Metformin decreases dementia risk

 A retrospective analysis of veterans found that metformin vs sulfonylureas was associated with a lower risk of subsequent dementia (see doi.​org/​10.​1212/​WNL.​0000000000004586 ).

Details:

--US veterans >65 yo with type 2 diabetes who were new users of metformin or a sulfonylurea, with no baseline dementia and without diabetes complications, were accessed through the VA clinical and administrative databases, with linkage to Medicare to help capture comorbidities

--mean age 73.5 years, mean A1c=6.8%. average follow-up 5 years; 10% black/87% white, 40% overweight/50% obese,

--more on sulfonylureas were older, normal weight, eGFR <60, CAD, heart failure, atrial fibrillation, PAD, TIA/stroke, cancer; and more on metformin had A1c <7%, hyperlipidemia, substance abuse

--analysis was done by propensity match scoring to decrease risk of confounding (this is a statistical method to attempt to equalize the groups by comorbidities, etc, to make them more alike and comparable, and decrease problem of confounding by indication). this methodology reduced the absolute standard difference between the groups to <10%, very important since the comorbidities differed so much between the groups

--17,200 new users of metformin vs 11,440 new users of sulfonylureas

Results:

--4,906 cases of dementia diagnosed; 2,177 (12.7%) in metformin users and 2,729 (23.9%) in sulfonylurea users. Minimal cognitive impairment (MCI) in 6.8% on metformin and 8.9% on sulfonylurea

--crude hazards ratio for dementia in those with metformin vs sulfonylurea:

    --33% lower, HR 0.67 (0.61-0.73) in those <75 yo

    --22% lower, HR 0.78 (0.72-0.83) in those >75 yo

--propensity-scored ratio for dementia in those with metformin vs sulfonylurea:

    --11% lower, HR 0.89 (0.79-0.99) in those <75 yo

    --4% lower​, HR 0.96 (0.87-1.05) in those >75 yo, nonsignificant

--subset analysis

    --24% lower risk in younger veterans with HgbA1c>7%, HR 0.76 (0.63-0.91)

    --14% lower risk in those with good renal function (eGFR>60), HR 0.86 (0.76-0.97)

    --13% lower risk in white veterans, HR 0.87 (0.77-0.99)

    ​--nonsignificant 10% lower risk in those >75yo and with A1c>7%

--use of insulin and TZDs were also higher in those on sulfonylureas (12.5%) vs metformin (3.5%). but, no change in outcome if exclude these patients

Commentary:

--dementia is very common in diabetics, up to 60% develop dementia. ?from some combo of hyperglycemia, hyperinsulinemia, increased oxidative stress, inflammation

--prior studies have suggested that metformin enhances neuronal survival, perhaps by activating the mTOR pathway and suppressing tau hyperphosphorylation and inflammation. 

--one potential confounder is he need to control for renal failure, since this would decrease treatment with metformin and probably lead to higher likelihood of dementia. this VA study did start in 2001, when sulfonylureas were the most commonly prescribed med (75%), but by 2012 it was metformin (87%). the ADA recommended metformin as first-line in 2006, though with stricter guidelines than now in terms of avoiding metformin in those with renal failure

--the difference in metformin-associated dementia in young vs old is unclear (none had baseline diagnosed dementia or MCI).  ??if the  pathophysiology of dementia is different in younger vs older patients?  ??if clinicians more likely to note dementia or MCI in younger patients, given this is a less typical/common group for dementia.

--of note, those with higher baseline A1c (and more at risk for dementia and other diabetic complications) benefited more from metformin

--the major weaknesses of this large study were that it was observational (so not able to attribute causality, just association), its reliance on a large database which might have had documentation errors, there was no systematic assessment of dementia/MCI, no women were in the study, residual confounding was not accounted for (eg use of other meds: anticholinergics, etc; use of illicit drugs; other comorbidities such as falls; baseline educational status; lifestyle issues; etc);  and the strengths of the study were the large size and that they really tried to look at people with relatively recent diabetes (excluding those with complications) and new to medications.

--a recent small study of 80 patients aged 55-90 who were overweight/obese with amnesic MCI but not on meds for diabetes (the exclusion criteria did change in the course of the study, but pretty much everyone was excluded who had A1c >6.5%, with some allowed up to 6.9% but not on meds), were put on metformin (most on around 1000mg qd) vs placebo for 12 months, finding improved memory recall in the metformin group. in this study those on metformin had a decrease in fasting insulin levels but no significant change in A1c levels (see Luchsinger JA. J Alzheimers Dis. 2016 ; 51(2): 501–514). 

So, this is yet another study promoting the positive effects of metformin, especially taken together with with the Luchsinger study.  Metformin has been the first-line diabetes med for a long time, with the added advantage of cardioprotection (which may be more important clinically than its impressive effects on A1c levels). And, as with several meds (statins, many antihypertensives), there is a large medication effect at very low doses (1000 mg daily is around 80+% as effective as 1000mg bid; and my experience is that 500mg daily is often quite sufficient for diabetes control). perhaps the main takeaway here, at least for me, is that it might be prudent to prescribe metformin earlier than i currently do.  prior studies, such as the Diabetes Prevention Trial, have found that lifestyle changes were more important than metformin in preventing the development of diabetes in a high-risk population. and lifesytle changes certainly are important, and in fact pivotal, in good treatment and prevention of diabetes. but these studies on dementia, though not completely conclusive (not large-scale rigorous RCTs), do promote earlier treatment with metformin.  the potential benefits of decreasing dementia likely overwhelm the real small risks of using metformin....

For older blogs on metformin, go to  http://gmodestmedblogs.blogspot.com/search/label/metformin . there are many prior blogs there, dealing with metformin prescribing in patients with renal, liver, heart disease.

http://gmodestmedblogs.blogspot.com/2014/10/heart-failure-microbiome.html  highlights a study showing that metformin seems to increase a specific bacterium in the gut microbiome (Akkermansia), which by itself improves glucose tolerance.

to be added to the regular email list, please contact me at gmodest@uphams.org