Statin use and diabetes. Is that a real problem?

Another article evaluated and confirmed the risk of diabetes in those on statins (see doi.org/10.1136/ bmjdrc-2017-000438 ). This one focused on patients who were at higher baseline risk of diabetes.

Details:

--the Diabetes Prevention Program (DPP) studied patients with glucose intolerance (and therefore at high risk of developing diabetes), finding overall that lifestyle changes was the most effective strategy, more so than metformin

--the Diabetes Prevention Program Outcomes Study of 3234 persons: mean age 50 and 20% >60yo, 50% white/20% African-American/15% Hispanic, BMI 34, FPG 107, 2-hr glucose after 75g glusose load was 165, insulinogenic index (a ratio of insulin vs glucose increase after the glucose load, a measure of insulin secretion) 109 (nonstatin) vs 97 (statin), fasting insulin 24, A1c 6, LDL 116 (nonstatin group) vs 139 (statin), HDL 45, TG 130 (nonstatin) vs 160 (statin), hypertension 24% (nonstatin) vs 35% (statin), CVD history 0.8% (nonstatin) vs 3% (statin), family history of diabetes 70%

--statin use was infrequent at baseline, and over the study was mostly simvastatin (40%) and atorvastatin (37%), and increased use was substantially higher after a diagnosis of diabetes. and those on statins were more likely to be older, male, but no difference by race/ethnicity

Results:

--at 10 years, the cumulative incidence of statin initiation prior to diabetes diagnosis was 33-37%, a  non-significant different among the different treatment groups in DPP

--statin use was associated with a 36% increased risk of diabetes, independent of the DPP treatment group, HR 1.36 (1.17-1.58)

    --no difference by baseline diabetes risk factors or potential confounders related to indications for statin therapy

    ​--in the fully adjusted model (which includes all of the likely suspects, as well as indications for statin use and updated diabetes risk factors), those developing diabetes associated with statin use at visit prior to developing diabetes (nonsignificant differences between groups):    

        --overall: 27%, HR 1.20 (1.08-1.50)

        --placebo: 20%, HR 1.20 (0.90-1.59)

        ​--lifestyle: 43%, HR 1.43 (1.06-1.94)

--the mean change in A1C levels were:

   --placebo: -.01 in placebo vs +0.04 in statin (p<0.01), difference of 0.05 percentage points

   ​--metformin: -.09 in placebo vs -0.03 in statin (p<0.01), difference of 0.06

    --lifestyle: -0.11 in placebo vs -0.05 in statin (p<0.001)​, difference of 0.06

--there was a general trend to increasing diabetes risk with length of statin therapy (excluding those patients on baseline statins), though this was only significant for those in the lifestyle group, p=0.007

--no difference in diabetes by statin potency: those in the "high potency" group (simvastatin, atorvastatin, rosuvastatin, cerivastatin) vs “low potency” (pravastatin, lovastatin, fluvastatin), and no difference by magnitude of LDL reduction (no data shown) [but, they did not look specifically at doses of the statins, or the LDL levels achieved by the different statins, which, as per prior blogs and especially with the TNT trial of atorvastatin 10 vs 80mg found either dose to be the equally efficacious in preventing cardiovascular events if achieving similar LDL levels, see blog https://mail.google.com/mail/u/0/#search/tnt+trial/14dd343ccc2d4bab ]. hence, my reluctance to accept the high vs low potency monikers, but instead look at the actual achieved LDL as the goal.

Commentary:

--although there are somewhat mixed results from the individual studies done to date, the general finding is that there is a 10-12% increased diabetes risk by using statins. (I should comment here that many people, including myself, initially thought that statins would prevent diabetes, given their being anti-inflammatory, anti-oxidative stress, and pro-endothelial function; it is always humbling to find that a very plausible conceptual mechanism can be wrong and reinforces yet again the need for well-designed studies.  Same for homocysteine, etc etc)

--this study was important in that it showed not just the expected finding of statin- associated diabetes, but suggested that there was a gradient: those at highest risk of developing diabetes (pre-diabetics) have an increased risk of diabetes development with statins.

--of course, this is still an observational study, not one where prediabetics were randomized to statin vs placebo, so it is only exploratory and not definitive: ie, observational studies can only show relationships and not determine cause. Though a few studies are notable

    -- http://gmodestmedblogs.blogspot.com/2016/06/statins-adverse-effects-again.html / is a blog which assesses the evidence for statins and myalgias, diabetes, and dementia

    -- http://gmodestmedblogs.blogspot.com/2015/09/low-ldl-and-diabetes-risk.html  assesses an article based on the Framingham Study Offspring study, finding that low LDL itself is associated with diabetes (ie, it is the low LDL levels themselves that are associated with diabetes, and perhaps the statins are just so good at lowering the LDL that the statins appear to be associated with the diabetes. this blog also comments that patients with variants of the HMG Co-A reductase gene have increased diabetes risk; these variants are associated with increased LDL receptors and lower LDL levels). And, in the above meta-analysis it would be useful to know if those with pre-diabetes who had a lower LDL not on statins were at higher risk of diabetes than the other high-risk diabetic patients, to see if these genetic variants perhaps came into play: ie, assess the diabetes risk based on their LDL levels independent of whether patients were on statins or not)

--and, many of these patients, being high-risk for diabetes, were more likely to cross the threshold for defining diabetes than lower risk groups. the real difference in A1c measurements (about 0.06 percentage points) suggests that on average, this pretty small (and ?? clinically really significant) difference in A1c levels may therefore translate into pretty large appearing differences in the "diabetes" diagnoses (36%) just because they crossed our imposed cutpoint defining diabetes. and people put on statins did have somewhat higher baseline blood sugar levels, making them more likely to get the "diabetes" diagnosis with these small further changes in blood sugars (the issue here is that we have an artificial cutpoint, eg A1c=6.5, for defining diabetes, yielding an all-or-none association. but the real difference of only .06 percentage points which catapults many of these high risk patients above this threshold and then getting the diagnosis of "diabetes" may really not be that clinically meaningful and would greatly exaggerate the association)

                                                                                                                                                                                                                                                        

 so, this study does add to most prior studies finding an increased risk for the development of diabetes, but i would suggest the following:

--the increase in diabetes may be more formal than real: the small increase in A1c levels  are likely only minimally clinically significant (the human body is more attuned to the physiologic changes than imposed human cutpoints in our definitions. and, by the way, the cutpoint of an A1c of 6.5 reflects the inflection point of a slight increase in retinopathy, which still is only a little more than that of 6.4). 

--the data on the positive effect of statin use in diabetes is pretty robust: studies finding 30% decreased relative risk of clinical cardiac events (and, since the absolute risk of heart disease is higher in diabetics, with 80% dying from atherosclerotic disease, the numbers of prevented cardiovascular outcomes is quite high, dwarfing the potential effects of statin-induced small glucose increases).

--American Diabetes Association recommendations: for patients aged 40-75 and low cardiovascular risk, "moderate-intensity  statin treatment should be considered in addition to lifestyle therapy", independent of their lipid levels [i do see occasional diabetic patients with quite low LDL levels, in the 70-80 range, where i am reluctant to use statins. And there are zero data showing benefit of statins for these patients, though there are data showing that lowering LDL from 100 to 70 is beneficial.  so, i still use LDL around 70 as my goal]

so, bottom line, statin benefits overall dramatically outweigh their risks for the vast majority of diabetic patients. and, though it is really important to keep advocating for lifestyle changes, we clinicians (i think) should be strongly encouraging our diabetic patients to take statins.  and, in terms of mortality benefit, for those patients who are reluctant to take yet another medication, i personally would usually prefer decreasing one of their other meds (including their 3rd or 4th diabetes med) than the statin, and i give my patients the choice but with my pretty strong recommendation.