November 15th, 2011

After AIM HIGH, What Future for Niacin? A CardioExchange Panel

With the publication and presentation of negative data from the AIM HIGH trial (see our news summary), CardioExchange convened a panel to answer this question:

What if any role remains for niacin after AIM HIGH?

Steven E. Nissen, MD

The AIM HIGH Study provides no evidence for benefit from niacin in the very patients in whom this drug is most commonly used. However, the differences in administration of LDL-lowering therapies in the two treatment groups represents a serious design flaw. Furthermore, the study was underpowered and stopped at a point where the lower limit of the 95% confidence interval could not rule out a modest benefit. The decision to stop the trial prematurely for futility was unwise and deprived the medical community of a more robust assessment of the benefits and risks of niacin treatment. Despite these flaws, it seems unlikely that niacin represents an overall benefit to patients and its routine usage should be discouraged.

JoAnne M. Foody, MD

Quite frankly, AIM-HIGH doesn’t change much for me, but rather underscores the continued necessary focus on LDL reduction with statins to improve CV outcomes. I will likely continue to use niacin (a) in combination with statins in very high risk individuals with far higher LDL levels than seen in AIM-HIGH and (b) alone in statin-intolerant patients or  individuals with extremely elevated Lp(a). None of these groups were included in this study.

Roger Blumenthal, MD

Niacin will still be useful as (a) a second-line therapy in patients whose LDL-C or non-HDL cholesterol is not at goal despite maximally tolerated statin therapy or (b) as a first-line agent in patients who cannot tolerate a statin. Niacin did not show benefit as an add-on therapy in AIM-HIGH when the baseline LDL-C was 70. It may be beneficial when the baseline LDL-C on maximally tolerated statin is 100.

I think that this study also tells us that there is little future role for serial carotid IMT testing to determine the best cholesterol lowering therapy. The ARBITER 6 – HALTS investigators were thrilled with a 0.014 mm improvement in carotid IMT vs. no change with ezetimibe. The AIM-HIGH clinical trial showed that this was associated with no benefit over a three year period, unfortunately. The insightful editorial that Erin Michos and I wrote about ARBITER 6 in the NEJM seems to have been right on the mark. I wonder if we will ever get an apologies from the few niacin zealots who were so critical of what we wrote. Vindication is a nice feeling, but I wish that the clinical trial results had been positive so that we could help our patients more.

William Boden, MD (PI of AIM HIGH)

An important therapeutic role clearly remains for niacin. What does AIM HIGH tell us then? If you are a patient with stable, non-acute cardiovascular disease who meets an indication for statin therapy AND who is able to achieve and maintain a very low level of on-treatment LDL-C as we attained in AIM-HIGH (in the low 60 mg/dL range), then there is no clinical benefit for HDL-C raising therapy with niacin to further reduce clinical events over and above what is achieved by statin therapy alone.

But how many patients fall into the above category? Data from 3 large data sets/registries (GE Healthcare, United Healthcare, NHANES) indicate that only 13% to 19% of high-risk CHD patients actually can achieve/maintain the very low levels of LDL-C we accomplished in our study. One can only apply the results of RCTs to the population that we actually enrolled in the trial and should not extrapolate the study findings to subgroups of patients who were not enrolled or those who, by trial design, were excluded (e.g., AMI or ACS patients).

AIM-HIGH also enrolled only a small percentage of women (15%) and an even smaller percentage of ethnic minorities (8%)—important subsets for whom these results may not apply.

Lastly, AIM-HIGH was plagued by an intensity and duration of prior statin therapy that may have made it difficult to discern an incremental clinical treatment effect, especially given the very low mean starting LDL-C in the 94% of trial patients who had been on a statin for at least 1 year (76%). Indeed, many of these patients had been on statin therapy for much longer (40% for 5 or more years). Such long-term therapy, superimposed on a background of optimal medical therapy and secondary prevention (e.g., ASA and antiplatelet drugs, beta-blockers, ACE inhibitors and ARBs), may have, in the aggregate, converted vulnerable coronary plaques destined to rupture (as the putative inciting event/trigger to clinical events) into stable, quiescent plaques where the  the soluble constituents of the necrotic core was “delipidated.” This is yet another reason we should be cautious in how we apply the results of AIM-HIGH to the broader population of patients that we did not enroll. Niacin still has a therapeutic role to favorably alter the lipid profile in high-risk patients and we need to reserve judgment on the long-term usage of this agent, pending the results of the ongoing HPS-2 THRIVE Trial.

[EDITOR’S NOTE: See also our coverage of the debate at the AHA news conference, a fellow’s take, and our earlier panel discussion on what stopping AIM HIGH means for the HDL hypothesis.]



9 Responses to “After AIM HIGH, What Future for Niacin? A CardioExchange Panel”

  1. As noted by commentators above and elsewhere, the “placebo” group in AIM-HIGH was mis-named in that they received HDL-raising doses of niacin, as well as more frequent ezetimibe and higher statin doses than did the “niacin-treatment” group. The rather meager power of AIM-HIGH was further compromised (fatally, I believe) by the very unfortunate early trial stop for futility.

    If, as some are claiming, prior statin therapy negates niacin benefit, then most contemporary niacin usage is inappropriate. This, however, is yet to be shown in AIM-HIGH, and in fact cannot be proven by post-hoc analyses.

    Most importantly, AIM-HIGH is much smaller both than the aggregate of pre-AIM-HIGH randomized trials of niacin (virtually all positive), and than HPS2-THRIVE (results unknown). Please note that HPS2-THRIVE was actually designed to ask the question on everyone’s mind–“does niacin help reduce CVD when added to a statin?” AIM-HIGH was NEVER designed to answer this question (please re-read the design paper) and for that reason, efforts to answer the question of niacin effects on CVD from AIM-HIGH data are doomed to failure from the outset.

    HPS2-THRIVE results should be available in less than 12 months. I urge all clinicians to politely ignore the tempest surrounding AIM-HIGH and postpone drawing any conclusions about niacin or making any significant changes in its use for 1 year, at which time HPS2-THRIVE promises to provide data to properly address the benefits and drawbacks of niacin as a statin adjunct (and that across a wide range of baseline lipid levels).

  2. Roger – your self congratulatory insults I’ll disregard in that they are amusing as they are uninformed, misguided and incorrect. A little more caution in your public commentary would be appreciated.

    The problems with AIM HIGH were cast when the study design was decided. The design was seriously confounded from the selection of patients previously treated with niacin, the switching of statins to uniform simva 40 (which for many was an uptitration), active management of LDL-C in both study arms in a failed attempt to isolate the HDL effects of niacin, and incorporation of niacin in the placebo. This fails the basic tenet of scientific experiments in which you try to isolate a single variable of interest, holding the rest constant. This completely distinguishes our ARBITER 2, and 6 trials from AIM HIGH from a study design standpoint rendering cross trial comparisons impossible. Couple these fatal flaws of AIM HIGH along with its limited power, short duration (even 4S was negative at 2 years of follow up), and reliance on soft endpoints, and you have a futile expense of $63M.

    This is not a criticism of Bill, or anyone else directly connected with the conduct of the trial, but reflects the “process” that led to the design of this trial which was over-cooked. So, the NHLBI had it correct in its original press release that this was a “futile trial”. One cannot conclude that the drug was futile. I’m open to the idea that we could use a high quality, properly designed trial in which the variable of interest (addition of niacin to statin monotherapy) is tested for its efficacy on CV outcomes. AIM HIGH is not that trial.

    The above is injury. The added insult is the faulty press release of May 2011 in which the stroke signal was highlighted, but turned out to be false positive. Beyond the fact that this influenced the DSMB to stop the trial early, it has worse effects on public trust of the government funded trials. Had a drug company touted false positive results in a press release, it would be required to retract it, and cover its tracks. A courageous NHLBI would issue a press release apologizing correcting their mistake by inappropriately alarming the american public on an unstable endpoint from incomplete results of a poorly designed trial.

    Competing interests pertaining specifically to this post, comment, or both:
    I have received honoraria from Abbott related to speaking using FDA compliant materials on the use of niacin.

  3. Lack of proof of value does not prove lack of value!

    How can such highly educated and respected clinicians continue to misunderstand this very very basic principle of statistics. The fact that a flawed study was stopped before results were obtained is in no way evidence against the use of niacin. To suggest that AIM HIGH is a reason to avoid niacin is

    When the correct study is undertaken and taken to conclusion, the results will show that niacin is a very powerful tool for the reduction of coronary events when used with omega-3 fatty acids plus low dose statin or bile salt sequestrates.

  4. Harvey Lerner, MD says:

    The New England Journal has undermined its journalistic integrity by printing such a sloppy, underpowered, and hyped-up inconclusive study. In a world where the impact of commercial interest on physician choices of treatment is so immanent and subtle, one might easily wonder whether there was behind the scenes lobbying.

  5. The controversy surrounding AIM-HIGH leaves me with some observations
    to ponder:
    1. From observational studies, we learn that HDL is cardioprotective
    irrespective of LDL levels, including LDL levels <100mg/dL.
    2. From post-hoc analyses of RCTs, we learn that this protective
    relationship holds even at LDL levels <70mg/dL (TNT) and <54mg/dL
    (JUPITER – using median cutoffs of apoA1, but not ApoA1 quartiles).
    3. So invoking the very low LDL levels achieved in AIM-HIGH as a
    possible explanation for the null finding with combination therapy
    seems like having it both ways – accepting data when it suits
    preconceived notions and rejecting it when it doesn’t!
    4. Trials should ideally be stopped only for unacceptable safety or
    futility and not for implausibly large treatment benefit (as observed
    in Polderman’s DECREASE II study, or arguably JUPITER). After 556 out
    of a planned 800 events were accrued during a mean f/u of 3 yrs
    (compared to planned 4.6yrs), the HR was virtually null, indicating a
    very low likelihood of the primary endpoint being met. Nearly 900
    patients (25% of the total cohort) had completed f/u at 4 yrs, and the
    KM curves were virtually superimposed! Hence, in my opinion, the study
    was appropriately stopped for “futility”. That ischemic stroke risk
    went in the wrong direction strengthened the DMSB decision to stop the
    trial. The published ischemic stroke rates are 27/1718 (1.6%) vs
    15/1696 (0.9%), RR 1.78, 95% CI 0.95-3.33. Simply stated another way,
    there is a 96% probability of increased risk of stroke =/>0% and a 85%
    probability of increased risk of stroke =/>25%. I wouldn’t feel
    “reassured” by these estimates!
    5. Thus far, the positive results with niacin have been observed in
    trials that used lipid biomarkers and surrogate markers (cIMT).
    Mortality in the CDP study was not significantly different at primary
    f/u, but was significantly superior at extended f/u (always
    challenging to prove that ‘randomization’ was maintained at extended
    So, the jury is still out with regards to niacin. The totality of
    evidence does not robustly support its ROUTINE use in clinical
    practice. Until the results of HPS2-THRIVE trial become available,
    niacin should be only used selectively in those that are unable to
    take statins or continue to have elevated lipids despite optimal
    statin treatment.

  6. Please reread Eliot Briton’s comments. A study poorly designed, using a therapeutic dose of the drug as the placebo, an intent to treat study where we still do not know how many stopped the niacin, and an increase in the statin dose and the use of zetia in the control group makes this study useless, not a basis for saying anything positive or negative about niacin.

    Throw out Aim-High and start again. In the meantime, keep using niacin as prior studies have demonstrated significant benefit.

    Competing interests pertaining specifically to this post, comment, or both:
    I use a lot of niacin and have remarkably few vascular events in my practice.

  7. Seth Martin, MD says:

    Fantastic commentary – many great points.

    Thanks to the AIM-HIGH investigators for running a challenging trial.

    Allen, in our Cardiology Today article, Roger and I made essentially the same points that you point out in your first paragraph – modest differential in lipid profiles with multiple moving parts in treatment protocol, limited power, short duration, and soft endpoints. Must say – the injection of healthy humor and friendly banter in academic discourse is refreshing – keeps things more interesting and exciting for us all.

    Also must say, neither ARBITER6-HALTS nor AIM-HIGH isolated HDL-C raising effects. Niacin raises HDL-C, but also lowers atherogenic lipids and lipoproteins. Therefore, as Roger and I pointed out in Cardiology Today, even if HPS2-THRIVE is a positive study, it does not prove the HDL-C raising hypothesis.

    HPS2-THRIVE adds niacin to a mid potency statin (simvastatin 40mg daily alone or with ezetimibe 10mg daily in a combination tablet). This study will not DIRECTLY address the key question in my mind which is the role of niacin or other agents on top of high potency statin therapy (high dose atorva or rosuva). That said, HPS2-THRIVE will be a superbly informative trial.

    As a taxpaying American, I must cling to the belief that AIM-HIGH wasn’t a complete waste of $63M…surely AIM-HIGH must teach us something…for example, while statins, niacin, and ezetimibe were used in dynamic fashion, in essence two overall treatment strategies were tested in AIM-HIGH; one more focused on niacin and the other more focused on ezetimibe; especially after divergence in ARBITER, aren’t we reassured that patients treated with both strategies did so well? Events were less than expected. That said, is it disconcerting that AIM-HIGH showed no signal of benefit whatsoever for niacin (as Sanjay says: KM curves superimposed)? Also, given a number of bright people are describing the trial as fatally flawed from the get-go, what does this teach us about the future “process” of trial design? How do we improve? Why were events so much lower than predicted leading to a need to modify the study protocol to expand the endpoint? Inaccurate prediction of event accrual is not unique to AIM-HIGH. It seems AIM-HIGH has generated worthy discussion not only about HDL-C, isolating trial interventions, and appropriate trial endpoints, but also the overall process of trial design.

    Perhaps if we work together to move the conversation forward, and improve, reports that our tax dollars were poorly spent will turn out to be greatly exaggerated. Like $16 muffins:

    Hopefully we can get a full breakfast out of all this, not just muffins.

  8. How do we design better trials? Before the NIH spends 63 million dollars, they should put the trial design to public (cardio exchange,,and other similar resources)scrutiny. Had they done this, AIM-HIGH might have had some value.

    NIH is now conducting a Vitamin D trial using what is likely a sub-therapeutic dose of vitamin D and to my knowledge not measuring levels. This study could have been dramatically improved by limiting enrollment to those with vitamin D deficiency and treating to blood levels between 50-70 however this will not be done. We will now spend millions and likely find a limited benefit if any benefit.

    Statin studies in low risk groups do not show value, this study might find the same result. This could have been dramatically improved with a better trial design.

    Competing interests pertaining specifically to this post, comment, or both:

  9. I agree with the comments of others concerning the flaws in the AIM-HIGH study. In addition, it is important to keep in mind the substantial literature on niacin effects that has preceded and followed AIM-HIGH. For example, Bays et al. (2011), combining data across four Phase III trials, showed that niacin/laropiprant significantly improved both LDL and HDL levels. Niacin was widely used prior to the development of statins and shown to be effective, then fell into disuse due to the flushing effect, and has more recently again come into widespread use. Niacin is perhaps the best current drug for raising HDL levels, and it appears that high HDL is at least as important as low LDL in protecting against cardiac events. Hence, we should not rely on a single study, such as AIM-HIGH, to provide final answers as to the efficacy of niacin. What we need now are studies designed to further assess the effects of niacin on all-cause mortality and cardiovascular events, since the existing literature on these issues is mixed.

    Competing interests pertaining specifically to this post, comment, or both:
    No conflicts of interest.