atorvastatin vs rosuvastatin: better for diabetes outcome??

A recent Korean study (the LODESTAR study) found that rosuvastatin was associated with lower LDL levels than atorvastatin, but rosuvastatin seemed have a higher risk for new onset diabetes requiring medications as well as cataract surgery (see statin rosuv vs atorv BMJ2023 in dropbox, or doi.org/10.1136/bmj‑2023‑075837) 

 

Details:

-- 4400 adults with clinically diagnosed coronary artery disease (stable ischemic heart disease or acute coronary syndrome) requiring statin therapy, in 12 hospitals in South Korea from 2016 to 2019, in an investigator-initiated, randomized, open label trial (i.e. not drug company sponsored)

-- mean age 65, 28% women, 14% current smoker, BMI 25

-- comorbidities: hypertension 67%, diabetes 33%, CKD 7%

-- cardiac history: percutaneous coronary intervention 56%, stroke 6%, acute MI in the last year 8%, unstable angina or revascularization in the past year 18%, MI more than one year ago 15%, unstable angina or revascularization more than a year ago 41%

    -- total of 74% of the patients had received their initial diagnosis or coronary revascularization more than one year previously

-- lipids at baseline: LDL 87 mg/dL, HDL 46 mg/dL, triglycerides 142 mg/dL

-- patients were randomized to receive either rosuvastatin or atorvastatin, dose of which was titrated to an achieved LDL level during the study: if > 70 mg/dL, then increase the statin, keep it the same if LDL was 50-70 mg/L, and decrease it if <50 mg/dL

 

-- primary outcome: three year composite of all-cause death, myocardial infarction, stroke, or any coronary revascularization

-- secondary outcome (safety endpoints): new onset diabetes (NOD), hospital admissions for heart failure, DVT/PE, endovascular revascularization for peripheral artery disease, aortic intervention or surgery, end-stage kidney disease, discontinuation of study drugs due to intolerance, cataract surgery, and a composite of laboratory-detected abnormalities

 

Results:

-- 4341 participants (99%) completed the trial

-- mean daily doses of meds at three years: rosuvastatin 17.1 mg, atorvastatin 36.0 mg

-- mean LDL during treatment: 70 mg/dL on rosuvastatin, 74 mg/dL on atorvastatin; p<0.001

    -- the use of “high-intensity statins” was not different in the 2 groups, it averaged around 72%

 

-- Primary outcome (cardiovascular composite): 189 participants (8.7%) being on rosuvastatin versus 178 participants (8.2%) on atorvastatin: HR 1.06 (0.86-1.30), p=0.58, not statistically significant

 

-- Secondary outcomes (safety):

    -- new onset diabetes requiring initiation of meds: 7.2% versus 5.3%, 39% increase, HR 1.39 (1.03-.87), p=0.03

    -- cataract surgery: 2.5% versus 1.5%, 66% increase, HR 1.66 (1.07-2.58), p=0.02

    -- no difference in the other safety endpoints measured.

 

Commentary:

-- the impetus for this study was to determine whether the clinical effects of statins was in fact a class effect of all statins or whether there were differences between individual statins, in this case atorvastatin and rosuvastatin

-- in this study, the researchers aimed for an LDL level between 50-70 mg/dL, increasing the statin dose if LDL >70mg/dL and decreasing it if <50mg/dL. Their target LDL was not fully achieved (it was 70 mg/dL for the rosuvastatin group and 74 mg/dL for atorvastatin)

    -- though it now seems that statins have increasing benefit without much increased risk even if the LDL is <20mg/dL: https://gmodestmedblogs.blogspot.com/2018/08/very-low-ldl-levels-benefit-without-harm.html . However, most of the data for this current conclusion was from studies of the really potent PCSK9 inhibitors, but this Korean study predated the PCSK9 trials

        -- it seems that the association of statins with diabetes may well be related to the HMG-CoA-Reductase function, which is the target of statin therapy, but not the target of PCSK9, or ezetimibe, as noted in https://gmodestmedblogs.blogspot.com/2023/11/pcsk9-inhibitors-vs-statins-and-diabetes.html.

-- therefore, in fact, it does make mechanistic sense that a more potent statin, such as rosuvastatin, might be associated with more glucose intolerance and diabetes than atorvastatin

-- this Korean study basically found no difference in the composite of cardiovascular and mortality outcomes between the meds, though the group on atorvastatin had more than twice the statin dose as those on rosuvastatin

    -- it is hard to get solid data on the relative LDL effect of atorvastatin vs rosuvastatin, in part because of pretty large variations in their effects in individuals. One study assessing their relative potency found that atorvastatin 40mg led to between a 42%-50% lowering of LDL, whereas rosuvastatin 40mg led to 56%-63% lowering. this suggests that there is about a 28% greater LDL lowering with rosuvastatin, though in this Korean study the ratio of rosuvastatin mg dose (17.1mg) vs rosuvastatin (36mg) is 2.1-fold: ie they did not increase the atorvastatin to relatively equipotent doses of rosuvastatin (see below: patients were put on ezetimibe instead). and atorvastatin can be increased to 80mg, leading to only a 7% difference with rosuvastatin 40mg (these numbers were from https://askdis.blogspot.com/2020/07/statins-administration-time-equivalence.html )

    -- so this Korean study cannot conclude that there was no difference in cardiovascular clinical outcomes from these 2 statins: they did not compare equipotent doses of these statins

-- they also found increased risk of cataracts with rosuvastatin, though there was not really much difference in the achieved LDL levels

    -- they commented that the increase in cataracts was presumably from “the greater LDL cholesterol lowering capacity of rosuvastatin might have prevented epithelial cell development within the crystalline lens”

-- in general, there is a concern about using “diabetes” as the clinical endpoint, with the cutpoint of A1c of  6.5% (as was used in this study). This cutpoint was established because it reflects the inflection point of an increase in microvascular complications (specifically retinopathy). BUT the most common cause of major morbidity and mortality is definitively cardiovascular causes, with around 80% of all mortality in patients with diabetes 

    -- in fact “prediabetes” is associated with a significant increase in cardiovascular disease, more significantly so in men, with increased cardiovasc risk even below an A1c of 5.5%: http://gmodestmedblogs.blogspot.com/2022/11/prediabetes-increases-risk-heart.html 

    -- so, using the marker of new-onset diabetes (NOD) may be misleading: 

        -- diabetes is a specific definition: eg A1c reaching the 6.5% mark, and prediabetes is a range of 5.7% up to 6.5%:  

            -- the rate of progression to diabetes in those with "prediabetes" is, not surprisingly, much higher in those closer to the 6.5% threshold 

            -- therefore, assessing the development of NOD may vary significantly depending on the clusters of patients involved (were their average A1c's closer to 6.4% vs 5.7%??). we do not know the A1c levels in this study, and they are likely to be similar in the 2 groups through their randomization in the study. But if they were close to 6.4%, even a very small increase in diabetes risk would hit their 6.5% cutpoint and appear that there was a dramatic increase in diagnosing NOD. And, nature being what it is, there seems to be a continuum of cardiovascular risk (ie, the cardiovascular risk of an A1c of 6.4% is probably not much different than 6.5%, though we put those A1c values into two very different buckets: not diabetes and diabetes) 

            -- it is therefore important in studies to actually assess the specifics of “prediabetes” and stratify the results accordingly 

            -- in two of the statin trials, patients with prediabetes developed diabetes while on the statins at a five-fold rate of those without prediabetes, over a five-year follow-up: https://www.sciencedirect.com/science/article/pii/S0735109714070375?via%3Dihub. ie, statins mostly bring out the diabetes in those with preexisting prediabetes. so, we are not really sure how much the formal diagnosis of NOD really matters 

            -- and in non-diabetic patients, one study found that statins increased fasting plasma glucose only from 97 to only 101 mg/dL: https://pubmed.ncbi.nlm.nih.gov/19188844/

      -- this all suggests that statins may well decrease glucose tolerance somewhat, this is supported mechanistically (as noted above), the effects of statins in nondiabetics is not so large, the vast majority of those developing NOD already have glucose intolerance, those persons are already at high risk of cardiovascular events though they may cross the 6.5% A1c threshold and be considered NOD, and they all should really have aggressive lipid management (nonpharmacological, and pharmacological as needed)

        -- but the real bottom line: to the extent that the rate of NOD is increased with statins, the consensus opinion based on evaluating the real risks and benefits of statins (and using the A1c of 6.5% as a cutpoint) is that the long-term risk of the potential major complications of diabetes from statins is far outweighed by the benefits of statins  

 

-- another potential conclusion of the study is that they did not find any significant difference in cardiovascular outcomes between rosuvastatin and atorvastatin. however, it is important to note that atorvastatin and rosuvastatin are different medications: the former is lipophilic and does cross the blood brain barrier, the latter is hydrophilic and does not do so. There could be positive or negative effects for atorvastatin since it does cross into the brain: since dementia, for example, is associated with neuroinflammation, atorvastatin might be beneficial by having its cerebral anti-inflammatory effects; conversely, there could be long-term negative effects by its being present in the brain, since atorvastatin could decrease levels of cholesterol, a necessity for neural cell membranes and in synapse formation and maturation (https://pubmed.ncbi.nlm.nih.gov/22269162/ ) .

    -- A meta-analysis in 2022 found that both lipophilic and hydrophilic statins decreased the risk of dementia equally, but in general the studies included were relatively short-term, and the instruments used to measure dementia were not so sensitive (see statins metanal dec risk dementia EurJPrevCardio2022 in dropbox, or doi:10.1093/eurjpc/zwab208)

 

-- One really big issue in this study was use of ezetimibe: at study entry, there were 259 people in the rosuvastatin group on ezetimibe (11.8%) vs 220 on atorvastatin (10.0%)

    -- but there were many more randomized to the atorvastatin group who were put on ezetimibe over the course of the study, such that by year 2-3, there were 402 who randomized to atorvastatin who were put on ezetimibe (18.8%) vs 252 randomized to rosuvastatin (11.8%), p<0.001. In fact, by six weeks into the study, there was already a highly statistically significant difference in ezetimibe use between the groups. This all despite recommendations in the study design to NOT use non-statin anti-lipid meds

        -- perhaps the reason that there were similar cardiovascular events in these two groups was a disproportionate use of ezetimibe in those on atorvastatin, perhaps because it is less potent than rosuvastatin (though the atorvastatin dose was only 36mg and could have been increased to 80mg...)

        -- ezetimibe might have influenced their results since it has pleiotropic effects beyond LDL lowering, including decreasing oxidative stress and decreasing platelet aggregation and activation

        -- and, given that ezetimibe decreases oxidative stress, perhaps it was the ezetimibe that was related to decreased diabetes diagnoses in the atorvastatin group, since oxidative stress is related to diabetes development (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5059829/ )

            -- there are no great studies on the effect of ezetimibe on glucose control. mouse studies have found decreased insulin resistance. In post-hoc analysis of the IMPROVE-IT study, patients on simvastatin vs simvastatin plus ezetimibe did not have any difference in NOD in those groups (though post-hoc analyses are not so statistically powerful, and the definition of NOD was the A1c of 6.5% cutpoint, with the caveats as noted above)

 

Limitations:

-- though the study had 12 hospitals enrolling patients, all were in South Korea and that may limit generalizability to other areas

    -- for example, their definition of NOD included “requiring initiation of antidiabetic medication” with no indication about what the criteria were for initiating diabetes medications (seems to have been left to clinician discretion). were clinicians more likely to initiate antidiabetic medications earlier in South Korea (perhaps instead of nonpharmacologic interventions) than in some of the other areas in the world? What meds did they use? We know some meds (e.g. insulin and sulfonylureas) may well be associated with increased cardiovascular events, perhaps distorting their primary outcome of cardiovascular results in favor of atorvastatin

-- this was a secondary prevention study, with patients having had documented cardiovascular disease, so the results may be different in patients without evidence of coronary artery disease but with hyperlipidemia

-- their definition of diabetes, as referenced above, is problematic in their conclusions. Since it seems that the most patients at risk for diabetes already have glucose intolerance, and glucose intolerance (“prediabetes”) itself confers significant cardiovascular risk. a better indicator of decreased glucose tolerance would be baseline and subsequent A1c levels throughout the study. And then we could understand the real impact of statins on diabetes risk (ie, did those developing NOD have a baseline of near 5.7% vs 6.4%, where going from 5.7% to NOD is a much more significant change in glucose tolerance than starting at 6.4%)

-- this was a short study, only three years, which might limit their conclusions, since coronary artery disease is a lifelong progressive process, and longer follow-up potentially could have different results

-- the comorbidities identified in the study are a bit problematic: hypertension and diabetes were binary variables (i.e. combining people with mild as well as severe diabetes or hypertension in the same groupings would be combining people with very different cardiovascular risks; and we know that defining smoking as current smokers may well miss people at cardiovascular high risk from passive smoking exposure, and even non-smokers who quit still have a residual increased risk of cardiovascular outcomes for 30 years)

    -- and what about the many other important comorbidities: either medical or psychosocial that might lead to diabetes or cardiovascular outcomes: https://gmodestmedblogs.blogspot.com/2023/10/update-ascvd-risk-factor-critique.html

-- as mentioned above, there was a disproportionate group of patients on atorvastatin who ended up being on ezetimibe, which potentially muddies the waters a lot

-- the use of composite outcomes as in this study is problematic: a composite makes it easier to have enough events to be statistically significant, but it gives equal value to mortality or stroke or MI with “any coronary revascularizations”. My guess is that most of us would prefer a revascularization over a stroke, or dying…

-- also the achieved LDL in this study was a bit higher (70 mg/dL for rosuvastatin and 74 mg/dL for atorvastatin) than what now seems to be the best LDL target in very high-risk patients: https://gmodestmedblogs.blogspot.com/2018/08/very-low-ldl-levels-benefit-without-harm.html .

    -- in this very high risk patient group, and especially in light of the higher achieved LDL than targeted, it was a bit odd that only 72% were on the high-intensity statins. We do not have individual-level data as to which participants had achieved what LDL level, but the fact that the average achieved LDL at best hugged the top of their targeted range (and the general sense at the time of the study was the LDL should be <70 mg/dL), it seems that lipid control was a bit suboptimal by either of the studied statins, limiting generalizability of the results even back at the time of the study (and now, many are targeting a much lower LDL than that)

 

So, it is a great idea to compare different statins’ outcomes, and not assume that all statins are interchangeable with a similar class effect. A few comments:

-- the study left open some very real questions about the specific comparison between atorvastatin and rosuvastatin:

    -- one big one was significantly increased use of ezetimibe in the atorvastatin group, leading to concerns that the ezetimibe might actually have played a role in both deceasing the atorvastatin dose administered as well as potentially decreasing the risk of diabetes

    -- another is that the medical culture in South Korea may not be generalizable to other areas of the world: this study was done between 2016 and 2019.  for example, were clinicians in South Korea using GLP-1 receptor agonists at that time, which decrease cardiovascular events? Or were they using insulin and sulfonylureas that might increase cardiovascular events?

    -- And, perhaps the biggest issue, as in the prior blog on PCSK9 inhibitors, is not including any granular data about glucose tolerance, only whether the patient hit the A1c target of 6.5% or not. It is reasonably clear that people developing diabetes on a statin are largely limited to those with glucose intolerance at baseline. And, this group with glucose intolerance is already at a significantly higher risk of cardiovascular events: statin use is quite often necessary, since glucose intolerance is often in combination with other commonly associated comorbidities (hypertension, obesity), together putting the patients at high cardiovascular risk

 

so what is to be done based on this study:

-- nothing, since there are so many concerns about their conclusions

    -- and, even if statins do increase the risk of NOD in the overall mix, they are pretty clearly beneficial overall

-- it is likely mechanistically that statins do increase glucose intolerance.

-- maybe we should be happy with prescribing statins at a lower dose and supplementing with ezetimibe??  this is worth studying, especially if we are trying to achieve ultimate cardiovascular protection in high risk patients

-- or maybe we should have an oral, cheap PCSK9 inhibitor 😊

 

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

  

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

  -- click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

  -- or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them

 

if you would like access to the dropbox for articles, go to https://www.dropbox.com/scl/fo/vj803z91w1trd471h9fj8/h?rlkey=klpxdjpdhcdt3sahnpirzz730&dl=0

 

please feel free to circulate this to others. also, if you send me their emails (gmodest@bidmc.harvard.edu), i can add them to the list