Endpoints in Metabolic Peptide Trials: HbA1c, MACE, Percent Body Weight

The short version

Pivotal trials of metabolic peptide drugs measure outcomes in a consistent vocabulary: HbA1c reduction for type 2 diabetes, percent body weight loss for obesity, and MACE (major adverse cardiovascular events) for cardiovascular outcomes. Each endpoint has a specific regulatory threshold for what counts as "clinically meaningful," a specific statistical method for analyzing it, and specific limitations on what it tells you about a drug's real-world value. This guide explains each measure in enough detail to read a Phase 3 result table and understand what the numbers mean.

HbA1c: the standard T2DM endpoint

What HbA1c measures

HbA1c is the percentage of circulating haemoglobin that has glucose covalently attached to the N-terminal valine of the beta chain (glycated). Because haemoglobin in red blood cells turns over approximately every 8-12 weeks, HbA1c reflects the average plasma glucose concentration over that period. A value of 6.5% or above is diagnostic of type 2 diabetes by American Diabetes Association and WHO criteria; treatment targets for most adults with T2DM are 7.0% or below, or individualized based on age, comorbidity, and hypoglycemia risk.

What a 1% HbA1c reduction means clinically

The landmark UKPDS analysis established that each 1 percentage point reduction in HbA1c is associated with approximately a 21% reduction in diabetes-related mortality, a 14% reduction in the risk of myocardial infarction, and a 37% reduction in the risk of microvascular complications (retinopathy, nephropathy, neuropathy).^[1] These associations are relative risk reductions in populations, not guaranteed individual outcomes, and they were established in the context of sulphonylurea and insulin therapy; whether the same relationships hold for GLP-1 receptor agonists and dual agonists is an open question, since these agents may reduce cardiovascular risk through mechanisms beyond glycemic control.

How HbA1c is reported in pivotal trials

In a Phase 3 T2DM trial, the primary endpoint result is typically a least-squares mean (LSM) change from baseline in HbA1c at the primary time point (usually 26-52 weeks), with a 95% confidence interval for the treatment difference versus placebo or comparator. The statistical analysis uses a mixed-model repeated measures (MMRM) or analysis of covariance (ANCOVA) model, with baseline HbA1c, treatment, stratification factors (e.g., baseline HbA1c category, background therapy), and time point as covariates.

In SURPASS-2, the primary endpoint was change from baseline in HbA1c at 40 weeks. Tirzepatide at 5, 10, and 15 mg produced LSM reductions of -2.01%, -2.24%, and -2.30% respectively, compared with -1.86% for semaglutide 1 mg. The between-group differences were -0.15%, -0.39%, and -0.45%, all statistically significant (p less than 0.001 for each).^[8]

Limitations of HbA1c as an endpoint

HbA1c does not capture glycemic variability, hypoglycemia frequency, or postprandial glucose spikes, all of which contribute independently to cardiovascular risk in some analyses. It is unreliable in haemolytic anaemia, haemoglobinopathies (e.g., sickle cell trait), severe iron deficiency, and after blood transfusion. It also does not change rapidly enough to detect short-term glycemic responses; continuous glucose monitoring (CGM) metrics such as time-in-range are increasingly used as secondary endpoints in trials where rapid glycemic dynamics are of interest.

MACE: major adverse cardiovascular events

The three components

The standard three-component MACE definition is: (1) cardiovascular death, (2) non-fatal myocardial infarction, and (3) non-fatal stroke. These three events are combined into a single composite outcome defined as time to first occurrence of any component. Some trials use four-component MACE (adding hospitalization for unstable angina), which produces more events but makes comparisons across trials more difficult because the fourth component is less severe than the first three.

Each event must be adjudicated by an independent clinical events committee (CEC) that reviews source documents (ECG, cardiac enzymes, imaging, death certificates) without knowledge of treatment assignment. Adjudication is essential because the definitions of MI and stroke require clinical judgment and could otherwise be applied differently across treatment arms.

Why the FDA requires CVOTs for T2DM drugs (since 2008)

Following a 2007 meta-analysis suggesting that rosiglitazone increased myocardial infarction risk despite improving HbA1c, the FDA issued guidance in 2008 requiring that all new T2DM drugs demonstrate cardiovascular safety in a dedicated outcomes trial.^[2] The requirement is non-inferiority: the upper bound of the 95% confidence interval for the hazard ratio for three-component MACE must be below 1.30 (pre-approval) or below 1.30 with an adequately powered superiority test post-approval for drugs with signals of possible cardiovascular benefit.

This requirement produced the cardiovascular outcomes trial programs that have defined much of the modern evidence base for GLP-1 receptor agonists:

• LEADER (liraglutide, 2016). 9,340 patients with T2DM and high cardiovascular risk. MACE HR 0.87 (95% CI 0.78-0.97), p = 0.01 for superiority.^[4]
• SUSTAIN-6 (semaglutide 0.5 mg and 1.0 mg sc weekly, 2016).3,297 patients. MACE HR 0.74 (95% CI 0.58-0.95), p less than 0.001 for non-inferiority, p = 0.02 for superiority.^[5]
• SELECT (semaglutide 2.4 mg sc weekly in obesity without diabetes, 2023). 17,604 patients with established cardiovascular disease and BMI 27 or above but without T2DM at baseline. MACE HR 0.80 (95% CI 0.72-0.90), p less than 0.001 for superiority. The SELECT trial extended the cardiovascular outcomes evidence to an obesity population not previously covered by a dedicated CVOT.^[6]

Reading MACE results: hazard ratios and confidence intervals

MACE is analyzed as a time-to-first-event outcome using a Cox proportional hazards model stratified by the major pre-specified stratification factors (e.g., prior CV event, renal function category). The result is expressed as a hazard ratio with a 95% confidence interval.

Interpreting the hazard ratio:

• HR = 1.0: the event rate in the active arm is identical to the control arm at each time point.
• HR = 0.80: the event rate in the active arm is 20% lower at any given time point (a relative risk reduction of 20%).
• HR = 1.30: the event rate in the active arm is 30% higher; if the upper bound of the CI is 1.30, the trial has met the non-inferiority criterion exactly.

The absolute risk reduction (ARR) and number needed to treat (NNT) are more meaningful to individual clinical decisions but require knowing the event rate in the control arm over the trial period. A drug with HR 0.80 in a trial where the control arm has a 10% MACE event rate at 3 years reduces the rate to approximately 8%, an ARR of 2 percentage points and an NNT of 50 over 3 years.

Percent body weight loss: the obesity endpoint

The FDA thresholds

The FDA's 2007 guidance for obesity drug development specifies two co-primary or dual-criterion approaches for demonstrating clinical meaningfulness:^[3]

1. Mean weight loss criterion. The active drug must produce a mean weight loss of at least 5% more than placebo after 1 year.
2. Responder criterion. A significantly higher proportion of patients in the active arm must achieve at least 5% weight loss compared with placebo. The FDA generally also looks for at least 35% of active arm participants to achieve 5% weight loss.

These thresholds were not arbitrary: population-level evidence shows that weight losses of 5% or more reliably produce clinically meaningful improvements in HbA1c, blood pressure, triglycerides, and HDL cholesterol in overweight or obese adults. Greater weight losses (10%, 15%, 20%) produce proportionally larger cardiometabolic benefits.

How percent weight loss is reported

The primary analysis in obesity trials uses a mixed-model repeated measures (MMRM) model for the continuous percent weight change from baseline, adjusted for baseline weight, treatment, stratification factors, and time. The result is the LSM percent weight change in each arm and the between-group difference with a 95% CI.

The responder analysis is a categorical outcome comparing the proportion of patients achieving each weight-loss threshold (5%, 10%, 15%, 20%) between arms, tested with a chi-squared or logistic regression test.

STEP 1 (semaglutide 2.4 mg for obesity, 2021) reported a mean weight loss of 14.9% with semaglutide versus 2.4% with placebo at 68 weeks, a between-group difference of 12.4 percentage points (95% CI 11.5-13.4, p less than 0.001). The proportion achieving at least 5% weight loss was 86.4% with semaglutide versus 31.5% with placebo.^[7]

Weight regain after stopping treatment

Percent body weight loss as a trial endpoint captures performance while on treatment. A consistent finding across GLP-1 receptor agonist obesity trials is that weight is substantially regained after discontinuation. The STEP 4 withdrawal trial demonstrated that approximately two-thirds of weight lost during semaglutide treatment was regained within one year of stopping the drug. This does not invalidate the primary endpoint, but it is essential context for interpreting the clinical durability of the benefit.

Time-to-event analysis versus difference in means

The choice of statistical analysis method is determined by the nature of the outcome, not by preference.

Difference in means (continuous outcomes)

HbA1c, body weight, blood pressure, and lipid values are continuous measurements. The treatment effect is expressed as the difference between the active arm mean and the control arm mean (or LSM from a regression model), with a 95% CI. The significance test is typically a t-test or MMRM F-test.

Time-to-event analysis (Kaplan-Meier curves and Cox model)

MACE, all-cause mortality, and hospitalization outcomes are analyzed as time to first event. The Kaplan-Meier (KM) estimator is a non-parametric method that estimates the cumulative probability of remaining event-free at each time point, accounting for participants who are censored (lost to follow-up or who have not experienced the event when the trial ends). KM curves are plotted as "survival" curves over the trial duration; the visual divergence or convergence of treatment and control arms tells you when the treatment effect begins to emerge.

The Cox proportional hazards model estimates the hazard ratio: the instantaneous rate of event occurrence in the active arm relative to the control arm, assumed to be constant over time (proportional hazards assumption). In large CVOT trials, the proportional hazards assumption is typically tested and satisfied; violations would require extended analysis or reporting of time-varying hazard ratios.

KM curves in CVOT publications show the cumulative incidence of MACE over time. In LEADER, the KM curves for liraglutide and placebo began to separate at approximately 1 year and diverged throughout the 3.8-year median follow-up, consistent with a persistent treatment effect. The final MACE event rate was 13.0% for liraglutide versus 14.9% for placebo, producing the HR of 0.87.^[4]

Reading the primary endpoint table: worked examples

pp = percentage points. LSM = least-squares mean. MMRM = mixed-model repeated measures. Cox PH = Cox proportional hazards. CI = confidence interval. HR = hazard ratio. Results from primary publications; see individual references.

Comparing across this table illustrates why the analysis method matters. SURPASS-2 and STEP 1 both report LSM differences for continuous outcomes: you can add units directly to understand the magnitude of effect. LEADER, SUSTAIN-6, and SELECT all report hazard ratios for time-to-event MACE outcomes: interpreting those requires knowing the background event rate in the control arm to convert the relative risk reduction to an absolute risk reduction.

Putting the endpoints in context

When reading a metabolic peptide trial result, a useful sequence of questions is:

1. Is the primary endpoint a validated surrogate (HbA1c, % weight loss), a hard composite (MACE), or an unvalidated biomarker? Validated surrogates support approval; hard composites support clinical confidence in patient-relevant benefit.
2. What is the absolute magnitude of the treatment effect? Statistically significant differences on continuous outcomes can be clinically trivial if the difference is small relative to measurement noise or the MCID (minimal clinically important difference).
3. For MACE outcomes: what is the absolute risk reduction and NNT? A hazard ratio of 0.80 in a low-event-rate population (annual MACE rate 2%) implies a smaller absolute benefit than the same HR in a high-event-rate population (annual MACE rate 8%).
4. Does the primary endpoint result align with the key secondary endpoints? If the primary endpoint is positive but the key secondary outcomes (including component-level MACE data) are inconsistent or trend in the wrong direction, the primary result is less convincing.