Carvedilol vs. metoprolol

In the continuing effort to demonstrate carvedilol’s superiority over metoprolol, comes a study in this week’s JAMA looking at metabolic side-effects of these two beta-blockers in diabetic hypertensives.

Patients with diabetes and hypertension, but without major cardiac disease, who were receiving ACE inhibitors or ARB’s, were first taken off their other anti-hypertensives (if any), but continued on the ACEI or ARB. They were then randomized to carvedilol or metoprolol tartrate, at increasing doses, until either blood pressure was adequately controlled or the maximum dose was attained. At that point, if BP was still not controlled, HCTZ and/or a calcium channel blocker was added to obtain satisfactory BP levels. The primary endpoint was change in HgbA1c levels; insulin levels and lipid levels were also looked at.

In both groups, adequate blood pressure control was achieved, with the same percentage of patients in each group requiring the addition of HCTZ (44%) and a calcium blocker (25%). The mean doses of study medications that were required were carvedilol 17.5 mg bid and metoprolol 128 mg bid. Approximately half the patients in each group required the maximum dose of the study medication: carvedilol 25 mg bid and metoprolol 200 mg bid.

Patients in the carvedilol group had better metabolic results overall. HgbA1c in the carvedilol group increased insignificantly from 7.21 to 7.23; in the metoprolol group it increased significantly by 0.15 from 7.19 to 7. 34 (a 2% change). The urinary albumin/creatinine ratio decreased by 14% in the carvedilol group but increased by 2.5% in the metoprolol group. Lipid changes were slightly in favor of carvedilol, as was insulin sensitivity.

The dose of metoprolol tartrate used in this study was very high. The target dose before adding another anti-hypertensive agent was 200 mg twice daily. Not surprisingly, there was more bradycardia in the metoprolol group. The same percentage of patients in each group required “rescue” antihypertensive treatment. Therefore, the chosen maximum doses of the drugs probably represent an approximately equal antihypertensive effect. Since carvedilol has both alpha and beta adrenergic blocking activity, whereas metoprolol is only a beta-blocker, it is not a complete surprise that a high dose of metoprolol was required to equal the antihypertensive effect of carvedilol. But in the real world, I doubt very much that practitioners would push metoprolol to such a limit, rather than adding a different medication. Which makes one wonder whether the metabolic advantage of carvedilol would have been as great if a more normal metoprolol dose had been chosen.

I was unable to find out the exact rationale behind the dose choices that were made. In the article, the authors state that “A detailed description of the study design and statistical methods has been published elsewhere”. But the reference, #17, indicates an article in press, which is not yet available.

It is interesting to note that in the COMET trial, comparing the beneficial effects of carvedilol to metoprolol in patients with CHF, the target dose of carvedilol was 25 mg bid, that of metoprolol was 50 mg bid. So, when looking at beneficial effects of these drugs, the investigators chose a rather low dose of the “competitor”, when looking at side-effects, they chose a mega-dose.


In the November 11 New York Times, Gina Kolata, referring to the CAMELOT study (just published in JAMA), states:

“A new study of heart disease patients finds that “normal” blood pressure may not be low enough. By reducing their pressure well below the levels suggested by national guidelines, patients had fewer heart attacks, strokes, cardiac arrests, hospitalizations for chest pain, procedures to open blocked coronary arteries, and deaths.”

Although strictly speaking correct, the above sentence is misleading, since it refers to a composite endpoint. Both cardiovascular deaths and overall deaths were insignificantly increased in the treatment arms compared to placebo. Anyone reading Kolata’s sentence would be forgiven for incorrectly concluding that deaths were also reduced. This is a typical example of the difficulty of accurately presenting study results in a bite-sized fashion for the general public.

As for the study itself, I’m not quite sure what it really does demonstrate.

Patients with angiographically documented coronary disease and a mean BP of 129/78 were randomized to daily amlodipine, enalapril or placebo. After two years, the composite endpoint noted above occurred in 23.1% of placebo patients, 20.2% of enalapril patients and 16.6% of amlodipine patients (significant for amlodipine vs placebo; trend not significant for enalapril vs placebo).

In their conclusion, the authors state that:

“These results suggest that the optimal blood pressure range for patients with CAD may be substantially lower than indicated by current guidelines.”

Blood pressure reduction was about 5/3 mmHg in both treatment groups vs placebo. If the main effect is via BP reduction, and BP was similarly reduced in both treatment groups, why was amlodipine more efficacious? The authors note that once daily enalapril taken in the morning may achieve an identical BP reduction later in the day, but drop off during the night. Perhaps, but just speculation. Much of the benefit of amlodipine was in reducing hospitalizations for angina. When a post-hoc analysis of mortality, MI and stroke was looked at, there was an identical but non-significant trend towards benefit in both treatment groups.

Interestingly, this study was published the same week as the PEACE trial, which appeared in the NEJM. That study looked at an ACE inhibitor, trandolapril, in patients with stable coronary disease and well controlled risk-factors. There was no benefit to the addition of the ACE inhibitor. In this study, blood pressure reduction was only about 3/1.5 mmHg compared with placebo.

Bottom line? Dropping blood pressure lower than what is currently considered normal in patients with CAD is probably good, although not dramatically so. Not a big surprise. Amlodipine reduces hospitalizations for angina. ACE inhibitors may not be the magic bullet the HOPE trial had suggested, if all other risk factors are very well controlled, or perhaps it depends on the specific ACE inhibitor used. Not clear.

I’m not sure how much useful information was added by these two studies.

A new patent for an old drug in a new population

In a much discussed (see Kevin MD and Medical rants) study presented at the AHA meetings in New Orleans and just being published in this week’s NEJM, a fixed dose combination of hydralazine and isosorbide dinitrate, BiDil, was found to significantly reduce morbidity and mortality in African American patients with congestive heart failure. BiDil (vs placebo) was added to standard therapy for CHF (including ACE inhibitors or ARB’s, aldactone and beta-blockers). The trial was halted early because of significant benefit of the active medication.

This trial is being actively discussed because it seems to be the first time that a drug is being promoted specifically for use in a racial group. Another aspect of this study which I find fascinating is its financial genesis. My understanding of the course of events is as follows.

Isosorbide dinitrate and hydralazine have been used for the treatment of heart failure in the past. The isosorbide-hydralazine combination pill was patented (in the 1980’s) for the treatment of congestive heart failure, but failed to provide significant benefits in large trials; as a result, the FDA would not approve it as a new medication. Subgroup analysis of the original trials, however, suggested benefit in black patients. The company NitroMed, which had acquired the rights to the combination, then applied for and received a new patent for it specifically for the treatment of heart failure in black patients. Unlike the original patent, which expires in 2007, the new one for the same medication is valid until 2020. Armed with this new patent, NitroMed sponsored the current trial. With the positive results just reported, the FDA is likely to approve BiDil for use in black patients.

My initial reaction to this whole story was that it represents a typical interaction between industry, the patent system and the FDA. There is a one-two punch with the patent system prolonging the drug company’s monopoly by granting a new patent for use in a subpopulation, which the FDA then protects by limiting approval of the medication to that subpopulation.

The other side of this coin, however, is that precisely this system led to a potential profit for NitroMed, which made sponsoring the trial financially attractive. The result is likely to be of benefit to many African Americans. A prime example of the strengths and problems with the pharmaceutical industry’s ties to regulatory agencies.

Vioxx, the FDA and the Lancet

On November 2, the FDA published the text of a study looking at the cardiovascular risk of Vioxx compared to Celebrex and other NSAIDS. It found a 3.7-fold increase in cardiovascular risk when high dose Vioxx (>25 mg/day) was compared to Celebrex, and a 1.5-fold increase when the standard dose of Vioxx was compared with Celebrex.

Some more information came out of this report, such as the lack of protective effective of conventional NSAIDS, but this was a large, retrospective, case-control study, subject to all of the limitations of non-randomized trials and thus not to be taken too seriously. In this case, the problems of confounding and bias which plague non-randomized studies are mitigated by the fact that two different COX-2 inhibitors are being compared. The choice of a specific COX-2 inhibitor is less likely to be related to cardiac risk than, say, the choice of a COX-2 inhibitor vs. a standard NSAID or no drug at all. Nevertheless, this sort of data should not be taken as conclusive proof of the safety of other COX-2’s.

Interestingly, the posting of this report by the FDA seems to have seriously ruffled feathers at the Lancet, which was reviewing it for publication. In an editorial, Richard Horton takes the FDA mightily to task for poor regulatory oversight, and then states:

“On Nov 2, 2004, the FDA tried to shore up its tarnished reputation by posting on its website an early version of a recently completed observational study into the safety of Vioxx. The report comes with a warning that it has “not been fully evaluated by the FDA and may not reflect the official views of the agency”. The FDA investigators estimate that over 27 000 excess cases of acute myocardial infarction and sudden cardiac death occurred in the USA between 1999 and 2003. “These cases”, they write, “would have been avoided had celecoxib been used instead of rofecoxib”. This study is presently under review at The Lancet. It is unclear why the FDA could not have waited for the fully evaluated report to have been scrutinised, revised, and published according to the norms of scientific peer review. Bypassing independent peer review smacks of panic in the FDA, which is under intense reputational pressure. And yet its decision to try to undermine the integrity of this work again shows that the agency’s senior management is more concerned with external appearance than rigorous science.”

Just when the Lancet is about to publish this timely report, the FDA goes and posts it on its website. Shades of Ingelfinger (prior publication elsewhere)! The Lancet is not amused.

Drug eluting stents and late thrombosis

Unlike restenosis, which is fairly common, stent thrombosis is a rare but much more dangerous complication after coronary stent placement. It usually occurs before endothelialisation has been completed. For bare-metal stents, this process takes a few weeks; the newer, drug-eluting stents inhibit restenosis by inhibiting fibroblast proliferation, but they also tend to delay the endothelialisation process. For this reason, patients who have had drug-eluting stents are prescribed aspirin and clopidogrel therapy for at least 6 months.

In this week’s Lancet is a Research Letter describing four cases of late stent thrombosis, occurring over a year after stent implantation and shortly after anti-platelet therapy was discontinued for various reasons. Two patients had had both drug-eluting and bare stents implanted; the thrombosis occurred only on the drug-eluting stents. Two of the patients had Cypher sirolimus-eluting stents and two had Taxus paclitaxel-eluting stents.

These numbers are, of course, minute compared to the large number of drug-eluting stents that have been placed, but they presumably represent only some of the actual cases that have occurred. Late stent thrombosis is thus a rare but potentially dangerous complication of drug-eluting stents, when anti-platelet therapy is discontinued.

Since drug-eluting stents reduce the incidence of restenosis significantly, they will continue to be used. Nevertheless, as the accompanying editorial points out, the issue of late stent thrombosis should make us wary of these stents in patients in whom discontinuation of antiplatelet therapy is likely to occur; more data from registries are needed; the duration of antiplatelet therapy may need to be prolonged; and shorter interruption for surgical procedures may need to be considered.

Interestingly, a press release from the Canadian Cardiovascular Congress 2004 that was referred to by EchoJournal indicates work being done on a stent coated with antibodies that attract endothelial cells from the bloodstream. This could conceivably speed up endothelialisation, a very attractive although not-yet-ready-for-prime-time approach.

Many new devices are sure to appear. Given the large sums of money being spent on drug-eluting stents, I find it very interesting that the authors report stent thrombosis in two each of the two main players in the market. A serendipitous coincidence, I assume.

Of course, lowering cardiovascular risk remains the best approach. You don’t need to worry about what kind of stent you’ll be getting if you don’t need a stent.

EBM and industry

This week’s BMJ is a theme-based issue on Evidence-Based Medicine. It contains the expected articles on how to judge whether or not EBM is living up to its promise, how best to implement it, and so on. There is one aspect of EBM, however, that is not addressed: its effect on the drug and device industries.

An implicit subtext of the evidence-based movement is that it helps counter the millions of dollars of industry propaganda and hype that wash over us every year. And EBM does, indeed, provide tools to judge drugs and interventions more objectively than we think the industry would like us to. But the drug and device industries have evolved along with (or ahead of) their customers.

Pharmaceutical companies have responded to EBM by carefully designing trials destined to apply to as wide a population as possible, while still obtaining (p<0.05) benefit. Then, armies of drug reps sally forth armed with reprints, while researchers are sent out to spread the gospel of statistical significance. EBM has made us particularly avid of hard data (while relegating clinical significance to a somewhat subordinate role). This emphasis on statistically significant data has been digested by industry and is now used to sell drugs and devices.

Evidence is good. Evidence-based methodologies are better. But they aren’t magic bullets. And EBM is a tool whose use is not restricted to pure and virtuous clinicians. Caveat lector.

Another case-control study ?

This week’s JAMA has a cohort study from the Netherlands looking at the risk of community-acquired pneumonia and use of gastric acid–suppressive drugs. The authors used a large, private practice clinical database to investigate a hypothesized linkage between the prescription of proton pump inhibitors or H2 receptor blockers and community acquired pneumonia. In order to adjust for the fact that many patients on these drugs are sicker, or more likely to have certain diseases, the authors performed a nested case-control examination that looked at recent acid-suppression therapy compared with previously discontinued therapy, adjusting for a large number of factors and diseases (such as age, sex, indication for the therapy, presence of diabetes, copd and stomach cancer).

The unadjusted relative risk of pneumonia for patients who had been prescribed acid-suppressant therapy vs. those who had not was quite high (4.5). In the matched, adjusted case-control cohort of recent therapy vs. past therapy, the adjusted odds ratio was 1.63, much lower, but still significant.

I believe it is impossible to properly adjust for all the reasons why pneumonia patients might have been prescribed acid-suppressant therapy shortly before getting pneumonia. Some of the factors that the authors controlled for are surely valid, but there are likely to be others. Even an individual practitioner’s style and personality could be linked with both the likelihood of prescribing these drugs and that particular patient population’s risk for pneumonia. Adequate adjustment is just not possible.

After multiple case-control and cohort studies showed significant cardiovascular benefit from estrogen replacement therapy, properly randomized trials demonstrated that this was not the case, forcing an incredible about-face. One would think that the estrogen replacement scandal would have put a major brake on performing and publishing these sorts of studies, yet they seem be flourishing. Has nothing been learned?

Vestibular rehabilitation for dizziness

In the October 19 issue of Annals of Internal Medicine is a study from the UK of the effectiveness of primary care-based vestibular rehabilitation for chronic dizziness. Primary care patients with a history of labyrinthine dizziness that had lasted at least two months were randomized in a single blind fashion to receive nurse-taught vestibular rehabilitation exercises or usual care; after three months the patients who had received the usual care were crossed-over to vestibular rehabilitation.

The rehabilitation exercises consisted of increasingly aggressive maneuvers designed to stress the vestibular system and provoke dizziness (such as turning the head with eyes open, then closed). These exercises were to be done once or twice daily. The details of the vestibular rehabilitation, and a detailed patient handout explaining the exercises, are available at the Annals website. Patients were evaluated using various questionnaires and forms, and with a device that measured postural stability.

The results indicated that patients did indeed improve more with the vestibular rehabilitation than without.

I found the results of this study to be frustrating. There are two things that I would have liked to find out: what percentage of patients in each group felt, say, “cured or almost completely cured” or “very substantially better” and whether or not there was any difference in the number of falls in either group. The authors bombard us with data, but do not answer these simple questions directly.

I am happy to hear that the Vertigo Symptom Scale was 13.3 in the usual care group and only 9.88 in the rehabilitation group. I am delighted that the Dizziness Handicap Inventory went from 35.88 to 31.09. I am impressed that the Short Form-36 (physical functioning) was 25.95 vs. 27.14. But what does all this mean? The authors do state that 67% of the treated group reported clinically significant improvement, compared with 38% of the usual care group, but what exactly constitutes clinically significant improvement? Did any of these patients actually fall, before or during the study? Nothing about this issue.

Since so many of my older patients complain of unsteadiness on their feet, I would love to recommend exercises to help them. Most likely, vestibular rehabilitation would be helpful and effective. But it would be nice to have some simpler, more comprehensible data about it. I understand that tables full of statistically significant numbers are necessary to get published, but it should be possible to sneak in a patient’s actual assessment of treatment efficacy somewhere.

Review of antiplatelet therapy

In this week’s JAMA are several articles dealing with stroke, stroke prevention and antiplatelet therapy. Tran and Anand review oral antiplatelet therapy in cerebrovascular disease, coronary artery disease and peripheral arterial disease. They looked at trials involving antiplatelet therapy in patients with documented vascular disease (stroke, TIA, coronary disease, peripheral arterial disease). Some of the main points that emerged are:

  • In patients with vascular disease, antiplatelet therapy leads to a risk reduction for vascular events of 20-30% vs. placebo.
  • Aspirin vs. plavix: The CAPRIE trial compared 325 mg of aspirin to 75 mg clopidogrel in almost 20,000 patients. Plavix was modestly superior to aspirin overall, with a risk reduction of 8.7%. The cost of plavix is much greater than that of aspirin, however, and the benefit is less than the benefit achieved by either drug alone over placebo (in the 20-30% range); the authors of this review suggest that, in most cases, either drug may be used for initial, secondary prevention.
  • Aspirin plus plavix: Greater bleeding risk. Not more effective than clopidogrel alone in the setting of stroke/TIA (MATCH trial); more effective than aspirin alone in unstable coronary syndrome patients; more studies ongoing in other CAD patients; recommended in patients after unstable angina and after percutaneous coronary interventions (aspirin indefinitely; clopidogrel for 12 mos, possibly longer).
  • Persantine (dipyridamole): One large trial of extended-release dipyridamole (ER-DP) in patients with stroke or TIA showed benefit of aspirin plus ER-DP above either drug alone. The theory is that ER-DP is superior to short-acting dipyridamole, which is the basis for the combination drug Aggrenox (aspirin plus ER-DP), but the evidence is based on this one trial, and other trials are still ongoing. No evidence yet for effectiveness in CAD patients, and theoretical concern about deleterious coronary vasodilation, particularly with short-acting dipyridamole.
  • Treatment failures on monotherapy (recurrent events): In general, patients on aspirin add or switch to clopidogrel; patients on clopidogrel add aspirin, but several variants depending on types of events. See article for details.

I found this review to be quite helpful. It applies mainly to the non-acute-coronary-syndrome patients. Another article in the same issue of JAMA deals with ACS patients in greater detail.

Traffic as a trigger for MI

In today’s NEJM is an article from Augsburg, Germany looking at exposure to automobile traffic as a trigger for myocardial infarction. The statistical methods employed were almost completely incomprehensible to me � I would ask anyone who understands the “statistical analysis” paragraph to email me an explanation. But the bottom line is that, during the hour before onset of an MI, patients were more likely to have been in traffic than would have been expected based on their previous days’ patterns. This association held for automobile, public transportation and bicycle traffic. It also held after a number of adjustments (including adjustment for “the stress of getting up in the morning”!)

The authors speculate that stress, noise and particulate air-pollution may all explain the association, but they concentrate almost exclusively on air-pollution as the triggering factor. In an accompanying perspective piece, Stone reviews the substrate-trigger theory of acute coronary events. The substrate is the atheromatous plaque. Triggers are those events that cause plaque to become unstable and thrombose, including inflammation, increased sympathetic stimulation, vasoconstriction and hypercoagulability. In discussing the current study, he, too, emphasizes the role of particulate air pollution, via inflammation, vasoconstriction and hypercoagulability.

I would be more inclined to view the stress of traffic as the principle trigger, rather than air pollution, but that�s just my gut feeling. I think it�s important to note that substrates are causal, whereas triggers are sometimes causal but sometimes affect timing more than anything else. Thus, if an event is likely to happen, the trigger may influence the “when” more than the “whether”. Treating a substrate is often more effective than suppressing all possible triggers. Reducing risk factors for atherogenesis is a better approach than attempting to avoid traffic, although who could argue against reducing air-pollution?