Obesity drugs are as good as they need to be at this point - the newest triple agonists are producing average weight loss in the high twenties, and the late-phase pipeline means more gains (well, losses) can likely be had. It’s greater efficacy than many patients need or even want. As patients can better choose treatments to meet their target efficacy, the competitive landscape shifts to tolerability - whether patients can actually live on them: at what dose, with what side effects, for how long. Tolerability questions are an important reason why placebo-controlled trials will be the standard for the next several years, and this creates a challenge for interpreting trial results.

Let’s return to the case from my last post of TRIUMPH-1, Lilly’s triple agonist retatrutide that read out positive late last year. The top line: 28.3% weight loss at 12mg versus 2.2% on placebo for those who stayed on the drug, but 25.0% versus 3.9% factoring in off-protocol behavior. Participants leave the treatment arm because they’re unhappy with the drug’s adverse effects, and in some cases because participants felt the weight loss was too large. Participants leave the placebo arm because they know they’re not getting the study drug, and in some cases they may start taking other marketed drugs or entering other trials to lose weight.

Over time, more treatments will become available, with a spectrum of efficacy and tolerability. So what would happen if that same trial was started now, or in a few years? We might expect that more placebo participants will either leave the trial once they realize they’re not getting the treatment or, worse, start taking one of the many other available treatments on the market. We might also expect those on treatment unhappy with the efficacy-tolerability the drug offers to leave the study in greater numbers, given more available options. Both contribute to a shrinking apparent treatment effect over time. This makes for a complicated picture for us to understand and compare treatments.

Obesity is Multi-Faceted

As we are reaching beyond the efficacy targets that most patients want, adoption and use of weight loss medicines will be driven by a broader set of properties and outcomes. Principally, these are safety, cost, dosing (route of administration, dose frequency), tolerability (adverse effects such as nausea, vomiting), quality of weight loss (e.g., muscle and bone density preservation), impacts on comorbidities, and the ease of maintenance dosing.

This is a multi-faceted view of patient benefit, and goes far beyond the idea that people just want to lose weight. For each drug or drug candidate, sponsors and patients will care where they fall across these properties. We should expect no clear single winner, and there will be a “frontier” of treatments that develops. Some patients may prefer large weight loss, dealing with adverse effects. Some may realize they are in it for the long haul, preferring options with demonstrated maintenance dosing. Some may switch over time, especially if options proliferate and cost is a factor.

This complexity speaks to the systemic nature of obesity. I’m used to indications like Alzheimer’s, which affects the whole individual and their caregivers in profound ways. Progress in treatments for Alzheimer’s is challenging, and so efficacy is the primary focus, but we can see paths to meaningful improvements. In obesity, those improvements are here now, and only improving. I think that effective treatments lead to another fascinating set of problems – how do you measure the value when the benefits are so multi-faceted?  There are fundamental things that improve when a person moves out of a weight range for obesity or morbid obesity. Their quality of life, and that of those in their family, can meaningfully improve, as well as their ability to work, otherwise contribute to society, not to mention their long-term outcomes for comorbid conditions. It reminds one of the benefits of migraine treatments, another scourge of the working adult (often women, as the incidence of chronic migraine is around 10:1 female-to-male) – what are the benefits when a working mother goes from 15 migraine days per month to just 6?

There is increasing urgency for regulatory innovation.

I want to emphasize that the job of regulators is largely decoupled from the commercial value of drugs. A lot of what we care about for obesity drugs is more in the commercial realm – they are expensive to develop, but modern treatments are generally safe and very effective. From the point of view of regulatory approval, as a sponsor you dot your i’s and cross your t’s with the pharmacology studies, CMC, and run big late-stage trials (3000 participants followed for a year on treatment and 1500 on placebo) and can get approval. From the commercial side, as competition increases, the challenges mount; if you don’t meet your target product profile — or if the perception is that you haven’t met expectations — even if you do get approval, the market may be brutal.

The cost of development and the competitive landscape likely means that there will be a lot more phase 2 trials where sponsors choose to stop development if they’re not meeting their TPP, since the return on investment in late-stage trials could be low. And although regulatory considerations are different, two parts present challenges: one is how efficacy is assessed, the other is the requirements on control arm sizes for registration.

Guidance requires efficacy to be assessed using what’s known as “treatment policy”, where trial participants are evaluated on their weight loss even if they change behavior during the study, such as dropping out due to adverse effects or leaving the trial due to placebo disappointment. This is designed to provide a real-world understanding of treatment efficacy, but what happens when the rates of dropout (or drop-in, when placebo patients start taking marketed weight loss drugs against the trial protocol) change through time, as we discussed above?  This is fine to give a changing real-world picture of relative efficacy, but it is (a) much more situational to the trial than real-world behavior, and (b) it decouples the commercial and clinical perception of efficacy compared to the regulatory one. Commercially and clinically, we want to be able to compare different treatments, especially as their number grows. This tension is arguably Not Good™. 

The sample size in Phase 3 obesity trials is determined by the safety exposure requirements, simply because the drugs are so effective at weight loss compared to placebo that, while there are a lot of nuances about the size of efficacy, whether they are better is not really in question. Enrolling 3000 participants on treatment and exposing them for a year is reasonable given the systematic nature of these drugs and their widespread use. We should ask whether we need 1500 placebo patients to evaluate safety – would 1000 suffice, or 500?  Given the abundance of data from patients on placebo in prior trials, we have a pretty good idea of the adverse event rates, and there are some interesting adaptive designs here. It would also make for an interesting option for a platform trial, although the ease of comparative effectiveness may be a deterrent in an extremely competitive space.

Digital twins for future obesity programs

Many of these challenges can’t be easily addressed by data alone. You need models to explore counterfactuals across treatment options, and to help interpret efficacy across a changing trial landscape. This is where we see the benefits of digital twins – the fact that they are comprehensive means that we can address multifaceted questions across efficacy, safety, tolerability, and quality of weight loss, as well as help answer counterfactual questions across treatments.

Obesity has outgrown the question of whether a drug works. The questions that matter now are how well it works, for whom, the holistic benefit, and advantages and disadvantages compared to other treatments. Trials alone can’t answer these questions, and digital twins can help. A comprehensive model of how each participant would likely have fared on placebo and across treatments lets us map out treatment effect estimates that are resilient as the trial landscape changes, and understand benefits across outcomes and over other treatments. That’s the frontier patients will actually be choosing from, and it’s the one sponsors need to see clearly when deciding whether a Phase 2 asset deserves a Phase 3. The drugs are doing their part; how we understand evidence needs to catch up, and digital twins are how we get there.

Sponsors are bringing these questions to us now. If your next obesity readout depends on who stays on treatment, we would welcome the conversation.