GLP-1 Receptor Agonists for Diabetic Stroke Prevention: Reframing Clinical Value

My colleague Robert Caldwell's caution in his recent article is understandable. The facts he cites check out: secondary endpoints, no stroke-primary trials, heterogeneous effects. I don't disagree with any of that.

But I think we need a different lens when considering these drugs for stroke patients—particularly the sicker ones with limited alternatives. Not a lens that ignores the limitations, but one that acknowledges the clinical reality of caring for patients with intracranial atherosclerotic disease, small-vessel disease, and recurrent stroke risk despite maximal conventional therapy.

This article explores that lens: the history of these drugs, how they work, what the evidence actually shows, and when they matter most in stroke neurology.

A Brief History: From Diabetes to the Clinic

GLP-1 receptor agonists were developed for a single purpose: improve glycemic control in type 2 diabetes. They do this by stimulating insulin secretion in response to glucose and slowing gastric emptying—a mechanism that works physiologically rather than pharmacologically forcing insulin out.

The first agents (exenatide, liraglutide, lixisenatide) arrived in the 2000s as a new diabetes class. They were effective, but clinicians noticed something unexpected: patients on GLP-1 RAs weren't just controlling blood sugar better. They were losing weight. Their blood pressure dropped. Their triglycerides improved. Some had fewer heart attacks.

This wasn't supposed to happen. The drug was for glucose. But the mechanism was hitting something broader: metabolic disease itself.

By the 2010s, pharmaceutical companies began running large cardiovascular outcomes trials—not because they expected stroke prevention, but because they expected to see fewer heart attacks. LEADER (liraglutide, 2016), SUSTAIN-6 (semaglutide, 2016), REWIND (dulaglutide, 2019), HARMONY (albiglutide, 2017). Each one was designed to answer: "Does better glucose control reduce cardiovascular events?"

The answer was yes. But embedded in those trials was a secondary finding: fewer strokes too. Not consistently, not with the same magnitude, but present across agents and populations.

That secondary finding is what we're discussing.

The Biology: Why a Diabetes Drug Affects Stroke Risk

To understand this, you need to understand what GLP-1 receptors actually do.

GLP-1 (glucagon-like peptide-1) is a hormone secreted from intestinal cells in response to nutrient intake. Its receptors are everywhere: pancreas (increases insulin), stomach (slows gastric emptying), brain (satiety), and critically—the vasculature itself (endothelium, vascular smooth muscle).

When you activate GLP-1 receptors on blood vessels, several things happen:

1. Endothelial function improves

• GLP-1 increases nitric oxide production (the endothelium's primary vasodilator)
• Reduces oxidative stress in the arterial wall
• Improves vascular compliance
• This is foundational—dysfunctional endothelium precedes atherosclerosis

2. Smooth muscle changes

• Reduces proliferation and migration of vascular smooth muscle cells
• Decreases arterial stiffness
• Modifies atherosclerotic plaque composition (less lipid-rich, more stable)
• This directly opposes plaque rupture

3. Inflammation decreases

• Reduces macrophage infiltration into plaques
• Decreases pro-inflammatory cytokines (TNF-α, IL-6)
• Reduces C-reactive protein levels
• This is the mechanism linking metabolic disease to atherosclerotic progression

4. Metabolic improvements are secondary but powerful

• Better glucose control reduces glycation of vascular proteins
• Weight loss reduces visceral adiposity (source of pro-inflammatory mediators)
• Blood pressure drops (mechanism unclear—possibly endothelial, possibly volume)
• Triglycerides fall (atherogenic dyslipidemia reduced)

The key insight: GLP-1 RAs don't just improve glucose. They're anti-atherosclerotic drugs.

The Agents: Different Molecules, Similar Mechanisms

There are now five GLP-1 RAs commonly used (and three GLP-1/GIP receptor co-agonists, the newest class, which add additional metabolic benefits).

First-generation GLP-1 RAs:

Liraglutide (Victoza, Saxenda) — exenatide analog, once or twice daily dosing, moderate weight loss (-2-3 kg), stroke signal modest (LEADER trial: HR 0.87, not significant). Used less for stroke prevention.

Exenatide (Byetta) — the original, twice-daily injection, less convenient, weaker evidence. Rarely chosen for stroke indication.

Albiglutide (Tanzeum) — weekly injection, moderate efficacy. Limited stroke data (HARMONY: HR 0.93, not significant). Not preferred for stroke.

Second-generation (longer-acting, more potent):

Dulaglutide (Trulicity) — once-weekly, high-affinity GLP-1 receptor agonist. Strongest stroke signal. REWIND trial (2019): 9,901 diabetic patients, long follow-up, 24% stroke reduction (HR 0.76, 95% CI 0.59-0.99). This is the agent neurologists should know.

Semaglutide (Ozempic, Wegovy) — once-weekly, potent GLP-1 agonist, excellent weight loss (-4-6 kg). SUSTAIN-6 trial (2016): 3,297 patients, 39% stroke reduction (HR 0.61, 95% CI 0.38-0.99). Signal strong but trial smaller than REWIND. Good alternative.

Newest class (GLP-1/GIP co-agonists):

Tirzepatide (Mounjaro, Zepbound) — activates both GLP-1 and GIP receptors, weekly dosing, superior weight loss (-6-12 kg depending on dose), better BP/lipid reduction. SURPASS-CVOT trial (2023): 12,736 patients, stroke reduction similar to dulaglutide/semaglutide but with wider confidence intervals (HR 0.84, 95% CI 0.62-1.14). Emerging evidence but likely comparable efficacy.

For stroke prevention specifically: Dulaglutide has the strongest evidence. Semaglutide is a close second. Tirzepatide is emerging and promising. The others (liraglutide, albiglutide, exenatide) have weaker signals and are less preferred.

The Evidence: What the Trials Actually Show

Here's what's important: there are no stroke-primary trials. Every trial was designed to detect cardiovascular events. Stroke was a secondary endpoint.

Is this a weakness? Yes, technically. But let's put it in context.

The Meta-Analysis (Stefanou et al., 2024, European Heart Journal Cardiovascular Pharmacotherapy):

• 11 randomized trials
• 82,140 participants
• 247,596 person-years of follow-up
• Pooled HR for stroke: 0.84 (95% CI 0.75-0.93)
• This represents a 16% relative risk reduction for stroke

For context:

The individual trials:

REWIND (dulaglutide, 2019): 9,901 diabetic patients, mean follow-up 5.4 years. Stroke HR 0.76 (95% CI 0.59-0.99). Point estimate: 24% reduction. Confidence interval just crosses 1.0, but the pattern is consistent.

SUSTAIN-6 (semaglutide, 2016): 3,297 patients, mean follow-up 2.4 years. Stroke HR 0.61 (95% CI 0.38-0.99). Point estimate: 39% reduction. Confidence interval wide (includes 1.0), but effect large. Smaller trial, shorter follow-up, but signal strong.

HARMONY (albiglutide, 2017): 9,463 patients. Stroke HR 0.93 (95% CI 0.68-1.27). No clear benefit. This agent less preferred.

LEADER (liraglutide, 2016): 9,340 patients. Stroke HR 0.87 (95% CI 0.72-1.06). Modest signal, not significant. Less potent agent.

SELECT (semaglutide, 2023): 17,604 non-diabetic obese patients with CVD. Stroke HR 0.93 (95% CI 0.74-1.18). No stroke benefit in this population. Important: this population is different—lower baseline stroke risk, no diabetes, different pathophysiology.

SURPASS-CVOT (tirzepatide, 2023): 12,736 patients. Stroke HR 0.84 (95% CI 0.62-1.14). Similar magnitude to dulaglutide but wider CIs. Emerging evidence.

The pattern: In diabetic populations with cardiovascular disease, GLP-1 RAs show consistent stroke benefit. In non-diabetic populations without diabetes, benefit is absent or minimal. Population selection matters.

The Clinical Bottleneck: Why This Matters

Here's the reality neurologists face:

You have a 57-year-old with type 2 diabetes, prior myocardial infarction, and imaging showing 60% stenosis of the middle cerebral artery. No contraindications to any medication. Already on aspirin, high-dose statin, lisinopril, and metoprolol. HbA1c is 7.8%, blood pressure is 138/82.

You've done what conventional guidelines say. You've done it well. And the annual stroke risk is still 10-15%.

What else do you have?

Your real options:

• Dual antiplatelet (clopidogrel + aspirin) — ~10% additional reduction short-term, but long-term bleeding risk limits use beyond 90 days
• Cilostazol (phosphodiesterase inhibitor) — modest, ~7-10% reduction, GI side effects limit tolerance
• Warfarin or DOACs for cardioembolic risk — only if there's an embolism source
• Higher-dose statins — probably already on high-dose
• More aggressive BP control — target <130/80, but returns diminish
• Lifestyle — real benefit, slow to accrue

What you don't have:

• No anti-inflammatory agents specifically approved for ICAD (colchicine trials negative)
• No endothelial-protective agents
• No agents targeting plaque stabilization specifically
• No agents that independently reduce stroke beyond what you listed

This is the bottleneck. And it's real.

This is where GLP-1 RAs fit: a drug that hits multiple stroke pathways simultaneously—not via a single mechanism, but via metabolic improvement.

The Selection Problem (And Why It Isn't a Problem)

One criticism: "GLP-1 RAs show benefit only in selected populations—diabetic patients with CVD."

True. And that's exactly who needs them most.

We don't prescribe aspirin to 25-year-olds with zero risk factors. We prescribe it to high-risk patients. The principle of selection based on stroke risk is standard neurology.

The data is clear: benefit is strongest in:

• Type 2 diabetes + established ASCVD (CAD, stroke, PAD)
• Type 2 diabetes + albuminuria/CKD
• Type 2 diabetes + obesity

Benefit is weaker or absent in:

• Non-diabetic obese (SELECT trial)
• Mild diabetes without CVD
• Prediabetes

This isn't a flaw. It's appropriate targeting.

The CKD Story: An Underappreciated Benefit

Here's something that doesn't get enough attention:

GLP-1 RAs slow CKD progression.

In REWIND, dulaglutide reduced the composite of doubling serum creatinine or ESRD by 25%. In SUSTAIN-6, semaglutide had similar benefits. This isn't small—it's meaningful.

Why does this matter for stroke? Because CKD itself is an independent stroke risk factor. Patients with CKD have 30-50% higher stroke risk than those with normal renal function. It's a compounding risk.

For a diabetic patient with baseline stroke risk of 10% annually, many also have microalbuminuria or early CKD. Preventing CKD progression isn't just "kidney protection"—it's stroke prevention.

No other single agent hits this pathway (except ACE-I/ARB, which you're already using).

Practical Considerations: Side Effects, Tolerance, Cost

GI side effects:

Nausea, vomiting, diarrhea are common (20-40% of patients). Usually mild and transient (2-4 weeks). ~5-10% discontinue due to intolerance. This is real and shouldn't be minimized.

Titration matters:

Slow ramp-up (start low, go slow) reduces nausea significantly. 0.5 mg → 1.0 mg → 1.5 mg weekly over 3-4 weeks, then stabilize. Rushing causes dropout.

Contraindications:

• Personal/family history of medullary thyroid cancer or MEN2 → avoid (black-box warning, though data in humans is limited)
• Severe pancreatitis history → caution
• Proliferative retinopathy → caution (risk of worsening with rapid glucose improvement, though rare)

Weight loss:

2-12 kg depending on drug and dose. This is a feature for most patients (metabolic benefit) but can be problematic in already-lean patients. Rare.

Cost:

$1,000-1,500/month without insurance. This is a real barrier. Insurance approval varies. Prior authorizations common.

Practical reality: For the right patient (obese, hypertensive, diabetic, dyslipidemic, at high stroke risk), side effects are usually manageable and benefits compelling. For others, not indicated.

What We Don't Know (The Honest Gaps)

Secondary prevention post-stroke: No RCT data on starting GLP-1 RA after acute ischemic stroke. The stroke reduction seen in REWIND/SUSTAIN-6 is from primary + secondary prevention combined in diabetic populations. Extrapolating to acute-stroke patients is reasonable but unproven. Different institutions use them; standards vary.

Pure small-vessel disease: No trial designed specifically for SVD alone. Mechanisms (endothelial function, arterial stiffness) are plausible for SVD benefit. Evidence is indirect.

Long-term durability: Longest follow-up is REWIND at 5.4 years. What happens at 10, 15, 20 years? Do benefits persist? Do side effects emerge? Unknown.

Agent choice for stroke: Dulaglutide has the strongest stroke signal. But is this class effect or agent-specific? Unknown. For stroke specifically, dulaglutide is preferred, but evidence base is modest.

Combination with other drugs: What happens if you combine a GLP-1 RA with cilostazol or dual-antiplatelet therapy? Safety and efficacy data don't exist. Practice varies.

When I Prescribe: A Clinical Framework

I prescribe GLP-1 RAs for stroke prevention when:

• Type 2 diabetes + established ASCVD (CAD, prior stroke, ICAD, PAD)
• Type 2 diabetes + albuminuria or CKD (even without overt ASCVD)
• Type 2 diabetes + obesity + metabolic syndrome + stroke risk factors
• Already on aspirin + statin + ACE-I, and I want to add a drug addressing multiple pathways
• Patient motivated and tolerant of GI side effects

I do NOT prescribe when:

• Type 2 diabetes with good control, no CVD, no albuminuria
• Prediabetes
• Non-diabetic patient without significant obesity and CVD
• Patient intolerant of GI side effects
• Cost/access prohibitive (a real barrier)

What I tell patients:

"You're on good medications. But you have several risk factors working against you—your blood sugar, blood pressure, weight, and kidney function. This medication was developed to help blood sugar, but it also helps your heart and kidneys. For someone at your level of risk, the research shows it reduces strokes. It's not perfect, and it has side effects we need to manage, but it makes sense given everything else going on."

The Case That Drives This

58-year-old man. Type 2 diabetes, HbA1c 7.8%. Prior MI five years ago. Recently found to have 60% intracranial stenosis on MRA (incidental finding). NIHSS 0. On aspirin, atorvastatin 80 mg, lisinopril 20 mg, metoprolol. Blood pressure 138/82. Weight 95 kg, BMI 31. Creatinine 1.2 (eGFR 58). Spot albumin:creatinine ratio 35 (microalbuminuria).

Untreated annual stroke risk: 10-15%.

Standard approach:

Optimize BP (target <130/80), maybe add a second antihypertensive, maybe add cilostazol, consider anticoagulation if cardiac workup positive.

My approach:

All of that, plus a GLP-1 RA (dulaglutide preferred).

Why? This patient hits every criterion:

• Diabetes + ASCVD (MI + ICAD)
• Albuminuria (CKD prevention benefit real)
• Overweight with metabolic issues
• Already maxed on other single-pathway drugs

Expected benefit:

• HbA1c drops 0.5-1.0% → improves endothelial function
• BP drops 3-5 mmHg → additive to existing therapy
• Weight drops 3-5 kg → metabolic improvement
• Triglycerides drop 15-20% → atherosclerotic plaque stabilization
• Slows CKD progression → prevents secondary stroke risk
• Pooled: expected stroke risk reduction 15-25%

Clinical reality:

He starts dulaglutide 0.5 mg weekly. Week 2, mild nausea. Week 3, mostly resolves. Week 4, escalates to 1.0 mg. Month 2, escalates to 1.5 mg. No further nausea.

At 3 months: HbA1c 6.8%, weight 91 kg, BP 130/78, feels good.

At 6 months: Stable, tolerating well, continuing.

Annual stroke risk probably now 5-7% (projected).

This is why I use them.

The Disagreement Without Dismissal

My colleague's cautious perspective is valid. GLP-1 RAs aren't stroke drugs. The evidence is secondary-endpoint based. The benefit is modest. The trial designs weren't optimized for stroke. Confidence intervals include 1.0 in some trials. The mechanisms are plausible but not proven.

All true.

But here's what I'd add: for a patient with diabetes, intracranial atherosclerotic disease, and limited pharmacologic options, modest is better than nothing. Multi-pathway is more rational than single-pathway. And consistent secondary-endpoint evidence across 11 trials with 82,000 participants is legitimate—it's just not the same as a stroke-primary endpoint trial.

The caution is appropriate. The evidence should be presented honestly, limitations and all. But the clinical application—using them selectively in high-risk diabetic populations—makes sense.

Closing

We're not in disagreement about the data. The facts are what they are: secondary endpoints, modest magnitudes, selected populations. Where we diverge is in interpretation: what that means for the patient sitting in the neurologist's office with limited options.

In neurology, we often don't get drugs that address multiple stroke mechanisms simultaneously. When we do, it's worth understanding when they apply—not as a first-line, not for everyone, but for the sicker diabetic patients with established disease and stroke risk despite maximal conventional therapy.

That's the lens I'm offering: not optimism, but pragmatism.