CABG from NEJM on Thanksgiving

After a weekend of gluttony, you’ll be happy to know that the invasive cardiology industry (any field that records roughly $50 billion in sales each year deserves to be dubbed an industry) has undergone major changes in recent years. If you have severely blocked heart arteries, you are much less likely to get bypass surgery today (where they remove a section of vein, usually from the leg, and use it to replace one or more blocked arteries leading to the heart). According to the Agency for Healthcare Research and Quality, the number of coronary artery bypass graft surgeries (CABG) declined to 308,000 in 2004, down 28 percent from 1997.

Instead, less invasive options like balloon angioplasty or stent insertions have taken up the slack. These procedures, known as percutaneous coronary interventions (PCI) because they use catheters inserted through the arm or groin to prop open the blocked artery, rose 36 percent to 791,000 procedures between 1997 and 2004.

Except for the fact that the number of people needing any of these options is still growing (largely due to the failure of Americans to take preventive measures like improving their diet, getting regular exercise, reducing stress, and quitting smoking), this shift in medical practice is a good thing. PCI is not only less invasive, it costs less and appears to save more lives.

A recent cost-benefit analysis by a group of Veterans Administration researchers (where would we be without the VA to conduct these studies?) compared the use of both approaches in high-risk patients. They randomly assigned 454 patients and found that those given PCI cost 18 percent less over five years compared to those given CABG ($81,790 versus $100,522). Moreover, their survival rate was slightly higher – 75 percent versus 70 percent.

Given these results, you’d think CABG would be disappearing even faster. But physicians and hospitals, especially those that depend on the high fees and lengthier hospital stays of the more expensive procedure, resist change. As Upton Sinclair reportedly said (If anyone can find me the original reference rather than Al Gore’s use of it in his movie, “An Inconvenient Truth,” please send it along): "It is difficult for those to see whose paycheck depends on them not seeing."

I was motivated to explore the CABG patch by a disturbing exchange of letters that appeared in the New England Journal of Medicine last week.

They involved the recent Food and Drug Administration advisory committee meeting that concluded the use of the blood clotting drug aprotinin, which is made by Bayer and known commercially as Trasylol (see this previous GoozNews), is safe. To understand the implications of this exchange, however, I have to spend a few moments explaining why this drug is used, since giving a blood-clotting drug to a person with blocked arteries is, as they like to say in the medical literature, counterintuitive.

The drug’s purpose, according to Bayer’s presentation at the meeting, is to reduce the need for blood transfusions during CABG operations. It also can reduce the number of people who have to go in for a second operation because the first hasn’t properly healed. Moreover, a small percentage of patients have adverse reactions to foreign blood factors, so reducing the need for blood transfusions can avoid this side effect.

While that all sounds positive, there has never been any evidence showing that any of these benefits actually improve patient survival. No matter, the FDA approved Trasylol in 1993 with the warning to physicians to watch out for the kidney damage that had appeared in a small percentage of patients in the company’s clinical trials. There is also concerns about patient “hypersensitivity” to the drug itself. Despite these warnings, and the fact that there are cheaper alternatives available, Bayer by last year had carved out a small but growing $250 million niche for Trasylol, largely in the invasive cardiology market.

In February of this year, however, two studies appeared in the NEJM that showed that Trasylol recipients had higher levels of kidney failure than patients who took alternative drugs to promote healing. One of the studies, by Dennis Mangano of the non-profit Ischemia Research and Education Foundation in San Bruno, Cal., also found that patients on Trasylol suffered higher mortality from additional heart attacks and strokes. Based on those two studies, the FDA issued a new warning and called for the September advisory committee meeting to see if the new evidence merited pulling Trasylol from the market.

In the run-up to the meeting, FDA reviewers asked Mangano for his group’s data. The chairman of the committee, William R. Hiatt of the University of Colorado School of Medicine, offered a long explanation in the current NEJM as to why the agency, and any researcher for that matter, would need access to the underlying data.

Mangano’s study, he explained, had come to very different conclusions than the other independent study that had just been published. It also differed from the new data that Bayer had submitted to the agency for the meeting. Since the Mangano study was an “observational” study, which means the careful randomization controls of a clinical trial weren’t present, there might have been other factors – like the possibility that patients in observed by Mangano were sicker than patients in the original clinical trials –- that influenced his results.

The FDA needed to see if Mangano had made appropriate adjustment factors. “Unfortunately,” wrote Hiatt:

Mangano did not give the FDA or the committee full access to his data, which would have allowed the agency to perform an independent review and analysis to validate his group's findings. Although there are many legitimate concerns with regard to the sharing of data (for instance, confidentiality and the need for informed consent), the lack of independent review greatly limited the committee's ability to draw conclusions from the study.

But the exchange of letters suggest that may not be how events transpired. According to Mangano:

The FDA asked that we present our data to them. We agreed and sent a team of IREF scientists to the FDA. The FDA and Bayer then asked that we send them our database, because they each wanted to combine our data with Bayer's data and then perform a number of unspecified analyses. We agreed to the FDA proposal, making three requests: that patient privacy and legal trust be preserved, that the methods used to mix our patient data with the Bayer patient data be specified, and that in addition to their analyses, the FDA independently analyze our data, comment on each finding of our study published in the Journal, and make the results known to the public. In May (4 months before the Advisory Committee meeting), frustrated by the delay and concerned that our data would not be reviewed, I abandoned all three requests and informed the FDA that we would provide our database and any and all source data for its review. Hearing no response, after several weeks, I wrote to the FDA and again asked for a review of our data. However, the FDA refused, and we were told that a review was not necessary. As I publicly stated to the Advisory Committee on September 21, 2006, we stand by our offer to the FDA to provide for review any and all data regarding our publication without constraint or condition.

A review of the transcript of the meeting confirms Dr. Mangano’s offer.

However, FDA reviewers R. Dwaine Rieves and Karen D. Weiss in their letter did not acknowledged that Mangano offered to make his data available without restrictions.

Dr. Mangano imposed restrictions, including his physical presence at the FDA while the FDA representatives inspected and analyzed the data, and limitations on the nature of the analyses the agency could perform. Furthermore, he required the FDA to make a determination of the validity of the published findings and conclusions rather than assess the methods alone. Under these terms, the FDA could not perform a substantive, independent evaluation of his statistical methods.

The advisory committee heard Mangano’s presentation, but in the end decided against pulling the drug from the market. I did not attend the meeting. But reviewing the transcript over the weekend was eye-opening. The panelists without exception ripped into Mangano’s methods. Specifically, they accused him of failing to properly adjust his data for various confounding factors that complicate every observational study. Then, John Flack of Wayne State University, who needed a conflict-of-interest waiver to participate in the meeting because he earned somewhere between $10,000 and $50,000 a year for serving on the speakers’ bureau of a Bayer competitor, asked point blank if Mangano would share his data.

Are you prepared now to basically let the FDA scientists analyze the data and go from there, because otherwise, I have serious reservations about what you are putting through here.

“I did,” Mangano replied. “I would give it to them tomorrow.” He then went on to explain why he was initially reluctant to cooperate with the FDA. The first two calls he received after his study appeared in the prestigious journal were from Bayer and FDA, in that order. He insisted on prior knowledge of the methods the FDA would use in analyzing his data because, he told the meeting:

If you are going to do a meta-analysis of everything and put everything into a pot, what is the design of that study, what are your endpoints, what are your questions, because you could pull out anything you want.

No one on the committee asked the FDA analysts present how they planned to use the data. The Bayer presenters – about a half dozen consultants helped analyze the company’s data – were similarly handled with kid gloves. Every public presenter – there were just three – spoke in favor of the company’s presentation. A epidemiologist from Indianapolis whose non-profit works closely with Eli Lilly was especially harsh toward Mangano. This epidemiologist, Stan Young of the National Institute of Statistical Science, also attacked the editors of NEJM for allowing the allegedly poorly-controlled Mangano study into the journal. “Where were the adults,” he asked.

At the end of the day, the committee voted unanimously to leave the status quo unchanged for the drug. Wrote Hiatt: “Having reviewed all the data available, the committee decided that there was insufficient evidence to require an additional warning on aprotinin's labeling and agreed that the clinical data supported an acceptable safety and efficacy profile for aprotinin.”

I know I’ve gone at length about this matter, but subsequent events cast a harsh light on the the mindset of the committee and the FDA officials at that meeting. A day after the meeting’s conclusions appeared on the newswires, a whistleblower inside one of Bayer’s contract research organizations called the agency to report that Bayer had conducted its own observational study, but its results and its data had not been given to the committee. That study also showed an increased risk from the use of Trasylol. The FDA quickly issued a milquetoast release saying it was “actively evaluating” the new information.

What’s the lesson here? For public health and safety groups that are pushing for better post-market surveillance systems to identify safety problems and to come up with good cost-effectiveness and comparative effectiveness data, be forewarned. Epidemiological “observational” studies are easily attacked if you have a mindset to do so.

Mangano and his colleagues around the world spent a decade coming up with the database that led to the study published last February in the NEJM. They did it without government or industry funds. (The $45 million study was funded by the foundation, which gets its money from an endowment that, according to Mangano’s testimony at the meeting, was created from excess funds generated by industry-backed studies.) They sampled 5,000 patients and collected over 7,500 datapoints for each. “This has been an entirely consuming effort of my life for the last 15 years,” Mangano said.

You’d think the FDA would work closely with such a researcher, giving him the assurances he needed to insure that they could get their hands on his data and do their own analysis. But at the end of his labors, the agency ignored him. And an FDA advisory committee, which included one consumer representative who never opened his mouth except to vote with the majority, ripped him a new asshole.

Posted by gooznews at November 27, 2006 08:47 AM