversity action plans has shifted how eligibility and site-selection strategy get built, requiring sponsors to specify enrollment targets for populations reflecting the disease's actual demographic distribution rather than defaulting to whichever population is easiest to recruit near existing trial sites.
For obesity-indication trials, that means enrollment targets informed by NHANES prevalence data across racial and ethnic groups, rather than a simple proportional quota applied without reference to underlying disease burden. Sponsors that build diversity targets into the site-selection algorithm — weighting site catchment demographics alongside enrollment velocity — reach representative enrollment without the late-stage scramble that otherwise follows a diversity action plan review.
Renal and hepatic impairment sub-cohorts, along with adolescent extension studies where the indication supports it, are increasingly built as parallel sub-protocols rather than bolted onto the main pivotal trial, which keeps the primary population homogeneous for the primary endpoint while still generating the sub-population safety data regulators expect at NDA submission.
Data Monitoring Committees and Interim Analysis Triggers
Interim analyses in trials of this size are typically triggered either by enrollment milestone (commonly at 50% enrollment) or by accumulation of a pre-specified adverse event count, with a formal alpha-spending function (O'Brien-Fleming or Lan-DeMets approaches are both common) governing how much statistical significance budget an interim look consumes.
Stopping rules need to be specified for both futility and overwhelming efficacy before the trial starts — not negotiated at the interim analysis meeting itself. A DMC operating without pre-specified stopping boundaries introduces exactly the kind of post-hoc decision-making that regulatory reviewers flag during NDA review.
Unblinding thresholds for the DMC should be documented separately from the sponsor-facing safety reports, with a firewall between the statistician generating the unblinded analysis and the sponsor team managing trial operations, to preserve the integrity of the ongoing blinded trial.
Regulatory Pathway and IND Considerations
A pre-IND (Type B) meeting with FDA typically has a target response window near 60 days from request, and protocol teams that use this meeting to align on endpoint selection, dosing rationale, and safety monitoring expectations before finalizing the protocol document avoid the more costly alternative of a protocol amendment after site activation.
Fast Track and Breakthrough Therapy designations remain available for molecules addressing a serious condition with an unmet medical need, but neither designation substitutes for a well-powered pivotal trial — they primarily affect the frequency and format of FDA interaction during development, not the evidentiary bar for approval.
REMS (Risk Evaluation and Mitigation Strategy) considerations for this drug class typically center on the GI and pancreatic risk profile carried through the label, and protocol-stage safety monitoring plans should be built with an eye toward the data package a REMS decision will eventually require, rather than treated as a separate downstream problem.
The Concrete Next Step
Before finalizing eligibility criteria on any new GLP-3 class protocol, the single highest-leverage action is running the proposed inclusion and exclusion criteria against a retrospective EHR cohort of the target population, or piloting them on a 20–30 patient run-in cohort, before those criteria are locked into the final protocol document. That step alone catches the majority of the screen-fail problems that otherwise surface three months into a multi-site enrollment period, when the fix is far more expensive than it would have been at the design table.
This article summarizes research and does not constitute medical advice. Consult a licensed clinician for diagnosis, treatment, or any decisions about medications or supplements.