GLP-1 for Prediabetes

Prediabetes is a number on a lab report, and one blood draw is enough to place you inside the band. The CDC sets the A1c band at 5.7 to 6.4 percent, with 5.6 and below counted as normal and 6.5 and above counted as diabetes. Fasting plasma glucose does the same job on a different scale: 99 mg/dL and below is normal, 100 to 125 mg/dL is prediabetes, and 126 or above is diabetes.

A two-hour oral glucose tolerance test reads 140 to 199 mg/dL across that same middle range.

Any single test in range is enough to put you there, and the three do not always agree, which is why a clinician often repeats whichever one flagged. The CDC puts the scale of it at more than 2 in 5 American adults. Prediabetes is common, it is defined by a threshold someone chose, and landing on one side of that line does not mean a drug is the next step. It means your glucose handling is drifting.

What the number does not tell you is your trajectory. An A1c of 5.8 that has been flat for six years is a different situation from a 6.3 that was 5.6 last year, and the trials below enrolled almost entirely at the higher-risk end. The ADA describes the highest-risk profile as fasting glucose at or above 110 mg/dL with an A1c at or above 6.0 percent, usually alongside a BMI of 35 or more. Where you sit inside the band changes the entire calculation.

No GLP-1 is FDA approved to treat prediabetes, and the honest version of this page says that before anything else. The current Wegovy label covers weight reduction in obesity, or in overweight with at least one weight-related condition, alongside cardiovascular risk reduction and a liver indication. Zepbound covers obesity or overweight with a weight-related condition, plus obstructive sleep apnea. Saxenda and the oral Foundayo label read the same way. The word prediabetes appears in none of those indication statements.

What is approvable is the route most take anyway. If your BMI is 30 or above, or 27 or above with a weight-related condition, weight management is an on-label conversation, and prediabetes is one of the conditions that qualifies. The distinction is not pedantic. A prescription written for prediabetes alone is off-label prescribing, which a clinician may lawfully do, but it changes what evidence stands behind it and what an insurer will agree to pay for.

Metformin sits in the same position, which surprises most readers. The metformin tablet label is indicated only as an adjunct to diet and exercise to improve glycemic control in type 2 diabetes, and prediabetes appears nowhere in it either. So the real contrast is not approved against unapproved. It is one old drug a major guideline recommends off-label for prevention on 21 years of data, against a class the same guideline declines to recommend for that purpose.

The numbers are genuinely large. In SCALE Obesity and Prediabetes, 2,254 adults with prediabetes and obesity were randomized 2 to 1 to liraglutide 3.0 mg or placebo for 160 weeks. Diabetes was diagnosed in 26 of 1,472 on the drug, about 2 percent, against 46 of 738 on placebo, about 6 percent, and time to onset ran 2.7 times longer. Only half the enrolled group completed, and withdrawn participants were not followed up. These are averages, and your own response can vary widely.

SURMOUNT-1 ran the same question on tirzepatide: 2,539 adults with obesity, of whom 1,032 also had prediabetes, treated for 176 weeks. Type 2 diabetes was diagnosed in 1.3 percent on tirzepatide against 13.3 percent on placebo, a hazard ratio of 0.07. Mean weight change at 176 weeks was 12.3, 18.7 and 19.7 percent below baseline across the 5, 10 and 15 mg arms, against 1.3 percent on placebo. Those are trial averages under trial conditions, with structured lifestyle support running in both arms.

Pooling helps and hurts. A 2026 meta-analysis of eight randomized trials covering 14,564 participants put the odds of returning to normal glucose at 4.62 times placebo, with semaglutide at 4.87 and liraglutide at 5.43, while exenatide showed no effect. Heterogeneity ran 92 percent and the authors graded the overall quality of evidence as low. They also flag the gap that matters here: every included trial enrolled adults who had obesity as well as prediabetes.

This is the part most pages skip. The STEP 1 extension followed 327 participants for 52 weeks after semaglutide 2.4 mg was withdrawn. Among those with prediabetes at baseline, 93.6 percent had a normal A1c at week 68 on the drug, against 41.5 percent on placebo. By week 120, a full year off treatment, that read 43.3 percent against 34.0 percent. The gap did not vanish, but most of it did, and the authors label the whole extension exploratory.

Weight moved the same way. The semaglutide arm lost 17.3 percent by week 68 and regained 11.6 points by week 120, netting 5.6 percent. SURMOUNT-4 tested it with a randomized withdrawal instead: after a 36-week open-label lead-in averaging 20.9 percent loss, continuing tirzepatide produced another 5.5 percent while switching to placebo produced a 14.0 percent regain, a 19.4-point difference.

Read together, the two trials say one thing. The glucose benefit tracks the weight, and the weight tracks whether you are still taking it.

SURMOUNT-1 added 17 weeks off treatment after its 176-week phase. Diabetes had been diagnosed in 1.3 percent on tirzepatide against 13.3 percent on placebo during treatment; after the washout it read 2.4 percent against 13.7 percent. Seventeen weeks is not 52, so that is not the STEP 1 comparison, and neither trial followed anyone for a decade. What none of this supports is the idea that a year on a GLP-1 resets your metabolism and you walk away clean. Plan the exit before the start.

The Diabetes Prevention Program randomized 3,234 adults with elevated glucose to intensive lifestyle intervention, metformin 850 mg twice daily, or placebo. Over a mean 2.8 years, diabetes incidence ran 11.0, 7.8 and 4.8 cases per 100 person-years in the placebo, metformin and lifestyle arms. Lifestyle cut progression 58 percent and metformin 31 percent, and lifestyle was significantly better than metformin. Number needed to treat over three years was 6.9 for lifestyle and 13.9 for metformin. Averages again, from a structured program with coaching built in.

Durability is where this evidence has no competitor. The 15-year DPPOS follow-up still showed 27 percent and 18 percent reductions for lifestyle and metformin, and the ADA's 2026 prevention chapter carries the lifestyle figure to 24 percent at 21 years. No GLP-1 trial has run near that long. The same chapter reports a dose-response worth memorizing: every kilogram of weight lost in the DPP tracked a 16 percent lower risk of progression over 3.2 years.

The ADA's own position is blunt. It states there are currently no long-term data supporting any drug other than metformin for the sole purpose of preventing type 2 diabetes, while adding that GLP-1-based therapy for weight management in overweight or obesity is highly beneficial and should be considered. Its grade A recommendation names metformin, especially between ages 25 and 59 with a BMI of 35 or above, fasting glucose at or above 110 mg/dL and A1c at or above 6.0 percent, or after gestational diabetes.

Where you sit in the band should drive the decision more than the word prediabetes does. The trials above enrolled adults with obesity, mostly in the mid-30s for BMI: the STEP 1 extension group averaged a BMI of 37.6, and 59.6 percent had prediabetes at baseline. If that describes you, the evidence is directly about you. If your BMI is 24 and your A1c is 5.8, no GLP-1 trial enrolled you, and the 2026 pooled analysis says so outright when it calls for research in normal-weight prediabetes.

Lifestyle-first is not a consolation prize, and the low end of the band is where it earns the call. The ADA's first-line recommendation for everyone at risk, graded A, is a structured program targeting 5 to 7 percent weight reduction plus at least 150 minutes a week of moderate activity. That is the arm that outperformed the drug in the only trial that ran both, and it is the arm with 21 years of follow-up behind it.

One thing prediabetes is not is harmless waiting. The ADA notes it is associated with heightened cardiovascular risk in its own right, so the honest framing here is risk management rather than a countdown to a diagnosis. It is also why the cardiovascular indication on some of these labels matters to some readers and not others. A clinician who takes your blood pressure, lipids and family history seriously is doing more for you than one who reaches for a pen.

The doses in these trials are specific and worth knowing before a consultation. The DPP used metformin 850 mg twice daily, which the tablet label reaches by starting at 500 mg twice daily or 850 mg once daily with meals and stepping up, with a 2,550 mg daily ceiling and no use below an eGFR of 30. Gastrointestinal effects drive most early stopping, and taking each dose with food is the standard mitigation.

On the GLP-1 side, SCALE used liraglutide 3.0 mg daily, the STEP 1 program used semaglutide 2.4 mg weekly, and SURMOUNT-1 used tirzepatide at 5, 10 and 15 mg weekly. All of them arrive there through a fixed monthly escalation running roughly four to five months, and that escalation exists to blunt nausea rather than to find a minimum effective dose. A clinician decides the schedule and whether to hold at a step.

The doses now marketed as a light touch for prediabetes are not the doses that produced these results. If a plan starts you at a fraction of the studied dose and keeps you there, nothing on this page describes what happens next. That is not automatically wrong, but it is unstudied, and you should hear it named as unstudied rather than sold as gentler.

The ADA also advises periodic B12 assessment on long-term metformin, with annual monitoring past four years.

Treating a number rather than a disease changes the math on side effects. The Wegovy and Zepbound labels carry a boxed warning about thyroid C-cell tumors observed in rodents, with contraindications for a personal or family history of medullary thyroid carcinoma or MEN 2. Gastrointestinal effects are the routine ones: nausea, vomiting, diarrhea and constipation, and they are the usual reason treatment stops. In SCALE, serious adverse events ran 15 percent on liraglutide against 13 percent on placebo across 160 weeks.

That comparison sits differently when you do not have a disease. Accepting a tolerability burden and a boxed warning to treat established diabetes is one trade. Accepting the same to move an A1c of 5.9 is another, and reasonable clinicians land in different places on it. Metformin's profile is better characterized after decades: gastrointestinal effects that usually fade, B12 depletion over years, and a hard renal floor at an eGFR of 30. Neither drug here is free of downside.

One more asymmetry deserves naming. The trials showing the largest prevention effects also ran the longest, and that is a long commitment for a lab value that has not become a diagnosis. SURMOUNT-1 treated for 176 weeks and the SCALE prediabetes trial ran 160, and neither answers what a decade looks like. A clinician weighing that with you out loud is doing the job. An intake form that never raises it is not.

This part changed recently enough that most pages are wrong about it. Medicare began covering Foundayo tablets, Wegovy injection or tablets, and the Zepbound KwikPen for weight management on July 1, 2026, at a $50 copayment per fill. Eligibility runs on BMI: 35 or above; or 30 or above with certain heart, blood pressure or kidney conditions; or 27 or above with at least one qualifying condition, and prediabetes is named on that list. That is the practical answer to the label problem. Prediabetes is not an indication on any GLP-1 label, but it is a qualifying weight-related condition that can unlock coverage at a BMI of 27. The whole demonstration ends on December 31, 2027, and every fill needs prior authorization, handled by a single central processor rather than by your own plan, with a decision inside 72 hours.

Three exclusions matter more than the BMI arithmetic. Single-dose Zepbound pens and vials are excluded, so only the KwikPen qualifies. The pathway exists only for someone who cannot already get a GLP-1 through their Part D plan, so you have to be enrolled in a standalone Part D plan or a Medicare Advantage plan with drug coverage to use it at all. And a diagnosis of type 2 diabetes, moderate-to-severe sleep apnea or fatty liver disease rules you out of it entirely. The Bridge is a short-term demonstration. It is the interim Part D pathway ahead of CMS's larger BALANCE model, which reached Medicaid in May 2026 and says in its own words that it "will not guarantee coverage for any individual." Read that as approval being something you have to win rather than something the diagnosis hands you.

Without that pathway the gap is enormous. At CMS pharmacy acquisition cost effective 2026-08-19, the Wegovy 2.4 mg pen is $435.36 per mL, about $326 a weekly dose or roughly $1,300 for four. The Zepbound KwikPen at 15 mg a dose is $280.03 per mL, near $168 a dose. Foundayo tablets run about $20.70 each. Metformin 850 mg is $0.02388 a tablet, so the exact DPP regimen is about $1.45 for 30 days. That is acquisition cost rather than what a pharmacy charges you.

A remote visit can be a real evaluation or an order form, and the difference shows up in the first five minutes. A clinician doing this properly will know your A1c or fasting glucose, your BMI, your blood pressure, your lipids and your kidney function before choosing anything. They will say out loud whether the prescription is on-label weight management or off-label for prediabetes. A service that cannot tell you which one it is writing is not evaluating you.

Ask what the exit looks like. Given how the STEP 1 extension and SURMOUNT-4 data read, the plan for stopping matters more than the plan for starting, and any answer that dodges it should worry you. Ask whether metformin and a structured prevention program were considered and why they were set aside, since the ADA still places both ahead of this class for prevention specifically. A good answer might be that your BMI makes weight management the better target. A non-answer is a signal.

Watch for the reverse tell too. A service that declines to prescribe when your numbers do not support it is telling you something useful about its judgment. None of this is medical advice, and the prescribing call belongs to a clinician.

Look for clear pricing that includes the titration months, a prescribing clinician licensed in your state, follow-up labs rather than a single intake form, and a written plan for stopping.