Search Annals: Advanced search

Podcast Transcript - October 16, 2007

[ Return to Podcast Home Page ]

Hello, and welcome to this week’s Annals of Internal Medicine audio summary for our October 16, 2007 issue. I’m Michael Berkwits, Deputy Editor at Annals.

We have another great issue for you this week, with articles on aspirin and venous thromboembolism, duration of treatment for H. pylori, hypercortisolism as a potential cause of idiopathic osteoporosis, and new standards for reporting observational research. And, we’ll be revisiting the issue of the safety of rosiglitazone, including a replay of my interview with two authors who presented their analyses of safety data to the FDA Advisory Subcommittee when they met in July to determine what to do about the drug.

But first, an in-depth summary of this week’s articles.

Our lead article is about the effect of low dose aspirin on venous thromboembolism. Aspirin is nowhere near as effective as anticoagulants for preventing VTE, but it is more effective than placebo in people at high risk for venous thromboembolism. Its effect in the lower-risk or no risk populations has never really been studied.

In this week’s report, lead author Dr. Robert Glynn and his coauthors from the Brigham and Women’s Hospital in Boston used data from Women’s Health Study, a large trial that randomized almost 40,000 women to receive 100 mg of aspirin every other day or placebo, to assess the effect of low dose aspirin on the incidence of venous thromboembolism, a prespecified secondary outcome. The original trial report showed that aspirin reduced women’s risk of ischemic stroke but had no significant effect on cardiovascular events.

This week’s report has 3 main findings.

First, about 480 women developed venous thromboembolism over about 10 years, which were mostly cases of DVT, but included 100 cases of PE, and 76 instances of both DVT and PE. These events were nearly evenly split between the aspirin and placebo groups, however, with no clinically or statistically significant differences between groups.

Second, there were no significant differences even within multiple subgroups, such as women with idiopathic events and those with thrombotic diatheses, such as Factor V Leiden or prothrombin mutations.

Finally, as detailed in the original 2005 trial report, the women who took aspirin were more likely to develop peptic ulcers and bleeding, including, from head to toe, epistaxis, GI bleeding, hematuria, and easy bruising.

On the basis of these findings, the authors conclude that 100 mg of aspirin every other day has little effect on preventing venous thromboembolism in healthy women. They acknowledge that they looked only at symptomatic venous thromboembolism, and that the dose in the study may have been too low to have had a protective effect, but they make the good point that even if higher doses were more effective, they would certainly also lead to more adverse events. And while no one is really prescribing aspirin solely to prevent venous thromboembolism, the authors suggest that the clinical relevance of the negative findings lies in the demonstration that aspirin isn’t harmless even when taken at a low dose every other day, and that clinicians who might take some reassurance that patients taking aspirin for other reasons are protected against venous thromboembolism, shouldn’t. But all the nails aren’t yet in the coffin for the use of aspirin in venous thromboembolism prevention. It may have a role for secondary prevention in patients in resource-poor settings, where warfarin and coagulation monitoring are generally unavailable, and two ongoing trials, one in Europe and one in Australia, are evaluating if, after an appropriate initial period of warfarin, aspirin might adequately prevent venous thromboembolism recurrence but still be safer than warfarin. For sources for those trials, see references 24 and 25 of this week’s article.

Our CME article this week is a meta-analysis of triple therapy for the eradication of Helicobacter pylori infection.

Triple therapy, comprising a proton pump inhibitor, clarithromycin, and either amoxicillin or, in pen-allergic patients, metronidazole, is the most commonly recommended first-line treatment for H. pylori infection, but the optimal duration of treatment remains controversial. Two previous meta-analyses have concluded that 14 days of triple therapy is modestly better than 7 days of treatment, but the more recent of the analyses was published in 2003, and 9 trials have been published since.

In this week’s article, lead author Lorenzo Fuccio of the University of Bologna and his coauthors updated the literature with their own literature search and meta-analysis. They searched for trials that compared different treatment durations and found 21 trials that met their criteria.

Their analysis had four main findings

First, the methodologic and reporting quality of most of the trials included in the analysis was low.

Second, the frequency and relative risk of H. pylori eradication was slightly higher with 14 compared to 7 or 10 days of treatment, but the effect was detectable only among patients taking amoxicillin, and not metronidazole, and the effect disappeared when the analysis was restricted to the 4 trials that were of high quality.

Third, the relative risk of H. pylori eradication was slightly higher with 10 compared to 7 days of treatment in the subgroup of patients with non-ulcer dyspepsia. In contrast, no difference in effect by treatment duration was detectable in the subgroup of patients with peptic ulcer disease.

Finally, diarrhea and changes in taste were the most frequently reported side effects, but there were no detectable differences in the frequency of these or other adverse events by treatment duration.

Taking the findings and the quality of the reviewed trials into account, the authors conclude that longer treatment durations are unlikely to be better than 7 days of triple therapy. They acknowledge that the conclusion diverges from those of the previous 2 meta-analyses, but emphasize that those analyses did not fully account for trial quality, publication bias, or adverse effects of therapy. They call for higher quality studies, studies of triple therapy using metronidazole, and studies that stratify by indication, i.e., by treatment for ulcer disease and non-ulcer dyspepsia. And they close by saying that their analyses demonstrate that longer treatment durations preclude the need for a second course of therapy in only 5% of individuals at most, so that cost-effectiveness studies will also be required before treatment beyond 7 days can be recommended.

On an editorial note, the issue of treatment duration is separate from that of whether treatment should be given concurrently or sequentially over the treatment period, a question which was addressed in a randomized trial published in the April 17, 2007 issue of Annals. One interpretation of evidence from trials comparing sequential to continuous therapy is that 5 days of a PPI and amoxicillin followed by 5 days of a PPI and clarithromycin is clearly more effective than 7 days of continuous triple therapy. The issue is addressed in an editorial in the September 22, 2007 issue of The Lancet, which highlights the need for trials to compare sequential therapy using different regimens to continuous therapy of different durations using different available regimens. So clearly, the work in this area continues.

Subscribers to Annals can receive CME credits by answering 2 quiz questions about this article. Just go to cme.annals.org, and click “My CME” or register to sign in.

This week’s issue also has two articles on the safety of rosiglitazone. The safety of the drug has been questioned since it was reported to increase the risk of myocardial infarction in a meta-analysis earlier this year. In the first of this week’s two articles, authors George Diamond, Leon Bax, and Sanjay Kaul point out that the statistical test used to compare outcomes in that original meta-analysis may have been inappropriate because of imbalances in the number of patients assigned to treatment arms.

Listeners who heard the summary of this article and my interview with two of its authors when they were released online in early August can go directly to the next segment of the summary by advancing to about the 18:40 marker on their mp3 or audio player.

The authors emphasize that the analysis did not use data from trials in which no one had an MI in either study arm. Trials without events still provide some useful information about the frequency of those events, they say, especially when they’re combined with information from other trials. The authors reanalyze the trial data using data from the trials with no events and find that the risk estimate for myocardial infarction is less than that reported in the original meta-analysis, and no longer statistically significant. Their primary point is that the risk estimate for MI, and the degree of uncertainty in that risk estimate, is highly dependent on which statistical test one uses, and how one accounts in the analysis for trials where no one experienced an MI. Their primary claim is that the available data to do not say anything definitive one way or the other about the risk for myocardial infarction with rosiglitazone.

To better understand these findings, I spoke with two of the authors of the article, Dr. George Diamond and Dr. Sanjay Kaul. Dr. Diamond is a Senior Research Scientist emeritus at the Cedars-Sinai Medical Center in Los Angeles, and Dr. Kaul is in the Division of Cardiology at Cedars-Sinai Medical Center and the David Geffen School of Medicine at the University of California in Los Angeles. The two are frequent contributors to Annals, and the coauthors of numerous other papers looking at the methodological underpinnings of cardiovascular research.

An editorial reminder for listeners: this interview was recorded in August, the week after the FDA Advisory meeting, and the authors’ upcoming comments on pioglitazone were made before publication of two meta-analyses, one in the September 12, 2007 issue of JAMA, and one in the September 29, 2007 issue of Lancet, suggesting that pioglitazone reduces cardiovascular mortality and increases heart failure, but not heart failure mortality.

Q: Dr. Kaul and Dr. Diamond thank you for joining me.

SK: Good morning.

GD: Morning

Q: You write in your paper that the risk of cardiovascular disease for patients with diabetes taking rosiglitazone is uncertain and that the available evidence does not justify the need for urgent evaluation of its safety. How do you come to those conclusions?

SK: Well there are three key elements that contribute to our conclusions. First, most of these trials were not designed to address safety concern as primary endpoint. The majority of the trials were around six months in duration, the outcome events were rather sparse and the treatment arms were unbalanced with respect to sample sizes as well as zero events, with more zero events occurring in the control treatment arm. Second, there was sufficient clinical heterogeneity in patient populations, design protocols and treatment regimens that might arguably preclude pooling of the studies. Third, the method used by Nissen and Wolski in their meta-analysis tends to overestimate risk, especially if there is an imbalance in zero events that favors the control treatment arm. So corrected analyses in this case tend to reduce risk estimates and the statistical certainty around it compared to uncorrected analyses. All of these issues contribute to a greater uncertainty in risk estimates than was reported in the original meta-analysis.

GD: The low level of outcome events and the high level of clinical heterogeneity produced substantial uncertainty in this case. The available evidence establishes neither an increased nor decreased level of risk. We simply need better data and more refined methodology to properly adjudicate the association between rosiglitazone and cardiovascular risk.

Q: So did the FDA Advisory Committee get it wrong by voting 20 to 3 on July 30th that rosiglitazone increases risk for MI?

GD: Yes and no. More likely than not the risk is increased but the magnitude of the increase is likely to be small and its clinical importance remains open to question. The committee appears to appreciate this, explaining why they also voted 22 to 1 not to take the drug off the market. The two votes represent a reasoned understanding of the inherent uncertainties.

SK: The advisory panel’s vote has to be interpreted in the proper context. There were two critical questions. The first question that was asked was, “Is there any evidence of increased risk?” Yes. We agree. “Is it certain enough to warrant a withdrawal?” No. This happens to be no different than the stance we put forward in our paper. In the end, we feel it was the appropriate decision. Even though the vote is non-binding, the FDA can still decide not to follow the advisory panel’s recommendation.

Q: Your analyses were performed before the FDA advisory meeting. Did we learn anything from that meeting that would change your conclusions or your analyses?

SK: There are two new observations that we made during the FDA deliberations. One, we were made aware of certain clinical subgroups that may be at higher risk with rosiglitazone therapy, and this is those patients who were already on insulin therapy and it appears that the combination therapy of rosiglitazone plus insulin may pose a higher risk. The second point that we observed were that there were three meta-analyses that were discussed at the FDA meeting, all pointing towards an increased risk in myocardial ischemic events with rosiglitazone. However, in the FDA meta-analysis, as well as the GSK meta-analysis, when you stratify according to serious myocardial ischemic events as well as non-serious myocardial ischemic events neither of the two meta-analyses reported a statistically significant increase in the risk of serious myocardial ischemic events. Which is consistent with what we found, that even though the risk estimates were greater than odds ratio of one for myocardial infarction, they were no longer statistically significant.

Q: Where do heart failure and death from heart failure fit into this discussion given evidence from trials that rosiglitazone increases heart failure risk?

GD: Well rosiglitazone does increase heart failure risk. It’s already contraindicated in patients with heart failure and it’s not clear if any of the deaths in the meta-analysis are directly attributable to the development of heart failure in patients who entered the trial without any previous history of having the condition.

SK: It is true heart failure patients appear to be at a higher risk of developing complications with rosiglitazone. In fact amongst the 40 small trials evaluated in the Nissen and Wolski meta-analysis the highest number of heart attacks and cardiovascular deaths in the rosiglitazone arm occurred in one trial in diabetic patients with heart failure. We therefore questioned whether this trial should have been combined with other trials which predominately included only low risk patients. Similar observations were made in the FDA meta-analysis where they observed nearly a doubling of the risk for myocardial ischemia with rosiglitazone in patients with congestive heart failure.

Q: Is the risk of pioglitazone any different from that of rosiglitazone?

GD: We just don’t know. FDA will be performing its own analysis in the coming months.

SK: A similar meta-analysis has been performed with pioglitazone by the sponsors. However it has not been vetted by the FDA statisticians. There appears to be substantial differences between the rosiglitazone and pioglitazone clinical trials, all of which can have material impact on outcomes, thereby raising questions about the comparability of the two glitazones. Having said this, in the PROACTIVE trial that evaluated pioglitazone in high risk patients, neither any benefit nor any risk was observed in pioglitazone. Both drugs increase fluid retention which can potentially result in heart failure exacerbation. Thus both have been recommended by the FDA to receive black box warnings for CHF exacerbation. And there are some minor differences with respects to changes in lipid profile. There is less of an elevation in LDL with pioglitazone compared to rosiglitazone and there is greater elevation in HDL with pioglitazone. And there is an elevation of triglycerides with rosiglitazone compared with a drop with pioglitazone. However the clinical relevance of these differential changes in lipid profile with these agents is not completely known.

Q: So what would you tell a primary care provider caring for a patient with type 2 diabetes in their offices?

GD: I would emphasize that there are two goals of therapy in the diabetic. The maintenance of glycemic control, and the application of proven appropriate preventive measures. Glycemic control appears to be more important to the prevention of the microvascular complications of diabetes where lipid management is more important to the macrovascular complications. It is important to pay equal attention to both of these goals.

SK: I would leave it up to the judgment of my diabetes specialist colleagues as to what their choices of therapeutic agents would be to accomplish microvascular goals. I have been told repeatedly by my colleagues and it was very clear to me at the FDA deliberations that diabetes specialists and their patients want more not less tools to tackle the looming diabetes epidemic.

Q: Dr. Kaul and Dr. Diamond thank you so much for talking to me.

SK: Thank you very much.

GD: Thank you.

That was Dr. George Diamond, the authors of an article this week, about the risk of myocardial infarction with rosiglitazone.

Other articles in this week’s issue include a cross-sectional study of the prevalence of knee buckling or giving way in a random sample of adults in Framingham, MA, demonstrating that the symptom is not restricted to patients with anterior cruciate ligament or meniscal tears, but is present in about 12% of the population; and that it seems to be associated with knee osteoarthritis, higher BMI, weaker quadriceps; and with functionally significant disability, including limitations in work activities.

We have a cross-sectional study of subclinical hypercortisolism in a population of patients referred for the evaluation of possible osteoporosis, demonstrating that about 10% of those patients who had T scores < 2.5 and vertebral fractures had subclinical hypercortisolism caused by adrenal or pituitary adenomas. And we have an accompanying editorial discusses that asks, and answers the question about whether physicians should now be screening patients with osteoporosis for hypercortisolism.

And Richard Baron’s follow-up to his enormously popular 2005 report on the challenges his medical practice faced when it implemented an electronic health record. This week’s article details the successes and frustrations of using the system for the deceptively simple task of reminding more women in the practice that they needed mammograms.

And following in the footsteps of CONSORT, QUORUM, and MOOSE guidelines that have set and improved the standards of reporting clinical research, we have papers announcing a new acronym, STROBE, and a new set of guidelines for the reporting of observational studies. This is an impressive effort, and while the print journal this week has only the reporting guidelines, listeners are also encouraged to go to the on-line version at www.annals.org to find the background document. Spend time with that, and it will give you a great education about the perils of performing and interpreting much of the research that is published in Annals and in all clinical journals.

And finally, we have an Update in Endocrinology, the 6th of 10 updates appearing this year based on presentations given at the American College of Physicians annual Internal Medicine 2007 session held in April in San Diego.

Well that’s it for today.

For full details of all of this week’s articles, please consult your print journal, or go to www.annals.org.

Technical support for this summary was provided by Andrew Langman, Neil Kohl, and Beth Jenkinson.

Special thanks to Kevin Stahl and all our friends at WHYY, public radio and television of Philadelphia, who helped produce this podcast.

Check back in 2 weeks for a complete summary of our regularly scheduled November 6, 2007 issue.

I’m Michael Berkwits, and thanks for listening.

[ Return to Podcast Home Page ]