Chelation therapy

What happens

EDTA is a molecule with arms that grip a metal ion; the name chelate comes from the Greek word for a crab’s claw. Whatever is bound leaves the body via the urine.

In the large trials it was not a pure EDTA solution: the 500 milliliters contained disodium EDTA, ascorbate, B vitamins, electrolytes, procaine and heparin. 40 infusions were given, initially weekly, then at longer intervals.

Where it is undisputed

For proven heavy metal poisoning, chelation is established. The US drug authority puts it this way: approved chelating agents are prescription-only and approved for specific indications, for example lead poisoning and iron overload. Anyone with lead poisoning is chelated.

Nothing that follows changes that. It is about a different claim: that the same procedure clears calcified blood vessels or serves general detoxification.

The trial that raised hopes

The idea is over 50 years old and not far-fetched, because lead and cadmium are linked to cardiovascular disease. In 2013 a publicly funded trial on it was published: 1,708 patients after myocardial infarction, 134 centers, 40 infusions versus placebo, double-blind.

The combined endpoint occurred in 26 percent of those treated and 30 percent on placebo, hazard ratio 0.82. This 18 percent relative reduction is the figure that has been quoted everywhere since.

It has two catches. First, the statistics: because of the interim analyses, the threshold for significance was 0.036, and the result was 0.035. Second, the course: 289 participants, 17 percent, withdrew their consent, unevenly distributed with 115 on EDTA and 174 on placebo. Overall mortality did not change: 87 versus 93 deaths.

The subgroup that carried everything

A subgroup with diabetes, 633 participants, had been prespecified. There things looked different: the endpoint occurred in 25 versus 38 percent, hazard ratio 0.59, and the effect withstood correction for multiple subgroups. To prevent one event, 6.5 people had to be treated for 5 years.

In the 1,075 participants without diabetes, no effect was seen, hazard ratio 1.02, and the difference between the two groups was statistically significant. A strong signal, and the authors themselves wrote that it justified a confirmatory trial but not routine use.

The confirmatory trial

Precisely this subgroup was retested, again with public funding: 88 centers, 1,000 randomized people with diabetes and a history of myocardial infarction, 959 of whom received at least one of the 40 weekly infusions. The trial had more than an 85 percent chance of finding a 30 percent reduction; median follow-up was 48 months.

The result: 35.6 versus 35.7 percent for the combined endpoint, hazard ratio 0.93 (0.76 to 1.16), p equal to 0.53. Cardiovascular death, infarction or stroke: 18.4 versus 19.7 percent. Death from any cause: 17.4 versus 17.6 percent, 84 deaths each.

The blood lead level reliably fell from 9.03 to 3.46 micrograms per liter, p below 0.001; with placebo from 9.3 to 8.7. The chain is complete, and at the end of it there is no clinical benefit. A mechanism is not an outcome, and this can rarely be read off so directly.

The second trial arm

Both trials had a second half: high-dose vitamins and minerals with 28 ingredients versus placebo. In the confirmatory trial, the endpoint occurred in 175 participants in both groups, hazard ratio 0.99.

The combination of infusion and vitamins, which had looked best in the first trial, was also no better than double placebo. For the question of general detoxification, this means: two popular approaches were tested at the same time, and neither held up.

What this means for people without symptoms

Both trials studied people after myocardial infarction, the second additionally only those with diabetes. Anyone who is free of symptoms and has been told they have a heavy metal burden will find nothing in these figures about their own situation.

For this group there is no study with hard endpoints, neither positive nor negative. The distinction is uncomfortable in both directions: the use there has not been refuted, but it has not been demonstrated either. What remains is a medical intervention without a tested benefit for one’s own situation.

Reports from practices that offer the procedure describe more resilience and better well-being. They remain unspecific, come from the commercial environment of the treatment and do not weigh as much as two randomized trials with hard endpoints.

What the studies show

TACT, JAMA 2013

Double-blind, placebo-controlled trial with factorial design, 1,708 patients aged 50 and over at least 6 weeks after myocardial infarction, 134 centers, 40 infusions. Primary endpoint: death, reinfarction, stroke, vascular procedure or hospital admission for angina. Result 26 versus 30 percent, hazard ratio 0.82, p equal to 0.035 with a threshold of 0.036. No effect on overall mortality; 289 participants withdrew their consent, unevenly distributed across the groups.

Diabetes subgroup of TACT, Circulation Cardiovascular Quality and Outcomes 2014

Prespecified subgroup with 633 participants, 322 on EDTA and 311 on placebo. The endpoint occurred in 25 versus 38 percent, hazard ratio 0.59, significant even after correction for multiple subgroups; the number needed to treat was 6.5 over 5 years. In the 1,075 participants without diabetes, there was no effect, hazard ratio 1.02. The authors called for a confirmatory trial and warned against routine use.

TACT2, JAMA 2024

Double-blind, placebo-controlled trial at 88 centers, 1,000 randomized, 959 participants with at least one infusion, all with diabetes and a history of myocardial infarction, 40 weekly infusions, median follow-up 48 months, more than an 85 percent chance of detecting a 30 percent reduction. Primary endpoint: 35.6 versus 35.7 percent, hazard ratio 0.93, p equal to 0.53. Overall mortality 17.4 versus 17.6 percent. The lead level fell from 9.03 to 3.46 micrograms per liter. The endpoint was not met.

Where the data stop

A benefit in atherosclerosis has not been demonstrated, and after two large trials with hard endpoints that is a robust statement. Something else has not been tested: use in people without symptoms with an alleged heavy metal burden. There is no study with hard endpoints for this, and that means not refuted but untested. It also remains open why the diabetes subgroup of the first trial looked so clear; that it was not confirmed in the trial built specifically for it is the best available answer, but not an explanation. Plus a structural point: in the first trial, 17 percent withdrew their consent, unevenly distributed across the groups, and no one knows what became of these people.

Status, approval and legal

Approved chelating agents are prescription-only and approved for specific poisonings, for example lead poisoning and iron overload; the US drug authority states this. There is no approval for coronary heart disease. In Germany, chelation therapy as a treatment for atherosclerosis is not a recognized service of statutory outpatient care; it is offered as a private service. There are explicit warnings against chelation products for home use, whether as food supplements, nasal sprays or suppositories: they are not approved for any disease and can have serious side effects. An infusion is in any case a medical intervention.

Safety

The most important serious side effects are a drop in calcium levels and kidney damage; in the first large trial, 100 of those treated, that is 11.9 percent, had serious adverse events, and 15 percent stopped the infusions because of side effects, similarly often in both groups. Deaths are also documented: the US Centers for Disease Control and Prevention investigated 3 cases from the years 2003 to 2005, all due to calcium deficiency with cardiac arrest, in a 2-year-old child, a 5-year-old child and a 53-year-old adult. In one case, the calcium-free salt had been given instead of the calcium-containing one because the brand names were mixed up. This is not a side effect risk in the usual sense but a mix-up risk. On top of that there is a harm that appears on no list: anyone who starts a series of infusions may postpone the treatment that works. Anyone considering it should first have it clarified whether there is any poisoning at all.

Sources

• Lamas et al., JAMA 2013 — TACT: EDTA chelation after myocardial infarction, 1,708 patients
• Escolar et al., Circ Cardiovasc Qual Outcomes 2014 — diabetes subgroup of TACT, 633 patients
• Lamas et al., JAMA 2024 — TACT2: EDTA chelation in diabetes and prior myocardial infarction, 959 treated patients
• Ujueta et al., JAMA Internal Medicine 2025 — vitamin arm of TACT2, 1,000 participants
• Lamas et al., American Heart Journal 2022 — study design of TACT2 with power above 85 percent
• CDC, MMWR 2006 — 3 deaths from calcium deficiency after chelation therapy, 2003 to 2005
• NCCIH — Chelation for Coronary Heart Disease: What You Need To Know, as of August 2024
• FDA — approved chelating agents are prescription-only, for example for lead poisoning and iron overload
• G-BA (German Federal Joint Committee) — guideline on methods in statutory outpatient care