Sunday, May 27, 2007

Who Benefits from Glaxo's Drug Disaster?

Former chair of the FDA's Cardiovascular & Renal Drugs Advisory, Dr. Steven Nissen, whose challenges to the safety of several prescription medicines—both best-selling, Vioxx (rofecoxib), and drugs awaiting approval, Pargluva (muraglitazar)—limited or derailed their use, now has turned his sights on Avandia (rosiglitazone), a $2.2 billion diabetes drug from GlaxoSmithKline (GSK-$52.43).

In a
meta-analysis of 42 trials just published in The NE Journal of Medicine, the Cleveland Clinic cardiologist concluded, that as compared with placebo or with other anti-diabetic regimens, treatment with rosiglitazone was associated with a significant increase in the risk of myocardial infarction, and with an increase in the odds ratio of death from cardiovascular causes (not statistically significant).

Critics address these findings as another example of gaps in the FDA review process, specifically the system of post-marketing surveillance of drug safety, such as its reliance on voluntary reporting by drug manufacturers themselves, which by some estimates occur only about 10% of the time.

Sidney Wolfe, director of Public Citizen's Health Research Group, called the Avandia situation another example of why the FDA's division for evaluating postmarketing safety of drugs should be removed from under the auspices of the agency's Center for Drug Evaluation and Research and made an independent branch in the agency.

Nissen said that “patients and providers should consider the potential for serious adverse cardiovascular effects of treatment with rosiglitazone for type two diabetes.”

These findings represent a particular health concern because more than 65% of the deaths amongst diabetic patients are attributed to heart disease.

“Diabetes Drug Called a Potential Death Risk”USA Today

Albeit this story has been covered ad nauseam, the 10Q Detective is posting an article on Nissen’s analysis because we believe that the public—once again—has been done an injustice by newsrooms’ inflammatory coverage of the [publicity-seeking] cardiologist’s findings.

Remember the Maine
To hell with Spain.

Like the
yellow journalism that set us on a collision course to war with Spain in 1898, many journalists are sensationalizing Nissen’s opinions to sell newspapers and/or magazines—with little regard for those most likely to be impacted by the findings—type 2 diabetes mellitus patients currently taking Avandia (and who might make unilateral decisions to stop taking a drug controlling there blood sugars, without notifying their doctors).

You can't serve the public good without the truth as a bottom line. ~~ Watergate journalist Carl Berstein

In June 2004, more than 3,000 people
surveyed by the respected Pew Research Center expressed skepticism toward news outlets and those who run them. More than half (53%) agreed with the statement "I often don't trust what news organizations are saying." Nearly as many (48%) believe people who decide on news content are "out of touch."

The 10Q Detective argues that the Avandia controversey epitomizes why circulation continues to decline at national dailies—like The New York Times, LA Times, and The Washington Post. Trust issues and lazy reporting are the corrosive causes of subscription losses—not reader flight to the Internet.

Rarely were the statistical flaws/ endemic biases in Nissen’s methodology mentioned in press accounts:
  1. Restriction of coverage issue: Nissen’s trial selections were not free of bias--trials of Avandia in which patients had no adverse cardiovascular events in either group were excluded from analyses;
  2. The pooled results were culled from a group of trials that were not originally powered to detect cardiovascular outcomes;
  3. It has been observed that the conclusions of a meta-view can be shown to be different from a subsequent, larger, randomized clinical trial. Nissen’s observed results were based on a relatively small number of events (86 myocardial infarctions in the rosiglitazone group and 72 in the control group); and,
  4. Fallibility of broad applicability for Nissen’s observed conclusions. Nissen said—and the press ran with his one comment—that rosiglitazone was associated with an increase in the risk(s) of cardiovascular morbidity and mortality.

The mean age of the subjects was approximately 56 years and the mean baseline glycated hemoglobin level was about 8.2 percent. We note that reporters failed to ask a salient query, such as, “how long were the blood sugars of the observed patients uncontrolled?”

The 10Q Detective reminds our readers that (a) diabetics are more prone to CHD and other cardiovascular events and (b) the longer the trial participants had the disease, the greater the risk of an observed CVD event!

A more widely recognized source of potential bias that can affect every type of medical study, including a meta-analysis, is a financial one. As an example, a researcher authoring any type of study would routinely tend to have a personal bias favorable to the company's product underwriting the study.

Nissen did disclose that he consults for, and receives research support to perform clinical trials through the Cleveland Clinic Cardiovascular Coordinating Center from Eli Lilly (LLY-$59.36) and Takeda—marketers of Avandia’s key (and only competitor) in the
thiazolidinedione class, the $2.6 billion seller, Actos (pioglitazone).

It is not our intent to impugn Nissen’s deserved reputation. Nonetheless, few—if any!—news outlets reported his potential conflict of interest, or any suspicion that the study was done in order to achieve a desired result.

Why did not one reporter mention that the interpretation of a meta-analysis is potentially subject to an author’s inherent bias? When an advocate of a particular position conducts a meta-analysis, it is more likely to be biased in concordance with the author's previously advocated opinion, rather than refute one’s prior conclusions. [It is well known within the pharmaceutical community that Nissen has been a vocal critic of Avandia for years.]

Facts are stubborn, but statistics are more pliable. ~~ Author/Humorist Mark Twain (1835 – 1910)

On reading the NEJM meta-view, we were troubled with Nissen’s comment that the observed risks of rosiglitazone might not be a "class effect" of thiazolidinediones:

“Pioglitazone is a related agent also widely used to treat type 2 diabetes mellitus. However, unlike rosiglitazone, pioglitazone has been studied in a prospective, randomized trial of cardiovascular outcomes, called Prospective Pioglitazone Clinical Trial in Macrovascular Events (PROACTIVE). The primary end point, a broad composite that included coronary and peripheral vascular events, showed a trend toward benefit from pioglitazone (hazard ratio, 0.90; P=0.095). A secondary end point consisting of myocardial infarction, stroke, and death from any cause showed a significant effect favoring pioglitazone (hazard ratio, 0.84; P=0.027).”

In plain English, “a trend toward benefit” is not material. Specifically, the primary end point occurred in 21% of patients on pioglitazone, compared with 25.5% of the patients on placebo. Pioglitazone's total risk reduction of 10% was not considered statistically significant. The study was powered to show a 20% reduction in the composite of primary end-point events.

Albeit the secondary end point—death, nonfatal heart attack, and stroke—was statistically significant, Nissen failed to mention heart failure occurred more frequently in the pioglitazone group (10.8% of patients) than in the placebo group (7.5% of patients), with 5.7% versus 4.1% being hospitalized.

In our view, the Avandia affair illuminates, too, the ignorance of many of Wall Street’s (purportedly) ‘best and brightest” health-care analysts.

"We certainly do not see any negative spillover to Actos," said Merrill Lynch pharmaceuticals analyst Masatake Miyoshi in a note to clients. "If anything, we believe that evaluation of Actos' effectiveness in reducing cardiovascular events and its safety will only improve," he added.

As the precise mechanisms underlying the actions of thiazolidinediones (TZDs) are largely unknown, it is a quantum leap to say that one drug in the class is safer than the other.

Contrary to Nissen’s aforementioned “class effect” comment, there are no published, double blind-trials directly comparing the safety and efficacy of Avandia and Actos as monotherapy or in combination with other anti-diabetic agents. And, both TZDs can cause fluid retention, which may exacerbate or lead to heart failure.

To his credit, Nissen did note, “the rapidity and magnitude of the apparent hazard was not consistent with an effect produced by lipid changes alone.” In other words, the build-up of unstable plaque in one’s arteries is a life-long event—not the result of 26 weeks of drug-induced, lipid raising therapy.

Again, the press was silent on this latter point—choosing instead to link Avandia [itself] directly to CVD risk and death.

It seems to me that just in the ratio that our newspapers increase, our morals decay. The more newspapers, the worse morals. Where we have one newspaper that does good, I think we have fifty that do harm. We ought to look upon the establishment of a newspaper of the average pattern in a virtuous village as a calamity. –Mark Twain

Editor, David J. Phillips does not hold any financial interest in the companies mentioned in this posting. The 10Q Detective has a Full Disclosure policy.


Anonymous said...

Diabetes Care. 2005 Jul;28(7):1547-54.

Here is a peer reviewed randomized trial evaluating efficacy.

Steve said...

Misreading Avandia

The recent ruckus over the drug rosiglitazone (Avandia) has been portrayed as another case of Big Pharma foisting a dangerous drug on the public while the overworked FDA can’t keep up.

Search Google for Avandia and Vioxx, and you’ll come up with more than three-quarters of a million hits, including ads from law firms ready to litigate.

There is an enormous difference between the two situations. In the case of Vioxx, it was alleged that otherwise healthy people suffered sudden cardiac disease and death after using the pain reliever. This new study, published in the New England Journal of Medicine on May 21, alleges that Avandia causes an unacceptable rise in risk of heart attack. This—at least from the evidence presented—is hard to believe.

There are several ways to read the statistics that “support” this study.

Those in the Avandia meta-study were not only diabetics, but diabetics whose disease was initially out of control, with average blood sugar levels approaching three times those of non-diabetics and blood sugars at the ends of the studies undisclosed. The complications of diabetes—in this case heart disease—are caused by the prolonged elevation of blood sugars.

With a little math, we can easily conclude, for example, that metformin is about 4.75*-fold more likely to cause more cardiac deaths than insulin. We can also conclude that placebo is about 4.4**-fold more likely to cause more cardiac deaths than insulin. Which should lead us to the conclusion that since insulin is the most effective of these medications at lowering blood sugars, it is likewise more effective at reducing diabetic complications such as the risk of cardiac death.

The other drugs are considerably less effective in doing that job—but the sulfonylureas, while more effective than metformin or rosiglitazone in the short term, are problematic because they can over time, burn out remaining insulin-producing beta cells.

The real story behind these figures is that there are a lot of health care professionals who are not doing their jobs. Anyone with an initial (as in this study) HgbA1c level of 8.2 percent (equating to sustained average blood sugar levels of about 228 milligrams per deciliter) should be treated with insulin (and a low carbohydrate diet) since no combination of the other drugs can get such blood sugars normalized. The other drugs can have marginal effects in reducing blood sugar levels, but are not nearly so effective as insulin.

Blood sugars of diabetics can be normalized. The real ‘scandal’ is that very few health care professionals think that diabetics have the right to the same blood sugars as non-diabetics. They therefore leave their patients wide open to an unnecessary “potential for serious adverse cardiovascular effects.”

Send your comments to
* 5.37/1.13 = 4.75
** 5.37/ 1.22 = 4.4
Richard K. Bernstein, MD, FACE, FACN, FACCWS Mamaroneck, New York

Dr Bernstein is author of DIABETES SOLUTION, Little Brown, 2007
Reprinted from

David J. Phillips said...

Nice summation of Avandia & adverse events w/ other classes of hypoglycemic agents. My only issue--even though I am not a big fan of SUFH class--it has NOT been clinically proven that SUFH class produces "beta-cell burnout."

Anonymous said...

the burnoout of beta cells, although not unequivocal, can be seen with hte results of the ADOPT trial

David J. Phillips said...

There is a school of thought that postulates that beta-cell 'burnout' might be the result of years of high blood glucose levels [undiagnosed & untreated for years prior to treatment with SUFH class]

Robert said...

Dave, you did a BRILLIANT job summarizing both the weaknesses of the Nissen analysis (which was a political tool, not a scientific pursuit) and the penchant of the media to tee this up as "another Vioxx" Comments made by others were also informative. We at made our own contribution to the deconstruction of Dr. Nissen's work, most notably the fact that using meta-analysis to construct an exploratory hypothesis most always leads one in the wrong direction.

We will link to your post. Big fan of the site in general!

Bob Goldberg
Center for Medicine in the Public Interest