Five-year GRADE trial finds liraglutide and insulin glargine hold A1C best when added to metformin

In the GRADE trial, 5,047 US adults with type 2 diabetes were randomly assigned to add one of four drugs to the metformin they already took: insulin glargine U-100, a long-acting insulin; glimepiride, a sulfonylurea pill; liraglutide, a GLP-1 receptor agonist given by injection; or sitagliptin, a DPP-4 inhibitor pill ^[1]. Participants had been diagnosed with type 2 diabetes less than 10 years earlier and started with A1C levels between 6.8% and 8.5% ^[1]. They were followed for a mean of 5.0 years; 19.8% were Black and 18.6% were Hispanic or Latinx ^[1]. The trial was funded by the National Institute of Diabetes and Digestive and Kidney Diseases and others, and registered as NCT01794143 ^[1]. Results were published September 22, 2022.

The main question was how long each drug could keep A1C below 7.0%. A1C was measured quarterly, and the primary outcome was a reading of 7.0% or higher that was later confirmed ^[1]. That happened at a rate of 26.5 per 100 participant-years with glargine and 26.1 with liraglutide — close to each other and lower than glimepiride at 30.4 and sitagliptin at 38.1 ^[1]. The difference across the four groups was statistically significant (P<0.001) ^[1]. Results for the secondary outcome, a confirmed A1C above 7.5%, followed the same pattern ^[1]. The trial found no material differences by sex, age or race and ethnic group, though among participants who started with higher A1C levels, glargine, liraglutide and glimepiride appeared to have an even larger advantage over sitagliptin ^[1].

Side effects differed. Severe hypoglycemia was rare in every group but significantly more frequent with glimepiride, occurring in 2.2% of those participants, compared with 1.3% with glargine, 1.0% with liraglutide and 0.7% with sitagliptin ^[1]. Participants taking liraglutide reported more gastrointestinal side effects — nausea, vomiting, bloating or stomach pain, or diarrhea at least once a week in the 30 days before a quarterly visit — and lost more weight than those in any other group ^[1].

A companion report tracked longer-term complications. Over the same mean 5.0 years, there were no material differences among the four groups in high blood pressure, abnormal cholesterol, kidney measures such as albuminuria and estimated filtration rate, or diabetic peripheral neuropathy, which occurred at an overall rate of 16.7 events per 100 participant-years ^[2]. The groups also did not differ in major adverse cardiovascular events (overall rate 1.0), hospitalization for heart failure (0.4), death from cardiovascular causes (0.3) or death from any cause (0.6) ^[2]. For any cardiovascular disease, rates were 1.9 with glargine, 1.9 with glimepiride, 1.4 with liraglutide and 2.0 with sitagliptin; compared with the other three treatments combined, the hazard ratio was 0.7 (95% CI, 0.6 to 0.9) for liraglutide and 1.2 (95% CI, 1.0 to 1.5) for sitagliptin ^[2]. The authors called those possible differences and noted the hazard ratios were not adjusted for multiple comparisons ^[2].

Why it matters for patients

The comparison addresses a common decision. The comparative effectiveness of glucose-lowering drugs added to metformin for keeping A1C at target had been uncertain ^[1]. GRADE put four options through the same five-year test in the same population, including a GLP-1 receptor agonist of the same drug class as semaglutide.

The practical trade-offs showed up in different places. Liraglutide and glargine each held A1C under 7% better than glimepiride or sitagliptin, and liraglutide came with the most weight loss — but also the most gastrointestinal complaints ^[1]. Glimepiride was the only drug with a significantly higher rate of severe hypoglycemia ^[1]. Sitagliptin had the lowest rate of severe hypoglycemia, at 0.7%, but the highest rate of losing A1C control, at 38.1 per 100 participant-years ^[1].

The five-year complication results cut both ways. Nothing separated the four drugs on heart attacks, heart failure, kidney or nerve outcomes, or death, which means the choice in this trial came down to glucose control, weight and side effects rather than proven protection against complications ^[1][2]. The possible cardiovascular signal favoring liraglutide was described by the authors as tentative, not established ^[2].

What the results do not cover

The trial enrolled adults with type 2 diabetes of less than 10 years' duration who were on metformin with A1C between 6.8% and 8.5%, so its findings describe that group ^[1]. It did not test semaglutide (Ozempic, Wegovy, Rybelsus) or tirzepatide (Mounjaro, Zepbound), and no comparison with those drugs has been reported here. The trial also did not compare costs, and average weight-loss amounts for each group were not disclosed in the published results.