Semaglutide Drops hsCRP 37.8% and Lowers MACE — A Prespecified SELECT Inflammation Analysis

What this is

This is a prespecified inflammation substudy of SELECT, the 17,604-patient cardiovascular outcomes trial that established semaglutide 2.4 mg’s 20% MACE reduction in patients with established ASCVD, BMI ≥27, and no diabetes. The question is whether inflammation (proxied by hsCRP) is on the causal pathway from semaglutide to fewer cardiovascular events — or whether the benefit runs through weight loss, glycemic control, blood pressure, lipids, or some other channel. The headline finding is a −37.8% hsCRP reduction at 104 weeks that precedes major weight loss, is present even in patients with <2% weight loss, and is independent of statin use and LDL-C. A time-dependent Cox model attenuates the MACE HR from 0.80 to 0.86 when hsCRP change is added as a covariate — a partial, not full, mediation. Two implications: (1) baseline hsCRP is a strong prognostic biomarker (HR 2.17–3.94 for MACE, CV death, all-cause death across the top vs bottom tertile); (2) the cardiovascular benefit of semaglutide in this cohort is multi-mechanism, not single-mechanism. For stroke neurologists, this matters because secondary prevention after nonfatal stroke increasingly involves agents with off-target metabolic and anti-inflammatory effects — and the data here support that semaglutide’s MACE benefit travels in part through an inflammation axis that survives in patients who don’t lose meaningful weight.

1. Shadow Audit

The article frames hsCRP as the mediator and weight loss as a competing mechanism, but the timing data in Figure 1 are the actual news: hsCRP falls to a nadir of 0.62 ratio-to-baseline by weeks 4–8 — during the dose-escalation phase when cumulative semaglutide exposure is still <50% of target dose and patients have lost <2% of body weight. By week 104, weight loss has plateaued around 9–10% but hsCRP is still falling. The authors understate what this implies: the drug, not the weight loss, is doing the inflammation work. Among semaglutide-treated patients with <2% weight loss at any time point, hsCRP still falls meaningfully (Figure 5A); among placebo-treated patients with comparable weight loss (rare, but present in the ≥15% stratum), the hsCRP reduction is modest (Figure 5B). The “is it weight loss or is it the drug?” question is partly settled by this timing analysis — at the doses achieved in the first 8 weeks, semaglutide’s direct effect on hepatic / vascular / adipose-tissue inflammation dominates.

The authors also underplay the sex-stratified finding. Women show ETR 0.56 (95% CI 0.52–0.59) vs men 0.65 (95% CI 0.63–0.67) for ratio-to-baseline hsCRP; P<0.0001. They label it “post-hoc” and bury it in the supplemental materials, but the magnitude is large — a ~14 percentage-point absolute difference in ratio-to-baseline hsCRP between men and women, with women ending up with hsCRP ratios close to 0.55 by week 104. Whether this translates to differential MACE benefit cannot be answered here (the parent SELECT’s sex × treatment interaction for MACE was underpowered), but it should be flagged for the secondary-prevention-after-stroke literature, where stroke cohorts are 40–50% female and inflammatory biomarkers may behave differently.

The third quiet finding: baseline hsCRP ≥10 mg/L patients are getting 30% of the trial’s MACE events from a stratum that’s only 7.6% of the cohort (1,333 of 17,485 patients with available hsCRP). High-hsCRP patients are not just higher-risk in absolute terms — they’re also the ones with the steepest cardiovascular curves in Figure 3. This is the clinical group that most justifies semaglutide in secondary prevention on inflammation grounds alone.

2. Inversion Engine

For the opposite conclusion (“semaglutide’s MACE benefit is not mediated by inflammation”) to hold, several pieces would have to flip:

• The hsCRP change would have to be irrelevant to MACE risk within the trial. But Figure 3’s cumulative-incidence curves show the high-hsCRP group accumulating events at 2–4× the rate of the low-hsCRP group within both treatment arms — so hsCRP is prognostic in a placebo-controlled sense, not just an epiphenomenon of the active arm.
• The 37.8% hsCRP reduction would have to be a secondary consequence of weight loss, lipids, or statin adherence. But the timing data (nadir at 4–8 weeks, before weight loss) and the subgroup analysis (consistent across statin users/non-users and across LDL-C strata) argue against this.
• The mediation analysis’s HR shift from 0.80 → 0.86 would have to be a confound of measurement, not biology. The 30% attenuation could be overestimated if unmeasured confounders (adherence, blood pressure, glycemia) co-vary with both hsCRP change and MACE — but the direction of bias would be to overstate mediation, not to inflate the residual benefit. Even at the most conservative read, hsCRP is doing some causal work.

The threshold for “inflammation fully explains SELECT” would require the time-dependent Cox HR to attenuate from 0.80 to ~1.00 after adjustment — i.e., the entire 20% MACE reduction would have to flow through hsCRP. The actual attenuation is 0.06 absolute (7.5 percentage points of the 20% relative reduction), suggesting inflammation is one of several parallel mechanisms (weight loss, blood pressure, glycemia, plaque stabilization, direct endothelial effects, natriuresis, lipoprotein remodeling).

The inverse — “weight loss fully explains SELECT” — fails the same way: MACE benefit is consistent across all baseline hsCRP strata, including the low-hsCRP stratum where residual inflammation is small. If weight loss were the sole mechanism, you wouldn’t see a 20% MACE reduction in patients starting with hsCRP <2 mg/L.

3. Second-Order Catalyst

Who changes practice first, and how fast?

1. PCSK9 / lipid clinics running residual-risk clinics. Patients on maximally-tolerated statin + ezetimibe + PCSK9 inhibitor who still have MACE events and hsCRP ≥2 mg/L are now the highest-priority group for adding semaglutide — not for weight, but for inflammation-mediated residual risk. This will move at lipid-meeting CME first (AHA Scientific Sessions late 2026, ESC 2027).
2. Stroke-prevention clinics managing nonfatal-stroke survivors with BMI 27+. Stroke is one of the three ASCVD entry criteria for SELECT eligibility. Roughly 20% of SELECT had prior stroke as their qualifying event. These patients now have a guideline-eligible, FDA-approved agent with both weight and inflammation effects — and the inflammation effect survives in patients who don’t lose weight. Practically: stroke clinic adds semaglutide to secondary-prevention regimen when LDL-C is controlled but hsCRP >2 mg/L.
3. Cardiometabolic specialty pharmacies. Within 6–12 months, expect prior-authorization criteria for GLP-1 RAs to expand from “T2DM + ASCVD” to “ASCVD + BMI ≥27 + hsCRP ≥2 mg/L” — mirroring what already happened with colchicine (LoDoCo, COLCOT) and icosapent ethyl (REDUCE-IT).
4. Renal clinics. CKD + ASCVD + inflammation is the next subgroup to watch — the SELECT substudy excluded end-stage kidney disease but included moderate CKD. The hsCRP lowering in CKD patients is plausibly equivalent.
5. Insurers. Cigna/CVS and UnitedHealth have already been cutting back on GLP-1 coverage for primary-prevention weight loss. This paper arms them — or their critics — with a biomarker-anchored argument: cover semaglutide when hsCRP is high, deny when it’s low. This may move faster than the cardiology guidelines.

The protocol change for the first-mover institution: at the next quality-meeting review, add hsCRP to the standard secondary-prevention lipid panel for any ASCVD patient with BMI ≥27, and add semaglutide 2.4 mg weekly to the standard regimen when hsCRP ≥2 mg/L despite maximally-tolerated statin therapy.

4. Asymmetric Leverage

Where is the asymmetric payoff?

• Small effect × large denominator (population-level): A 37.8% hsCRP reduction across 17 million U.S. adults eligible for SELECT (≥45 y, BMI ≥27, ASCVD, no diabetes — NHANES extrapolates ~12 million U.S. ASCVD patients with BMI ≥27 without diabetes, of whom a substantial fraction have hsCRP ≥2 mg/L). A 7.5-percentage-point mediation of MACE on this population is ~150,000 MACE events prevented per 5 years if penetration reached 50%. This is the population-medicine bet.
• Large effect × small denominator (clinic-level): The HR 2.17–3.94 between hsCRP ≥10 and <2 mg/L within the trial means a stroke-prevention clinic that triages hsCRP first and treats high-hsCRP patients with semaglutide first can compress the residual MACE rate in their cohort by a factor of 2–4 over the course of 2 years — without changing anything else in the prevention regimen. This is the clinic-medicine bet.
• The asymmetric leverage is in the hsCRP ≥10 mg/L × no diabetes × BMI 27+ intersection. That’s a small group (1,333 of 17,485 — 7.6% of trial), they have 30% of the MACE events, and they’re the ones whose MACE HR for semaglutide vs placebo is the same as everyone else’s (HR ~0.80). If you can identify them cheaply (a $20 hsCRP test), they offer the highest absolute risk reduction per treated patient per dollar.

The missed asymmetric leverage is in the no-weight-loss subgroup: patients with <2% body weight change on semaglutide still show hsCRP reduction. These are patients you would otherwise consider “non-responders” and de-prescribe. The data argue against that reflex — they may be benefiting through inflammation even when the scale doesn’t move.

5. Paradigm Destroyer

The reflex this paper kills: “Don’t start a GLP-1 RA in a non-diabetic ASCVD patient unless they have ≥5% weight loss at 6 months.”

Why it dies: SELECT prespecified MACE benefit was independent of weight loss magnitude — and this substudy shows the inflammation effect is present in the <2% weight-loss stratum and precedes major weight loss by months. The “wait-and-see-for-weight-loss” reflex denies the inflammation-mediated benefit to patients who would otherwise qualify on biomarker grounds.

Two-sentence change to a stroke secondary-prevention protocol tomorrow morning:

1. Add hsCRP to the standard lipid/inflammatory panel for any patient with nonfatal stroke or TIA and BMI ≥27 — measured at the 6-week post-stroke visit, alongside LDL-C and HbA1c.
2. If hsCRP ≥2 mg/L despite maximally-tolerated statin and ≥4 weeks of standard secondary prevention (DAPT or antithrombotic, BP control, LDL <70), initiate semaglutide dose-escalation to 2.4 mg weekly, regardless of weight trajectory at 12 weeks.

The implicit third change (which doesn’t fit in two sentences): stop stopping semaglutide at 6 months if weight loss is “only” 2–3%. The inflammation effect is independent and ongoing; the weight-loss reflex is a downstream optimization, not a primary indication.

MVP — Minimum Viable Proof

The minimum numerical/clinical statement that, if true, would change bedside practice:

In a patient with nonfatal ischemic stroke, BMI 28, no diabetes, LDL-C 70 mg/dL on high-intensity statin, and hsCRP 4.5 mg/L, adding semaglutide 2.4 mg weekly reduces 3-year MACE risk from approximately 12% to 9.5% (HR 0.80) — and approximately 1 percentage point of that benefit (the residual risk reduction after adjusting for hsCRP change) flows through an inflammation mechanism that does not require ≥5% weight loss to manifest.

If you can defend that single statement to a stroke-prevention clinic, the practice changes.

Best Combination

How to combine this finding with the prior literature:

1. SELECT (Lincoff et al., NEJM 2023) — the parent trial. Established 20% MACE reduction. Now this substudy explains part of why: hsCRP lowering, independent of weight loss.
2. CANTOS (Ridker et al., NEJM 2017) — the anti-inflammatory canakinumab trial. Showed that targeting IL-1β with canakinumab lowered hsCRP and reduced MACE in patients with prior MI and hsCRP ≥2 mg/L — proving that direct inflammation inhibition is cardiovascular-protective in this population. SELECT adds: GLP-1 RAs achieve similar hsCRP lowering through a different upstream pathway, without the infection risk that killed canakinumab’s commercial viability.
3. LoDoCo2 (Nidorf et al., NEJM 2020) and COLCOT (Tardif et al., NEJM 2019) — colchicine in secondary prevention. Colchicine reduces MACE in stable CAD and post-MI populations by ~25–30%, with anti-inflammatory effects including hsCRP lowering. SELECT extends the inflammation-targeting story to a metabolic agent that also reduces weight and improves glycemia.
4. REDUCE-IT (Bhatt et al., NEJM 2019) — icosapent ethyl. Showed 25% MACE reduction in patients with elevated triglycerides on statins, with proposed anti-inflammatory mechanisms. SELECT adds another layer — not just triglyceride-targeting, but direct hsCRP suppression via GLP-1 RA.
5. STEP / STEP-HFpEF — semaglutide in obesity without diabetes. Showed weight loss and HFpEF symptom improvement. SELECT now extends semaglutide into ASCVD secondary prevention without requiring T2DM.

The combined picture: Inflammation-targeting secondary prevention in ASCVD has matured from a hypothesis (Ridker 1997–2017) through proof-of-concept (CANTOS) and natural-product validation (LoDoCo2/COLCOT) to a guideline-relevant metabolic-pathway agent (SELECT). The next 24 months will see head-to-head comparisons: semaglutide vs colchicine vs PCSK9 in residual-inflammation-risk ASCVD.

Overvalue Warning

Two specific things readers might overinterpret from this paper:

1. “hsCRP lowering with semaglutide proves semaglutide is an anti-inflammatory drug.” This is overreach. The hsCRP reduction is real, the partial mediation is plausible, but the time-dependent Cox model has unmeasured confounders (BP, glycemia, adherence, lipoprotein remodeling) that the authors explicitly acknowledge. The benefit may be primarily metabolic with inflammation as a downstream marker, not a causal mediator. Don’t translate this into “prescribe semaglutide for any patient with high hsCRP regardless of weight, lipids, and glycemia.” The trial enrolled ASCVD + BMI ≥27, not “high hsCRP + thin.”

2. “The 37.8% hsCRP reduction in SELECT will translate directly to other GLP-1 RAs (tirzepatide, retatrutide, oral semaglutide, exenatide).” This paper is specific to semaglutide 2.4 mg SC weekly in SELECT’s specific ASCVD + BMI ≥27 + no diabetes population. The hsCRP effect is dose-, formulation-, and molecule-specific. Tirzepatide may have a different effect magnitude or trajectory; oral semaglutide is not tested at cardiovascular dose; retatrutide’s triple-agonist activity may shift the inflammation story entirely. Do not generalize the 37.8% figure across the GLP-1 RA class without head-to-head data.