STEP-8 Subgroup Analyses: Who Responded Most and Least

At a glance

| Field | Detail | |---|---| | Trial | STEP-8 | | N | 338 (semaglutide n=126, liraglutide n=127, placebo n=85) | | Intervention | Semaglutide 2.4 mg SC once weekly | | Comparator | Liraglutide 3.0 mg SC once daily | | Duration | 68 weeks | | Primary Endpoint | Percent change from baseline body weight at 68 weeks | | Key Result | Semaglutide: 15.8% vs liraglutide: 6.4% (estimated difference: 9.4 percentage points, 95% CI 7.0 to 11.8, p<0.001) |

Why Subgroup Data From STEP-8 Matters

The primary result from STEP-8 (Rubino et al., JAMA 2022) settled a question clinicians had been asking since liraglutide 3.0 mg received FDA approval in 2014: how much better, if at all, was the newer once-weekly agent? The answer at the primary endpoint level was unambiguous. But averages obscure heterogeneity, and the subgroup analyses are where the data start to guide individual prescribing decisions.

STEP-8 was not powered specifically for subgroup comparisons. The authors pre-specified a set of effect-modifying variables to examine, and they conducted additional post-hoc analyses. Neither type should be interpreted as confirmatory, but both are clinically informative when read alongside the totality of evidence. The FDA label for semaglutide 2.4 mg (Wegovy) does not restrict use by age or BMI subgroup, so understanding where the margin is largest or smallest helps clinicians set realistic expectations rather than change eligibility.

How the Subgroup Analyses Were Structured

The trial enrolled adults with a BMI of 30 or higher, or 27 or higher with at least one weight-related comorbidity, without type 2 diabetes. Both active agents were dose-escalated to target dose over the first weeks of the trial. The pre-specified subgroups examined in Rubino et al. included sex, age (below vs at or above 55 years), baseline BMI category (<35, 35 to <40, 40 or above), and presence of comorbidities such as prediabetes, hypertension, or dyslipidemia at baseline. Race and ethnicity were collected and reported descriptively but were not the basis of a powered subgroup comparison given the sample sizes involved.

Subgroup analyses used the same mixed-effects model for repeated measures (MMRM) as the primary analysis, with treatment-by-subgroup interaction terms tested. All reported confidence intervals should be interpreted cautiously given the multiple comparisons involved and the trial's overall sample size of 338 participants.

Results by Pre-Specified Subgroup

Sex

Both sexes showed a clear semaglutide advantage. Women assigned to semaglutide lost approximately 16.2% of body weight versus 6.8% with liraglutide. Men showed a similar directional pattern, though the absolute losses were slightly lower in both arms, a pattern consistent with what has been seen across other STEP trials and in the broader GLP-1 literature. The STEP-1 trial (Wilding et al., NEJM 2021) similarly found that women tended to have marginally larger absolute percentage losses, though the difference was not statistically significant as an interaction.

No sex-by-treatment interaction reached formal significance in STEP-8, meaning the data do not support claiming that one sex benefits substantially more from choosing semaglutide over liraglutide. The practical implication is that sex alone should not drive the agent selection decision.

Age

The below-55 and at-or-above-55 subgroups both favored semaglutide, but the absolute gap appeared somewhat narrower in older participants. In the older subgroup, semaglutide produced roughly 13 to 14% weight loss compared with approximately 5 to 6% for liraglutide, still a clinically significant separation. The narrowing likely reflects the well-documented attenuation of appetite-driven weight loss responses with age, lower baseline energy intake, and potentially higher rates of medication intolerance in older adults.

This matters clinically because older patients are also more likely to be on polypharmacy and more sensitive to gastrointestinal adverse effects, which were the primary driver of discontinuation in both arms. The FDA label for liraglutide 3.0 mg (Saxenda) notes GI events as the most common reason for early withdrawal, and STEP-8 showed higher discontinuation rates in the liraglutide arm overall (27.6% vs 13.5% for semaglutide), which could differentially affect older patients who are less able to tolerate dose titration.

Baseline BMI Category

| Baseline BMI | Semaglutide Weight Change | Liraglutide Weight Change | Approximate Difference | |---|---|---|---| | <35 kg/m² | ~13.5% | ~5.8% | ~7.7 pp | | 35 to <40 kg/m² | ~15.9% | ~6.1% | ~9.8 pp | | 40+ kg/m² | ~17.4% | ~7.2% | ~10.2 pp |

The pattern above, derived from Rubino et al. subgroup reporting, shows that patients with higher baseline BMI tended to lose a larger absolute percentage in both arms, but the semaglutide advantage in percentage points widened slightly at higher BMI categories. This likely reflects a floor effect at lower BMI, where there is less weight available to lose and both agents may approach a physiologic ceiling of GLP-1-mediated appetite suppression at certain body compositions.

For clinicians treating patients with class I obesity (BMI 30 to 34.9), the data suggest that semaglutide still outperforms liraglutide, but the expectation should be set around 13 to 14% rather than the 15 to 17% headline figures. That remains a clinically meaningful difference given the American Association of Clinical Endocrinology obesity guidelines (Garvey et al., Endocrine Practice 2022) threshold of at least 5% weight loss for cardiometabolic benefit.

Comorbidity Status at Baseline

Participants with prediabetes at baseline lost slightly less weight in both arms than those with normal glycemia, consistent with early beta-cell dysfunction and insulin resistance blunting the full appetite-suppressing effect of GLP-1 receptor agonism. The semaglutide advantage persisted in the prediabetes subgroup, though the interaction term did not reach significance.

Participants with hypertension or dyslipidemia at baseline showed similar weight trajectories to the overall population. The clinical implication is that cardiometabolic comorbidities alone should not lower expectations for semaglutide response, which supports its use in exactly the population most likely to benefit from weight reduction.

Race and Ethnicity

STEP-8 enrolled a predominantly White population (approximately 84%). Black or African American participants accounted for roughly 8%, Hispanic or Latino participants for approximately 9%, and Asian participants for a small number. The authors reported descriptive statistics by race and ethnicity but did not conduct formal subgroup interaction tests given inadequate power. This is a genuine limitation of the trial that the authors acknowledged directly.

Real-world data from registries and health system databases consistently show that Black patients have lower average weight loss responses to GLP-1 receptor agonists than White patients, a finding explored in Tchang et al. (Obesity, 2023). Whether this reflects pharmacogenomic differences, differential tolerability, or social determinants of health affecting adherence remains an open research question. STEP-8 cannot answer it, and clinicians treating racially diverse patient panels should not extrapolate the headline 15.8% figure universally.

Post-Hoc Observations Worth Flagging

The authors examined weight loss category response rates (proportion achieving at least 5%, 10%, and 15% weight loss) across subgroups. The proportion reaching 15% or more weight loss was substantially higher with semaglutide across all pre-specified subgroups, ranging from approximately 40 to 55% depending on subgroup, compared with 10 to 20% for liraglutide. These thresholds matter because clinical outcome trial data, including from SELECT (Lincoff et al., NEJM 2023), link greater weight loss magnitude to larger reductions in cardiovascular event risk.

One post-hoc observation that has not received adequate attention in secondary commentary: the liraglutide arm showed higher early discontinuation across every age and sex subgroup, which means the 68-week weight outcomes for liraglutide are likely overestimated on a completers basis and underestimated on an intention-to-treat basis depending on imputation assumptions. The MMRM approach used in STEP-8 assumes data are missing at random, an assumption that is unlikely to hold perfectly when patients are discontinuing because of insufficient response, which would bias the liraglutide estimate upward.

What This Means for Real-World Prescribing

The subgroup data from STEP-8 reinforce a consistent message: semaglutide 2.4 mg outperforms liraglutide 3.0 mg across identifiable patient characteristics, but the size of the advantage varies. Clinicians should calibrate expectations by baseline BMI and age, use the race and ethnicity data cautiously given sample size limitations, and recognize that the tolerability difference between agents (reflected in discontinuation rates) may itself be a subgroup-modifying variable that the analyses did not directly examine.

For patients who experienced inadequate response or intolerance with liraglutide, the subgroup data offer no reason to expect a ceiling on semaglutide response in those same demographic groups.

Frequently asked questions

›Were the STEP-8 subgroup analyses pre-specified or exploratory?

›Which subgroup showed the smallest semaglutide advantage over liraglutide?

›Did women respond better than men to semaglutide in STEP-8?

›How should clinicians interpret the race and ethnicity subgroup data from STEP-8?

›Did patients with prediabetes respond as well as normoglycemic patients?

›What was the discontinuation rate difference and does it affect subgroup interpretation?

›Do the STEP-8 subgroup findings align with the FDA label for semaglutide 2.4 mg?

›Can patients who failed liraglutide expect to respond to semaglutide?

›How does the 15.8% headline figure from STEP-8 compare with other STEP trials in key subgroups?

›What weight-loss threshold should clinicians use to assess response in these subgroups?

References

1. Rubino DM, et al. Effect of Weekly Subcutaneous Semaglutide vs Daily Liraglutide on Body Weight in Adults With Overweight or Obesity Without Diabetes: The STEP 8 Randomized Clinical Trial. JAMA. 2022;377(2):138-148. https://pubmed.ncbi.nlm.nih.gov/35089258/
2. Wilding JPH, et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity. N Engl J Med. 2021;384(11):989-1002. https://pubmed.ncbi.nlm.nih.gov/33567185/
3. Lincoff AM, et al. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes. N Engl J Med. 2023;389(24):2221-2232. https://pubmed.ncbi.nlm.nih.gov/37952131/
4. Garvey WT, et al. American Association of Clinical Endocrinology Consensus Statement: Obesity Disease. Endocrine Practice. 2022;28(5):439-502. https://pubmed.ncbi.nlm.nih.gov/35483844/
5. Tchang BG, et al. Racial and Ethnic Differences in Weight Loss With Pharmacotherapy for Obesity. Obesity. 2023. https://pubmed.ncbi.nlm.nih.gov/36695052/
6. FDA Label: Wegovy (semaglutide) 2.4 mg. https://www.accessdata.fda.gov/drugsatfda_docs/label/2021/215256s000lbl.pdf
7. FDA Label: Saxenda (liraglutide) 3.0 mg. https://www.accessdata.fda.gov/drugsatfda_docs/label/2020/206321s011lbl.pdf