Yesterday we commented with concern on some non-disclosure in the recent JUPITER (Crestor- rosuvastatin) trial report as follows: (the figures have been corrected since the JUPITER report gave some stats as percentage but others as incidence per year):
JUPITER NON-DISCLOSURE: in the huge statin JUPITER trial of Crestor 20mg/d, over the mean of 1.9yrs in 17803 patients mean age 66yrs BMI 28.3kg/sqm, 48% with metabolic syndrome, starting with LDLC below 180mg 3.4mmol/L :
mean LDLC was reduced from about 2.8 to 1.4mmol/L;
there was 50% reduction in defined total cardiovascular endpoints ie stroke, AMI, hospitalization for revascularization or angina, or death from stroke or myocardial infarction.Thus incident deaths by heart failure or arrhythmia or pulmonary embolism, or death from bleeding on anticoagulants, were not reported as primary events or under CVD deaths!
40% reduction in cancer deaths,
but only 20% reduction in all-cause mortality-
and (in fine print), 25% increase in incident diabetes p0.01.
These figures show implausibly higher reductions than other primary statin prevention trials: eg Mills’ concurrently published metanalysis Canadian Universities’ Study showed reduction of only 7% in all-cause mortality, 23% reduction in heart attack, but only 11% reduction in major cardiovascular events – for perhaps obvious reasons eg:
1. JUPITER was allowed to run less than half the usual major statin trial time; and
2. possibly stricter exclusions for cardiovascular events; and
3. inexplicable exclusions of major causes, cases of CVD events and deaths.. which deaths were lumped under non-CVD …
So if in JUPITER cancer deaths fell by 40% and major vascular events by 50%, the much shorter mean duration of the trial, and the major exclusions, explain why CVD events showed such great reduction, despite only 20%. reduction in total deaths. It is puzzling that not even the original full paper nor the editorial reveals the number of deaths from cardiovascular causes, nor the non-CVD non-cancer causes– but the small reduction in total mortality suggests that (unless there was some other undisclosed non- CVD- non-cancer cause of mortality), the fall in CVD mortality must have been far lower than 20%. All previous major trial reports- of statins, HRT, metformin etc- have tabulated both total CVD deaths and all the causes of death- why not in JUPITER?
It is common cause that statins, which materially interfere with liver function, will materially interfere with anticoagulation and thus make antioagulation stabilization more hazardous. .
Yesterday the lead author Dr Ridker kindly replied immediately to these queries as follows:
Dear Dr Burman,
Within JUPITER, the definition of “cardiovascular deaths” was extremely strict, essentially requiring hospitalization and records. Thus, many “true cardiovascular” deaths (and any that would have occurred out of hospital) are unfortunately classified as “non-cardiovascular”. Thus, to avoid confusion, we reported all cause mortality. For what it is worth, for deaths adjudicated as clearly cardiovascular, the same approximate 20 percent hazard reduction is seen that was seen for death overall. Most of the deaths avoided are thus truly cardiovascular, not cancers.
If we look more closely at the results published in the JUPITER trial, the results now look even more disturbing:
In a mean of only 1.9years, there were understandably only 35 deaths from cancer on Crestor versus 58 off it – a reduction of 40%, but placebo cancer incidence of only about 0.34%pa. Average expected lifetime mortality from cancer at this age would be about 4.5%, but in this cohort in whom there were no signs of cancer at the start, negligible cancer mortality would be expected in only 1.9years as the common cancers have far longer genesis time.
So, deducting the cancer deaths from all deaths (198 on Crestor vs 247), there were 163 non-cancer deaths on Crestor vs 189 off it – a reduction of only 14%.But JUPITER wont say what cause each of the “non- cancer” deaths was certified. .
The previous major trials of statins ran for a mean of 5 years, and showed either no change or some worrying trend to more cancers and more non-natural deaths on statins.
By comparison, the Jupiter trial for a mean of only 1.9years seems to be the only medium-term (let alone long term) trial of rosuvastatin vs placebo – another trial of Crestor vs placebo of more than a year is not apparent on Pubmed abstracts..
Rosuvastatin has been reported on Pubmed only since A new statin: a new standard. Olsson AG.Am J Manag Care. 2001.
So there is apparently no medium-term (with a mean approaching 4 to 5 yrs) placebo-controlled trial of rosuvastain, nor long-term experience, let alone longterm trials, on which to judge it’s risk: benefit ratio. But the frequent occurrence of adverse effects of other statins have been documented for decades.
If as Dr Ridker says, the definition of “cardiovascular deaths” was extremely strict, thus many “true cardiovascular” deaths (and any that would have occurred out of hospital) are unfortunately classified as “non-cardiovascular”.. For what it is worth, for deaths adjudicated as clearly cardiovascular, the same approximate 20 percent hazard reduction is seen that was seen for death overall, then one must deduce that many of the non-cancer (and even cancer) deaths may in fact have been vascular. In the absence of the results for suicidal, accidental and other non-cancer deaths, it may be suspected that many of the non-cancer deaths – eg outside the core trial hospitals- were in fact cardiovascular, although classified as other because of the strict CVD death criteria. So the Crestor-related reduction in CVD deaths may well have been far less than 20%, despite the reduction in other events..
Since JUPITER also does not disclose the changes in girth or weight in the two trial arms, or what added other systemic therapy these patients required on the trial, the fact that there was also 25% increase in incident diabetes also suggests that apart from lipidemia, Crestor if anything worsened rather than improved metabolic syndrome. there was likely even higher conversion rate to metabolic risk factors eg ie rise in body fat and waist girth on Crestor.
But since (like the Women’s Health Initiative) the JUPITER trial was stopped prematurely despite there being no major reduction in all-cause or CVD mortality or demonstrated increased risk, and no disclosure of what the non-cancer deaths were from, no conclusion can be reached about the risk-benefit ratio of Crestor for LONG TERM prevention– which is what primary prevention (where compliance is notoriously low) is about- ESPECIALLY with very low baseline LDLC BUT also raised CRP. .At best we see that reduction in number of deaths deaths was matched by similar number of new diabetes cases.
We know that like Crestor in JUPITER, metformin in polycystic ovary syndrome also halves CRP. But as we have reported previously below, unlike Crestor or any statin, metformin in appropriate titrated dose has an excellent safety record, almost halves morbidity and mortality in diabetics, and about halves new diabetes in preventative studies as well as reversing weight gain.
This unjustified positive media spin for Crestor (but ignoring metformin) is particularly worrying since Crestor has been in use only since 2001 ie for about 7years; and since the drug was tested also on many South Africans and is marketed in RSA apparently for primary prevention. Remember that it is not a decade since cerivastatin Baycol crashed due to drug-related deaths – after the same time in use as Crestor has been .Yet Astra-Zeneca reports Crestor sales of about $3.5billion for the past year, their third-biggest raincheck on gross annual sales of close to $30billion.