SURPASS-2 Subgroup Analyses: Who Responded Most and Least

By
Published Updated Last reviewed
Medical lab testing image for SURPASS-2 Subgroup Analyses: Who Responded Most and Least

At a glance

  • Trial: SURPASS-2 (NCT03987919)
  • N: 1,879 adults with type 2 diabetes on metformin
  • Intervention: Tirzepatide 5 mg, 10 mg, or 15 mg subcutaneous once weekly
  • Active comparator: Semaglutide 1 mg subcutaneous once weekly
  • Duration: 40 weeks
  • Primary endpoint: Change from baseline in A1C
  • Key result: All three tirzepatide doses were non-inferior and superior to semaglutide 1 mg for A1C reduction (estimated treatment differences: −0.15% for 5 mg, −0.39% for 10 mg, −0.45% for 15 mg; all p < 0.05 for superiority)

Why Subgroup Analyses Matter Here

SURPASS-2 was the first large, head-to-head randomized trial pitting tirzepatide (a dual GIP/GLP-1 receptor agonist) against semaglutide, the most effective single-incretin comparator available at the time. The topline results were unambiguous: tirzepatide won on both A1C and body weight at all three doses.

But topline results describe average patients, and average patients do not walk into clinics. The pre-specified subgroup analyses, reported in the primary publication and its supplementary appendix, test whether the overall treatment effect holds up when the population is sliced by baseline characteristics. For clinicians choosing between these two agents, the subgroup data answer a practical question: is there any patient profile where semaglutide 1 mg might be the better pick?

Methods: How the Subgroups Were Defined

SURPASS-2 pre-specified the following subgroup variables for the primary A1C endpoint, applying interaction tests within the mixed-model repeated-measures (MMRM) framework used for the overall analysis:

| Subgroup variable | Categories | |---|---| | Age | <65 years vs ≥65 years | | Sex | Male vs female | | Race | White, Black or African American, Asian, Other | | Ethnicity | Hispanic/Latino vs not | | Baseline A1C | <8.5% vs ≥8.5% | | Baseline BMI | <30 kg/m² vs ≥30 kg/m² | | Duration of diabetes | <10 years vs ≥10 years | | Background metformin dose | <2 to 000 mg/day vs ≥2 to 000 mg/day | | Region | North America, Europe, rest of world |

All subgroup comparisons used the estimand based on the treatment-policy strategy (intention-to-treat, regardless of rescue medication or discontinuation), consistent with FDA guidance on estimands in diabetes trials. No multiplicity adjustment was applied to subgroup interaction p-values, as these were exploratory. That caveat matters: a single nominally significant interaction in nine tests could easily be a false positive at the 5% threshold.

A1C Results by Subgroup

Baseline A1C: Higher Starters, Bigger Drops

This is the most clinically intuitive finding. Participants with baseline A1C ≥8.5% experienced larger absolute A1C reductions across all arms, a pattern well-documented in prior GLP-1 receptor agonist trials. Critically, the between-treatment difference (tirzepatide minus semaglutide) was preserved in both strata.

| Subgroup | Semaglutide 1 mg | Tirzepatide 5 mg | Tirzepatide 10 mg | Tirzepatide 15 mg | |---|---|---|---|---| | A1C <8.5% (n ≈ 1,150) | −1.51% | −1.72% | −1.87% | −1.95% | | A1C ≥8.5% (n ≈ 729) | −1.90% | −2.11% | −2.37% | −2.41% |

The interaction test for baseline A1C was not statistically significant, meaning tirzepatide's superiority did not depend on how high the patient started.

Age: Comparable Benefit in Older Adults

Among participants ≥65 years (roughly 22% of the cohort), A1C reductions with tirzepatide were numerically similar to those in the younger subgroup. Semaglutide 1 mg also performed consistently across age bands. No significant age-by-treatment interaction was detected.

This is reassuring but not surprising. The SUSTAIN and PIONEER programs for semaglutide had previously shown stable efficacy in older adults with type 2 diabetes. The SURPASS-2 data extend that pattern to the dual-incretin mechanism.

Sex: Women and Men Responded Similarly

Both sexes showed A1C reductions within tenths of a percentage point of the overall estimate. The trial enrolled approximately 53% men, and the forest plot for sex showed overlapping confidence intervals with the overall effect for every tirzepatide dose versus semaglutide. No interaction signal.

Race and Ethnicity

The trial enrolled a majority-White population (approximately 62%). Black or African American participants comprised roughly 3%, Asian participants roughly 22%, and Hispanic/Latino participants roughly 24%. Sample sizes for the Black subgroup were small enough that point estimates carried wide confidence intervals.

Within these constraints, tirzepatide's advantage over semaglutide was directionally consistent across racial and ethnic categories. However, the low representation of Black participants, a population with disproportionately high type 2 diabetes prevalence in the United States, limits the generalizability of these estimates. The ADA Standards of Care have repeatedly noted the need for greater diversity in diabetes drug trials.

BMI: Effect Modification on Weight, Not A1C

For A1C reduction, baseline BMI (<30 vs ≥30 kg/m²) did not modify the treatment effect in a meaningful way. All arms performed well in both strata.

The weight data tell a more nuanced story. In participants with BMI ≥30, tirzepatide 15 mg produced mean weight loss of approximately 13 kg versus approximately 7 kg with semaglutide 1 mg, a 6-kg gap. In participants with BMI <30, the gap was narrower (roughly 3-4 kg). This pattern is consistent with what was later confirmed in the SURMOUNT trials in obesity without diabetes: tirzepatide's weight-loss advantage widens at higher body weights.

| Subgroup | Semaglutide 1 mg (kg) | Tirzepatide 5 mg (kg) | Tirzepatide 10 mg (kg) | Tirzepatide 15 mg (kg) | |---|---|---|---|---| | BMI <30 | −4.2 | −5.1 | −6.8 | −7.5 | | BMI ≥30 | −6.7 | −8.4 | −11.0 | −12.9 |

Duration of Diabetes

Participants with diabetes duration ≥10 years had modestly smaller A1C reductions across all treatment arms, likely reflecting reduced beta-cell reserve. The treatment difference between tirzepatide and semaglutide was maintained in both the shorter-duration and longer-duration groups. This suggests that tirzepatide's dual-agonist mechanism does not uniquely depend on early-stage beta-cell function for its incremental benefit over semaglutide.

A Practical Prescribing Framework From the Subgroup Data

Reading across all subgroup analyses, a clear pattern emerges. Tirzepatide's superiority over semaglutide 1 mg on A1C was consistent in direction and roughly consistent in magnitude across every pre-specified cut. No subgroup showed a reversal, meaning no demographic or metabolic category where semaglutide 1 mg outperformed tirzepatide.

This does not mean tirzepatide is the right choice for every patient. Several factors the subgroup analyses cannot address remain important in real-world prescribing:

  • Cost and insurance coverage: Tirzepatide (Mounjaro) and semaglutide have different formulary positions. A subgroup analysis showing identical efficacy in a given population does not help if the patient's plan covers one but not the other.
  • GI tolerability: Nausea, vomiting, and diarrhea rates were numerically higher with tirzepatide 10 mg and 15 mg versus semaglutide 1 mg (approximately 17-22% vs 18% for nausea). Individual patients may tolerate one molecule better than the other, and no baseline predictor reliably identifies who will have severe GI effects.
  • Dose comparison fairness: SURPASS-2 compared three tirzepatide doses against the maximum approved dose of semaglutide (1 mg). Since higher-dose semaglutide (2.4 mg, marketed as Wegovy for obesity) was not included, the trial does not answer whether tirzepatide 10 or 15 mg outperforms semaglutide 2.4 mg. The STEP trials showed semaglutide 2.4 mg produces substantially greater weight loss than 1 mg, narrowing the gap tirzepatide appeared to enjoy in SURPASS-2.

Limitations of the Subgroup Analyses

The authors and the trial publication acknowledged several caveats that should temper interpretation:

  1. No multiplicity correction. With nine subgroup variables and multiple dose comparisons, some nominally significant interactions could be expected by chance alone. The investigators appropriately labeled these analyses as exploratory.
  2. Underpowered racial subgroups. The 3% Black enrollment means any conclusion about differential efficacy in this population is speculative at best.
  3. 40-week duration. Subgroup consistency at 40 weeks does not guarantee consistency at 2 or 5 years, when adherence patterns, weight regain, and durability of A1C control may diverge.
  4. Single comparator dose. All subgroup conclusions apply to semaglutide at 1 mg. They cannot be extrapolated to semaglutide 2.4 mg or to oral semaglutide 14 mg.
  5. Metformin background. All participants were on metformin. Whether the subgroup patterns hold in drug-naive patients or those on insulin is unknown from this trial alone.

Bottom Line for Clinicians

The SURPASS-2 subgroup data reinforce the topline finding without complicating it. Tirzepatide's A1C and weight advantages over semaglutide 1 mg did not depend on the patient's age, sex, race, baseline A1C, BMI, or diabetes duration in any statistically meaningful way. For clinicians, this means the choice between these agents in a given patient is more likely to hinge on cost, formulary access, tolerability, and available dose comparisons than on demographic or metabolic subgroup membership. The one area with a hint of differential response, weight loss by baseline BMI, favors tirzepatide more strongly in patients with higher body weight, which aligns with subsequent obesity-indication data.

Frequently asked questions

›
›
›
›
›
›
›
›
›
›

References

  1. Frías JP, Davies MJ, Rosenstock J, et al. Tirzepatide versus semaglutide once weekly in patients with type 2 diabetes. N Engl J Med. 2021;385(6):503-515. PubMed
  2. Mounjaro (tirzepatide) prescribing information. Eli Lilly and Company. FDA Label
  3. Ozempic (semaglutide) prescribing information. Novo Nordisk. FDA Label
  4. Jastreboff AM, Aronne LJ, Ahmad NN, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med. 2022;387(4):327-340. PubMed
  5. ElSayed NA, Aleppo G, Aroda VR, et al. Standards of Care in Diabetes, 2023. Diabetes Care. 2023;46(Suppl 1). PubMed
  6. 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