This topic was created for members to discuss the AIM HIGH Trial and its implications.
New Topic
Post Reply
Flag/UnflagI’ll throw out 5 or 6 hypotheses about how the AIM-HIGH results might be explained:
1. The "HDL hypothesis" is wrong. That is, HDL itself may not be anti-atherogenic in humans. But HDL clearly is anti-atherogenic in mice and rabbits, and the clinical epidemiology for HDL is strong.
2. When LDLC is low enough, HDL doesn't matter any more. But the epidemiologic relationship still holds up in intensive trials such as TNT. I suspect that even in AIM-HIGH, we will continue to see the correlation between low HDL and events to hold.
3. Lesions might improve with niacin treatment, but something else could happen to promote atherothrombotic events. Overall these effects could balance in a 4-year study. Niacin inhibits fatty acid release from adipocytes, and for some time each day more glucose may be required to meet energy needs. This could require a counterregulatory hormone response, including epinephrine, perhaps promoting events.
4. Patients may differ in their response to good and bad (?) effects of niacin. An analogy would be exercise. Everyone believes that exercise is good for the heart and the arteries, but events can also occur at the time of intensive exercise. Nobody would recommend intensive exercise shortly after acute coronary syndrome when plaques are still unstable.
5. Rapid reverse cholesterol transport out of atherosclerotic lesions might have a bad effect in some cases. I’ve seen patients with Achilles tendon xanthomas develop pain and inflammation of the tendons after intensification of their cholesterol regimen, including the addition of niacin. Something similar could occur in arteries.
6. A small number of strokes may have occurred if niacin increased atrial fibrillation in AIM-HIGH (this is not yet determined), as it did in the Coronary Drug Project, where 4.7% of patients assigned to niacin and only 2.9% assigned to placebo had "definite or suspected" atrial fibrillation. Atrial fib can certainly lead to ischemic stroke.
These hypotheses are not mutually exclusive.
What a disappointment, if this holds up.
Can anyone shed light on the power statistics? Could this be a "false negative"? What does stopping the trial early do to the power stats or confidence limits? Maybe it just takes longer to see curve separation? The following is the study design:
Statistical considerations
The trial will be analyzed under the principle of intention to treat. All participants will be followed-up to a common termination date or death (estimated average follow-up 4.6 years; minimum follow-up 30 months and maximum approximately 7 years).
As originally designed, the primary end point was specified as a composite end point consisting of CHD death, nonfatal MI, ischemic stroke, or hospitalization for high-risk acute coronary syndrome. Under this design, with an estimated sample size of 3,300, the trial had 99% power to detect a 25% reduction in primary outcome events among patients randomized to the combination of extended-release niacin + simvastatin (combination therapy), assuming a 0.089 hazard rate in patients randomized to placebo + simvastatin (standard therapy), using an unadjusted one-sided P value of .025. The interim analysis plan allowed for one adjustment to sample size if the overall primary event rate was lower than expected. Because of a lower-than-expected overall primary end point rate, protocol amendment 5, adopted April 2010, modified the primary end point to the expanded composite described above. Expansion of the trial sample size and extension of follow-up without change to the primary endpoint were considered but, for budgetary reasons, not implemented. This modification was proposed by the AIM-HIGH Executive Committee (online Appendix A), which was blinded to by-treatment-arm results, and approved by the National Heart, Lung, and Blood Institute (NHLBI) following the recommendations of an Independent Review Committee appointed by the NHLBI, which was also blinded to by-treatment-arm outcome data.
The interim analysis plan allows for 2 examinations of the efficacy data by the Data and Safety Monitoring Board when approximately 50% and 75% of the projected primary events have occurred.
Very unexpected results. Many many patients including family members are on a combo therapy with statin and Niacin. Granted you most if not all of them I have been following them up for more than 15 years with very controlled LDL-P and all still events free. The results of AIM-High were shocking but it bring me to couple of thoughts:
1. Is the results are a manifestation of scentific rigorous RCT ( Randomized controll studies ) sponsered bt the NIH vs drug company sponsered RCT ?
2. Possible that the niacin in the AIM-High ( Niaspan - abbott ) might have a different therapeutic effect than the niacin used in the CDP, class, HATS , extra studies ?
3. The results , in my opinion bring the RESIDUAL RISK concept into a question ? If the LDL-C and LDL-P were tightly controlled as it was in AIM-High in all the previous statin trial , will the residual risk theory stated over and over as it is close to 65%, will hold true ? and if it is true, now we are facing a new challenges about how to deal with it ?
I am in agreement with Dr. Blanchet's perspective on the TheHeart.org "Forum":
|
|
|
 |
|
 |
 |
 |
|
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
|
|||
|
|
||
|
|
|
|||
|
|
|
|
|||
|
|
|
|
|||
A better design of an AIM-High type study.
Rather than randomize patients based on lipid parameters, a poor risk stratifier to be sure, I would propose a study where the patients were first separated with lipid parameters then risk assessed by coronary calcium imaging. After accurately assessing risk thusly, the different quintiles of coronary calcium could be randomized to niacin vs no niacin with the statin. A much smaller study would reach clinical significance when the treated and control populations are equally risk matched.
I suspect that when the details of AIM-High are revealed, we will see that with this small a number of patients, there will be enough heterogeneity of risk in the treatment and control group that the whole study will be determined to be a colossal waste of resources. Perhaps it is not too late to now do CAC studies on the AIM-High study population and therefore cut to the chase regarding assessment of global baseline risk.
Additional disclosure: I work at a facility that uses EBT-CAC and carotid ultrasound to assess baseline MI risk and direct therapy. I use more than an average amount of niacin, I see very very few heart attacks and ischemic strokes.
Author's disclosure (May 28, 2011)
Three years ago, I attended a dinner meeting or two sponsored by Abbott.
Sorry for the redundancy,.. the multiple entries appear to be flaws in the 2 sites software,.. I merely cc'd & pasted onece,.. not several times. My apologies,...
I agree with John Guyton, MD, about the concept that when LDL is low, perhaps HDL is no longer relevant. In AIM HIGH, there was no difference in the incidence on coronary events, whilst where was a slight increase in the number of strokes, when the statin group was compared to the statin/niacin group. In 2000, I published an article in the Journal of Cardiovascular Risk analyzing eight published angiographic regression trials (JCR, 2000;7:415-423) in which I noted that my risk factor prediction graph was associated with stabilization/regression of coronary plaque (angiographically) in 75% of cases when the goal was reached. I use a ratio between LDL and HDL to define dyslipidemia as part of the predictive tool, but if HDL is very low then the ratio will always be high and moreover since POSCH was not structured to control blood pressure and since systolic blood pressure is part of the predictive graph, I decided to look at LDL by itself. I found that when whatever therapy was employed, if LDL was brought anywhere below 80 mg/dl (2.0 mmoles/L), then angiographic stabilization/regression occurred in 93% of cases. Since angiographic stabilization/regression is associated with a dramatic fall in clinical atherothrombotic disease (ATD) events, the answer to AIM HIGH would appear to be self-evident. I also find Dr Guyton's thoughts on niacin and atrial fibrillation to be of interest.
I am going to side with Dr. Blanchet's conjecture, off the Heart.org FORUM:
|
 |
|||
 |
 |
 |
||
 |
|
|
 |
|
 |
 |
 |
||
 |
|
|
||
|
|
||||
|
|
|
|
|||
|
|
|
June 26, 2011 02:59 (EDT) |
|
||
|
|
|
|
|||
|
|
|
|
|||
|
|
|||
|
|
|
||
|
|
|
|
|
|
|
|
||
|
|
|
||
Yes this is truely food for thought, as clinicians we need to be careful we don't fall into the "true believer" category and continue to go with the best evidence we have. Occasionally ( more often, however, than we'd like to admit), Our "best guess" at what is best for our patients proves wrong. When this happens we see practitioners split into two groups: those that go with the evidence and the "true believers" who stick to their guns despite the evidence ( I call this "faith based" medicine).
Whatever the reason, based on what we currently know after AIM HIGH, sustained release Niacin prescriptions are now more difficult to defend. It should be noted that what was studied WAS NOT Niacin vs no therapy but niacin IN ADDITION TO a statin. It may well be that niacin alone does offer some benefit as shown in earlier (pre-statin) trials.
This, along with the relative lack of benefit so far with ezetimibe in addition to a statin (or alone for that matter), and the paradoxical findings with torcetrapib (although those are being attributed to "off target adrenal effects" ) SHOULD be making all serious scientists ask, are we on the wrong track completely? Is it possible that LDL(C or P) and ABO-B are great surrogate markers for statin efficacy but NOT directly causal.
