Loosening A1c goal??? is low A1c really the problem???

The American College of Physicians just released a "Guidance Statement Update" revising their A1c targets for nonpregnant adults with type 2 diabetes (see dm a1c target AIM2018 in dropbox, or

http://annals.org/aim/fullarticle/2674121/hemoglobin-1c-targets-glycemic-control-pharmacologic-therapy-nonpregnant-adults-type .

Details:
--Guidance statement 1: personalize glycemic goals on the basis of a discussion of benefits and harms of pharmacotherapy, patient preferences, patient general health/life expectancy, treatment burden, and costs
--Guidance statement 2: aim to achieve A1c goal of 7-8% in most patients
--Guidance statement 3: consider de-intensifying drug therapy in those with A1c <6.5%
--Guidance statement 4: treat patients to minimize symptoms of hypoglycemia, and "avoid targeting an A1c level in patients with a life expectancy less than 10 years due to advanced age (80 years or older), residence in a nursing home, or chronic conditions (such as dementia, cancer, ESRD, or severe COPD or CHF) because the harms outweigh the benefits in this population"

Commentary:
--the basis of these guidelines is their review of "5 large, long-term RCTs comparing intensive glucose management (A1c 6.3-7.4%) vs less intensive (7.3-8.4%) in adults with baseline age of 53-66 yo”. they found that the main effect of intensive control is "small reductions in the risk for microvascular surrogate events", ie retinopathy or nephropathy
--will briefly review these studies, though my strong bias (as will become evident) is that the argument for lower A1c targets are reasonable: the results are pretty consistent for microvascular problems (though mostly surrogate markers, such as retinopathy or albuminuria).  The over-riding issue, however, is that diabetics die from macrovascular disease (80+%). And many diabetes drugs that make the A1c look good may actually increase cardiovascular disease:
    --rosiglitazone: the data are quite clear showing its adverse cardiovascular effects
    --insulin: most studies have also found either no benefit or actual cardiovascular harm from insulin
        --many epidemiologic studies have found that increased insulin levels, independent of blood sugar, are associated with increased cardiovasc disease; eg see http://circ.ahajournals.org/content/97/10/996 . Insulin has many potentially adverse/cardiotoxic effects: studies have found that it can stimulate platelet adhesion, can be prothrombotic (a major regulator of plasminogen activator inhibitor 1, PAI-1), pro-inflammatory, associated with endothelial dysfunction, even induces HMG-CoA reductase activity (the enzyme blocked by statins)
        --and, exogenously administered insulin can lead to hyperinsulinemia (which can be worse with the combo of insulin and long-acting sulfonylureas)
    --sulfonylureas can also increase cardiovascular disease, esp in comparison to metformin.
        --dm metf betterthan SU in ckd jgim2018 in dropbox, or Marcum ZA J Gen Intern Med 2018; 33(2):155–65) which accessed the VA database and found that in those with CKD, initiating metformin vs sulfonylurea led to 20-41% reduction of mortality, depending on the eGFR range
        --dm SU in women inc CAD diabcare2014 in dropbox, or DOI: 10.2337/dc14-1306 which reviews the Nurses’ Health Study, finding that the longer the use of sulfonylureas, the higher the risk of cardiovascular disease (3.27-fold risk in those on combo metformin plus sulfonylurea vs just metformin)

--dm dec mortal metf vs sulf diabcare2002 in dropbox, or Johnson JA. Diabetes Care. 2002; 25:2244–2248, for review of Saskatchewan Health database of 12,272 new users of oral diabetic meds, finding that after 5 years, there was a 40% decreased mortality in those on metformin monotherapy vs sulfonylurea monotherapy, and a 34% decreased all-cause mortality in those on the combo metformin plus sulfonylurea vs sulfonylurea monotherapy.

--so, a recurrent theme in the following review is that the intensive group with "better" A1c levels are typically on more of these not-so-heart-friendly drugs. ie, the problem may well be the drugs used in the studies and not the A1c achieved. And, the background epidemiological studies suggest that there is an increased cardiovascular risk with glucose intolerance, on the order of an 80% increase (a couple finding that even and A1c in the 5.5-6 range is associated with increased cardiac events in men in particular), and that this increased cardiovascular risk becomes more profound with increasing A1c levels (eg, see ARIC study: Selvin, NEJM 2010)

The trials on which the ACP recommendations are based:
--ACCORD trial (n=10,251): mean age 62, diabetes for 10yrs, baseline A1c=8.1%. achieved A1c of 6.4% in the intensive group vs 7.5%. trial terminated early after 3.5 years because the intensive group had: 22% increase in all-cause mortality, 35% increase in cardiac mortality, and 3-fold increase in severe hypoglycemia (though there were fewer nonfatal MIs). BUT:
    --those in the intensive group were on more rosiglitazone (91% vs 58% on TZDs, and almost all were rosiglitazone), or on insulin (77% vs 55%)
    --AND, post-hoc analysis showed that the difference in mortality was explained by which group the patient was assigned to (intensive vs standard). The mortality benefit was actually a continuous one, increasing across the A1c range of 6-9 (see dm ACCORD posthoc analysis diabcare 2010 in dropbox, or Riddle MC. Diabetes Care 2019; 33:983).  ie, it was not the A1c achieved so much as that adding more and more meds in the attempt to lower the A1c seemed to be associated with the bad outcomes
--ADVANCE trial (n=11,140): mean age 66, diabetes for 8yrs, baseline A1c=7.5%. achieved A1c of 6.5% vs 7.3%. Duration of trial 5 years. No reduction in major macrovascular events, all-cause mortality or cardiovasc deaths (all showed trend to being better with intensive therapy). but 14% benefit on microvascular events. 2-fold increase in severe hypoglycemia in intensive group
    --intensive group: 93% on a sulfonylurea vs 59% in control group;  insulin in 40% intensive and 24% nonintensive, TZD in 17% vs 11% (no mention of which one used). so, again, intensive therapy relied on potentially cardiotoxic meds
--UKPDS trial 33 and 34 (n=3867 and 753):  I will not go into detail/pretty complex studies to describe, and they had a more lax methodology and used older and no longer available meds, such as chlorpropamide (very long-acting sulfonylurea). But the gist of the studies: in the first one, 10 years of intensive control led to 12% reduction of any diabetes-related endpoint (5.1 events per 1000 patient-yrs), largely due to reduction in composite of microvascular outcomes. no difference in diabetes-related deaths, all-cause mortality, MI, stroke or amputation. These patients were folded into the second study (which was a metformin study in obese diabetics), finding that overall those on metformin had a 32% relative risk reduction for any diabetes-related endpoint, 42% reduction for diabetes-related death, and 36% reduction in all-cause mortality (absolute reductions of 5 fewer diabetes-related deaths and 7 fewer all-cause deaths per 1000 patient-yrs). and these reductions were greater in those on metformin than if on intensive therapy with sulfonylureas or insulin
    --these UKPDS trials were the first ones to find a significant macrovascular and death benefit from metformin, even on extended follow-up, so were important landmark studies. (ie: giving metformin, a heart-friendly drug, was beneficial). But they do not add significantly to our understanding of diabetes treatment otherwise, given older drugs no longer used and a less strict methodology.
--VADT trial (n=1791): mean age 60, diabetes for 11.5 yrs, baseline A1c=9.4, achieved Aic of 6.9% vs 8.4%. followed 5.6 years: intensive care associated with decrease in albuminuria but no difference in major cardiovascular events or deaths.    
    --baseline use of insulin 52%, increased to 89% in intensive vs 74% nonintensive. also increase in TZDs (53 vs 42%), and 7.8 vs 3.4 kg increase in weight
    --Rosiglitazone was the TZD used

so, to me the issue here is a paradigm shift in diabetes management:
--i do think the data are pretty robust that lowering the A1c does decrease surrogate markers for microvascular complications, which is very important to do even if they do not often lead to death
--but the big issue is what meds are being used. it is clear that insulin, and to lesser extent sulfonylureas, lead to more hypoglycemia, especially if one is aiming for a low A1c level.  and, i think the ACCORD trial shows that lowering A1c is good, but not in the climate of aggressively flogging patients with meds (especially non-heart friendly ones). And several of the studies above used lots of the cardiotoxic rosiglitazone.
--so, my approach has evolved to:
    --using metformin as first line (not much controversy here)
    --trying to use a more cardioprotective 2nd line drug (ie, my ideal combo is metformin plus GLP-1 agonist)
        --pioglitazone was cardioprotective in one study (PROactive), but not another (TOSCA-IT)
        --as noted in many of my blogs, i really like the GLP-1 agonists, because:
            --they target a specific normal "incretin" which stimulates glucose-mediated insulin release but is decreased in diabetics, the GLP-1 agonist injections have dramatic effects on blood sugar control (i have several patients with A1c in the 10+ range on insulin, who get down to the 6-7 range often with decreasing and some not even using insulin), they don't cause hypoglycemia, patients lose weight, and they are well-tolerated other than some GI symptoms which often resolve. i have had few patients who decline any injections, though i try to be clear that this is NOT insulin (many fear that), that it protects the heart (at least the somewhat flawed liraglutide study found that), that it can be a once a week injection with a small and pretty painless needle. see http://gmodestmedblogs.blogspot.com/2017/08/liraglutide-new-indication-for.html  for liraglutide blog
            ​--though DPP-4 inhibitors do increase GLP-1 levels, they block a ubiquitous enzyme system (DPP-4) and deactivate many different bioactive peptides, (ie they do not have a specific target), they do not affect A1c levels much, and they may actually increase hospitalizations for heart failure (see http://gmodestmedblogs.blogspot.com/2015/06/dpp-4-inhibitors-and-cardiovascular.html  ) and a safety alert by the FDA (see http://gmodestmedblogs.blogspot.com/2016/04/diabetes-dpp-4-inhibitors-and-risk-of.html  ). 
            --and i have real concerns about the SGLT-2 inhibitors, which not only have really bad adverse reactions (urosepsis, diabetic ketoacidosis with some blood sugars <200, increased amputations/fractures (eg see hhttp://gmodestmedblogs.blogspot.com/2016/05/another-fda-alert-about.html  ) and do not target a specific, deficient diabetes pathway.
                    -- in the emapagliflozin and canagliflozin studies there was increase in TZD use, but it is not mentioned which one (even after my scouring the supplementary material). see http://gmodestmedblogs.blogspot.com/2016/06/canagliflozin-decreases-macrovasc.html  for a critique of canagliflozin and http://gmodestmedblogs.blogspot.com/2015/12/empagliflozin-good-and-bad.html ​ for empagliflozin.  Also increased insulin and sulfonylurea use in the control groups
--and, bottom line, i am very hesitant to change the A1c goal as per the ACP suggestions, since these recommendations were based upon old studies involving intensified control with meds that were non-cardioprotective and some harmful. The issue is not simply the A1c, but what meds are used to lower it: I would elevate the issue of using heart-friendly meds as a primary consideration.

so, my reiteration of the recommendations would be more like:
    --Guidance statement 1: personalize glycemic goals on the basis of a discussion of benefits and harms of pharmacotherapy, patient preferences, patient general health/life expectancy, treatment burden, and costs [no different from theirs]
    --Guidance statement 2: aim to achieve A1c goal of 6.5-7 if can be done with cardioprotective meds (especially metformin, then a  GLP-1 agonist). and always work with the patient to optimize nonpharmacologic management, specifically low glycemic index diet, weight management, and lots of exercise. if this goal can be achieved with the addition of small amounts of other meds, that is reasonable. if not, a goal in the 7-8% range as per the ACP is probably fine (basically, avoid flogging patients with meds which are not cardioprotective)
    --Guidance statement 3: would consider de-intensifying meds if A1c <6.5 (as per ACP), but not if the patient were young, on cardioprotective meds, and tolerating them well.  i do have several people on metformin and GLP-1 agonist who had  suboptimal responses to metformin alone, but with the combo get A1c in the low 6 range without a problem.  and i leave them there. the perspective here is that A1c in the high 5/low 6 range, especially in men, is associated with an 80% increase in cardiovasc mortality, and this increases as the A1c increases.  so since they are tolerating the meds, why not leave the patients on them since it might decrease cardiavasc events even further??? untested but quite possible
    --Guidance statement 4: this guideline is an offshoot of the first one. i totally agree with personalizing goals (where age, life expectancy and comorbidities are taken into account), and avoiding  hypoglycemia (which doesn't happen with metformin or GLP-1's)