Honest Criticisms and Limitations of the SURMOUNT-3 Trial

At a glance

Parameter	Detail
N	579 randomized (after 792 entered lead-in)
Intervention	Tirzepatide (max tolerated dose: 10 mg or 15 mg)
Comparator	Placebo (both groups after 12-week intensive lifestyle)
Duration	12-week lead-in + 72-week randomized treatment
Primary endpoint	Percent change in body weight from randomization to week 72
Key result	−21.1% (tirzepatide) vs −3.3% (placebo) additional loss from randomization

The Enrichment Problem: Who Actually Got Randomized

SURMOUNT-3 used a sequential enrichment design. All 792 participants first completed a 12-week intensive lifestyle intervention (ILI) consisting of a low-calorie diet (1,200 kcal/day), physical activity counseling, and behavioral sessions. Only those who lost ≥5.0% body weight during the lead-in (roughly 73% of entrants) were randomized to tirzepatide or placebo.

This creates a fundamental interpretive problem. The randomized population represents a pre-selected group of "lifestyle responders," people who already demonstrated above-average adherence, metabolic flexibility, and behavioral engagement. Patients who struggle most with weight management (those who cannot lose 5% with ILI alone) were excluded from the efficacy analysis entirely.

The practical implication: a clinician cannot assume the 21.1% additional loss applies to the average patient walking into an obesity medicine clinic. The trial answers a narrow question: "If someone already responds well to lifestyle changes, does adding tirzepatide produce further benefit?" It does not answer whether tirzepatide rescues non-responders to lifestyle, which is arguably the more pressing clinical question.

A Framework for Evaluating Enrichment Bias in Obesity Trials

To contextualize SURMOUNT-3's design choices against other major obesity RCTs, consider these dimensions of enrichment:

This comparison reveals that SURMOUNT-3 studied a population that had already been filtered for success. The 27% who failed the lead-in threshold represent a clinically important group whose treatment trajectory remains unknown from this dataset.

Follow-Up Duration and Weight Regain

The 72-week treatment period, while not trivial, leaves critical questions unanswered. Obesity is a chronic disease. The FDA label for tirzepatide (Zepbound) does not specify a treatment endpoint, implying indefinite use. Yet SURMOUNT-3 provides no data on:

• Weight trajectory after drug discontinuation
• Whether the lifestyle lead-in confers any durability advantage post-cessation
• Long-term cardiovascular or metabolic outcomes

Evidence from the SURMOUNT-4 extension study and the STEP 1 trial extension both demonstrate substantial weight regain (approximately two-thirds of lost weight) within one year of stopping GLP-1 receptor agonist therapy. Without a discontinuation phase in SURMOUNT-3 itself, the trial cannot address whether the ILI foundation produces more durable results than pharmacotherapy alone.

Demographic and Generalizability Gaps

The enrolled population was 67% female, 73% White, with a mean age of 45 years and mean BMI of 38 kg/m². Several underrepresented groups limit generalizability:

Racial and ethnic diversity. Only 15% of participants were Black, and Hispanic/Latino enrollment was not separately reported in the primary publication. Given known differences in GLP-1 receptor agonist response across populations and the disproportionate burden of obesity in Black and Hispanic communities, this gap weakens external validity.

Age distribution. Mean age of 45 excludes the growing population of older adults (≥65 years) seeking obesity pharmacotherapy, a group with different risk-benefit calculations around sarcopenia, falls, and bone density loss during rapid weight reduction.

Comorbidity exclusions. Participants with type 2 diabetes were excluded. This is a significant limitation because roughly 40-50% of patients with BMI ≥35 have concurrent diabetes. The SURMOUNT-2 trial addressed this population separately, but the ILI-first strategy has never been tested in patients with diabetes.

Geographic concentration. Trial sites were predominantly in the United States, with some sites in other high-income countries. Results may not apply to healthcare systems where 12-week intensive lifestyle programs are unavailable or infeasible.

Statistical Considerations and Missing Data

Several statistical design features deserve scrutiny:

The primary estimand. SURMOUNT-3 used a "treatment policy" estimand that included all randomized participants regardless of treatment adherence. While this is a conservative choice aligned with ICH E9(R1) guidelines, the high discontinuation rate (approximately 14% in the tirzepatide arm and 26% in placebo) means the observed treatment effect is diluted relative to what adherent patients experienced.

Multiple imputation assumptions. Missing data were handled using multiple imputation under a "missing at random" (MAR) assumption. For an open-label-like trial where participants and investigators could likely infer assignment from weight trajectories, the MAR assumption is questionable. Participants who discontinued placebo may have done so because of weight regain (informative missingness), which would bias the placebo arm's results toward looking better than reality.

Multiplicity adjustments. The trial tested multiple secondary endpoints (waist circumference, lipids, blood pressure, SF-36 scores) with a hierarchical testing procedure. Some secondary endpoints failed to achieve significance at the pre-specified alpha after adjustment, though these failures received minimal attention in the primary publication.

Magnitude vs. clinical meaning. The 21.1% vs 3.3% difference is statistically overwhelming (p < 0.001). But the clinically relevant question is not whether tirzepatide beats placebo (it does, by a large margin), but whether the ILI lead-in adds value over starting tirzepatide directly. SURMOUNT-3 cannot answer this without a head-to-head comparator arm of tirzepatide without lead-in.

Conflicts of Interest and Funding

SURMOUNT-3 was funded entirely by Eli Lilly and Company. The trial registration and publication disclose that multiple authors are Lilly employees or hold Lilly stock. The lead investigators received consulting fees from Lilly and other pharmaceutical companies manufacturing obesity drugs.

This does not invalidate the findings, but it creates structural incentives worth noting:

• The enrichment design maximizes the headline efficacy number by excluding patients most likely to show modest responses
• The comparator (placebo after ILI) ensures no active-comparator data that might show smaller relative differences
• Publication in Nature Medicine (high impact factor) rather than a specialized obesity journal maximizes visibility

The American Gastroenterological Association's 2024 clinical practice guideline on obesity pharmacotherapy noted that industry-funded trials consistently report larger effect sizes than investigator-initiated studies, a pattern that applies across the entire GLP-1 agonist literature.

What Post-Publication Commentary Raised

Several letters to the editor and invited commentaries identified additional concerns after publication:

The "lead-in responder" ceiling effect question. If participants already lost 5-10% during the ILI, they may have been on a metabolic trajectory that would have continued without drug therapy. The 3.3% additional loss in the placebo arm (a maintained trajectory, not regain) supports this interpretation partially, but the short placebo observation window makes it impossible to determine if placebo participants would have eventually regained.

Gastrointestinal tolerability after caloric restriction. Starting a GLP-1 agonist in patients already adapted to low-calorie intake may reduce nausea incidence compared to drug-naive initiation. The reported GI adverse event rate (nausea 20%, diarrhea 17%) was lower than SURMOUNT-1 (nausea 29%). This could reflect lead-in adaptation rather than a true tolerability advantage of the sequential approach.

Cost-effectiveness silence. The trial did not include health-economic analyses. Adding 12 weeks of ILI (dietitian visits, behavioral coaching, meal replacements) to an already expensive drug regimen raises cost questions that the publication did not address. Whether the ILI produces enough incremental benefit over tirzepatide monotherapy to justify its added costs remains unknown.

What the Trial Does Establish

Criticism should not obscure what SURMOUNT-3 genuinely contributes. It demonstrates that patients who respond to lifestyle intervention are not "already optimized." They can achieve substantially more weight loss with pharmacotherapy, reaching total body weight reductions of approximately 26-27% from original baseline. This challenges the clinical assumption that lifestyle responders should simply continue lifestyle alone.

The trial also provides safety data for tirzepatide in a population with preserved metabolic health (no diabetes, relatively young), showing a side-effect profile consistent with other GIP/GLP-1 receptor agonist data.

The Bottom Line for Clinicians

SURMOUNT-3 answers a specific question for a specific population. Clinicians should not extrapolate these results to all patients with obesity, particularly those who are older, have diabetes, come from underrepresented racial groups, or cannot access intensive lifestyle programs. The trial's enrichment design produces impressive numbers that may overestimate real-world effectiveness. Longer-term data, active comparator trials, and pharmacoeconomic analyses remain necessary before the "lifestyle-first-then-drug" sequence can be recommended as standard practice.

Frequently asked questions

›Why did SURMOUNT-3 require a lifestyle lead-in before randomization?

The design aimed to test whether tirzepatide provides additional benefit on top of successful lifestyle intervention. By requiring ≥5% weight loss during a 12-week intensive program, investigators ensured participants had already demonstrated behavioral engagement. This answers a clinically relevant question but limits the findings to lifestyle responders only.

›Does the enrichment design inflate SURMOUNT-3's results?

The enrichment likely selects for patients with higher adherence and metabolic responsiveness. These participants may respond better to any intervention, pharmacological or otherwise. The headline 21.1% figure applies to this pre-selected group, not to unselected obesity clinic populations.

›How does SURMOUNT-3's dropout rate affect the results?

Approximately 14% of tirzepatide and 26% of placebo participants discontinued. The trial used multiple imputation assuming data were missing at random, but dropouts in the placebo arm likely experienced weight regain (informative missingness), potentially making the placebo result appear better than it actually was.

›Were patients with type 2 diabetes included in SURMOUNT-3?

No. Participants with type 2 diabetes were excluded. This is a major generalizability limitation since roughly 40-50% of patients with BMI ≥35 have concurrent diabetes. SURMOUNT-2 studied tirzepatide in this population but without the lifestyle lead-in design.

›Who funded SURMOUNT-3 and does it matter?

Eli Lilly funded the trial entirely, and several authors are Lilly employees or stockholders. Industry funding does not invalidate results, but the design choices (enrichment, placebo comparator, no active comparator arm) align with maximizing demonstrated efficacy rather than answering pragmatic clinical questions.

›Does SURMOUNT-3 show what happens after stopping tirzepatide?

No. The trial had no discontinuation phase. Evidence from related trials (SURMOUNT-4, STEP 1 extension) shows roughly two-thirds of lost weight returns within 12 months of stopping GLP-1 receptor agonist therapy. Whether the lifestyle foundation in SURMOUNT-3 would mitigate this regain remains unknown.

›How racially diverse was the SURMOUNT-3 population?

The trial enrolled 73% White and only 15% Black participants. Hispanic/Latino enrollment was not separately reported prominently. Given disparities in obesity prevalence and known population-level differences in drug metabolism, this limits confidence in applying results across racial and ethnic groups.

›Can SURMOUNT-3 tell us if lifestyle lead-in is better than starting tirzepatide immediately?

No. The trial lacked a comparator arm of tirzepatide without lead-in. Without this head-to-head comparison, clinicians cannot determine whether the 12-week ILI adds meaningful value over simply beginning tirzepatide at the first visit.

›Were gastrointestinal side effects lower in SURMOUNT-3 than other tirzepatide trials?

Yes. Nausea occurred in about 20% of participants versus 29% in SURMOUNT-1. This may reflect adaptation from the low-calorie lead-in diet rather than a genuine tolerability advantage of the sequential strategy, though the trial was not designed to distinguish between these explanations.

›What is the total weight loss from original baseline in SURMOUNT-3?

Combining the lead-in loss (approximately 6-7%) with the additional 21.1% from randomization, total weight reduction from original baseline reached approximately 26-27% in the tirzepatide group. This is among the largest reductions seen in any obesity pharmacotherapy trial to date.

References

1. Wadden TA, et al. Tirzepatide after intensive lifestyle intervention in adults with overweight or obesity: the SURMOUNT-3 randomized clinical trial. Nat Med. 2023;29(11):2909-2918. https://pubmed.ncbi.nlm.nih.gov/37840095/

2. U.S. Food and Drug Administration. Zepbound (tirzepatide) prescribing information. 2023. https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/217806s000lbl.pdf

3. Aronne LJ, et al. Continued treatment with tirzepatide for maintenance of weight reduction in adults with obesity: the SURMOUNT-4 randomized clinical trial. JAMA. 2024;331(1):38-48. https://pubmed.ncbi.nlm.nih.gov/38078870/

4. Jastreboff AM, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med. 2022;387(3):205-216. https://pubmed.ncbi.nlm.nih.gov/35658024/

5. Garvey WT, et al. Tirzepatide once weekly for the treatment of obesity in people with type 2 diabetes (SURMOUNT-2). Lancet. 2023;402(10402):613-626. https://pubmed.ncbi.nlm.nih.gov/37385275/

6. ICH E9(R1) addendum on estimands and sensitivity analysis in clinical trials. Ther Innov Regul Sci. 2021;55(3):482-492. https://pubmed.ncbi.nlm.nih.gov/33729538/