The Iron-Heart Hypothesis: Title and subTitle BreakSearch for the Ironclad Evidence

The iron-heart hypothesis first put forth by Sullivan¹ in 1981 suggests that increased body iron stores are a risk factor for coronary heart disease (CHD) and thus that iron depletion through phlebotomy or other means can reduce risk. The hypothesis, which was based on markedly lower incidence of CHD in premenopausal women (who lose iron through menstruation) compared with men and postmenopausal women, is intuitively appealing and seems to be well-grounded in biochemistry.

Iron is an essential mineral and an important component of the oxygen-transporting and storage proteins hemoglobin (in red blood cells) and myoglobin (in muscles).² Conversely, free ferrous iron in the body is a catalyst for the formation of hydroxyl radicals, powerful prooxidants that attack cellular membranes, proteins, and nucleic acids.³ Iron-catalyzed lipid oxidation has been implicated in increased formation of the circulating oxidized form of low-density lipoprotein cholesterol, which is thought to be more atherogenic than the native form.

In addition to enhancing oxidative stress, increased iron stores are believed to adversely affect cardiovascular disease through other mechanisms, including alteration of endothelial function, decreased vascular reactivity, and reperfusion injury by iron-induced free radicals.⁴ In rabbit animal models, iron overload accelerated the progression of atherosclerotic lesions,⁵ whereas an iron-deficient diet reduced atherosclerotic lesions in mice deficient in apolipoprotein E.⁶

Testing the iron-heart hypothesis in humans has focused on epidemiologic associations between biomarkers of body iron status and risk of cardiovascular disease. Serum ferritin, the iron-storage protein for which levels are elevated with iron overload and proportionally reduced with iron depletion, is considered the best biomarker for long-term iron stores. Salonen et al⁷ conducted the first prospective cohort study to suggest a positive association between serum ferritin concentration and risk of CHD in a Finnish population. After adjustment for established cardiovascular risk factors, this small study showed a more than 2-fold increased risk of acute myocardial infarction among men having serum ferritin levels of 200 ng/mL or greater compared with those having lower serum ferritin levels. These promising results, however, have not been confirmed in most subsequent prospective cohort studies.^{8 - 10}

Epidemiologic research on blood donations has also generated mixed results, with the largest cohort study to date finding no association between frequency of blood donations and CHD incidence in men.¹¹ The 1996 discovery of HFE gene mutations responsible for most cases of hereditary hemochromatosis (an autosomal recessive genetic disorder of excess iron accumulation in the body)¹² has led to the use of genetic markers of iron stores (ie, heterozygosity for the C282Y mutation in the HFE gene as a marker of lifelong moderate iron overload) in epidemiologic studies. In contrast to biomarkers, genetic markers of iron overload can be measured exactly and are not influenced by such factors as inflammation, recent blood loss, diet, and use of medications (eg, aspirin). However, results from studies on genetic markers of iron overload and cardiovascular disease have been conflicting, and a recent large study from Denmark found no increased risk of CHD among carriers of the C282Y mutation or individuals who had compound heterozygosity for the C282Y and H63D mutations.¹³

These inconsistent and largely negative epidemiologic findings have fueled the debate over the role of iron in the pathogenesis of atherosclerosis and the value of iron reduction in prevention of cardiovascular disease. For this reason, results have been eagerly awaited from the first randomized controlled clinical trial assessing iron-store reduction and hard end points. In this issue of JAMA, Zacharski and colleagues¹⁴ report outcomes from a multicenter, randomized, controlled, single-blinded trial of the effects of iron reduction through phlebotomy on mortality and morbidity.

This trial was conducted between May 1999 and April 2005 in the Department of Veterans Affairs Cooperative Studies Program and enrolled 1277 patients with symptomatic peripheral arterial disease. Nearly all patients were men, more than 80% were white, and their mean age was 67 years. During a mean follow-up period of approximately 4.5 years, there were no significant effects on either primary (all-cause mortality) or secondary (death plus nonfatal myocardial infarction or stroke) study end points. In addition, reduction of iron stores had no effect on incidence of myocardial infarction or stroke. However, a post hoc subgroup analysis found significant interactions between treatment and age, ie, younger patients (43-61 years) randomly assigned to undergo iron reduction had a significant decrease in primary as well as secondary end points. These benefits were not observed in older patients.

The results from this well-designed and executed trial generate more questions than answers. As with any negative study, the first question is whether the research was adequately powered. Due to lower than expected recruitment and higher than expected noncompliance rates, the investigators estimated that they had only 68% power to detect a 30% reduction in mortality. Thus, despite a relatively large sample size and long duration of follow-up, the study was underpowered.

The second question is did the intervention work? The investigators carefully developed a simple and safe (it avoided iron deficiency) method of reducing body iron stores through graded phlebotomy, and adherence was reasonably good. On average, patients assigned to undergo iron reduction had 72% of the calculated amount of blood removed over the course of follow-up. In control patients who underwent sham phlebotomy, mean ferritin levels did not change; in the treatment group, however, there were significant reductions in mean ferritin levels, from 121.4 ng/mL to 79.7 ng/mL. A subsidiary analysis that excluded participants with less than 50% compliance did not change the results, suggesting that the magnitude of iron reduction had no effect on outcomes.

A third question is whether the study patients were a suitable population in which to test the hypothesis. Because all participants had advanced peripheral arterial disease, and most had some form of cardiovascular disease, this study can be considered a secondary prevention trial. It is understandable that high-risk patients were selected to achieve reasonable duration and sample size, but it is unclear whether the findings can be generalized to healthy individuals.

The investigators also highlight the interaction between age and iron-reduction treatment. Although these results are intriguing, they should not be overinterpreted because the analyses were not planned a priori and were based on a small proportion of the total end points. The interaction test between age group and treatment for total mortality was only marginally significant (P = .04); thus, the statistical significance is unlikely to remain following adjustment for multiple testing. The same caution should be applied to interpretation of the interaction between current smoking and treatment. Although it is biologically plausible that iron reduction may benefit only younger people or current smokers, the only way to confirm these newly generated hypotheses is to replicate the findings in another trial.

This study does not provide definitive proof that the iron-heart hypothesis is false, but the overall results are consistent with largely negative epidemiologic evidence accumulated so far. Although a trial in younger patients would require a much larger sample size and longer duration of follow-up, it is the next logical step. Rather than a trial in younger patients alone, however, it seems more sensible to conduct an adequately powered trial in postmenopausal women across a wide range of age groups, because the initial iron-heart hypothesis was based on sex differences as well as premenopausal and postmenopausal differences in CHD risk.

Because most cohort studies on iron and CHD have been negative, it is possible to argue that investing more time and money on further epidemiologic analyses of iron and cardiovascular disease is unnecessary. However, most prior studies have been small and have not explored potential subgroup effects. This can be performed in an updated meta-analysis of published cohort studies or in de novo analyses of much larger cohorts. The relationships between increased body iron stores and other chronic conditions, such as diabetes and cancer, have also generated much interest and will continue to be investigated in epidemiologic and mechanistic studies.

Recently, genetic epidemiologic studies of iron stores and chronic diseases have been expanded to include the analyses of gene-environment interactions.^{15 - 16} In addition, the new tools of proteomics can provide a direct measure of iron-carrying protein expressions in tissues such as coronary arteries,¹⁷ although such measurements are not feasible in large epidemiologic studies. Unlike ferritin, which only provides a measure of bound iron, new biomarkers, such as plasma non–transferrin-bound iron, provide a measure of a labile or free form of iron that may play a more direct pathologic role in oxidation-mediated tissue damage. However, a recent Dutch study found no relationship between plasma non–transferrin-bound iron and risk of CHD.¹⁸

Like any other theory in cardiovascular medicine, the iron-heart hypothesis will undergo many “trials and tribulations” before it is proven or refuted. Fortunately, it is not necessary to wait for additional data to implement effective strategies that can prevent coronary arteries from getting “rusty.” Even though the question of whether reduced iron stores and CHD risk are causally linked remains unanswered, there is solid evidence that regular exercise and maintaining a healthy weight can reduce both iron stores and risk of CHD.