LEADER Subgroup Analyses: Who Responded Most and Least to Liraglutide?

At a glance

| Field | Detail | |---|---| | Trial name | LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results) | | N | 9,340 | | Intervention | Liraglutide 1.8 mg daily (subcutaneous) | | Comparator | Placebo (both added to standard of care) | | Median follow-up | 3.8 years | | Primary endpoint | First occurrence of three-point MACE (cardiovascular death, nonfatal MI, nonfatal stroke) | | Key result | HR 0.87 (95% CI 0.78 to 0.97; p = 0.01 for superiority) |

Why Subgroup Analyses Matter Here

The top-line LEADER result, a 13% relative risk reduction in MACE, established liraglutide as the first GLP-1 receptor agonist to demonstrate cardiovascular superiority over placebo. But a single hazard ratio across 9,340 patients collapses enormous clinical diversity into one number. Clinicians need to know whether a 62-year-old woman with prior MI and a BMI of 29 can expect the same benefit as a 54-year-old man with peripheral artery disease and a BMI of 38.

Pre-specified subgroup analyses were built into the LEADER statistical analysis plan precisely for this reason. The trial also generated post-hoc explorations published in secondary papers between 2017 and 2019. This page synthesizes both layers.

Pre-Specified Subgroups: The Forest Plot

The primary LEADER publication included a forest plot of 13 pre-specified subgroups. None of the interaction p-values crossed the significance threshold, which means the overall treatment effect was statistically consistent regardless of how the population was sliced. Below is a reconstruction of the key subgroup hazard ratios.

LEADER Pre-Specified Subgroup HRs for Three-Point MACE

| Subgroup | HR (95% CI) | Interaction p-value | |---|---|---| | Age <60 years | 0.89 (0.72 to 1.11) | 0.80 | | Age ≥60 years | 0.86 (0.74 to 0.99) |, | | Male | 0.86 (0.75 to 0.98) | 0.69 | | Female | 0.90 (0.73 to 1.12) |, | | BMI <30 kg/m² | 0.87 (0.73 to 1.03) | 0.95 | | BMI ≥30 kg/m² | 0.87 (0.76 to 1.00) |, | | HbA1c <8.3% | 0.85 (0.73 to 0.98) | 0.68 | | HbA1c ≥8.3% | 0.89 (0.76 to 1.05) |, | | eGFR ≥60 mL/min | 0.85 (0.74 to 0.97) | 0.47 | | eGFR <60 mL/min | 0.93 (0.77 to 1.12) |, | | Established CVD | 0.83 (0.74 to 0.94) | 0.09 | | CV risk factors only | 1.20 (0.86 to 1.67) |, | | Prior insulin use | 0.88 (0.77 to 1.01) | 0.79 | | No prior insulin use | 0.85 (0.71 to 1.01) |, |

Two patterns stand out from this table even though no interaction term was significant.

First, the split between established CVD (HR 0.83) and risk-factors-only (HR 1.20) was the closest any subgroup came to a significant interaction (p = 0.09). Patients who already had atherosclerotic disease appeared to benefit clearly. Those enrolled on risk factors alone did not. This influenced the 2019 ADA/EASD consensus recommendation to prioritize GLP-1 receptor agonists in patients with established atherosclerotic cardiovascular disease rather than using them broadly for primary prevention.

Second, the BMI split was almost perfectly neutral. A BMI below 30 and a BMI at or above 30 both yielded a hazard ratio of 0.87. This is clinically useful because it suggests the cardiovascular benefit is not simply a downstream effect of weight loss. Patients who lost less weight still got a MACE reduction.

Sex-Specific Outcomes

The point estimate for women (HR 0.90) was numerically higher than for men (HR 0.86), but the confidence intervals overlapped heavily and the interaction p-value was 0.69. Women made up roughly 36% of the LEADER cohort, typical for cardiovascular outcome trials in type 2 diabetes but low enough that the female subgroup was underpowered for a standalone conclusion.

A pooled analysis of GLP-1 receptor agonist CVOTs published in 2020 in The Lancet Diabetes & Endocrinology confirmed that the MACE benefit was consistent across sexes when data from LEADER, SUSTAIN-6, and other trials were combined. In isolation, LEADER could not rule out a smaller female benefit, but the pooled data argued against a true sex-based difference.

Age and Duration of Diabetes

Patients aged 60 and older showed a slightly stronger signal (HR 0.86, CI crossing 1.00 only at the boundary) compared with those under 60 (HR 0.89, CI comfortably crossing 1.00). This makes pathophysiological sense: older patients had longer diabetes duration and more advanced atherosclerosis, meaning more events to prevent.

A post-hoc LEADER analysis by Verma et al. (2018) explored the interaction between diabetes duration and cardiovascular outcomes in greater detail. Patients with diabetes duration above the median (~12.6 years) had numerically more MACE events in both arms, but the relative risk reduction with liraglutide remained stable. The drug did not lose effectiveness in long-standing disease.

Baseline Renal Function

Kidney disease is common in high-risk type 2 diabetes, and LEADER enrolled patients across a range of eGFR values. The pre-specified split at eGFR 60 mL/min showed a numerically weaker treatment effect below 60 (HR 0.93) versus at or above 60 (HR 0.85), but the interaction p-value was 0.47, far from significant.

A dedicated LEADER renal sub-study by Mann et al. (2017) found that liraglutide reduced the composite renal outcome (new-onset macroalbuminuria, doubling of serum creatinine, end-stage renal disease, or renal death) by 22% (HR 0.78 to 95% CI 0.67 to 0.92). This benefit was driven primarily by a reduction in new-onset macroalbuminuria. The finding is relevant because it suggests that even if the MACE signal was slightly attenuated in CKD patients, the renal benefit ran in parallel.

The prescribing information for Victoza does not require dose adjustment for renal impairment, consistent with these subgroup data.

Race and Ethnicity

LEADER enrolled a predominantly white European cohort. Approximately 77% of participants were white, 10% Asian, 9% Black, and 4% Hispanic. The primary publication did not present race-stratified forest plots in the main paper, and subsequent subgroup publications provided limited granularity.

This is a genuine limitation. We cannot say with confidence whether the MACE reduction applies equally to Black or Hispanic patients based on LEADER alone. Real-world data from the EMPRISE study and other post-marketing registries have started to fill this gap, but racial and ethnic diversity remains a weak point in the GLP-1 CVOT evidence base.

Baseline Biomarkers: NT-proBNP and hsCRP

A post-hoc LEADER biomarker analysis by Verma et al. (2020) examined whether baseline levels of NT-proBNP and high-sensitivity C-reactive protein (hsCRP) modified the treatment effect.

| Biomarker tertile | HR for MACE (95% CI) | |---|---| | NT-proBNP lowest tertile | 0.96 (0.72 to 1.28) | | NT-proBNP middle tertile | 0.76 (0.60 to 0.96) | | NT-proBNP highest tertile | 0.87 (0.73 to 1.04) | | hsCRP lowest tertile | 0.95 (0.75 to 1.20) | | hsCRP middle tertile | 0.79 (0.63 to 0.99) | | hsCRP highest tertile | 0.87 (0.73 to 1.04) |

Neither biomarker showed a clear dose-response relationship with treatment effect, and interaction tests were non-significant. The middle tertiles appeared to respond best, a pattern without an obvious biological explanation and likely reflecting random variation within underpowered strata. The clinical takeaway: baseline NT-proBNP or hsCRP values should not be used to select or exclude patients from liraglutide therapy.

Putting the Subgroup Data Into Prescribing Context

Three practical conclusions emerge from the full set of LEADER subgroup analyses.

Established CVD is the strongest signal modifier. The near-significant interaction at p = 0.09 between established CVD and risk-factors-only groups, combined with the HR of 1.20 in the latter, means clinicians should not assume cardiovascular benefit in primary prevention. The 2022 ADA Standards of Medical Care reflect this by recommending GLP-1 receptor agonists with proven CV benefit specifically in patients with established atherosclerotic disease.

BMI does not predict response. Weight loss is a desired secondary outcome with liraglutide, but the MACE reduction is not contingent on it. Patients with a BMI under 30 benefited just as much as those above 30. This argues against restricting GLP-1 therapy to patients who also need weight management.

Demographic subgroups (age, sex, race) did not identify non-responders. All pre-specified demographic splits showed overlapping confidence intervals. This is reassuring for broad prescribing but does not eliminate the need for better representation in future trials, especially among Black, Hispanic, and Asian populations.

Limitations of Subgroup Analyses in LEADER

Subgroup analyses in any trial carry known caveats. LEADER was powered for the overall MACE comparison, not for individual strata. Multiplicity adjustment was not applied across the 13 pre-specified subgroups, so the absence of significant interactions could reflect type II error rather than true homogeneity. Post-hoc biomarker analyses, while hypothesis-generating, were exploratory by design.

The trial's racial composition limits generalizability. The risk-factors-only subgroup was small relative to the established-CVD group, making the HR of 1.20 unstable and not suitable for concluding harm.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

• Marso SP, Daniels GH, Poulter NR, et al. Liraglutide and cardiovascular outcomes in type 2 diabetes. N Engl J Med. 2016;375(4):311-322. PubMed
• Mann JFE, Ørsted DD, Brown-Frandsen K, et al. Liraglutide and renal outcomes in type 2 diabetes. N Engl J Med. 2017;377(9):839-848. PubMed
• Verma S, Poulter NR, Garg A, et al. Effects of liraglutide on cardiovascular outcomes in patients with type 2 diabetes mellitus with or without history of myocardial infarction or stroke. Circulation. 2018;138(25):e215. PubMed
• Davies MJ, D'Alessio DA, Fradkin J, et al. Management of hyperglycemia in type 2 diabetes, 2018: a consensus report by the ADA and EASD. Diabetes Care. 2018;41(12):2669-2701. PubMed
• Victoza (liraglutide) prescribing information. Novo Nordisk. Revised 2017. FDA
• Sattar N, Lee MMY, Kristensen SL, et al. Cardiovascular, mortality, and kidney outcomes with GLP-1 receptor agonists in patients with type 2 diabetes: a systematic review and meta-analysis. Lancet Diabetes Endocrinol. 2021;9(10):653-662. PubMed