When the news sounds too good… statin, the new wonder drug
“It's a breakthrough study. It's a blockbuster. It's absolutely paradigm-shifting.” — Dr. Steven E. Nissen of the Cleveland Clinic in Ohio (Washington Post, November 9, 2008)
The news has been nothing short of breathless, as every major news publication has come out with a story on the clinical trial of the statin rosuvastatin (Crestor). Each news story has been more sensational than the last, describing the study’s results as dramatic, huge, spectacular and unprecedented and statins as the wonder drug of hope. The practice of preventive medicine will be forever transformed as a result of this landmark study and the clinical guidelines will be rewritten because of its extraordinary findings, we’ve heard.
The study, presented at a meeting of the American Heart Association and simultaneously published online at the New England Journal of Medicine, was said to provide powerful evidence that millions of healthy Americans can cut their risks for heart attacks and mortality in half by taking this statin, even if they have low cholesterol levels and no other risk factors for heart disease. It is also being reported as proving that a new blood test (C-reactive protein) can identify those who need to be on statins and now give doctors a reason to treat as many as 10 million healthy adults who they would never have had a reason to treat with statins before.
The clinical trial, Justification for the Use of Statins in Primary Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER), was a phase III, multi-center, randomized, placebo-controlled, double-blind clinical trial of 17,902 healthy seniors, average age 66. It examined the cardiovascular events and deaths among those on 20 mg rosuvastatin/day compared to those taking a placebo. (ClinicalTrials.gov #NCT00239681)
JUPITER was conducted at 1,315 medical centers in 26 countries. It was designed to run for four years and continue until 520 confirmed primary endpoints had been documented, said its authors, in order to provide a statistical power (90%) to detect a 25% reduction in the rate of the primary end point, with a two-sided statistical significance level p=0.05. But earlier this year, on March 29, 2008, the pharmaceutical company sponsor, AstraZeneca, announced that it was stopping the clinical trial after just 1.9 years because of “unequivocal evidence of a reduction in cardiovascular morbidity and mortality among patients treated with rosuvastatin compared with those treated with placebo.”
Stopping a major clinical trial halfway through is a big deal, as it invalidates any credible statistical analysis of the benefits and effectiveness of a treatment, and makes safety and longer-term risks impossible to evaluate.
A recent study of cancer trials led by Dr. Giovanni Tafuri, M.D., with the Italian Medicines Agency in Rome, reported that the number of clinical trials being stopped early, just as they start to show a benefit, has increased dramatically over recent years. Of the 14 randomized controlled clinical trials published in 2005-7 that had been prematurely stopped, all when they’d begun to show a benefit, 79% were used to support an application for marketing approval (through the Food and Drug Administration or European Medicines Agency). “This suggests a commercial component in stopping trials prematurely,” said co-author, Dr. Giovanni Apolone, M.D., head of the Laboratory of Translational and Outcome Research at the Mario Negri Institute for Pharmacological Research in Milan. Ending studies early risks overestimating the treatment’s effects, and too many are being stopped early without strong evidence of superior benefits, these authors concluded. It leaves insufficient data to make a credible statistical analysis of a drug’s effectiveness. “If a trial is evaluating the long-term efficacy of a treatment,” they wrote, “short-term benefits, no matter how significant statistically, may not justify early stopping. Data on disease recurrence and progression, drug resistance, metastasis, or adverse events, all factors that weight heavily in the benefit/risk balance, could easily be missed.”
Today, we learned that the actual (absolute) difference in mortality between the statin and control groups after nearly 2 years in the JUPITER trial was only 0.25% and incidences of heart attacks and strokes differed by a mere 0.35% a year. Looking more closely at the design of this trial, its methodology, selection of participants, attrition, findings and adverse events; as well as the evidence of C-reactive protein (CRP) and safety data of Crestor to date, suggests that all of the amplified claims being heard deserve a much more cautious look.
Selection of participants — results not typical
“This is a new way to prevent cardiovascular events in an entirely new population that we have been missing… something for the asymptomatic older person for the prevention of heart disease that now has greater benefit than anything we have ever had in the past.” — Dr. Douglas Weaver, president of the American College of Cardiology (Time Magazine, November 9, 2008)
While this study is being said to provide evidence for preventing heart disease in millions of seemingly healthy asymptomatic adults, the participants in this study were not at all representative of the general population, even of healthy people without high cholesterol. In fact the selection process in this trial was so rigorous, it screened out 8 of 10 seniors recruited, including anyone with anything that might “compromise safety or the successful completion of the study.” Its results certainly cannot be extended to most people in the clinical setting.
Adults were excluded from participating if they had any history or evidence of cardiovascular disease; if they had any inflammatory condition such as arthritis, lupus or inflammatory bowel disease; diabetes; hypothyroidism; high blood pressure; any evidence of liver or kidney dysfunction (including abnormal lab values); had a history of cancer in past 5 years; were on hormone replacement therapy; had a history of alcohol or drug abuse; if they’d ever been on lipid-lowering meds or were taking any immunosuppressant medications or oral glucocorticoids; if their triglyceride levels were >500mg/L or LDL-cholesterol levels >130mg/dL; and another 856 people for unstated reasons. The participants’ CRP levels also had to be ≥2mg/L (which describes two-thirds of seniors their age). The men had to be 50 years of age or older, and the women 60 years of age and older.
From 89,890 people initially screened, 23,231 then underwent a four-week trial on the placebo, meaning anyone exhibiting potential compliance problems could be screened out. The authors excluded another 1,521 during this run-in test. Finally, 17,802 people (38.2% women) were randomly assigned to receive daily doses of rosuvastatin or a placebo.
While the authors said they only lost 81 people in follow-up during the 1.9 years, compliance rates were unusually low. At the time the study was terminated, only 75% of participants were taking their study pills, they reported. Having nearly 15% of participants a year stop taking their pills is higher than any other statin trial and if this rate had continued, there would have been few people left participating after five years. The authors don’t reveal which arm of the trial these drop-outs occurred in, but people so rigorously screened and motivated to participate in a clinical trial don’t independently stop taking their pills without some reason. This lends added precaution in interpreting the side effects reported in this study, especially in light of the safety problems reported to the FDA, as we’ll examine.
Results with a critical eye
“This is unprecedented,” said Dr. Steven Nissen, chair of cardiovascular medicine at Cleveland Clinic in Ohio, as quoted by Time Magazine. “I have never seen a result of this magnitude reduction in risk.” The results were significant enough to stop the study three years early.
The only thing we’ve heard in the news have been the changes in odds ratios or relative risks, not the more clinically meaningful absolute differences in all-cause mortality and measurable cardiovascular events. In fact, the actual death rates among this unique cohort of seniors were so extraordinarily low, those on the statin differed by only 0.25% (1 death per 100 in the treatment arm, versus 1.25/100 in the placebo arm).
The total number of deaths (cumulative deaths in Chart D) were so low no difference is even visible.
Only when they created a magnified graf, enlarging the y axis and condensing the x axis, could any difference be seen at all. Then, the trajectory of deaths among those on the statins, had the study continued, would have exceeded the control group. But the study was stopped just as the lines met.
Note, too, that the 2-year data only reported on less than half of the study population — about 4,300 people in each arm of the trial. Most of the people reported on in this JUPITER trial had only taken the statin for under a year. This is not reassuring evidence of the long-term effectiveness of a drug being proposed for healthy people to take for life.
The cardiovascular-related deaths were 0.45% among those on the statin versus 0.85% among the placebo group. This is being reported as the statins cutting heart disease deaths in half. But what isn’t being reported is that the deaths from other causes were 27% higher in the statin group, leaving a marginally statistical difference in overall death rates. More unsettling, the report failed to report the causes for any of the 186 noncardiac-related deaths.
While this JUPITER study population was healthy, asymptomatic people with low risks for heart disease, this trial reported benefits quickly and contradict other rosuvastatin trials done for primary-prevention or for high-risk patients taking the statin for considerably more years. Those have failed to find a notable benefit of this statin in reducing cardiovascular events, deaths or all-cause mortality, according to Dr. Michel de Lorgeril at the School of Medicine of Grenoble University in France pointed out in the clinical discussions at the NEJM. In the 3-year CORONA clinical trial on heart failure patients, for example, the statin reduced LDL-cholesterols and CRP levels as targeted, but failed to find any benefit in cardiovascular deaths, heart attacks or strokes. The GISSI-HF trial of all-age adults with chronic heart failure also failed to show any benefit in clinical outcomes after about 4 years on rosuvastatin.
Regarding the cardiovascular events reported by JUPITER, they included hospitalizations and procedures, which are more subjective measures at the discretion of a physician and make the results more challenging to evaluate as supporting reductions in cardiovascular events. The total numbers of fatal heart attacks and strokes, for example, were no different between the study arms (a total of only 12 deaths in each arm of the trial). The incidences of clinically-diagnosed nonfatal heart attacks and strokes after 1.9 years were also exceedingly low — a total of 0.7% versus 1.4% of the participants in the statin and control arms, respectively. Numbers too under powered to credibly extrapolate to real-world clinical practice.
Even including all of the cardiovascular events, procedures and hospitalizations reported by the authors, 120 people in this study were treated for 1.9 years to prevent a single event, according to Dr. Mark A. Hlatky, M.D., of Stanford University School of Medicine in California.
Clinical practice guidelines are not usually changed (ideally, at least) until a study’s results have been independently replicated and the benefits have been shown to exceed the risks. This is especially critical before putting millions of healthy, asymptomatic people on a drug for the rest of their lives. Given the unimpressive absolute reductions in overall deaths or cardiovascular events, weighing the potential risks and safety consequences, and costs become more important…
The JUPITER authors said that the “reported serious adverse events were similar in the rosuvastatin and placebo groups.” Even if we disregard the fact that one in four of the study participants had already stopped taking their study pills for some reason, we might question how a biologically inert placebo would elicit nearly the same side effects and at the same levels as an active chemical. But the findings among the serious events they selected to monitor and report suggest that problems might already have been developing among those taking the statin. While not yet statistically significant, those on the statin were reporting slightly more muscular weakness and pain, kidney and liver disorders, and more patients had elevating lab values (creatinine, alanine aminotransferase, and hematuria, which are indications of possible kidney and liver damage). One case of nonfatal rhabdomyolysis in a patient taking rosuvastatin, diagnosed after the trial had been suspended, was also reported. These adverse events are especially noteworthy, given how carefully the study participants were specifically screened to eliminate those with any indications or potentials of such problems.
Why was that?
AstraZeneca, the maker of Crestor, has had a history of safety problems with this statin. In 2001, when it first filed its FDA application to market rosuvastatin, it was delayed when the company had to halt worldwide pre-approval clinical trials due to serious adverse events of kidney damage, kidney failure and rhabdomyolysis among people taking 80 mg/day and the FDA requested more clinical data from the company. It was later approved in lesser strengths, but in 2005, the FDA issued a Public Health Advisory on Crestor and revised the package insert Warning notice, as a result of phase 4 studies reporting muscle toxicity (myopaty/rhabdomyolysis) and kidney damage and kidney failure, especially at the 40 mg dosage. The risks were greater among those over age 65 and in those patients who are candidates for statin therapy, according to the FDA, such as people with diabetes, hypertension, atherosclerosis and heart failure.
Dr. Eliot A. Brinton, M.D., Director of the Metabolism Section of Cardiovascular Genetics, and Associate Professor at the University of Utah School of Medicine in Salt Lake City, in Medscape Cardiology discussed why the safety of statins is so important.
Safety problems with cerivastatin (Baycol, Bayer) ultimately led to it being taken off the market and the same concerns have resurfaced with rosuvastatin, he said. “Medication safety is always an issue, but especially so with the statins, for several reasons. First, …statins are the single most prescribed category of agents, in dollar value, in the United States today. Second, they are prescribed for long periods of time… A third and very important factor is that statins are most commonly used in middle-aged or elderly patients, who tend to be taking many other medications for other problems. This heightens safety concerns,” he said. Clinical trials typically report lower rates of these complications than seen in the real world because eligibility criteria exclude patients with potential problems, whereas healthcare providers cannot typically exclude those patients in clinical practice.
There was one statistically significant adverse effect of the statin reported in this JUPITER trial. There was a 25% higher number of newly diagnosed cases of diabetes among the statin group compared to placebo (270 cases versus 216 cases, 3.0% versus 2.4% of participants, respectively). In other words, the actual higher number of people who got diabetes cancelled out the fewer number who had a nonfatal cardiovascular event. There were also small, but significant elevating levels of glycated hemoglobin among those on Crestor, although the authors stated that “the incidence of newly diagnosed diabetes and worsening glycemic control have been reported in previous trials of pravastatin, simvastatin, and atorvastatin,” too.
The bottom line, by suspending this trial prematurely, any ability to evaluate the long-term safety risks over benefits for healthy people has been hindered. Yet, almost as if predicted by Dr.Tafuri and colleagues several years ago, the New York Times reported today: “The company plans to use results from the Jupiter study to seek FDA approval to widen its claim about Crestor’s effectiveness.”
A new risk factor?
With JUPITER, cardiologists are finally acknowledging that the cholesterol hypothesis is unsupported. As Dr. James Stein, M.D., from the University of Wisconsin Medical School in Madison told Heartwire yesterday, he praised the JUPITER study for exposing that current therapeutic LDL-cholesterol levels are arbitrary, but more importantly, a poor indicator of cardiovascular risk. “Many patients with heart attacks have normal LDL-cholesterol values,” he said, adding that doctors and patients have been lulled into a false sense of security with normal LDL cholesterol levels.
This brings us to the most controversial aspects of this trial: assertions that it has produced powerful evidence that a single blood test can spot seemingly healthy people who are at increased risk for a heart attack or stroke. A new health risk index.
As lead JUPITER investigator Dr. Paul M. Ridker, M.D., with the Center for Cardiovascular Disease Prevention at Brigham and Women's Hospital in Boston, and colleagues wrote: “Measurement of high-sensitivity C-reactive protein, an inflammatory biomarker that independently predicts future vascular events, improves global classification of risk, regardless of the LDL cholesterol level. We have previously shown that statin therapy reduces high-sensitivity C-reactive protein levels and that among healthy persons, patients with stable coronary disease, and those with the acute coronary syndrome, the magnitude of the benefit associated with statin therapy correlates in part with the achieved high-sensitivity C-reactive protein level."
The news this weekend has been filled with quotes from cardiologists in breathless excitement about this new health index that seems to support statin use for vast numbers of new patients:
“For the cardiology world, discovering a major new risk factor as well as an effective treatment is like hitting a walk-off home run to win the World Series.” — Dr. Eugene Braunwald at Brigham and Women's Hospital (Medical News Today, November 10, 2008)
“JUPITER should dramatically change prevention guidelines, the bottom line here is simple — if your hsCRP is high, you should be on statin therapy regardless of your cholesterol level. This is an approach we can start using tomorrow.” — Dr. James Willerson, Director of the Texas Heart Institute in Houston (Medical News Today, November 10, 2008)
“This takes prevention to a whole new level. Yesterday you would not have used a statin for a patient whose cholesterol was normal. Today you will.” — Dr. W. Douglas Weaver, president of the American College of Cardiology. (Washington Post, November 9, 2008)
“This changes medical practice in a major way. People are going to flock to their doctors to get their CRP measured, and if it's elevated they will say, 'Here's this drug you can take,' We'll save many lives and a lot of money.” — Dr. Steven E. Nissen of the Cleveland Clinic in Ohio (Washington Post, November 9, 2008)
“Although we have known for several years that elevated CRP was a risk factor for heart disease, until now we did not have evidence that targeting CRP could reduce disease burden,” said Dr. Nissen. In fact, “we must now begin to think of CRP as an accelerator of disease activity, not just a marker associated with high risk.” He didn’t just say that. He said it more than three years ago in a 2005 Cleveland Clinic press release. He was lead investigator for the Reversal of Atherosclerosis with Aggressive Lipid Lowering (REVERSAL) trial suggesting CPR be added to therapeutic biomarkers. What has not been explained in mainstream media this weekend, is that he and Dr. Ridker have led the push for clinical guidelines calling for testing and reducing CRP levels, even among people with low cholesterol levels. As Dr. Ridker told attendees at the American Heart Association meeting, the “potential public health benefits are huge.”
“These are findings that are really going to impact the practice of cardiology in the country,” said Dr. Elizabeth G. Nabel, director of the National Heart, Lung and Blood Institute. She was also widely quoted thie weekend as supporting CRP testing and saying they will be reviewing the data before deciding how to revise federal guidelines on CRP testing and statin treatment guidelines.
It is premature to change clinical practice guidelines on this study. The key problem with such suggestions is that JUPITER provided no such evidence at all for CRP testing, of CRP as a measure of cardiovascular risk, or for the use of CRP values in the prevention or clinical management of healthy people to reduce cardiovascular events or reduce premature deaths.
Why? JUPITER was a trial of statin treatment, not of CRP testing. To make any conclusions of the clinical merits of CRP levels would require a randomized, clinical trial comparing patients with high and low lab values.
“The trial did not compare subjects with and those without high-sensitivity C-reactive protein measurements, nor did it compare the use of high-sensitivity C-reactive protein with the use of other markers of cardiovascular risk. It also did not ascertain whether subjects with a high-sensitivity C-reactive protein level of less than 2.0 mg per liter would benefit from treatment,” wrote Dr. Hlatky. “There is increasing recognition that laboratory and screening tests need to be evaluated according to their effects on clinical management and outcomes, not just risk levels,” he said.
Not only that, but CRP has already been shown to not be the cause of heart disease! Just last week in the New England Journal of Medicine, for instance, Dr. Borge Nordestgaard, M.D., DMSc, professor and chief physician at Copenhagen University Hospital in Denmark, and colleagues, reported their large investigation to specifically see if people with naturally high CRP levels have more heart disease and strokes. They first measured the CRP levels in more than 10,000 people and found high levels associated with 30% and 60% higher risks for strokes and heart disease, respectively. But they then specifically looked at CRP genes and measured CRP levels in nearly 32,000 people, and compared those who actually had heart disease and strokes, compared to people who remained disease free. “People with the most active CRP genes were at no higher risk for heart disease and stroke than were people with the least active CRP genes,” they wrote. CRP is a risk factor, but doesn’t have a causal role in heart disease.
This means CRP does not cause heart disease, said cardiologist Dr. Heribert Schunkert, M.D., director of Germany's Luebeck University Hospital and professor of cardiology at the University of Leicester in England. “It [the Denmark study] is pretty definitive,” he said. “This is a nail in the coffin for the idea that CRP is a causal factor in heart disease,” said Dr. Thomas A. Pearson, M.D., Ph.D., MPH, senior associate dean for clinical research at the University of Rochester Medical Center. He had led a National Panel on CRP Testing finding that there is no evidence to date to support that making a treatment decision based on CRP improves survival. “You don’t treat risk markers,” he said.
Even earlier studies were unable to support the hypothesis of CRP as predictive of heart disease. Dr. Michael Miller, M.D., at the University of Maryland in Baltimore and colleagues analyzed NHANES data and found that high CRP in the absence of traditional risk factors is exceedingly rare and present in only one in 2,000 adults in the U.S. Given the prevalence of heart disease, this suggests that “high CRP is a biomarker rather than a direct promoter of coronary heart disease,” he said, and of limited clinical utility as a screening tool.
More recently, researchers reported that CRP levels added minimally to risk predictions and failed to predict which elderly people in the multi-national PROSPER study benefited by statins for primary or secondary prevention, during 3.2 years of follow-up.
Doctors Steven Woloshin, M.D., and Lisa Schwartz, M.D., at the Veterans Affairs Medical Center in White River Junction in Vermont, also countered suggestions that CRP is a valuable risk marker for cardiovascular disease. They analyzing CRP levels among a representative sample of 21,004 U.S. adults measured in NHANES from 1999 through 2002. They found levels of CRP ranged from 0.1 to 296 mg/L, with a mean of 4.3. “Recently, researchers have begun to use a CRP level of 2 mg per liter or greater as the threshold for defining high cardiovascular risk,” they wrote. “The proportion of people with a level of 2 mg per liter or higher is substantial at all ages (e.g., 41 percent among those 20 to 29 years of age versus 62 percent among those 80 years of age or older). Half of U.S. adults have levels greater than 2 mg per liter, highlighting the need for caution in accepting this threshold as the definition of high risk before there is clear evidence of benefit from randomized trials.”
C-reactive protein is a protein that can rise with short-term infections, injuries or inflammatory process anywhere in the body. Elevated levels are found associated with everything from anger and stress to arthritis, cancer, lupus, connective tissue disease, inflammatory bowel disease, pneumonia, tuberculosis, with oral contraceptive and pregnancy, heart attacks, surgery, trauma, burns, strenuous exercise, and countless conditions.
Marketing or the best science?
Prevention sounds good, but the JUPITER study wasn’t a CRP study, so why is it being interpreted as one and led to calls to test every healthy adult for CRP and put anyone with levels above 2 mg/L — half of every adult in the country — on statins for life? When claims go beyond the science, then other possible influences warrant a look.
The lab test itself costs $50-$80 apiece, according to Business Week and AstraZeneca's stock climbed 45% after JUPITER was halted last March. Crestor costs $1,400 a year, according to Forbes. If publicity of this trial increases the number of people taking statin, business analysts estimate that AstraZeneca’s already $3.5 Billion in annual sales for Crestor will double over the next five years. But if the clinical guidelines are changed, suggesting as is being said that 7 to 10 million more American adults would warrant statins; to meet federal guidelines (those performance measures), that could potentially mean $14 Billion a year for the drug company.
According to the JUPITER disclosure: “The trial was financially supported by AstraZeneca. The sponsor collected the trial data and monitored the study sites but played no role in the conduct of the analyses or drafting of the manuscript and had no access to the unblinded trial data until after the manuscript was submitted for publication.” The authors’ full disclosures are below.*
The lead author, Dr. Ridker at Brigham and Women’s Hospital co-invented a CRP test, with the patent held by Brigham and Women’s Hospital and patent rights have been licensed to Dade Behring and AstraZeneca. Dr. Ridker told the Washington Post that he and his hospital receive royalties from the CRP test. The newspaper reported that that was no reason to doubt the study’s findings.
I can’t even do the math for the potential royalties on doing a CRP screening test on half of the adults in this country. But add that to the costs for all of the labwork for monitoring patients on statins for myopathies, kidney and liver damage. Then, there are the costs of caring for the 300,000 new cases a year of diabetes this study suggests… Add that to our country’s healthcare and government public health expenditures that are already in crisis, glutted by waste and ineffective preventive health interventions.
This illustrates the importance of science and basing public health policies and clinical guidelines on the soundest clinical evidence, with proven benefits that outweigh the costs and risks for patients. The price of not doing so are too great.
© 2008 Sandy Szwarc