PIONEER-6 Subgroup Analyses: Who Responded Most and Least

At a glance

Parameter	Detail
Trial	PIONEER-6 (NCT02692716)
N	3,183 (1,591 oral semaglutide, 1,592 placebo)
Intervention	Oral semaglutide 14 mg once daily
Comparator	Placebo (both added to standard of care)
Duration	Median 15.9 months
Primary endpoint	Time to first MACE (CV death, non-fatal MI, non-fatal stroke)
Key result	HR 0.79 (95% CI 0.57 to 1.11); non-inferiority confirmed (p < 0.001)

Why Subgroup Analyses Matter Here

PIONEER-6 was designed as a non-inferiority cardiovascular outcomes trial (CVOT), not a superiority study. With 137 primary MACE events across 3,183 patients and a median follow-up under 16 months, the trial was statistically powered to rule out excess cardiovascular harm, not to detect benefit in the overall population or in individual subgroups (Husain et al., NEJM 2019).

That matters for interpretation. Subgroup analyses in PIONEER-6 cannot prove that oral semaglutide reduces MACE in any single stratum. What they can do is identify consistency (or inconsistency) of the treatment effect across clinically relevant patient characteristics, information that directly shapes prescribing decisions.

How the Subgroup Analyses Were Structured

The PIONEER-6 investigators pre-specified subgroup analyses in the statistical analysis plan before unblinding. Subgroups were defined by:

• Age: <65 years vs. ≥65 years
• Sex: male vs. female
• Race: white vs. non-white (Black, Asian, other)
• Region: North America, Europe, rest of world
• Baseline BMI: <30 kg/m² vs. ≥30 kg/m²
• Baseline HbA1c: <8.0% vs. ≥8.0%
• Baseline eGFR: <60 vs. ≥60 mL/min/1.73 m²
• Established CVD at baseline: yes vs. no
• Duration of diabetes: <15 years vs. ≥15 years

Interaction p-values were calculated for each subgroup to test whether the treatment effect differed significantly between strata. All analyses used Cox proportional hazards models with treatment as the sole factor (Husain et al., NEJM 2019).

The HealthRX.com Subgroup Response Matrix

The table below consolidates point estimates and confidence intervals from the pre-specified forest plot. Because PIONEER-6 reported 137 total primary events, individual subgroup event counts are small, and confidence intervals are wide. We flag subgroups where the point estimate most clearly favored semaglutide (HR < 0.70) and where it was closest to neutral or trended toward harm (HR > 1.0).

Source: Forest plot, Supplementary Appendix, Husain et al., NEJM 2019. Approximate subgroup sizes reconstructed from published percentages. No interaction p-value reached statistical significance.

Who Responded Most

Three subgroups showed the most favorable point estimates for oral semaglutide on MACE:

Patients aged 65 and older. The HR of 0.68 in this stratum was the strongest signal in the age analysis. About 56% of trial participants were 65 or older, reflecting the enrichment strategy that required participants to have either established cardiovascular disease (age ≥50) or cardiovascular risk factors (age ≥60). Older patients carried a higher absolute event rate, which made even modest relative risk reductions more clinically meaningful. This pattern mirrors findings from the injectable semaglutide CVOT, SUSTAIN-6, where the treatment effect was also numerically larger in older participants (Marso et al., NEJM 2016).

Patients with longer diabetes duration (≥15 years). An HR of 0.70 is consistent with the hypothesis that GLP-1 receptor agonists may confer greater cardiovascular protection in patients with more advanced metabolic disease. Longer diabetes duration correlates with higher cumulative glycemic burden, more extensive vascular damage, and greater baseline risk. The ADA Standards of Care recommend GLP-1 RAs with proven CV benefit for patients with established atherosclerotic cardiovascular disease (ASCVD), a population that overlaps heavily with this subgroup.

Patients with higher baseline HbA1c (≥8.0%). The point estimate of 0.71 suggests that patients with worse glycemic control at entry may have derived more benefit. This could reflect a direct glucose-lowering effect on CV risk, or it could simply indicate higher baseline event rates enabling a larger observed effect. Oral semaglutide reduced HbA1c by approximately 1.0 percentage point more than placebo in the PIONEER program (FDA label, Rybelsus).

Who Responded Least

Patients under 65. The HR of 1.07 shows a point estimate on the wrong side of 1.0, though the confidence interval (0.60 to 1.90) is extremely wide and includes the overall trial estimate. Younger patients had fewer events overall, limiting statistical power. This does not indicate harm. It indicates insufficient data to draw any conclusion in this stratum.

Patients without established CVD. Only about 15% of PIONEER-6 participants lacked established cardiovascular disease at enrollment (they qualified via risk factors alone). With an HR of 1.12 (0.49 to 2.55) and very few events, the analysis is essentially uninformative. The SELECT trial of subcutaneous semaglutide 2.4 mg later demonstrated CV benefit in a broader population including patients without prior CV events but with overweight/obesity (Lincoff et al., NEJM 2023), though that trial used a different dose and formulation.

Shorter diabetes duration (<15 years). An HR of 0.92 is close to null. These patients had lower absolute event rates, which reduces power and compresses observed effects.

Interaction Testing: No Significant Heterogeneity

None of the pre-specified interaction tests reached statistical significance. This means the trial provided no evidence that the treatment effect differed across any subgroup pair. The practical implication: the overall HR of 0.79 is the best single estimate of the treatment effect regardless of patient characteristics examined.

This is a common finding in CVOTs with moderate event counts. The EMPA-REG OUTCOME trial (empagliflozin) and LEADER trial (liraglutide) similarly showed consistent subgroup effects without significant interaction terms (Zinman et al., NEJM 2015).

Limitations of These Subgroup Data

Low event count. Only 137 primary MACE events powered the analysis. Subgroup strata with 20 to 40 events cannot meaningfully detect effect modification. Confidence intervals spanning from 0.4 to 2.5 tell us almost nothing.

Short follow-up. Median 15.9 months is brief for a cardiovascular outcomes assessment. The trial was designed for regulatory non-inferiority, not for definitive subgroup-level conclusions. Longer follow-up might shift point estimates in any direction.

No race-specific granularity. The published subgroup analysis collapsed race into "white" and "non-white." Black participants comprised roughly 6% of the trial population, Asian participants about 11%. Neither group was large enough for stand-alone analysis, a recurring limitation in CVOT design.

Standard-of-care variability. Both arms received background therapy per local guidelines, meaning statin use, antihypertensive regimens, and antiplatelet therapy varied by region. Regional subgroup analyses partly capture this variation, but individual medication interactions were not modeled.

Post-hoc analyses remain exploratory. Any subgroup finding from PIONEER-6 should be treated as hypothesis-generating. The SELECT trial and ongoing real-world evidence studies offer better-powered platforms for confirming subgroup signals.

What This Means for Real-World Prescribing

The consistency of PIONEER-6 subgroup results supports a broad indication for oral semaglutide in patients with type 2 diabetes and cardiovascular risk. No subgroup showed a safety concern. The numerically stronger signals in older patients, those with established CVD, and those with longer diabetes duration align with current ADA and ESC guideline recommendations to prioritize GLP-1 RAs in high-risk patients.

For younger patients without established CVD, the data are neither reassuring nor concerning. They are simply underpowered. Clinicians selecting oral semaglutide for these patients should base decisions on metabolic endpoints (glycemic control, weight) rather than expecting proven CV benefit from PIONEER-6 alone.

The absence of racial/ethnic subgroup granularity remains a gap. Prescribers treating diverse populations should recognize that cardiovascular safety was demonstrated in the overall trial but not confirmed at the level of individual racial groups.

Frequently asked questions

›Were PIONEER-6 subgroup analyses pre-specified or post-hoc?

The primary subgroup analyses (age, sex, race, BMI, baseline HbA1c, eGFR, established CVD, diabetes duration, region) were pre-specified in the statistical analysis plan before unblinding. Some additional exploratory analyses were conducted post-hoc.

›Did any subgroup show a statistically significant benefit from oral semaglutide?

No. No individual subgroup reached statistical significance for superiority on MACE. The trial was powered for non-inferiority in the overall population, not for detecting benefit within subgroups.

›Was oral semaglutide harmful in any subgroup?

No subgroup showed a statistically significant increase in MACE risk. Some strata (age <65, no established CVD) had point estimates near or slightly above 1.0, but confidence intervals were extremely wide and included the overall trial estimate.

›Why did older patients appear to benefit more?

Patients aged 65 and older had higher baseline cardiovascular event rates. With more events in this stratum, the point estimate is more precise. Higher absolute risk also means the same relative reduction translates to a larger absolute benefit.

›How does PIONEER-6 compare to SUSTAIN-6 in subgroup findings?

Both trials showed consistent treatment effects across pre-specified subgroups with no significant interaction terms. SUSTAIN-6 used injectable semaglutide (0.5 or 1.0 mg) and showed superiority for MACE (HR 0.74), while PIONEER-6 used oral semaglutide 14 mg and demonstrated non-inferiority (HR 0.79). Subgroup patterns were broadly similar.

›Were Black or Asian patients analyzed separately?

No. The published subgroup analysis used a binary classification of white vs. non-white. Black patients made up roughly 6% and Asian patients about 11% of enrollment, numbers too small for independent MACE analyses.

›Does baseline BMI affect the cardiovascular response to oral semaglutide?

In PIONEER-6, the MACE point estimates were similar for BMI <30 (HR 0.82) and BMI ≥30 (HR 0.78), with overlapping confidence intervals. There was no evidence of effect modification by BMI category.

›Did renal function affect outcomes in PIONEER-6 subgroups?

Patients with eGFR <60 and ≥60 mL/min/1.73 m² had nearly identical point estimates (HR 0.78 vs. 0.80). Oral semaglutide showed consistent cardiovascular safety regardless of baseline renal function in this analysis.

›Should clinicians choose oral semaglutide over injectable based on these subgroup data?

PIONEER-6 subgroup analyses do not directly compare oral to injectable formulations. The choice between oral semaglutide (Rybelsus) and injectable semaglutide (Ozempic) should consider patient preference, adherence factors, dosing requirements, and available efficacy data from head-to-head trials like PIONEER-7.

›Has the SELECT trial changed how we interpret PIONEER-6 subgroups?

SELECT (2023) demonstrated cardiovascular superiority for subcutaneous semaglutide 2.4 mg in patients with overweight/obesity and established CVD but without diabetes. This provides stronger evidence for CV benefit in a broader population, though at a higher dose and different formulation than PIONEER-6 studied.

References

1. Husain M, Birkenfeld AL, Donsmark M, et al. Oral semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2019;381(9):841-851. PubMed
2. Marso SP, Bain SC, Consoli A, et al. Semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2016;375(19):1834-1844. PubMed
3. Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. Semaglutide and cardiovascular outcomes in obesity without diabetes. N Engl J Med. 2023;389(24):2221-2232. PubMed
4. Rybelsus (oral semaglutide) prescribing information. Novo Nordisk. FDA Label
5. Zinman B, Wanner C, Lachin JM, et al. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med. 2015;373(22):2117-2128. PubMed
6. American Diabetes Association. Standards of Care in Diabetes, 2024. Diabetes Care. 2024;47(Suppl 1). ADA Standards