The SOUL trial results, in one line: among 9,650 adults with type 2 diabetes and heart or kidney disease, 12.0% of those taking a daily semaglutide tablet (Rybelsus, up to 14 mg) had a heart attack, stroke or cardiovascular death over about four years, compared with 13.8% on placebo, a look-alike pill with no drug in it. That's a 14% lower relative risk, driven mostly by fewer non-fatal heart attacks.
Those numbers put a heart-risk use on the Rybelsus label in October 2025. What SOUL can't do is describe a different product, and that includes compounded semaglutide.
What question did SOUL ask?
By 2019 the tablet had passed a safety test: PIONEER 6, a trial of 3,183 people, was built to rule out extra heart risk, not to prove a benefit. SOUL asked the harder question. Does the pill actually lower the risk of serious heart events?
Nobody could assume it would. The weekly injection had been linked with fewer heart events in SUSTAIN-6, but a tablet delivers far less drug: about 0.4% to 1% of a 14 mg dose reaches the blood, against about 89% of an injected one, figures the SOUL paper itself cites.
Who was in the SOUL trial, and for how long?
Everyone was 50 or older with type 2 diabetes and an HbA1c (a blood test of average sugar over about three months) between 6.5% and 10.0%. Each also had artery disease in the heart, brain or legs, or chronic kidney disease, meaning an eGFR (an estimate of how well the kidneys filter) below 60. People with end-stage kidney disease or on long-term dialysis were left out.
From June 2019 to March 2021, 444 sites in 33 countries randomly split 9,650 people into two equal groups (McGuire et al., NEJM 2025; registry record NCT03914326). The average participant was 66, and about 4 in 10 had kidney disease. Only 28.9% were women, and 2.6% were Black.
Everyone kept their usual care, which for about 1 in 4 included an SGLT2 inhibitor, another diabetes drug class with heart and kidney benefits. The tablet stepped up from 3 mg to 7 mg, then 14 mg. Follow-up averaged 47.5 months.
What did SOUL measure? MACE, explained
The main outcome was MACE, short for major adverse cardiovascular events: the first of a cardiovascular death, a non-fatal heart attack or a non-fatal stroke. Each person counted once, and stayed in their assigned group even if they stopped the pills.
Behind it sat a queue of confirmatory outcomes tested in strict order, starting with a five-part kidney composite and then cardiovascular death. If one step failed, formal testing stopped there.
SOUL trial results: the heart numbers
The main outcome happened to 579 of 4,825 people on the tablet and 668 of 4,825 on placebo. The hazard ratio, which compares how quickly events built up in each group, was 0.86 (95% confidence interval 0.77 to 0.96). That's the 14%.
In plain counts the groups ended 1.8 percentage points apart. The paper's own three-year estimate was 2.0 points, roughly one event prevented for every 50 people treated.
| Outcome, about four years | Oral semaglutide | Placebo | Hazard ratio (95% CI) |
|---|---|---|---|
| Heart attack, stroke or cardiovascular death | 12.0% | 13.8% | 0.86 (0.77–0.96) |
| Non-fatal heart attack | 4.0% | 5.2% | 0.74 (0.61–0.89) |
| Non-fatal stroke | 3.0% | 3.3% | 0.88 (0.70–1.11) |
| Cardiovascular death | 6.2% | 6.6% | 0.93 (0.80–1.09) |
| Five-part kidney composite | 8.4% | 9.0% | 0.91 (0.80–1.05) |
| Death from any cause | 10.9% | 12.0% | 0.91 (0.80–1.02) |
Only the heart-attack row clearly separated. Stroke and cardiovascular death leaned the same way, but their ranges cross 1.0, meaning no clear difference. The kidney composite missed (P = 0.19), so cardiovascular death was never formally tested. At two years, HbA1c was 0.56 percentage points lower than on placebo and weight 2.95 kg (about 6.5 lb) lower.
What side effects showed up in SOUL?
SOUL didn't count every mild symptom. Per the current label, it collected serious adverse events, events that made someone stop the drug, and events of special interest, so this table can't show who felt queasy for a week.
| Safety measure | Oral semaglutide | Placebo |
|---|---|---|
| Any serious adverse event | 47.9% | 50.3% |
| Serious stomach or bowel disorder | 5.0% | 4.4% |
| Stopped the drug because of an adverse event | 15.5% | 11.6% |
| Stopped because of stomach or bowel problems | 6.4% | 2.0% |
| Acute pancreatitis | 0.4% | 0.4% |
| Gallstones (label figure) | 1.1% | 0.9% |
| Gallbladder inflammation (label figure) | 1.1% | 0.7% |
Serious events were slightly less common on the drug, mostly because the placebo group had more heart problems and infections. The cost showed up as quitting: stomach and bowel trouble led about 1 in 16 people on the tablet to stop, against 1 in 50 on placebo.
What critics and later studies said
The authors named the limits themselves: few women, few Black participants, and a kidney question left unsettled. An editorial by Jialal and Olatunbosun added that the benefit rested mostly on non-fatal heart attacks, that the findings don't simply transfer to the typical clinic patient, and that the newer, lower-dose tablet was never tested. It also asked for a closer look at a small excess of tumor reports on the tablet in the supplementary tables.
A 2025 meta-analysis of ten trials with 71,351 people found long-acting GLP-1 medicines lowered MACE by 14%, with no meaningful difference between pills and injections. Among SOUL participants who already had heart failure, a 2026 secondary analysis reported fewer heart-failure events on the tablet (hazard ratio 0.78), a pattern rather than proof.
As of September 23, 2026, the day this article was written, two newer papers add detail. A July 21, 2026 analysis found the heart benefit looked larger in people with a higher starting HbA1c or a bigger HbA1c drop, but didn't vary with body weight. An August 7, 2026 pooled analysis combined SOUL with two injectable trials, 30,787 people in all, and reported a lower risk on a kidney-and-heart composite (hazard ratio 0.84). Pooling doesn't change SOUL's own kidney result, and SOUL and the pooled paper were both funded by Novo Nordisk, which makes Rybelsus.
SOUL and the Rybelsus label
On October 17, 2025, the FDA signed a supplement letter adding a new use to Rybelsus: reducing the risk of cardiovascular death, non-fatal heart attack or non-fatal stroke "in adults with type 2 diabetes mellitus who are at high risk for these events."
As of September 23, 2026, the day this article was written, the label on DailyMed reads "Revised: 1/2026" and covers Rybelsus (3, 7 and 14 mg) plus a newer formulation sold as Ozempic tablets (1.5, 4 and 9 mg). Both carry the heart indication, though SOUL itself tested Rybelsus, and the label says the two aren't interchangeable milligram for milligram. The same heart-risk wording reached Mounjaro (tirzepatide) in an August 27, 2026 letter based on SURPASS-CVOT, which compared it with another GLP-1 medicine rather than placebo.
What SOUL can't say about compounded semaglutide
SOUL's answer belongs to the product it tested, a branded tablet whose absorption depends on its own formula and a fasting routine. Through Promise, semaglutide is dispensed as a compounded medication, which is different from an FDA-approved product: the formulation offered here is not FDA-approved, and compounded medications aren't reviewed by the FDA for safety, effectiveness or quality. Promise's product page lists it only as a compounded injectable. SOUL did not test that preparation, so its heart numbers don't carry over. A licensed provider may still prescribe a compounded formulation; that decision is between you and your doctor.
How a prescriber reads SOUL
For someone who resembles the trial (over 50, type 2 diabetes, heart or kidney disease), SOUL is solid evidence about one branded tablet. It isn't a personal forecast, and it says little about people without that disease. The practical questions: which product is actually on the table, and how does it fit with medicines already doing heart and kidney work?
SOUL is also one trial among several: SELECT tested the weekly 2.4 mg injection in people with heart disease and overweight but no diabetes. The tablet's history is on our Rybelsus approval page, and how oral semaglutide works covers the pill itself.
At Promise, a licensed provider reviews every request, and not everyone qualifies. The prescriber sets any dose, and SOUL is one piece of evidence in that conversation.
