A once-daily pill that mimics the gut hormone GLP-1 was put directly against a widely used diabetes drug from a different class, and it won on blood sugar. At its top dose it did more than tie. It beat the comparator outright.

The pill is safiglipron, the drug formerly known by its code name HRS-7535, from Jiangsu Hengrui Pharmaceuticals in China. The comparator was dapagliflozin, an SGLT2 inhibitor sold as Farxiga that lowers blood sugar by making the kidney spill extra glucose into the urine. The trial, called OUTSTAND-2, was published in Nature Medicine ↗ in early October. It randomized 810 adults in China whose type 2 diabetes was not controlled on metformin, across 98 sites, to one of three safiglipron doses (30, 60, or 90 milligrams) or to dapagliflozin 10 milligrams, for 32 weeks.

Why a head-to-head matters here

Most GLP-1 drugs that people know, the weekly shots like semaglutide ↗ (Ozempic and Wegovy) and tirzepatide ↗ (Mounjaro and Zepbound), are peptides. A peptide is a short chain of amino acids, the same building blocks proteins are made of, and the gut digests them on contact. That is why those drugs have to be injected. Safiglipron is not a peptide. It is a small molecule built to switch on the same GLP-1 receptor while surviving the stomach, so it can be swallowed.

What made OUTSTAND-2 unusual is the comparator. Drug trials usually test a new agent against a placebo, a dummy pill, because that is the lowest bar to clear and the easiest to run. OUTSTAND-2 skipped the dummy and went straight at a real competitor. By the trial authors' account it is the first double-blind Phase 3 study to put an oral small-molecule GLP-1 drug directly against an SGLT2 inhibitor, the two pill-based classes that doctors actually choose between after metformin stops working.

What the numbers say

HbA1c is the standard blood-sugar yardstick, a measure of how much sugar has stuck to hemoglobin over the prior few months. A lower number is better, and a difference of a few tenths of a percentage point is clinically meaningful at scale.

Safiglipron lowered HbA1c by 0.22 to 0.40 percentage points more than dapagliflozin did, depending on the dose. All three doses met the trial's non-inferiority goal, meaning safiglipron was at least as good as the established drug. The 90 milligram dose cleared the harder bar and passed the superiority test, meaning it was statistically better, not just no worse.

The gap shows up again in how many people hit target. Between 54.9 and 63.5 percent of safiglipron patients got their HbA1c under 7.0 percent, the common treatment goal, against 36.5 percent on dapagliflozin. For the stricter target of 6.5 percent or below, the figures were 38.9 to 48.6 percent on safiglipron versus 16.1 percent on dapagliflozin. More than three in five patients reached goal on the pill that mimics GLP-1, against barely more than one in three on the SGLT2 drug.

Weight loss, the headline people associate with GLP-1 drugs, was close to a wash. Safiglipron patients lost about 2.35 to 4.17 percent of body weight, dapagliflozin patients about 3.60 percent. Both classes trim weight, and at these doses over 32 weeks neither pulled clearly ahead.

The cost side

The blood-sugar edge came with the familiar GLP-1 tax. Gastrointestinal side effects, mostly nausea and related gut complaints, were more common on safiglipron than on dapagliflozin and climbed with dose. The share of patients who stopped the drug for adverse events ran about 3.9 to 4.0 percent on safiglipron against 1.5 percent on dapagliflozin. That is the same tradeoff injectable GLP-1 drugs carry, and it is the number a prescriber weighs against the extra blood-sugar drop when the stronger top dose is on the table.

Hengrui funded the trial and announced topline results in July. The company already has a Phase 2 record on the same molecule that this site has covered: an obesity trial where the middle dose beat the high dose on weight ↗, and a diabetic-kidney trial where it cut urine protein on top of finerenone ↗. OUTSTAND-2 is the first time the molecule has been measured against a live competitor rather than a placebo, and the answer, on glucose, is that the pill held its own and then some. Whether that translates into the harder outcomes that matter most, heart attacks, strokes, kidney decline, is a question 32 weeks of HbA1c cannot answer. Those trials are the next ask.