Liraglutide and Insulin Come Out on Top for Glycemic Control in GRADE study

When choosing a second therapy to add to metformin, certain antidiabetic agents edged out others, according to findings from the GRADE study.

The head-to-head comparison of four classes of medications assessed multiple metabolic, cardiovascular (CV), and safety outcomes for sulfonylurea glimepiride (Amaryl), the DPP-4 inhibitor sitagliptin (Januvia), insulin glargine (Lantus, Basaglar, Toujeo), and the GLP-1 receptor agonist liraglutide (Victoza) in adults with type 2 diabetes.

When added to pre-existing metformin, insulin glargine and liraglutide performed the best for glycemic control, with the lowest number of patients developing an HbA1c at or above the target range of 7%, which was the study's primary endpoint:

• Insulin glargine: 67%
• Liraglutide: 68%
• Glimepiride: 72%
• Sitagliptin: 77%

Compared with insulin glargine or liraglutide, glimepiride and sitagliptin performed significantly worse at maintaining HbA1c levels under 7%. Sitagliptin also performed significantly worse than glimepiride in maintaining glucose control, according to David Nathan, MD, of Massachusetts General Hospital in Boston. Preliminary findings of the long-term study were presented at the American Diabetes Association (ADA) virtual meeting.

Looking at the first year of combination treatment, those on liraglutide had the tightest glucose control, ending month 12 with an HbA1c of 6.7%. By the end of the year 4 of treatment, those on insulin glargine and liraglutide maintained the lowest average HbA1c of the group.

For timing, those on sitagliptin had the shortest time to seeing an HbA1c reading about target range at 697 days (1.9 years). Those on glimepiride had an average time to an A1c at or above 7% at 810 days (2.2), followed by insulin glargine at 861 days (2.4 years), and liraglutide at 882 days (2.4 years).

Of note, the majority of participants at some point had an HbA1c of 7% or higher (71%).

Similar patterns were seen for the study's secondary metabolic outcome, the number of patients who developed an HbA1c of 7.5% or higher. Significantly fewer patients on insulin and liraglutide progressed to this, representing only 39% and 46% of the groups, respectively. However, insulin was significantly better than liraglutide at keeping levels below this threshold. Half of those on glimepiride and 55% of those on sitagliptin developed an A1c of 7.5% or higher at some point during follow-up.

"I think one of the most prominent findings of GRADE was the rate at which A1cs rose -- I mean we're really not at a point with the drugs that we have, including combination therapy, where we know how to reduce what appears to be almost an inexorable rise in A1c over time," Nathan emphasized.

The results are long-awaited as planning for the multicenter project began over a decade ago. After securing funding, screening for the study started in 2013, with recruitment lasting nearly 4 years. Data collection wrapped up in May 2021 so about 10% of the reported CV events during the study have been adjudicated, leading the researchers to stress the "preliminary" nature of these findings.

SGLT-2 Inhibitors

A major limitation of the study is the lack of inclusion of SGLT-2 inhibitors, noted David R. Matthews, DPhil, BM, BCh, of the University of Oxford in England. He said the researchers should be kicking themselves for that omission in hindsight. Matthews is not involved in the study.

Nathan explained GRADE only included medications that were FDA approved, and with which there was sufficient clinical experience, in order to avoid any potential safety concerns, as was seen with the thiazolidinedione class of agents. When GRADE was in the planning stages, SGLT-2 inhibitors were very new, with canagliflozin (Invokana) first approved in 2013, he stated.

"Frankly, if we had added another drug -- whether it was a [thiazolidinedione] or SGLT-2 inhibitor -- it would've increased our comparisons to five groups...the size of the study would have doubled and we probably would have had to delay the study for a year or 2 years," Nathan said.

Matthews highlighted that some of the key takeaways of GRADE included the fact that glimepiride tended to work well for glucose control early on, but later had a high failure rate. Sitagliptin also didn't perform particularly well in patients with an HbA1c of 7.3% or higher on metformin alone. However, he praised liraglutide for both its CV protection and weight loss benefit, calling the weight loss seen with this agent "iconic."

More GRADE Details

Beyond glycemic control, the researchers looked at between-group differences in weight loss, and liraglutide did offer the biggest weight-related benefit, with patients losing an average of 8.8 lbs (4 kg) after the first year of treatment. Those on sitagliptin also lost weight over the course of 4 years of treatment. Those on insulin glargine did not gain weight and remained stable through treatment.

As for CV outcomes, there wasn't a statistically significant difference between the four classes of medications for major CV event, hospitalization for heart failure, and all-cause mortality. However, liraglutide trended towards fewer events versus the other three agents. Liraglutide was significantly better than sitagliptin when it came to the development of any CV disease.

For microvascular outcomes, there were no significant differences seen between the groups for moderately or severely increased albuminuria; development of an eGFR below 60 mL/min/1.73 m²; or distal sensory polyneuropathy.

For safety, there were no significant differences between the groups in terms of serious adverse events or the rate of pancreatitis or pancreatic cancer. However, severe hypoglycemia was slightly higher with glimepiride, but not significantly different. Not surprisingly, liraglutide patients experienced a higher rate of gastrointestinal side effects that are commonly seen with GLP-1 receptor agonists.

The total GRADE cohort included 5,000 U.S. adults (average age 57) with type 2 diabetes who were already on 1,000 to 2,000 mg of immediate-release or extended-release metformin at the time of randomization.

The cohort was fairly diverse, with 66% being white, 20% Black, 4% Asian, and 3% American Indian/Alaska Native. Of this, about 19% were ethnically Hispanic/Latino. About 64% were male, with an average baseline BMI of 34 and HbA1c of 7.5%.

Basal Insulin

ADA symposium co-chair Julio Rosenstock, MD, of the Dallas Diabetes and Endocrine Center, said that a pleasant surprise to come out of GRADE was how well basal insulin performed, with low rates of hypoglycemia -- only 1.5% of patients experienced severe hypoglycemia -- and noteworthy weight stability.

GRADE co-investigator Steven E. Kahn, MD, of VA Puget Sound Health Care System in Seattle, pointed out that GRADE determined that basal insulin can be added "early on" to metformin. He explained that, in many parts of the world, generic drugs must be used for cost reasons, "so now we've got sulfonylureas and basal insulin -- if you can get it at a reasonable price -- that look like very good options to add to metformin."

Comment