Insulin vs GLP-1 agonists for patients failing oral med treatment

As many of you know based on my prior blogs, I have largely switched from using insulin to prescribing a GLP-1 agonist as my second line treatment for diabetes, after metformin (though I still use the short-acting sulfonylurea glipizide after metformin in patients who decline injection therapy, then try to use a GLP-1 agonist as the next step). A recent blog (see http://gmodestmedblogs.blogspot.com/2016/06/liraglutide-decreases-cardiovascular.html  ) looked at using liraglutide vs placebo in older diabetic patients at high risk of cardiovascular events, finding that the GLP-1 agonist lowered A1c by 0.4 percentage points (the other meds were increased in the placebo wing to narrow the A1c gap), was associated with more weight loss, and had a 15% mortality benefit as well as a 22% decrease in major cardiovascular events, with benefit evident within 12-18 months. A recent retrospective review of the UK Health Improvement Network (10.5 million patients in 532 general practices) found benefit of adding a GLP-1 agonists over insulin (see dm glp1 dec adv rxn heart2016 in dropbox, or doi:10.1136/heartjnl-2015-309164). details:

--2003 patients were treated with either insulin or a GLP-1 agonist after failure of metformin/sulfonylurea dual-therapy (419 patients on MET+SU+insulin, and 1584 on MET+SU+GLP-1 agonist), with 5 years of follow-up

--mean age 53, 50% male, no difference in socio-economic status by Townsend deprivation index, but significant differences comparing insulin vs GLP1 groups in mean A1c of 9.9 vs 9.4, BMI 30 vs 40, BP 133/80 vs 136/83, aspirin in 14 vs 20%, BP meds in 37 vs 51%

--results:

                --overall rates of Major Adverse Cardiovascular Events (MACE, including mortality, and nonfatal MI or stroke), using propensity-scoring to mathematically adjust for baseline differences in the groups:

                                -- MET+SU+insulin: 231, for rate of 44.5/1000 person-years

                                -- MET+SU+GLP-1 agonist: 11, for rate of 7.7/1000 person-years

--so, a 73% reduction in the GLP-1 group [HR 0.27 (0.14-0.53), p<0.0001)]

                --no difference in A1c levels

                --insulin was associated with a significant increase in weight (+1.78 kg vs -3.93 kg)

                --there was a similar MACE outcome in those with BMI>30: 84 vs 11 MACE with a 69% MACE reduction; and in those with BMI>35: 30 vs 7 events, with a similar 69% MACE reduction.

   -- in subgroup analysis of the individual MACE events, only mortality was statistically significantly different, with a 79% decrease in the GLP-1 agonist group.

Commentary:

--this was a retrospective observational study, so has the limitations of all such studies: unable to prove causation because of differences in the groups and other potential confounders. As is evident above, the group on the GLP-1 agonists was significantly more obese, had a lower A1C but also a higher blood pressure and were more likely to be on aspirin or antihypertensive therapy. However, the overall results, with attempts to mathematically control for these differences, are in agreement with two large recent meta-analyses finding cardiovascular benefit of GLP-1 agonists, though these benefits are not consistent in all studies or meta-analyses over time.

--I have been concerned for awhile about potential adverse effects of insulin, at least at a cellular level, given its potential role as a growth factor (stimulating smooth muscle migration, proliferation and inflammation) as well as its procoagulant effects (stimulating circulating tissue factor-procoagulant activity), and it may even promote HMG CoA reductase activity (the one blocked by statins). And there are some observational/epidemiological studies suggesting increased cardiovascular events with increasing insulin levels even in nondiabetics.  GLP-1 agonists, in particular, seem to me to be particularly physiologic, restoring a normal enzyme which is depressed in diabetic patients (through the "incretin effect"), which leads to food-stimulated insulin secretion without significant hypoglycemia and pretty consistently with weight loss. I consider this data as suggestive that higher insulin levels potentially could increase cardiovascular risk, do not conclusively show that relationship, but this study, as well as the liraglutide study mentioned above, do support the conclusion that GLP-1 agonists are likely more cardioprotective.

--there has been concern about GLP1-associated pancreatitis and pancreatic cancer, though more recent large studies do not confirm these events.

--this study reflects a large number of patients in different actual primary care practices throughout the UK.

--so, my sense is that there are increasing data that the efficacy of GLP-1 agonists to decrease A1c levels is pretty similar to insulin, there are not the downsides of hypoglycemia or weight gain found with insulin, and there are several studies finding important cardiovascular and mortality benefits, consistent with some of the known physiologic effects of GLP-1 agonists vs insulin. And, my personal experience is that these meds are well-tolerated overall (major adverse effect being GI, or some soreness/lumps at site especially of some long-acting exenatide injections) and really do improve glucose control.