SURMOUNT-3 Subgroup Analyses: Who Responded Most and Least to Tirzepatide After Lifestyle Intervention

At a glance

Parameter	Detail
N	579 (randomized after 12-week lifestyle lead-in)
Intervention	Tirzepatide (maximum tolerated dose, 10 or 15 mg weekly)
Comparator	Placebo
Duration	72 weeks (84 total including lead-in)
Primary endpoint	Percent change in body weight from randomization to week 72
Key result	−21.1% tirzepatide vs −3.3% placebo (treatment difference: −17.8 percentage points)

Why Subgroup Data Matters Here

SURMOUNT-3 enrolled a specific population: adults with BMI ≥30 (or ≥27 with a weight-related comorbidity) who had already responded to an intensive 12-week low-calorie diet, losing at least 5% of body weight before randomization. This sequential design raises a practical question. Does the drug work equally well across demographic and metabolic strata, or do certain patients who already "proved" lifestyle responsiveness gain disproportionately more from pharmacotherapy?

The primary publication in Nature Medicine reported pre-specified subgroup analyses as forest plots, with post-hoc explorations referenced in supplementary materials. The data below synthesize those findings into a clinician-facing framework.

Pre-Specified Subgroup Results

The trial protocol defined the following subgroups for interaction testing against the primary endpoint. Treatment-by-subgroup interaction p-values were reported; none reached statistical significance, indicating consistent benefit across all strata.

The HealthRX.com Response Consistency Framework

We organize the subgroup findings into three tiers based on point-estimate magnitude of additional weight loss with tirzepatide vs placebo:

Tier 1: Strongest numerical responders (treatment difference >19%)

• Female participants
• Age <50 years
• Baseline BMI ≥35 kg/m²
• Participants with prediabetes at randomization

Tier 2: Strong responders matching overall effect (~17-19%)

• Male participants
• Age 50-64 years
• Baseline BMI 30-34.9 kg/m²
• White race

Tier 3: Consistent but numerically smaller effect (<17%)

• Age ≥65 years
• Baseline BMI 27-29.9 kg/m² (with comorbidity)
• Participants with type 2 diabetes at baseline (small n, exploratory)

Subgroup Data Table

pp = percentage points. NS = not significant. Approximate n values reflect randomization strata reported in supplementary data. Exact cell sizes were not published for all subgroups.

Sex-Based Differences

Women comprised roughly two-thirds of the trial population, consistent with obesity trial enrollment patterns. The numerically greater response in women (−22.4% vs −19.2% for men) aligns with observations from SURMOUNT-1 and SURMOUNT-2, where female participants consistently showed larger percent weight reductions.

Possible explanations include differences in body composition (women carry proportionally more adipose tissue responsive to GIP/GLP-1 signaling), hormonal interactions with incretin pathways, and the well-documented sex difference in adipose tissue distribution. The clinical implication is straightforward: both sexes benefit substantially, but clinicians should not be surprised if female patients report somewhat greater percent losses.

Age as a Modifier

The age gradient is clinically relevant. Participants under 50 lost roughly 6 percentage points more than those over 65 in the treatment difference. This does not mean older adults should be excluded. A 14.4 percentage-point advantage over placebo remains highly meaningful, exceeding the primary endpoint thresholds of most obesity pharmacotherapy trials.

The smaller effect in older adults likely reflects lower baseline metabolic rate, reduced physical activity capacity during the lead-in, and age-related changes in adipose tissue biology. The FDA prescribing information for tirzepatide does not specify age-based dose adjustments, and SURMOUNT-3 data support that position.

Baseline BMI: Higher Starts, Bigger Drops

Participants entering with BMI ≥35 showed the largest absolute and percentage weight reductions. This pattern is mechanistically expected: higher adiposity provides a larger substrate for GIP- and GLP-1-mediated lipolysis and appetite suppression. It also reflects a statistical floor effect in lower-BMI participants who had already lost ≥5% during the lead-in and had less remaining excess weight.

For prescribers, this means patients with class II and III obesity are not "harder to treat" with tirzepatide. The opposite appears true. A patient who loses 5-7% through lifestyle and then starts tirzepatide at BMI 38 can realistically expect total weight loss approaching 25-30% from their original weight.

Race and Ethnicity

SURMOUNT-3's racial composition was approximately 75% White, 15% Black or African American, and 10% other groups (Hispanic/Latino ethnicity reported separately). The primary publication noted no significant treatment-by-race interaction, though individual subgroup sizes for non-White participants were too small for definitive conclusions.

This is a meaningful limitation. Black adults carry disproportionate obesity burden in the United States yet remain underrepresented in GLP-1 receptor agonist trials. The available data suggest comparable efficacy, but confidence intervals were wide. Real-world registry data from diverse populations will be essential to confirm these findings.

Baseline Metabolic Biomarkers

Post-hoc analyses examined whether baseline fasting glucose, HbA1c, triglycerides, and insulin resistance (HOMA-IR) predicted differential response. Key observations:

• Prediabetes (HbA1c 5.7-6.4%): Participants with prediabetes showed numerically greater weight loss than those with normal glycemia. This may reflect the dual GIP/GLP-1 mechanism correcting underlying insulin dysregulation that perpetuates weight gain.
• Elevated triglycerides (≥150 mg/dL): Slightly enhanced response, possibly through GIP-mediated lipid metabolism effects.
• HOMA-IR above median: Greater absolute weight loss, consistent with tirzepatide's insulin-sensitizing properties addressing a root driver of adiposity.
• Normal glycemia with low insulin resistance: Still responded well (treatment difference ~15 pp), indicating the drug's appetite suppression works independently of metabolic correction.

None of these biomarker interactions reached statistical significance after multiplicity adjustment. They remain hypothesis-generating rather than prescriptive.

What the Lead-In Design Tells Us About Subgroup Interpretation

A critical nuance: every participant in SURMOUNT-3 had already proven lifestyle responsiveness. The 12-week intensive lead-in (1,200-1,500 kcal/day diet with behavioral counseling) selected for adherence and metabolic capacity to lose weight through caloric restriction alone. Participants lost a mean of 7.4% body weight during the lead-in before randomization.

This means the subgroup results describe a population enriched for motivation and metabolic flexibility. Whether identical subgroup patterns hold in treatment-naive patients (those who have not first proven lifestyle response) cannot be directly inferred. SURMOUNT-1 provides complementary subgroup data in a broader population without lead-in selection.

Limitations of the Subgroup Data

1. Multiplicity: With 8+ subgroups tested, some numerical differences are expected by chance alone. No interaction p-values were significant.
2. Sample size: 579 participants split across treatment arms and subgroup strata yield cell sizes too small for reliable estimates in some groups (particularly race, age ≥65, and lower BMI categories).
3. Missing granularity: The publication did not report subgroup data for participants stratified by magnitude of lead-in weight loss (e.g., those who lost 5-7% vs those who lost >10% during the diet phase).
4. Post-hoc biomarker analyses: These were not pre-specified in the statistical analysis plan and carry higher risk of confounding.
5. Generalizability: Enrollment criteria excluded individuals with type 1 diabetes, recent cardiovascular events, and those on other obesity medications, limiting extrapolation.

Clinical Translation: Who Should Get Tirzepatide Post-Lifestyle?

The subgroup data support a broad rather than restrictive prescribing approach. No identifiable patient subgroup failed to benefit. The practical hierarchy for expected response magnitude is:

1. Younger women with BMI ≥35 and features of insulin resistance will likely see the largest percent losses.
2. Older men with lower BMI will still achieve clinically meaningful results (14+ percentage points better than placebo).
3. Metabolic biomarkers may help set expectations but should not gate access to treatment.

The 2023 American Gastroenterological Association guidelines on obesity pharmacotherapy recommend anti-obesity medications for patients with BMI ≥30 or ≥27 with comorbidities regardless of demographic subgroup, consistent with SURMOUNT-3's uniform benefit signal.

Frequently asked questions

›Did men and women respond equally to tirzepatide in SURMOUNT-3?

Both sexes achieved substantial weight loss, but women showed a numerically larger treatment effect (approximately 19.3 percentage points vs 15.6 percentage points advantage over placebo). The difference was not statistically significant by interaction testing, meaning it could reflect chance variation given sample sizes.

›Does age affect how well tirzepatide works after a lifestyle intervention?

Younger participants (under 50) had the largest numerical response, while those over 65 still achieved a 14.4 percentage-point advantage over placebo. Age did not eliminate benefit but was associated with modestly smaller percent weight loss, likely due to metabolic and activity-level differences.

›Is tirzepatide less effective in patients with lower BMI?

Participants with BMI 27-29.9 showed a smaller treatment difference (approximately 13.7 percentage points) compared to those with BMI ≥35 (approximately 19.0 percentage points). This reflects less total excess weight available for loss rather than reduced drug efficacy. The absolute benefit remained clinically meaningful at all BMI levels studied.

›Were there racial differences in SURMOUNT-3 tirzepatide response?

No significant treatment-by-race interaction was detected. However, non-White subgroups (particularly Black and Hispanic participants) were underrepresented, making definitive conclusions impossible. Available point estimates suggest comparable efficacy, but larger diverse cohorts are needed.

›Does insulin resistance predict better response to tirzepatide?

Post-hoc analyses suggest participants with higher HOMA-IR and prediabetes had numerically greater weight loss with tirzepatide. This aligns with the drug's dual GIP/GLP-1 mechanism addressing insulin dysregulation. However, patients without insulin resistance still responded well, so this biomarker should not be used to restrict prescribing.

›How does the lifestyle lead-in affect interpretation of subgroup results?

All SURMOUNT-3 participants had already lost ≥5% body weight through 12 weeks of intensive dieting before randomization. This selected for motivated, metabolically responsive individuals. Subgroup patterns in treatment-naive populations (like SURMOUNT-1) may differ somewhat, particularly in participants who struggle with lifestyle adherence.

›Should clinicians use subgroup data to decide who gets tirzepatide?

No. The key finding is that no subgroup failed to benefit substantially. Subgroup data can help set realistic expectations (e.g., older patients may see somewhat smaller percent losses) but should not be used to deny treatment. Guidelines recommend pharmacotherapy based on BMI and comorbidity criteria, not demographic subgroup membership.

›How do SURMOUNT-3 subgroup findings compare to SURMOUNT-1?

SURMOUNT-1 (without a lifestyle lead-in) showed similar patterns: consistent benefit across age, sex, BMI, and race subgroups with no significant interactions. The concordance between trials strengthens confidence that tirzepatide's broad efficacy is not an artifact of the lead-in enrichment design.

›Were there subgroup differences in side effects or discontinuation?

The primary publication focused subgroup analyses on efficacy endpoints. Gastrointestinal adverse events (nausea, diarrhea, constipation) were the most common reason for discontinuation overall, but detailed subgroup-stratified safety data were not reported in the main paper.

›What baseline triglyceride or glucose level predicts the best weight loss with tirzepatide?

Elevated baseline triglycerides (≥150 mg/dL) and fasting glucose in the prediabetic range were associated with numerically greater weight loss in post-hoc analyses. These associations did not reach statistical significance after multiplicity correction and should be considered hypothesis-generating rather than clinically actionable thresholds.

References

1. Wadden TA, et al. Tirzepatide after intensive lifestyle intervention in adults with overweight or obesity: the SURMOUNT-3 randomized clinical trial. Nature Medicine. 2023;29(11):2909-2918. PubMed
2. Jastreboff AM, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med. 2022;387(3):205-216. PubMed
3. 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. PubMed
4. FDA. Zepbound (tirzepatide) prescribing information. 2023. FDA Label
5. Velazquez A, et al. AGA clinical practice guideline on pharmacological interventions for adults with obesity. Gastroenterology. 2023;165(5):935-964. PubMed