The results of the unethical and misbegotten Trial to Asess Chelation Therapy (TACT) are finally revealed, part 2

With very limited exceptions, chelation therapy is, as I said before in my somewhat Insolent opinion, is pure quackery. The sole exception is for real, documented cases of acute heavy metal poisoning that are known to respond to chelation, such as iron overload due to transfusion, aluminum overload due to hemodialysis, copper toxicity due to Wilson’s disease, acute heavy metal toxicity, and a handful of other indications. Basically, chelation therapy involves infusing chemicals that can bind to metal ions and make them easier for the kidneys to excrete. The problem is, there is no good basic science or clinical evidence to suggest that mercury from vaccine causes autism or that heavy metal overload causes atherosclerosis. Unfortunately lack of physiologic mechanism, plausibility, and clinical evidence that it doesn’t work hasn’t stopped chelation therapy from long ago having become a favored “alternative” medicine therapy that is used for atherosclerotic cardiovascular disease (for which it doesn’t work), autism (for which it doesn’t work and for which its rationale is based on the false idea that mercury from vaccines caused autism), and for general “detoxification” for what ails you, including autism, cancer, Alzheimer’s disease (which Hugh Fudenberg has blamed on the flu vaccine, a claim parroted with Bill Maher, of course!), and just about every ailment under the sun

Despite this extreme implausibility, randomized controlled studies showing that chelation is no better than placebo for cardiovascular disease, a veritable cottage industry of chelation therapy for cardiovascular disease long ago sprang up and, against all evidence, persists, fueled by extravagant claims likening chelation to a “Roto-Rooter for your arteries” and an alternative to angioplasty and coronary artery bypass surgery. The belief that chelating toxic metals out of people can treat cardiovascular disease has no basis in physiology, biology, or pharmacology, but it’s a major treatment modality favored by naturopaths and many other “alternative” practitioners. Given the infiltration of quackademic medicine into medical academia, it should not be surprising that its advocates promoted clinical trials of this disproven modality. In the early 2000s, they succeeded in the form of the National Institutes of Health (NIH) Trial to Assess Chelation Therapy (TACT), a five year phase 3 trial begun in 2003 to test office-based, intravenous disodium ethylene-diamine-tetra-acetic acid (Na₂EDTA) as a treatment for coronary artery disease (CAD). A few months ago, I commented extensively why the results of this misbegotten and unethical trial presented at the American Heart Association’s annual meeting in a session on late-breaking clinical trials in the form of two abstracts were not nearly as impressive as they were being spun. Kimball Atwood was just as unimpressed.

Kimball and I were just waiting for the results to be published in a real peer-reviewed journal in order to see what other tidbits are there. Then I learned that Gervasio Lamas, the Principle Investigator of the TACT presented more of the results of the study at the American College of Cardiology Meeting this weekend:

Background: We tested the cardiovascular effect of a high-dose oral multivitamin and mineral supplement as a treatment factor in the NIH-funded Trial to Assess Chelation Therapy (TACT). The factorial design eliminated potential confounding from uncontrolled concomitant use of oral vitamins and minerals in patients receiving EDTA chelation or placebo infusions.

Methods: The TACT vitamin study is a double-blind controlled trial comparing an oral high-dose multivitamin and mineral supplement with placebo. A committee of alternative medicine practitioners designed the active oral therapy for use in conjunction with chelation therapy. 1,708 patients were randomized. Inclusion criteria were age 50 or older, MI >6 months prior, and creatinine < 2.0. The active vitamins were a 28-component mixture. The primary endpoint is a composite of death, MI, stroke, coronary revascularization and hospitalization for angina. The analyses are intent-to-treat. Results: Patients were 65 (IQR 59,72) years old, 18% female, and 9% minority. The qualifying myocardial infarction occurred 4.6 (1.6, 9.2) years prior to enrollment. Patients were followed a median of 55 months. There was a high prevalence of diabetes (31%), of prior coronary revascularizations (83%), and use of aspirin (84%), beta-blockers (72%), and statins (73%). Fasting glucose was 102 mg/dL (92, 121), and low-density lipoprotein 89 mg/dL (67, 115). The effect of vitamin therapy will be assessed, as well as its effect in pre-specified subgroups. The event rate in all 4 factorial cells: active oral vitamins + active EDTA chelation infusions; placebo oral vitamins + active EDTA chelation infusions; active oral vitamins + placebo infusions; and placebo oral vitamins + placebo infusions will be reported and compared. Conclusion: The intent-to-treat analysis of the vitamin arm of TACT will show whether a very high dose of oral vitamins and minerals reduces the incidence of the composite primary endpoint, and whether the effect, if any, is additive to the effect of intravenous chelation therapy.

So basically, the TACT trial was set up according to a 2 x 2 factorial design:

• Chelation plus high oral high dose vitamin and mineral supplement
• Chelation placebo plus oral high dose vitamin and mineral supplement
• Chelation plus oral high dose vitamin and mineral supplement placebo
• Chelation placebo plus oral high dose vitamin and mineral supplement placebo

As I noted before, the regimen was described in detail in a recent publication. The vitamin supplements included doses ranging from 25% to 6,667% of the RDA for various vitamins. For example, the dose of vitamin C was 2,000% of the RDA; thiamin, 6,667%; and vitamin A, 500%. The previous presentation looked at the chelation therapy aspect of the study. This study looks at the oral high dose vitamin and mineral supplement treatments.

Basically, this presentation also looks to me like a technique of presenting what we in the biz like to call the “minimal publishable unit” (also known as the MPU). Instead of presenting the results of all four groups in the first presentation last year, Lamas is spreading out the love among multiple presentations. The first time around, it was the chelation arm and the quality of life results. This time around, it’s the vitamin/supplement arm. In and of itself, this isn’t anything remarkable or even wrong. It’s just playing the game, and far too many clinical investigators do it. In any case, the rationale for the vitamin arm of the trial given in Dr. Lamas’ slides consists of (1) the observation that chelation practitioners also use high doses of anti-oxidant vitamins and minerals in conjunction with intravenous chelation and (2) oral vitamins and minerals therefore constituted a potential confounder. The primary composite endpoint of the overall study was time to first occurrence of either death, MI, stroke, coronary revascularization, or hospitalization for angina. Of course, as I discussed last time, this sort of composite endpoint, in which all of these different but related endpoints are lumped together, is problematic, as different confounding factors can be amplified, or, as González et al concluded:

The use of composite end points in cardiovascular trials is frequently complicated by large gradients in importance to patients and in magnitude of the effect of treatment across component end points. Higher event rates and larger treatment effects associated with less important components may result in misleading impressions of the impact of treatment.

That same problem still exists and is still worth emphasizing. It’s not a problem unique to TACT (many cardiovascular trials use similar composite endpoints), but it is the sort of problem that makes barely statistically significant results, like the results of the chelation therapy arm of TACT, suspect if the trial used to obtain them used a composite endpoint. Lamas’ reporting of another MPU from this study is also rather annoying because it allows him to refer to his previous presentation has having definitively demonstrated that the TACT had found a definitive treatment effect due to chelation therapy when in fact what he had found might have been statistically significant but was almost certainly not clinically significant.

Cutting to the chase here, the results of this part of the study, which looked at the vitamin supplement arms of the study, were completely negative. There was no signal. Nada, zip. At five years, the composite endpoint showed 37% of patients int he control group had had cardiac events, while 34% in the vitamin groups had, with a p-value of 0.212, nowhere near statistical significance. Moreover, subgroup analysis of the components of the endpoint showed that none of the individual cardiac events that made up the end point were statistically significantly different. This part of the trial is about as negative as negative could be. Of course, this being the TACT trial, the same problems were operative here, including a high dropout rate (50%) for the high dose vitamin arm. Lamas also noted a high vitamin noncompliance rate, but he didn’t say what that noncompliance rate was.

He ended up concluding:

• High dose oral vitamins reduced the composite outcome by 11%, which was not statistically significant.
• When combined with EDTA chelation the benefit was additive and magnitude was statistically significant.

11% Yes, maybe, but the absolute value is 3%. Here’s the graph:

Lamas then shows another subgroup analysis for differences between various characteristics of the groups. All of these subgroup analyses were negative two—all save one, which found that patients who were not on statins showed a better hazard ratio due to high dose vitamins. What does this mean? Probably nothing. The authors checked six different subgroups; it’s not too surprising that one of them came up “positive.” As clinical investigators know, subgroup analyses are fraught with peril to the results of a study. They frequently turn up “positive” results. It’s not clear to me from the presentation whether these were post hoc or prespecified subgroup analyses, either, but the smell lik post hoc analyses.

I found this video rather interesting. It’s commentary by E. Magnus Ohman, MD, FACC of Duke University, who notes right off the bat that it’s “unusual” to have a 2×2 factorial design trial presented in two separate presentations. He points out that it’s a single trial. On the other hand he calls it “intriguing” that there was a statistically significant treatment effect between the placebo-placebo group and the high dose vitamins-chelation group. He points out that chelation is being done and will continue to be done regardless of the results of this trial and then opines that there “may be something to this.” He even says “it’s not wrong.”

Well, not really. There are so many problems with this trial that a small effect that is weakly statistically (p=0.16; prespecified p-value required for statistical significance, 0.036 in the original trial—see my original discussion) that its results should be taken with a huge grain of salt. One thing’s for sure, there’s no benefit to the vitamins. Let’s just put the results in context. First, consider these results in the context of what is claimed for chelation therapy. As I described a few months ago, giving lots of examples before, it goes beyond a weakly statistically significant improvement in a composite endpoint. No, quacks claim that chelation therapy can replace angioplasty, stenting, and coronary artery bypass, melting away atherosclerotic plaques, cleaning them out like a “Roto Rooter for your coronary arteries.” Even the most generous interpretation of this trial would concede that the benefits of chelation therapy, if they exist, are minimal. Even Dr. Lamas himself concedes that, stating that his study “do not support the use of high-dose vitamin and mineral therapy as an adjunct to optimal evidence-based medical therapy in patients with prior myocardial infarction.” No kidding.

I think I can speculate now about why the results were presented in three separate presentations, two at the AHA meeting and one at the ACC meeting, instead of presenting the study as one presentation. Think about it. If the study were presented as a single study, it would be unrelentingly negative, except for one composite endpoint and then only in diabetic patients. If all the results were aggregated together into one presentation, as they should have been, Lamas would have been forced to state in the same presentation that chelation doesn’t work except maybe in diabetics (in which the effect is still small and could be due to confounders), that it doesn’t improve quality of life, and now that the vitamin component of the therapy is useless. Having them as separate studies allows advocates to point to the positive part of the study as a separate presentation and ignore the other two parts of the trial. Is it cynical of me to say so? Perhaps. The other explanation is that Lamas wanted to maximize the number of presentations by presenting only MPUs, which, while lots of investigators do it, doesn’t exactly reflect well on those who do.