I did note that, and in my reply said that they (statins) don't treat the root cause.
However, I used the example of torcetrapib and clofibrate merely to point out that "reduction in cholesterol" is done most efficaciously by other compounds too. (I don't know if the overall mortality was because of alternate toxicities of the compounds in question, other fibrates are still under consideration).
I did not mean to imply that you endorsed either T or C, and I hope I didn't sound like I did. I may as well have accused you of promoting thalidomide...
The intent was to show that treating lipoprotein numbers was unproductive... ie that the lipoprotein numbers are merely a symptom [2]. More on this below in the coronary calcification note below..
> statins have the clearest risk-benefit profile of just about any drugs ever invented
I do not believe this to be true. For example, statins have well known effects in "worsening"[1] the lipoprotein profile (especially VLDL or Lp(A) -- can't find the reference at this moment)
If dyslipidemia is a symptom, what accounts for Coronary arterial disease? What about coronary calcification? For example look at this:
From the abstract: We did not observe a relationship between on-treatment LDL cholesterol levels and the progression of calcified coronary atherosclerosis.
>Finally, it is natural for medical textbooks to change
Of course I agree, with the proviso that the change is "as our knowledge changes". But what if the medical advice changes for political or commercial (drug company) reasons?
So, if (let's assume) lipoproteins aren't the problem, what is? Here's an intriguing possiblity:
[1] worsening = causing a lipoprotein profile that is positively correlated with cardiac death (correlation is not causation, there could be an underlying common factor etc. disclaimer goes here).
[2] an attempt to answer your question about where it is that people think medicine treats only the symptoms
Hmm, the point of torcetrapib was to increase HDL cholesterol, not reduce cholesterol. Most likely, it increased mortality via totally off target effects (see Rader's 2008 NEJM review on the topic).
Looking at coronary calcification is no better than looking at lipoprotein fractions; both are intermediate endpoints. I see no reason to trust calcification if you don't trust LDL. Reduction in LDL correlates so well with reduction in mortality (statin trials) that fibrates got licensed on those grounds without showing reduced mortality. That is a problem, but not the fault of statins. It's an FDA-got-bought-out-by-pharma sort of problem.
To reject the overwhelmingly data-backed LDL hypothesis, I'd need to see something much more compelling, I must say.
Edit: your pubmed article addresses insulin resistance. Hypercholesterolemia is usually independent of that. In fact, high LDL is not even part of the metabolic syndrome... It seems as though you and I are not so much in disagreement as we are talking about different things entirely. I've been focusing on high cholesterol.
>Reduction in LDL correlates so well with reduction in mortality (statin trials) that
Citation please.
From my other comment:
In fact, if you would be so kind as to refer to some studies where, as you say "Reduction in LDL correlates so well with reduction in mortality (statin trials)", I would be very grateful, because, afaik usually those studies found "reduction in cardiac mortality", but all-cause mortality figures are more important.
Harrison's 17th edition has an entire chapter summarizing all statin trials to date; there is a dose-response curve between LDL and mortality, with no lower limit. Or try a recent meta-analysis of statin trials looking specifically at all-cause mortality (PMID 20403481). Or see (PMID 19067719) if you want another statins + all-cause mortality meta-analysis.
In another post you pointed to one article asking about the risk of cancer in statin-treated patients. That was a scare a few years ago; now, in contrast, the NIH is funding statin trials for cancer prevention (specifically colorectal and skin cancers).
> both are intermediate endpoints. I see no reason to trust calcification if you don't trust LDL.
I respectfully disagree. Cardiac deaths, including strokes, heart attacks etc., have many causes. One of the leading proximate causes is stenosis due to coronary atherosclerosis (which causes infarcts in heart muscle over time), and the mechanism involved there is the development of plaque in the intima of the coronary arteries: this plaque consists of [1] calcium deposits, fibrinogen and lipoprotein. It seems high CRP, fibrinogen and possibly homocysteine trigger calcium deposition and lipoprotein/platelet aggregation, thereby leading to plaque formation. Does insulin predispose cells to secrete calcium [2] (this would exacerbate the problem of inadequate vasodilation) ?
[1] http://www.ncbi.nlm.nih.gov/pubmed/20551578 (yes, I am aware they found a low correlate with CRP and others, but that they're investigating indicates that the hypothesis exists)
In other words, I see CAC scores (Agatston scores) as far better markers of cardiac events.
> Reduction in LDL correlates so well with reduction in mortality (statin trials) that fibrates got licensed on those grounds without showing reduced mortality. That is a problem, but not the fault of statins. It's an FDA-got-bought-out-by-pharma sort of problem. To reject the overwhelmingly data-backed LDL hypothesis, I'd need to see something much more compelling, I must say.
In fact, if you would be so kind as to refer to some studies where, as you say "Reduction in LDL correlates so well with reduction in mortality (statin trials)", I would be very grateful, because, afaik usually those studies found "reduction in cardiac mortality", but all-cause mortality figures are more important (IOW, what if there are off-target effects with statins too, just not at the levels of clofibrate/torcetrapib?).
From the abstract:
> The J-curve association was observed between average TC or LDL-C concentrations and total mortality. Malignancy was the most prevalent cause of death. The health of patients should be monitored closely when there is a remarkable decrease in TC and LDL-C concentrations with low-dose statin.
Since they found a J-curve, IOW, reduction in LDL does not correlate well with reduction in mortality. If LDL caused heart disease one would expect a dose-dependent relationship. Also would you agree that if there is, for example a U-shaped curve, average reductions in mortality, P-values etc. mean nothing and are misleading? IOW, it's a case of the aggregation hiding a vast disparity in outcome?
It's a very interesting paper, and the full paper is available, which increases its value substantially.
> your pubmed article addresses insulin resistance. Hypercholesterolemia is usually independent of that.
Exactly. Hypercholesterolemia is independent of insulin resistance (IR), but IR (aka metabolic syndrome) is highly associated with cardiac mortality. IMHO this supports my point that in most cases, actual cholesterol numbers have very little to do with cardiac mortality (except for a few pathological cases of hypercholesterolemia as defined above some 500mg/dl etc).
> In fact, high LDL is not even part of the metabolic syndrome.
Again, that's exactly where I'm coming from. Metabolic syndrome (easiest identified by high triglycerides, low HDL, IR, intra-abdominal obesity [3]) is the "best" precursor/predictor for cardiac mortality.
> not so much in disagreement as we are talking about different things entirely.
To loosely paraphrase Feynman, where there is unexplained data is where science has the potential to advance most. I think (and so do others, I'm not a lone loony) that IR, metabolic syndrome, diabetes, etc. have basically a common trigger, which is carbohydrate/omega-6 induced metabolic dysfunction (systemic dysfunction, I should add) -- this is basically a summary of the second part of Taubes' book.
> I've been focusing on high cholesterol.
What do you mean by cholesterol exactly? What method of assay do you refer to, and what do you consider as the significant numeric values?
We have so many better markers now, ApoA/B ratios, HDL/Triglyceride ratios, homocysteine, inflammation factors and fibrinogen. Cholesterol by itself has been recognized as a completely inadequate explanatory factor, heck, everyone now knows that eating eggs and their cholesterol has no effect (other than a small beneficial one) on cardiac events.
Tangential at this point:
> Reduction in LDL correlates so well with reduction in mortality (statin trials) that fibrates got licensed on those grounds without showing reduced mortality. That is a problem, but not the fault of statins. It's an FDA-got-bought-out-by-pharma sort of problem.
I agree with you overall, that it seems fibrates sneaked in under the wire, but (conspiracy hat on) if the pharmas are devious enough for that, are they not devious enough to also massage data and corrupt researchers? :-)
One hopes not, but I am trying to point out that the fibrates fiasco (again, other fibrates are still, afaik, under consideration, though it's probably harder now that they have the clofibrate rap to contend with) is a symptom of the pervasive "treat the numbers" mentality. I wish they would adopt a "find the mechanism" attitude instead.
At one point you say that the metabolic syndrome is correlated with insulin resistance, and at another point you use IR as another word for the metablic syndrome. It's important to be clear, as you pointed out. (Btw, when I say "high cholesterol," I mean elevated LDL-C.) The correlation coefficient between ApoB and repeat LDL-C measurements is high enough as to make the (more expensive) ApoB not worth measuring except for research purposes; though I'd be happy to measure ApoB directly, doing so would change very little. It's certainly not "much better." There is actual research, and then there is clinical junk like taking ratios of different lipoprotein species; that doesn't help you get to the heart of the problem. And you are extrapolating out too far from the papers that you are citing; the medical literature merits a bit more of a cautious reading than I think you're giving it, IMHO. Making super-bold claims like "Metabolic syndrome ... is the "best" predcursor/predictor for cardiac mortality" from a paper entitled "Evaluation of some markers of subclinical atherosclerosis in Egyptian young adult males with abdominal obesity" is a perfect example of over-interpreting.
In conclusion, your level of belief of well-established science (such as statin trials; >100,000 people, total) is too low, and your willingness to extrapolate from very small studies (50 people, total) is too high. I would urge you to recalibrate your beliefs in proportion to the weight of each study, instead of in proportion to your desire for the study to be true. It's hard for all of us to do, but it's the hallmark of the scientific approach.
> Making super-bold claims like "Metabolic syndrome ... is the "best" predcursor/predictor for cardiac mortality" .. example of over-interpreting.
That is an unfair charge, the claim is not based solely on that study. Taubes reviews 50+ years of observations and research on this. Besides, that claim is my hypothesis, which is yet to be tested.
I in no way present it as anything other than a theory.
> In conclusion, your level of belief of well-established science (such as statin trials; >100,000 people, total)
Citation?
> is too low, and your willingness to extrapolate from very small studies (50 people, total) is too high.
I will grant that this is the case. It's because the large scale studies for the hypothesis I put forward do not exist yet. I would like to see NIH funded diet-intervention studies at the same scale, even n=200 would be a great thing.
> each study, instead of in proportion to your desire for the study to be true. It's hard for all of us to do, but it's the hallmark of the scientific approach.
Nice burn in the last sentence!
While I agree that my hypothesis has not been tested, it has not been negated either.
I would like to see well funded and well designed studies to settle the issue conclusively, therefore the charge about the scientific approach is a bit off the mark: I am not holding on to a belief in the face of overwhelming evidence against it, I merely postulate an alternative hypothesis which has the potential to neatly (in the Occam's razor sense) explain several inconsistencies in the current model (and should be tested for merely on that basis, if for no other).
The ApoB issue was raised as there were some pubmed papers that discussed the test and what advantages it had over others in predictive power. I cannot find the ref. right now, but it should be easy to search for. In any case, that was a tangential point I made and you spend a lot of time addressing it, while not addressing other tests like homocysteine levels etc.
MS is a catch-all term for several coincidentally occuring symptoms, as far as I can tell. IR is one of them, abdominal obesity is another, prevalence of LDL and high triglyceride counts are yet more markers. For all I know, IR is the major symptom, that's why at one point I refer to IR and then say "aka MS". Again, this point is tangential.
When you said "Cholesterol" I naturally assumed "Total Cholesterol". I note that you have clarified that. Also, LDL-C I presume includes the VLDL counts? That raises another possible issue: what if the symptoms observed in these studies actually indicate that VLDL is highly correlated with mortality, and that due to some confounding factor, the subjects in these studies have very high VLDL? In that case LDL would be unfairly blamed. Do you agree?
> Looking at coronary calcification is no better than looking at lipoprotein fractions; both are intermediate endpoints. I see no reason to trust calcification if you don't trust LDL. Reduction in LDL correlates so well with ... FDA-got-bought-out-by-pharma sort of problem.
You did not address the Agatston score point, which iirc was the main rebuttal I'd made to your earlier comment. As this discussion is threatening to run off on another tangent, I will end here.
> To reject the overwhelmingly data-backed LDL hypothesis, I'd need to see something much more compelling, I must say.
Perhaps GCBC may itself be compelling?
I will state that most of my stance is covered in GCBC (to the extent that you could say I got it from there) and it should be quite easy to test the alternative hypotheses laid out in that book. Perhaps we can revisit this thread when the issue is settled to our satisfactions and marvel at the different positions we took.
No, it's completely fair. We have objective criteria for evaluating "best", and expert opinion is Grade C evidence (different groups use different lettering conventions, but suffice it to say that expert opinion is not regarded highly since it is, in fact, not evidence).
LDL is not VLDL. LDL-C is different from VLDL measures. Since this isn't that popularly known, let me explain a bit: people rarely measure LDL-C directly. Instead, they assume that virtually all serum TG is carried by VLDL; they measure HDL; and then they infer LDL-C. It turns out that this inferred LDL-C correlates well enough with directly measured LDL-C for the vast majority of people that it hasn't been worth the cost of implementing a direct test except for research purposes (Canadians will differ on this).
I'm not exactly sure where you are going with the VLDL vs LDL discussion, however, since VLDL is the precursor to LDL. It is always a formal possibility that a confounder is modifying or even inverting the relationship between X and Y, but typically one would expect some evidence in support of that. The LDL-MI association holds up in every population tested - so unless this postulated confounder is present in every population, it's not likely. Plus, most people expect that both VLDL and LDL will correlate with MI risk, though the best studied one is LDL.
>You did not address the Agatston score point, which iirc was the main rebuttal I'd made to your earlier comment. As this discussion is threatening to run off on another tangent, I will end here.
My point isn't that the Agatston score is bad; it's that you're rejecting the intermediate endpoint with the best evidence (LDL-C, which has prospective randomized trials supporting it) and then endorsing an intermediate endpoint that's far less accepted (coronary calcification). I rebutted your Agatston score comment in a couple of ways previously and without actual evidence supporting its importance in the causal chain for MI, I see it merely as a biomarker (unlike LDL-C, which is causal).
>Perhaps GCBC may itself be compelling?
No. Since the GCBC is a book that expresses its author's expert opinion, I do not find it to be evidence. When I hear "evidence," I expect data, not opinion.
>The J-curve association was observed between average TC or LDL-C concentrations and total mortality. Malignancy was the most prevalent cause of death. The health of patients should be monitored closely when there is a remarkable decrease in TC and LDL-C concentrations with low-dose statin.
IOW remarkably low LDL concentrations increase death rates.
>IOW remarkably low LDL concentrations increase death rates.
Since you must have read the article, given that you're citing it, what were your thoughts on the following things that I found troubling:
(1) People were stratifed according to post-treatment TC, not according to any baseline measure, LDL-C, or an appropriate compound measure of risk. In other words, this is all post hoc, and highly suspect.
(2) The group with low on-treatment TC was overwhelmingly male (>60%, vs 30-40% in other groups) at baseline.
(3) The groups with low on-treatment TC had higher blood pressure (53-58% vs 39-48%) at baseline.
(4) The groups with low on-treatment TC were overwhelmingly more alcoholic (45% vs 28%) at baseline.
(5) The groups with low on-treatment TC were overwhelmingly more burdened by liver disease (11-16% vs 8%) at baseline.
(6) The groups with low on-treatment TC were overwhelmingly more likely to smoke (30% vs 15%) at baseline.
In other words, there are plenty of reasons at baseline! for the low TC group to exhibit increased risk of all-cause death. Perhaps the most interesting conclusion is that the statins appeared to protect these hypertensive male smokers from cardiovascular disease!
Your conclusion is almost certainly incorrect (and probably inverted, in fact), due to the confounders that I've enumerated above.
I did note that, and in my reply said that they (statins) don't treat the root cause.
However, I used the example of torcetrapib and clofibrate merely to point out that "reduction in cholesterol" is done most efficaciously by other compounds too. (I don't know if the overall mortality was because of alternate toxicities of the compounds in question, other fibrates are still under consideration).
I did not mean to imply that you endorsed either T or C, and I hope I didn't sound like I did. I may as well have accused you of promoting thalidomide...
The intent was to show that treating lipoprotein numbers was unproductive... ie that the lipoprotein numbers are merely a symptom [2]. More on this below in the coronary calcification note below..
> statins have the clearest risk-benefit profile of just about any drugs ever invented
I do not believe this to be true. For example, statins have well known effects in "worsening"[1] the lipoprotein profile (especially VLDL or Lp(A) -- can't find the reference at this moment)
If dyslipidemia is a symptom, what accounts for Coronary arterial disease? What about coronary calcification? For example look at this:
http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=S...
From the abstract: We did not observe a relationship between on-treatment LDL cholesterol levels and the progression of calcified coronary atherosclerosis.
>Finally, it is natural for medical textbooks to change
Of course I agree, with the proviso that the change is "as our knowledge changes". But what if the medical advice changes for political or commercial (drug company) reasons?
So, if (let's assume) lipoproteins aren't the problem, what is? Here's an intriguing possiblity:
http://www.ncbi.nlm.nih.gov/pubmed/17606264?ordinalpos=1&...
[1] worsening = causing a lipoprotein profile that is positively correlated with cardiac death (correlation is not causation, there could be an underlying common factor etc. disclaimer goes here). [2] an attempt to answer your question about where it is that people think medicine treats only the symptoms