SURMOUNT-1 Subgroup Analyses: Who Responded Most and Least to Tirzepatide

At a glance

| Parameter | Detail | |-----------|--------| | N | 2,539 (randomized 1:1:1:1) | | Intervention | Tirzepatide 5 mg, 10 mg, or 15 mg SC weekly | | Comparator | Placebo SC weekly | | Duration | 72 weeks | | Primary endpoint | Percent change in body weight from baseline | | Key result | −15.0% (5 mg), −19.5% (10 mg), −20.9% (15 mg) vs −3.1% placebo | | Population | Adults with BMI ≥30, or ≥27 with ≥1 weight-related comorbidity, without diabetes |

Why Subgroup Data Matters for This Trial

The headline numbers from SURMOUNT-1 are extraordinary: a mean 20.9% body weight reduction at the highest dose. But averages conceal variation. Clinicians selecting patients for tirzepatide, payers setting coverage criteria, and patients setting expectations all need granular data. The pre-specified subgroup analyses in the primary publication and its supplementary appendix provide that granularity.

SURMOUNT-1 was powered for overall efficacy, not formal subgroup comparisons, so no interaction p-values crossed correction thresholds for superiority claims. Still, effect-size patterns across strata tell a clinical story worth reading closely.

Pre-Specified Subgroup Definitions

The trial protocol and supplementary appendix defined the following subgroup factors for the primary estimand (treatment-policy estimator):

• Sex (male vs female)
• Age (<50 vs ≥50 years; also <65 vs ≥65)
• Race (White, Black/African American, Asian, other)
• Ethnicity (Hispanic/Latino vs not)
• Baseline BMI (<35 vs ≥35 kg/m²)
• Baseline waist circumference (sex-specific cut-offs)
• Baseline prediabetes status (yes vs no)
• Region (North America vs rest of world)

Results by Subgroup: The HealthRX.com Response-Differential Framework

Below we organize findings into three tiers: factors that showed a clear response gradient, factors with modest directional trends, and factors showing no meaningful differential.

Tier 1: Clear Response Gradient

Sex

Women assigned to tirzepatide 15 mg lost approximately 22.0% of body weight versus roughly 17.5% in men at the same dose. This differential persisted across all three dose levels. The gap is consistent with prior GLP-1 receptor agonist trials (semaglutide 2.4 mg STEP program data) and likely reflects differences in body composition (higher baseline fat mass percentage in women) rather than differential drug pharmacology. Women also started with a somewhat higher mean BMI in this cohort, yet lost a greater relative percentage.

Baseline BMI Stratum

Participants entering the trial with BMI <35 kg/m² achieved greater relative weight loss (percentage terms) than those with BMI ≥35. At the 15 mg dose, the <35 subgroup mean was approximately −22.5% versus −19.8% in the ≥35 stratum. This pattern recapitulates findings from liraglutide and semaglutide trials: patients closer to the obesity threshold tend to lose a higher fraction of body weight. In absolute kilograms, the heavier subgroup lost more, but percentage loss (the clinically reported metric and the primary endpoint) favored those with lower baseline BMI.

| Subgroup | Tirzepatide 15 mg (% weight change) | Placebo (% weight change) | Treatment difference | |----------|--------------------------------------|---------------------------|---------------------| | BMI <35 | −22.5% | −3.4% | −19.1 pp | | BMI ≥35 | −19.8% | −2.8% | −17.0 pp | | Female | −22.0% | −3.5% | −18.5 pp | | Male | −17.5% | −2.5% | −15.0 pp |

Tier 2: Modest Directional Trends

Age

Participants <50 years showed a numerically greater weight reduction than those ≥50. The difference was small (roughly 1 to 2 percentage points at 15 mg). Among the smaller subgroup aged ≥65, point estimates remained clinically significant but confidence intervals widened considerably given limited sample size (n ≈ 80 in that stratum). These findings align with the FDA prescribing information for Zepbound, which does not recommend dose adjustment for age.

Baseline Prediabetes

Participants without prediabetes lost slightly more weight than those with prediabetes at all three doses. At 15 mg, those without prediabetes lost approximately −21.5% versus −19.5% in the prediabetes subgroup. The hypothesis: insulin resistance and hyperinsulinemia may blunt incretin-mediated appetite suppression to a small degree. This observation is consistent with the larger differential seen in the SURMOUNT-2 trial, which enrolled participants with type 2 diabetes and reported mean weight loss of 14.7% at 15 mg (lower than the 20.9% in SURMOUNT-1's non-diabetic cohort).

Race and Ethnicity

The trial enrolled predominantly White participants (approximately 70%). Black participants (approximately 14%) showed numerically similar weight loss to White participants at 15 mg. Asian participants (small n) showed point estimates that were directionally higher but with wide confidence intervals. Hispanic/Latino participants responded similarly to non-Hispanic participants. No interaction was statistically significant. Limited racial diversity constrains interpretation.

Tier 3: No Meaningful Differential

Region (North America vs Rest of World)

Response was nearly identical across geographic regions at all doses. This argues against significant lifestyle or dietary confounders driving differential responses.

Baseline Waist Circumference

After accounting for BMI stratum, waist circumference added minimal explanatory value for weight-loss response.

Post-Hoc and Exploratory Analyses

Several post-hoc analyses from the SURMOUNT-1 dataset have since been published:

Categorical weight loss thresholds by subgroup. At 15 mg, 56.7% of the overall population achieved ≥20% weight loss (primary publication, Figure 2). Broken down by sex, a higher proportion of women crossed the 20% threshold than men. Among those with BMI <35, nearly two-thirds reached this milestone.

Cardiometabolic biomarker improvement by baseline risk. Participants with elevated baseline triglycerides, waist circumference, or blood pressure saw greater absolute improvement in those specific parameters, even though percentage weight loss was similar or slightly lower. This suggests tirzepatide's metabolic benefits are partly independent of weight trajectory.

Body composition (DXA substudy). A subset (n = 255) underwent DXA scanning. Approximately 33% of total weight lost was lean mass across all subgroups, consistent with the expected ratio during caloric restriction. No subgroup lost a disproportionate fraction of lean mass, though this substudy was underpowered for formal subgroup comparisons.

Clinical Translation: What This Means for Prescribing

The subgroup data supports several practical observations for clinicians:

1. Women will likely see larger percentage reductions than men at equivalent doses. Setting expectations accordingly prevents premature discontinuation in male patients who achieve a "mere" 15% loss.

2. Patients with lower baseline BMI get closer to normal weight. A patient starting at BMI 32 who loses 22% of body weight may reach BMI 25. A patient starting at BMI 45 who loses 19% remains at BMI 36. Dose escalation discussions should account for starting weight.

3. Prediabetes or insulin resistance does not preclude strong response. The attenuation is modest (1 to 2 percentage points). Clinicians should not withhold tirzepatide from patients with prediabetes based on expected reduced efficacy.

4. Age alone does not warrant dose adjustment. The Zepbound label confirms no geriatric dose modification is needed. The slight numerical attenuation in older adults may partly reflect higher dropout or greater tolerability-driven dose reductions.

5. Racial data are insufficient to draw differential prescribing conclusions. The trial underrepresented Black, Asian, and Hispanic populations relative to obesity prevalence in those communities. SURMOUNT-3 and SURMOUNT-4 provided some additional diversity, but dedicated studies in underrepresented populations are still warranted.

Limitations of the Subgroup Data

The investigators and subsequent commentators have acknowledged several constraints:

• No multiplicity correction. Subgroup analyses were not adjusted for multiple comparisons. All findings are hypothesis-generating.
• Unequal subgroup sizes. The ≥65 age group, Asian participants, and several other strata had fewer than 100 participants, limiting statistical power.
• 72-week snapshot. Subgroup-specific durability beyond 72 weeks requires data from extension studies (SURMOUNT-3, SURMOUNT-4).
• Missing granularity. The publication does not report interaction terms for combined strata (e.g., men over 65 with prediabetes). Such intersectional analyses would require individual patient data access.
• Dropout differential. Discontinuation rates were 14.3% (5 mg), 16.4% (10 mg), 15.1% (15 mg), and 26.4% (placebo). Subgroup-level dropout rates were not fully reported, making per-protocol estimates in smaller strata less reliable.

Comparison to Other GLP-1 Subgroup Data

The patterns observed in SURMOUNT-1 are largely congruent with STEP 1 semaglutide 2.4 mg data, where women outperformed men and lower baseline BMI predicted greater percentage loss. The magnitude of tirzepatide's effect is larger at every subgroup level, reinforcing that the dual GIP/GLP-1 mechanism provides additional efficacy without altering the subgroup response hierarchy.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Jastreboff AM, Aronne LJ, Ahmad NN, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med. 2022;387(3):205-216. PubMed
2. Wilding JPH, Batterham RL, Calanna S, et al. Once-weekly semaglutide in adults with overweight or obesity (STEP 1). N Engl J Med. 2021;384(11):989-1002. PubMed
3. Garvey WT, Frias JP, Jastreboff AM, 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. Zepbound (tirzepatide) prescribing information. U.S. Food and Drug Administration. 2023. FDA Label
5. Wadden TA, Hollander P, Klein S, et al. Weight maintenance and additional weight loss with tirzepatide (SURMOUNT-4). JAMA. 2024;331(1):38-48. PubMed