GLP-1 and Inflammation: Laboratory Markers Versus Clinical Outcomes

Semaglutide studies have found both lower inflammatory markers and fewer major cardiovascular events. The harder question is how much one explains the other.

Imagine a headline saying that a treatment “reduces inflammation.” A reader might reasonably picture less pain, fewer disease flare-ups or protection from heart attacks. The study behind the headline may instead have measured a blood protein. Those outcomes are related research questions, but they are not interchangeable.

Semaglutide offers an instructive example because the evidence includes both laboratory changes and a major clinical-outcome trial. Exploratory analyses of the STEP trials reported lower C-reactive protein. SELECT subsequently demonstrated fewer major cardiovascular events in a specific high-risk population, and a later analysis examined whether changes in inflammation might help explain that benefit.^[1]^[2]^[3] A careful account needs to preserve all three findings without merging them into one claim.

What an inflammation marker measures

C-reactive protein, or CRP, is an established biomarker of inflammation. The STEP report notes that it is commonly elevated in people with overweight or obesity. High-sensitivity CRP, abbreviated hsCRP, refers to high-sensitivity measurement of the same protein; the later SELECT analysis used this marker to investigate cardiovascular risk and treatment-associated changes.^[1]^[2]

A biomarker is a measurable biological signal. Lowering that signal is a real observed result, but “less CRP” is more specific than “less inflammatory disease.” It does not, on its own, tell us whether a person has less pain, whether a particular condition improved, or whether a cardiovascular event was prevented.

There is also a difference between a prognostic marker and a causal mechanism. A prognostic marker helps predict what happens later. Showing that people with higher CRP have greater risk does not establish that changing CRP itself will change that risk. The SELECT secondary analysis explicitly examines these relationships, making this distinction central to interpreting its findings.^[2]

STEP found substantial CRP reductions

Verma and colleagues' 2022 report analyzed three randomized, double-blind, placebo-controlled STEP trials lasting 68 weeks. Participants had overweight or obesity; STEP 2 included people with type 2 diabetes, while STEP 1 and STEP 3 did not. Change in CRP was a prespecified secondary endpoint, although the publication also included post hoc analyses, planned after the original trial design.^[1]

At week 68, the estimated treatment differences in CRP versus placebo were −44% in STEP 1, −39% in STEP 2 and −48% in STEP 3. All three differences were statistically significant. These are measured effects on a laboratory endpoint, rather than an inference drawn only from weight loss.^[1]

The reductions occurred alongside weight loss and correlated with changes in body weight, waist circumference and metabolic measures. The authors described a potential anti-inflammatory role for semaglutide in obesity.^[1] That is a reasonable research interpretation. It does not mean the paper demonstrated treatment of every condition described as inflammatory, or established what fraction of the CRP change came from each biological pathway.

SELECT also measured events that matter to patients

The primary SELECT report, published in 2023 by Lincoff and colleagues, moved beyond an inflammation marker. It randomized 17,604 adults aged at least 45 who had established cardiovascular disease and a body-mass index of at least 27, but no history of diabetes. Participants received the trial's semaglutide preparation or placebo.^[3]

The primary endpoint was a composite: death from cardiovascular causes, nonfatal heart attack or nonfatal stroke, analyzed as time to a first event. Over a mean follow-up of 39.8 months, an event occurred in 569 of 8,803 participants receiving semaglutide, or 6.5%, versus 701 of 8,801 receiving placebo, or 8.0%. The hazard ratio was 0.80, with a 95% confidence interval of 0.72 to 0.90.^[3]

This is clinical-outcome evidence from a large randomized trial. It would be inaccurate to summarize the research as only a change in a blood test. The benefit belongs to this trial population, preparation and follow-up period, rather than to all people or every product using GLP-1 terminology.

The rounded event percentages differ by 1.5 percentage points. The hazard ratio corresponds to a 20% lower relative hazard over follow-up; it is not a 20-percentage-point absolute reduction. Keeping these quantities distinct helps readers understand the scale of the result without exaggerating it.^[3]

Benefits also do not erase tolerability findings. Adverse events leading to permanent discontinuation occurred in 16.6% of the semaglutide group and 8.2% of the placebo group.^[3] These are discontinuations because of adverse events, not counts of every adverse event experienced during the trial.

The later analysis connects CRP with cardiovascular risk

Plutzky and colleagues' 2026 prespecified secondary analysis examined hsCRP within SELECT. Higher baseline hsCRP predicted greater risk of major adverse cardiovascular events, often abbreviated MACE. The analysis reported that semaglutide reduced hsCRP by 37.8% at 104 weeks and reduced MACE risk across the baseline hsCRP groups examined.^[2]

Greater reductions in hsCRP were associated with greater weight loss. However, changes in hsCRP were evident by four and eight weeks, before major weight loss, and were also observed among people without weight loss. Reductions in hsCRP were prognostic of lower MACE risk. The authors' modeling suggested that decreased inflammation may have contributed in part to the cardiovascular benefits.^[2]

These findings make inflammation a plausible contributor worth investigating. They do not establish that CRP reduction caused all of the benefit, or that the clinical effect was wholly independent of weight change. The precise language matters: “associated with,” “prognostic” and “may have partially involved” describe different levels of inference from a demonstrated treatment effect on cardiovascular events.

Evidence level	Result measured	Interpretation limit
STEP biomarker	Lower CRP	No universal disease claim
SELECT primary	Fewer major CV events	Defined trial population
SELECT secondary	CRP linked to risk	Mechanism not proved

Randomization of treatment does not randomize the mechanism

Random assignment helps make the semaglutide and placebo groups comparable at the outset. That supports the primary comparison of cardiovascular outcomes. But participants were not separately randomized to a particular amount of CRP reduction while everything else was held constant. Weight, metabolism and inflammatory markers changed together in ways the secondary analysis sought to understand.^[2]^[3]

A mediation model asks whether an intermediate change may account for part of a treatment's effect. Its interpretation depends on assumptions about relationships among measured and unmeasured factors. Modeling can strengthen a mechanistic hypothesis; it does not reproduce an experiment that independently manipulates each pathway.

The observations about early CRP changes and people without weight loss are therefore informative without being definitive. They argue for investigating more than a simple one-variable explanation. They do not justify turning “some changes occurred without weight loss” into “all cardiovascular protection is independent of weight loss.”^[2]

Keep the endpoint and the product attached to the claim

A practical reading approach is to identify what was measured before interpreting the headline. CRP concentration, a cardiovascular event and an inferred biological pathway each require a different sentence. Also check whether a result was primary, secondary or exploratory, and which population produced it. These distinctions preserve favorable findings while keeping their meaning intact.

The trials studied specified semaglutide preparations. Their results do not characterize an arbitrary research vial's identity, quality or suitability for human use. Laboratory research materials should not be consumed or self-administered. A strong medicine trial cannot fill those product-specific evidence gaps, and these papers do not establish treatment of unrelated inflammatory diseases.

This article uses the original published abstracts of the STEP analysis, SELECT primary report and SELECT hsCRP analysis. Full-text methods, detailed mediation assumptions and complete safety tables were not reviewed. The cardiovascular benefit is demonstrated in the reported trial; the degree to which inflammation explains it remains a more qualified conclusion.