How to cheat with complicated calculations
The New England Journal of Medicine recently published the results of the SATURN trial. It was designed to study the effect of atorvastatin (Lipitor) vs. rosuvastatin (Crestor) on the volume of atheroma in a coronary artery. It was hoped that the volume would be reduced, demonstrating that high dose statins can decrease the burden of atherosclerosis.
Initially 1578 patients were selected for the trial, but after a run-in period of 2 weeks where they were treated with half-maximal doses of either atorvastatin or rosuvastatin, 193 patients were excluded. The rest were treated either with 80 mg atorvastatin or 40 mg rosuvastatin. After 2 years of treatment a further 346 patients had disappeared.
Before and after the trial the patients underwent intravascular ultrasonography to measure the lumen diameter and the total diameter of a coronary artery. Subtracting the lumen area from the total area of the artery is thought to reflect the total atheroma volume, (represented as the percentage atheroma volume). The primary endpoint was to measure the reduction in percent atheroma volume
After the treatment the lumen had increased on average by 0.99 % in the atorvastatin group, and by 1.22 % in the rosuvastatin group, and the per cent atheroma volume (eg. the area of the arterial wall) had decreased by 1.1 % and 1.3 % respectively.
There was also a secondary end-point. However, I must admit that I did not understand what it meant, and if anyone does, please explain it to me. Here is what the authors wrote:
A secondary efficacy end point, normalized total atheroma volume (TAV), was calculated as follows:TAVnormalized = Σ (EEMarea − lumen area x median no. of no. of images in pullbackimages in cohort. For TAV, the summation of the EEM area minus the lumen area is performed first. This value is divided by the number of images in the pullback and then multiplied by the median number of images in the cohort. The average plaque area in the pullback was multiplied by the median number of images analyzed in the entire cohort to compensate for differences in segment length between subjects. The efficacy end point of change in normalized TAV was calculated as the TAV at 104 weeks minus the TAV at baseline. Regression was defined as a decrease in PAV or TAV from baseline.
Anyway, the critical measure was the difference between the inner and outer area of the artery. Unfortunately, there is no evidence that the figure from this calculation reflects atheroma volume. For example, vascular dilation will increase the inner diameter, without having any effect on the thickness of the arterial wall. But this would result in an apparent decrease in atheroma volume. To further understand what I mean, read the following section from my book Fat and Cholesterol are GOOD for You!:
The anguish of angiography
Let us still have in mind, that a change of the coronary diameter is nothing but a surrogate outcome. It is assumed that a widening of a coronary vessel on an X-ray means less atherosclerosis and thus a better chance to avoid a heart attack, but this is only an assumption.
Artery walls are surrounded by smooth muscle cells. When such cells contract, the artery narrows. When they relax, it widens. Various factors may stimulate the smooth muscle cells of the coronary arteries. Most important, mental stress, anxiety, exposure to cold, and even a sustained handgrip may lead to contraction. The latter effect was studied six years earlier by Dr. Greg Brown, the same Dr. Brown who led the angiographic trial mentioned above (1). He found that a handgrip sustained for a few minutes was followed by a 35 per cent decrease of the vessel diameter.
Consider that the changes seen in the trials were only a few per cent on average. What do you think you would do yourself if somebody were to put a long catheter all the way from your groin up to your heart and into your coronary vessels? If you are not a stuntman or an astronaut I think that you probably would have gripped the nurse’s hand or something else very tightly.
Also, drugs which relax the coronary vessels, and which are used by almost all coronary patients, may have disturbed the study. In the trial Dr. Brown and his coworkers were aware of that problem. The use of such drugs was “duplicated as exactly as possible.” This can’t have been too easy because the level of any drug in the blood depends on a large number of factors, which are difficult to standardize. And Dr. Brown and his colleagues didn’t write anything about duplicating possible handgrips or anxious feelings because such duplication is, of course, impossible. So, any factor which may influence the state of the smooth muscle cells in the coronary vessels may have influenced the vessel diameter much more than the possible appearance or disappearance of a tiny amount of cholesterol.
There are more uncertainties. Dr. Seymor Glagov and his colleagues from University of Chicago studied the hearts of 136 deceased individuals and found that when vessels become sclerotic, they widen to compensate for the narrowing brought about by the deposition of cholesterol in their walls (2). In fact, this widening overcompensates for the deposition, until the cholesterol deposits occupy about 40 per cent of the area beneath the muscle wall. Only thereafter does the vessel become steadily narrower. In other words, an increase of vessel diameter may be due to disappearance of cholesterol in a highly sclerotic vessel, but also to a compensatory widening during the first stages of cholesterol deposition. How could the trial directors know whether the increase of vessel diameter was due to a disappearance of deposited cholesterol, or to a compensatory widening due to an appearance of deposited cholesterol or due to less anxiety at the second angiography?
1. Brown G and others. Reflex constriction of significant coronary stenosis as a mechanism contributing to ischemic left ventricular dysfunction during isometric exercise. Circulation 70:18-24, 1984
2. Glagov S and others. Compensatory enlargement of human atherosclerotic coronary arteries. The New England Journal of Medicine 316:1371-1375, 1987.
In short, the degree of arterial dilation is a massive and uncontrolled variable in the SATURN study. This problem could have been solved if the investigators had included a placebo group (a group of patients who unknowingly received an ineffective pill). However, “It was not considered ethically possible to measure disease progression in placebo-treated patients”, as they wrote.
There were other major problems with this study. The issue of drug related adverse events is extremely important. This was virtually dismissed within the paper. “Both agents had acceptable side-effect profiles”.
Can this be true. A more detailed review of adverse events reveals that “new proteinuria”, defined as an excretion of more than twice the amount of protein in the urine during the follow-up, in 1.7 and 3.8 %, respectively in the two groups. An increase in proteinuria is a measure of progressive damage to the kidneys. This trial only lasted two years, so we don´t know what would have happened in the longer term.
Equally it was stated that less than two per cent had laboratory signs of liver damage. However, liver damage was only recorded if the laboratory signs were at least three times higher than the upper limit of the normal range. And whilst less than two per cent had muscular damage, this was only reported if the laboratory signs were at least five times higher than the upper limit of the normal value. What do you think will happen with the liver and muscles of patients whose laboratory signs were “only” twice or four times higher, respectively?
In the end a further 22 % of the patients had disappeared. The reasons were said to be preference of patient (7.7% and 7.8 %), adverse effects (7% and 6.5 %), loss to follow-up (1.3% and 2.9 %) and noncompliance (2.3% and 1.9 %). What they meant with “preference of patient” I don´t know, but I am confident that “non-compliance” and perhaps also “loss to follow-up” represent those who could not tolerate the medication. Thus, whilst the authors claimed that ‘both agents had acceptable side effect profiles,’ the reality is that 12% could not tolerate these agents at the start of the study, and another 23 % dropped out – most likely to due to intolerable adverse events.
My summary of the SATURN study would be that it used a primary end point that has never been properly validated, and can be affected by a host of confounding variables e.g. stress . This variability could only have been controlled for by including a placebo arm, which was not done. Therefore, the result is rendered meaningless.
More importantly, it would appear that the burden of adverse events from using high doses of statin drugs is unacceptably high. It is likely that more than a third of patients will be unable to tolerate 80 mg Atrovastatin, or 40 mg of Simvastatin. All of this suffering in order to have an uncertain effect on a surrogate end-point, which may or may not mean anything.
Finally, I should mention that the study was supported by AstraZeneca, and most of the results were in favour of their drug rosuvastatin, although most of the differences were not statistically significant. I shall leave you with a list of the authors´ conflicts of interest:
Dr. Nicholls reports receiving consulting fees from Roche, Esperion, Merck, Omthera, Sanofi-Aventis, and Boehringer Ingelheim, serving as an unpaid consultant for Abbott, Pfizer, LipoScience, Novo Nordisk, AtheroNova, and CSL Behring, receiving grant support from Eli Lilly, AstraZeneca, Novartis, Anthera, LipoScience, Roche, and Resverlogix and lecture fees from AstraZeneca and Roche;
Dr. Ballantyne, receiving grant support from Abbott, Astra-Zeneca, Bristol-Myers Squibb, Genentech, GlaxoSmithKline, Kowa, Merck, Novartis, Roche, Sanofi-Synthelabo, and Takeda, consulting fees and honoraria from Abbott, Adnexus, Amarin, Amylin, AstraZeneca, Bristol-Myers Squibb, Esperion, Genentech, GlaxoSmithKline, Idera, Kowa, Merck, Novartis, Omthera, Resverlogix, Roche, Sanofi-Synthelabo, and Takeda and lecture fees from Abbott, AstraZeneca, GlaxoSmithKline, and Merck;
Dr. Barter, holding an advisory board position for AstraZeneca, Merck, Roche, CSL Behring, and Pfizer, receiving grant support from Merck, consulting fees from CSL Behring, and lecture fees from AstraZeneca, Kowa, Merck, Pfizer, and Roche;
Dr. Chapman, receiving grant support from Merck and Kowa, consulting fees from Merck and Pfizer, and lectures fees from Merck and Kowa;
Dr. Erbel, receiving grant support from the Heinz Nixdorf Foundation and the German Research Foundation and support for travel, accommodations, or meeting expenses from Biotronik, Sanofi, and Novartis;
Dr. Libby, serving as an unpaid consultant for Novartis, Johnson & Johnson, Amgen, and Roche, serving in unpaid leadership roles for clinical trials sponsored by AstraZeneca, GlaxoSmithKline, Novartis, Pfizer, Pronova, and Sigma Tau, and having previously received royalties from Roche for the patent on CD40L in cardiovascular risk stratification;
Dr. Raichlen, being an employee of and owning stock in AstraZeneca; and
Dr. Nissen, receiving consulting fees from Eli Lilly, grant funding from AstraZeneca, Pfizer, Novartis, Karo Bio, Novo Nordisk, Takeda, Resverlogix, and Omthera, and support for travel, accommodations, or meeting expenses from Novo Nordisk, Takeda, Karo Bio, Eli Lilly, Pfizer, Novartis, and Amgen.
It is a sad fact that this report has been accepted for publication in The New England Journal of Medicine considered the leading medical journal in the world. Either the editors and referees did not do their jobs, are not qualified for their tasks, or they have been blinded by Mammon.
Should children lower their cholesterol???
There are more miserable news. Recently an expert panel appointed by the National Heart, Lung and Blood Institute and endorsed by the American Academy of Pediatrics has published new guidelines according to which every child in the United States should be tested for high cholesterol between ages nine and 11 so steps can be taken to prevent heart disease later on.
Such crazy thoughts have been aired several times in the past. In a letter to The Lancet (published on January 1, 2000; a good start of the new millennium). I tried to explain why this is a most dangerous idea, but obviously the letter made no impact.
How can an expert panel produce such malicious recommendations, you may ask. Because they are paid by the industry, of course. Read for instance the article from the non-profit organisation Integrity in Science