Iron: focus on epidemiology and current studies relevant to CVD and all-cause mortality

Although, I believe that iron is more of a player in carcinogenesis than cardiovascular disease (CVD) I would like to do a brief review of the latter anyway. So far, the best evidence suggests that modest iron excess may promote atherosclerosis but not other types of CVD. But does it really?

1. Ferritin associates with all-cause mortality - does it matter?

2. Evidence against the iron-atherosclerosis hypothesis: ferritin & CVD incidence
3. Hemochromatosis: a special casse

4. Special epidemiology - beyond ferritin and CVD incidence

5. What are the problems with epidemiology?
6. The solution: best biomarker, best study design

7. Other studies of interest - biogerontology
8. References

1. Ferritin associates with all-cause mortality - does it matter?
A recent study by Ellervik et al. (2014) found a positive association of ferritin with total mortality (n~9000, 23y follow-up). They also performed a "meta-analysis", but in the end only three studies were included (4).
"...we conducted a meta analysis of our own study and 2 additional previous population-based studies on risk of increased total mortality according to ferritin concentrations in quartiles or tertiles, indirectly testing the hypothesis that using quartiles or tertiles for risk prediction hides an increased risk because the highest quartiles or tertiles often include part of the reference interval for ferritin concentrations, thereby diluting risk estimates."

Indeed, given these tertiles any significant result is unlikely using this approach:
"lowest tertile T1 (range 1–55 ug/L), second tertile T2 (56–126 ug/L), and highest tertile T3 (127–1524 ug/L)."
The normal range is 12-150 ng/mL or up to 300 for males. For comparison, male participants in the largest iron reduction study ever performed reached 50-70ng/ml and lower seemed better (10).

The critical discussion of the Ellervik study by several letters and editorials reminded me just how complicated iron-CVD epidemiology is (4a-4e). Below follow a few key Points.

2. Evidence against the iron-atherosclerosis hypothesis: ferritin & CVD incidence
Other authors, like von Haehling et al., argue the case that low iron status is harmful to CV health, especially to heart failure patients (3):
Indeed, a hypothesis published in 1981, suggesting that iron has 'cardiotoxic' effects,14 has since been called into question, and studies have shown that iron deficiency is important not only in patients with HF, but also in patients with other cardiovascular diseases, such as coronary artery disease (CAD) or pulmonary hypertension, and potentially in those undergoing cardiac surgery. 

They cite the following systematic reviews of epidemiologic studies (5-8), some of which are quite powerful and provide useful discussion (ref. 7, n=17 studies, n>150k). I do take one issue with their review, though. Whatever case von Haehling et al. may have, it seems dishonest not to cite Zacharski and the randomized, controlled FeAST study (10, 11). FeAST has shown a trend for reduced CVD and highly decreased cancer incidence after phlebotomy. This should be absolutely essential knowledge for anyone working (or planning to work) in the field.

A recent meta-analysis by Das et al. (7) focused on CHD/MI incidence in relation to ferritin, total iron binding capacity or serum iron. The authors mention some key limitations of their study: "...for example, early subclinical atherosclerosis may have caused transferrin saturations to drop, inferring reverse causality. A further limitation is that comparison was done between the top tertile and bottom tertile of body iron status. This could potentially mask different effects of mild iron deficiency/anaemia on CHD risk versus the effect of mild iron overload on CHD risk."
Under point "5. What are the problems with epidemiology?" I will discuss why epidemiology is so error-prone.

Keeping the limitations in mind, it is still interesting to note that the analysis by Das and colleagues (7) foung higher serum iron and transferrin saturation to be protective. Serum ferritin and total iron binding capacity was unrelated to CHD. The biology is not quite clear but they speculate that:
"having a higher level of iron is a proxy for having a normal haemoglobin level, which could potentially be protective for CHD. However, not enough of the biology is known between iron and CHD to discern whether haemoglobin status is a true confounder or a mediator."

3. Hemochromatosis: a special casse
Sullivan, the godfather of the iron-heart hypothesis, notes that the data linking relatively severe iron overload to atherosclerosis is mixed (12; cf. 18, 19 for original studies).

Hence Sullivan speculates that:
The very low hepcidin levels seen in homozygous hemochromatosis are associated with systemic iron loading because reduced hepcidin levels permit unregulated ferroportin mediated transfer of iron from intestinal epithelial cells into the systemic iron pool. The more extreme the degree of hepcidin deficiency, the more severe the level of parenchymal iron load, but also the more extreme the macrophage iron retention deficit [which is supposed to be protective]. (12)

So we could also state it as follows. Phlebotomy:
decreased Fe stores -> less hepcidin -> more macrophage FPN -> reduced macrophage Fe beneficial
less hepcidin -> increased Fe stores -> more macrophage FPN -> reduced macrophage Fe beneficial

3a. A brief update on mouse models
Support for the iron-atherosclerosis hypothesis from mouse models is mixed at best. This most recent study from 2013 by Kautz et al. was negative and provides a brief discussion of the literature (13). The mouse study was devised as a test of Sullivan's hypothesis of macrophage-iron involvement. Of note, almost all studies ever performed have been done in the apoE model in B6 mice. Many CVD drugs looking good in the apoE mouse have failed in the clinic. Nevertheless people won't give up yet. At a recent conference I've seen data from another mouse study of iron overload suggesting a contribution to CVD, but these mice had unphysiologically high levels of iron in the vasculature. Given these problems, I am not sure if mouse models are the way to go.

4. Special epidemiology - beyond ferritin and CVD incidence
Other markers of iron metabolism that have been studied in relation to CVD endpoints include:  free iron, hemoglobin/hematocrit (outdated), ferritin, total iron binding capacity, transferrin saturation (TSAT), serum iron, blood donation and dietary iron intake

Blood donation (16) was studied and briefly reviewed in a very interesting paper from 2013. In the introduction they write:
The effect of iron stores on CHD has been studied in the context of voluntary blood donation, [but]with inconsistent results.[10-15]
To overcome this problem  Germain et al. (16) instead used an elegant quasi-randomized, natural experiment design and they find no differences in CHD incidence, consistent with data from the relatively well-adjusted HPFS (one of the Harvard health professionals studies)

In CKD there is some indication of 'lower is better' or J-shaped (i.e. lowest is neutral) respones of iron "intake" from dialysate and/or ferritin Levels (9a, 9b).

Transferrin saturation (TSAT), Ellervik et al. 2014 (4) write:
Our results are in accordance with previous evidence that increased transferrin saturation is associated dose-dependently with increased total mortality (23–25 ).

Stack et al. (15) reviewed the data as well and are more cautious:
The prognostic importance of TSAT for total and cardiovascular mortality has recently been evaluated in several large population studies [4,5,7,10]...Taken together, these would suggest that the TSAT ratio is an independent predictor of cardiovascular death risk in the general population and adds incremental prognostic value beyond that of existing Framingham risk factors...[But]The lack of a consistent pattern in the pattern of association of TSAT with cardiovascular death might suggest that this biomarker behaves differently when evaluated in different clinical populations, a feature that would threaten its clinical utility. 

Interestingly, Das et al. (7) saw the opposite in their meta-analysis but they studied CHD/MI as an endpoint.

Supplementation. As per the Iowa Women’s Health Study Supplementation could be harmful (17)

Other markers
, from a recent 2014 review by Basuli et al. (14):
In some early prospective studies, a weak association between high blood hemoglobin and hematocrit and risk of CHD was noted [but they are] not good surrogates for body iron status...Several recent studies have shown that serum ferritin is independently associated with preclinical measures of vascular diseases...
assess ... the relationship between catalytic iron and heart disease. Catalytic iron is the iron that is not bound to transferrin or ferritin and is available to take part in chemical reactions to produce oxidant products. This can be measured using a bleomycin detectable iron assay (BDI; von Bonsdorff et al., 2002). Results from such studies are equivocal.

5. What are the problems with epidemiology?

The key problems are reverse causation, residual confounding and study power.

A. Vegetarian diets and healthy diets with low levels of bioavailable iron cannot be adjusted for completely. This will exaggerate the benefit of low iron levels (e.g. low ferritin would be more likely in vegans).

B. Inflammation:
the "finding, that high serum ferritin is associated with increased mortality ...  may simply be because ferritin is an acute-phase reactant, and individuals affected by various chronic diseases often have a chronic inflammatory state that includes raised ferritin as part of the inflammatory biomarker signature." (4c)

C. Is there a link between ferritin and bodyweight, anorexia, cachexia, BMI, early undiagnosed disease and weightloss preceding disease?
On the one hand: "Severe protein-energy malnutrition is significantly associated with increases in serum ferritin levels" but what about the small changes preceeding many types of diseases by decades?

It is known that the BMI literature is fraught with problems due to reverse causation by early disease-associated weight-loss and it would be interesting to find out if this also applies to the ferritin literature.
For instance: "Low ferritin levels often mean an iron deficiency is present. This can be caused by long-term (chronic) blood loss ... not enough iron in the diet, or bleeding inside the intestinal tract (from ulcers, colon polyps, colon cancer, hemorrhoids or other conditions)."

Hence it appears at least plausible that malnutrition, disease and weightloss in the early stages could decrease ferritin, whereas it would increase at leater stages. Nevertheless, at first glance I could not find much concrete evidence for this proposition, so it would seem most likely that epidemiology would exaggerate the benefits of low ferritin (see points A and B).

D. Study power
I use the term loosely for a combination of two problems. First, the use of tertiles and second 'regression dilution bias', which dilutes the real effects (if any) due to the fluctuation of the studied marker (e.g. ferritin).

6. The solution: best biomarker, best study design

The best epidemiologic study is a randomized controlled trial. This is the only solution. Given the epidemiologic and preclinical data anyone's guess is as good as mine. Although, naturally my hunch is to give the heighest weight to the FeAST RCT and read the epidemiology ex post. The key problem is, this works well for cancer, because the FeAST data was convincing and highly significant, but the CVD result was a mere trend, i.e. not significant. It requires a lot of handwaiving to tell a story of how iron promotes CVD and death. The animal data is mixed, the hemochromatosis data is mixed, epidemiology with classic biomarkers is flat out contradictory; epidemiology with unconventional markers and endpoints is somewhat more supportive, however, the data from a quasi-randomized study of blood donors again contradicts the idea of a link.

Note, that this is common in science. Opinions differ when the data is conflicting. That's why expert opinion is nearly worthless compared to well-designed studies - which we still lack for the iron-CVD and iron-cancer hypothesis.

Meanwhile, I believe the following biomarkers are promising. Transferrin saturation (TSAT) (15), free iron and tissue iron (improved MRI techniques?).

7. Other studies of interest - biogerontology
Today I am not going to review the recently published studies of iron and aging (1, 2), but I recommend anyone interested to read them.

8. References (some key studies in bold)

Mendelian randomization

1. James, S. A., Roberts, B. R., Hare, D. J., de Jonge, M. D., Birchall, I. E., Jenkins, N. L., ... & McColl, G. (2015). Direct in vivo imaging of ferrous iron dyshomeostasis in ageing Caenorhabditis elegans. Chemical Science, 6(5), 2952-2962.

2. Ma, S., Lee, S. G., Kim, E. B., Park, T. J., Seluanov, A., Gorbunova, V., ... & Gladyshev, V. N. (2015). Organization of the Mammalian Ionome According to Organ Origin, Lineage Specialization, and Longevity. Cell reports, 13(7), 1319-1326.

3. von Haehling, S., Jankowska, E. A., van Veldhuisen, D. J., Ponikowski, P., & Anker, S. D. (2015). Iron deficiency and cardiovascular disease. Nature Reviews Cardiology, 12(11), 659-669.

4a. Clin Chem. 2014 Nov;60(11):1419-28. doi: 10.1373/clinchem.2014.229013. Epub 2014 Aug 25.
Total and cause-specific mortality by moderately and markedly increased ferritin concentrations: general population study and metaanalysis.
Ellervik C1, Marott JL2, Tybjærg-Hansen A3, Schnohr P2, Nordestgaard BG4.

4b. Ferrotoxic Disease: The Next Great Public Health Challenge

4c. Is Ferrotoxicity a New Great Public Health Challenge?
Breimer & Nilsson

Clin Chem. 2015 Apr;61(4):669-70. doi: 10.1373/clinchem.2014.237784. Epub 2015 Feb 2.
In reply.
Zacharski LR1.


5. Ma, J. & Stampfer, M. J. Body iron stores and coronary heart disease. Clin. Chem. 48, 601–603 (2002).

6. Danesh, J. & Appleby, P. Coronary heart disease and iron status: meta-analyses of prospective studies. Circulation 99, 852–854 (1999).

7. Das De, S., Krishna, S. & Jethwa, A. Iron status and its association with coronary heart disease: systematic review and meta-analysis of prospective studies. Atherosclerosis 238, 296–303 (2015).

8. Kang, P., Liu, T., Tian, C., Zhou, Y. & Jia, C. Association of total iron binding capacity with coronary artery disease. Clin. Chim. Acta 413, 1424–1429 (2012).

9. Iron indices: What do they really mean?
9b. Kidney Int. 2015 Jan;87(1):162-8. doi: 10.1038/ki.2014.275. Epub 2014 Jul 30.Data from the Dialysis Outcomes and Practice Patterns Study validate an association between high intravenous iron doses and mortality. Bailie et al.

10. Zacharski LR, Chow BK, Howes PS, et al. Reduction of iron stores and cardiovascular outcomes in patients with peripheral arterial disease. JAMA 2007;297(6):603-610.

11. Decreased Cancer Risk After Iron Reduction in Patients With Peripheral Arterial Disease: Results From a Randomized Trial.

12. Sullivan. Are Hemochromatosis Mutations Protective
Against Iron-Mediated Atherogenesis?

13. Kautz, L., Gabayan, V., Wang, X., Wu, J., Onwuzurike, J., Jung, G., ... & Nemeth, E. (2013). Testing the iron hypothesis in a mouse model of atherosclerosis. Cell reports, 5(5), 1436-1442.

14. Epidemiological associations between iron and cardiovascular disease and diabetes.
Basuli D, Stevens RG, Torti FM, Torti SV.
Front Pharmacol. 2014 May 20;5:117. doi: 10.3389/fphar.2014.00117. eCollection 2014. Review.


Germain, M., Delage, G., Blais, C., Maunsell, E., Décary, F., & Grégoire, Y. (2013). Iron and cardiac ischemia: a natural, quasi‐random experiment comparing eligible with disqualified blood donors (CME). Transfusion, 53(6), 1271-1279.

17. Mursu J, Robien K, Harnack LJ, Park K, Jacobs DRJr. Dietary Supplements and mortality rate inolder women: the Iowa Women’s Health Study. Arch Intern Med 2011;171:1625–33.

Atherosclerosis. 2014 Oct;236(2):292-300. doi: 10.1016/j.atherosclerosis.2014.07.002. Epub 2014 Jul 24.
Hemoglobin, iron metabolism and angiographic coronary artery disease (The Ludwigshafen Risk and Cardiovascular Health Study).

Lian J, Xu L, Huang Y, Le Y, Jiang D, Yang X, Xu W, Huang X, Dong C, Ye M, Zhou J, Duan S.
Gene. 2013 Sep 15;527(1):167-73. doi: 10.1016/j.gene.2013.06.034. Epub 2013 Jun 20.
Rovers, M. M., Grobbee, D. E., Marx, J. J., Waalen, J., Ellervik, C., Nordestgaard, B. G., ... & Boer, J. M. (2008). Mutations in the HFE Gene and Cardiovascular Disease Risk An Individual Patient Data Meta-Analysis of 53 880 Subjects. Circulation: Cardiovascular Genetics, 1(1), 43-50.