Design for efficiency: Employ trial strategies to meet stakeholder needs

This article is part of Parexel's "Navigating to 2030" playbook series on differentiating next-generation obesity therapies. This series offers strategic insights across trial design, regulatory considerations, clinical operations, and patient retention strategies to support sponsors in this rapidly evolving and competitive market.  

Home Section 1 Section 2 Section 3 Section 4 Section 5

Testing GLP-1s and other innovative products in cardiometabolic and endocrine clinical trials presents unique statistical and design challenges. At Parexel, in conducting dozens of such trials over the last five years, we have found the following strategies helpful:  

Proper handling of missing data is critical to obesity trials. 

Obesity trials can have high dropout rates if patients are unable to tolerate GI side effects or if, due to a lack of weight loss, they suspect they are in the placebo arm. Also, side effects may cause patients in the active arm not to comply with the dosing-up titration schedule even though they remain in the trial. These factors can affect the completeness or the interpretation of the key efficacy assessments collected in the trial. The FDA’s draft guidance developing products for weight reduction recommends using estimands in clinical trials with special attention to handling missing data.¹    

An estimand is a precise description of the treatment effect that a study aims to estimate. It is crucial to ensure that the study's design, analysis, and interpretation are aligned. An essential part of the estimand is handling “intercurrent” events such as treatment discontinuation and rescue medication that can affect the interpretation of the results.

While the FDA’s draft guidance recommends using a treatment policy estimand, the clinical trial objective in Phase 2 may be better aligned with an “efficacy” estimand where treatment effects are assessed as though participants remained adherent to treatment and did not start other treatments. Missing data handling needs to be aligned with the targeted estimand and multiple imputations using data based on “retrieved dropouts” will likely be required. For these methods, it is important to continue following up with participants who discontinue treatment. 

The sensitivity of the results should be assessed using tipping point analyses. At Parexel, we are experienced with these approaches.

Include differentiating secondary endpoints in Phase 2 trials.

Standard primary endpoints for diabetes and obesity trials include hemoglobin A1C levels (a measure of blood sugar), percent change in body weight, and body mass index (BMI). However, a host of secondary endpoints, including insulin resistance, inflammatory biomarkers, cardiovascular outcome measures, biomarkers related to hunger and satiety, and patient-reported outcomes (PROs), can provide more detail about a product’s risk-benefit profile. BMI has well-documented limitations—including potentially overestimating body fat in athletes and underestimating body fat in older adults—and newer measures, such as body roundness index (BRI), waist circumference, and waist-to-hip ratio, should be evaluated. These innovative endpoints could be incorporated into Phase 2 trials when they might help differentiate a product.

Recently, we worked with a sponsor to capture secondary body composition endpoints, including visceral fat and muscle volume, in their Phase 2 study. They were choosing between bone density and magnetic resonance imaging to measure them. At Parexel, we have deep experience with imaging tests such as these in early-stage studies and, therefore, could advise the sponsor on the pros and cons of each. We advised the company to use MRI because it captures more detail about body composition. These data will give them an early readout on differentiating endpoints to include in the pivotal trial design.

Consider designs with innovative elements. 

To accelerate development, an interim analysis during the Phase 2 trial(s) can be included to inform the planning of the Phase 3 program. Go/No Go criteria based on current estimates of the Phase 2 trial efficacy may be defined for quick decision-making about ongoing clinical evaluation. 

Early to mid-stage study designs could also incorporate Bayesian approaches. Bayesian methods that borrow information from clinically relevant external evidence, such as previous RCTs, can reduce the sample size, including the ratio of randomization to the placebo arm. A Goldilocks design is a Bayesian adaptive design that utilizes predictive probability to adaptively select a trial's sample size based on accumulating data.²  Broad patient selection criteria can facilitate more diverse enrollment, considering factors like gender, ethnicity, and BMI. For example, FDA guidance suggests including a sample of patients with Class 3 or severe obesity (BMI>/= 40 kg/m2). Enrolling study participants more representative of the target patient population could lead to better trial outcomes and more generalizable results, which payers prefer. One recent study of 246 randomized controlled trials in obesity found that a majority of them over-enrolled white females aged 40 years or older with low- to moderate-risk obesity and under-enrolled severely obese people, non-Whites, and males.³ 

Decentralized clinical trials surged during the COVID pandemic, but their utilization has since declined, partly because some patients prefer face-to-face contact with clinical trial sites and staff.  However, targeted decentralized techniques like telehealth visits may increase patient enrollment and retention by easing the burden of participating in a clinical trial and allowing sponsors to reach specific individuals typically under-represented in clinical trials, such as people living in rural areas.⁴ 

Evaluate connected devices and sensors for data collection.

Digital technologies that capture data passively, such as wearable or connected devices, can facilitate patient recruitment, reduce the burden of trial participation, and improve participant retention. However, those devices must be validated for sensitivity and reliability, ease of use, real-time data collection capability, and ability to measure meaningful metrics. 

Leveraging connected wearable devices and sensors in obesity trials to capture continuous objective data could reduce the need for larger patient cohorts by boosting recruitment and retention, thus offering a competitive edge to smaller biotechs. Connected wearable devices and sensors also facilitate complete data capture. For example, all the GLP-1s currently approved for diabetes, overweight, and obesity are required to be taken in combination with diet and exercise. Passive and continuous capturing of exercise metrics via an activity tracker (often a watch-like device) could minimize missing information and provide a wealth of data on patient compliance with a trial’s treatment regimens.

Wearables, sensors, and implants combined with AI-driven analysis and reporting could also lead to the development of novel digital biomarkers, deliver personalized dosing, or trigger alerts to provide patients with emotional and behavioral counseling to encourage adherence. 

Resources

1. Obesity and Overweight: Developing Drugs and Biological Products for Weight Reduction, FDA Draft Guidance Document (January 2025).
2. Not too big, not too small: a goldilocks approach to sample size selection, Journal of Biopharmaceutical Statistics (2014).
3. Sex, race, and BMI in clinical trials of medications for obesity over the past three decades: a systematic review, The Lancet Diabetes & Endocrinology (June 2024).
4. Strengths and opportunities to clinical trial enrollment among BIPOC, rural dwelling patients in the northwest United States: a retrospective study, Frontiers in Pharmacology (January 24, 2024).

Return to Insights Center