The relationship between serum ferritin levels and electrocardiogram characteristics in acutely ill patients

Abstract

BACKGROUND:

Cardiac arrhythmias are common comorbidities among acutely ill patients admitted to hospitals. An abnormal iron metabolism may contribute to the abnormalities in the conduction and propagation of action potentials through myocardium.

OBJECTIVE:

To determine whether serum indexes of iron metabolism correlate with electrocardiogram (ECG) changes.

METHODS:

In the present retrospective, pilot chart review, serum levels of iron, ferritin, Na⁺, K⁺, Mg²⁺, Ca²⁺ and total iron-binding capacity in 77 hospitalized patients with acute illness were correlated with ECG variables.

RESULTS:

The serum ferritin level correlated strongly (r=0.49) with QT/QTs interval. There were three subjects with QT prolongation (longer than 450 ms) within the high serum ferritin (576 ng/mL or greater) group versus subjects with low ferritin. Multiple regression analysis showed that serum ferritin level and serum iron level contributed to the variance in the QT/QTs prolongation. No other correlation between the studied serum markers and ECG characteristics were found.

CONCLUSION:

The present study suggests that serum ferritin and iron levels affect the QT interval in a variety of medical conditions, possibly contributing to the emergence of fatal cardiac arrhythmias.

Iron plays an important role in oxygen delivery, free radical production and immunity (1). Cardiomyopathy associated with faulty storage of excess iron is described in several hereditary and acquired conditions (2–5). In the heart, iron is deposited predominantly in myocardial cells, rather than in the interstitium (6,7). This leads to impaired generation and propagation of electrical impulses at the level of the myocardial membrane (7–10). It has been suggested that excessive intracellular iron interferes with electrical function of the heart, either by generating large amounts of free radicals or by causing selective dysfunction of Na⁺ channels (7,9,11). Aberrant function of Na⁺ and K⁺ channels contributes to the etiology of prolonged QT syndrome, ventricular tachyarrhythmias and atrial fibrillation (8,12–15). Additionally, an abnormal impulse propagation, such as delayed impulse conduction, is important in the development of a variety of bradyarrhythmias (8,11,16). These effects are most likely independent of concomitant structural abnormalities of the heart, myocardial hypertrophy or cardiac ischemia – pathologies frequently observed in a siderotic heart (1,6,11,17,18).

Excessive iron storage is common in various clinical conditions such as alcoholism, hepatitis C and other inflammatory syndromes (1,2,4,19). In acutely ill patients, iron turnover can be severely affected because some of the proteins involved in its metabolism belong to the acute phase response protein family (1,20). Experimental data suggest that the propagation of the action potential is abnormal in cardiomyocytes overburdened with iron (7,9). Data from human subjects suggest that excessive iron storage is associated with cardiac arrhythmias (11,21). However, the effect of abnormal indexes of serum iron metabolism was not studied in the general population of acutely ill subjects. Because abnormalities of iron metabolism are common in acute illness, the increase in prevalence of fatal cardiac arrhythmias could result from high or low iron levels (1,2,4,20). We hypothesized that acquired iron metabolism abnormalities associated with an emergence of an acute medical pathology have an impact on the frequency of abnormalities seen in the electrocardiogram (ECG), independent of electrolyte disturbances and administration of proarrhythmogenic drugs.

METHODS

The retrospective chart review was approved by the Institutional Review Board of Griffin Hospital, Connecticut, USA. Using the hospital’s database, patients hospitalized between 2001 and 2005 who had serum iron and ferritin levels measured (n=2748) were identified. From this database, demographic parameters (age, sex) and serum levels of Na⁺, K⁺, Mg²⁺, Ca²⁺, iron, total iron-binding capacity (TIBC) and ferritin were obtained. Serum Ca²⁺ levels were corrected for albumin concentration. ECG results were retrieved and reviewed for accuracy. QTs interval was calculated according to methods suggested by Sagie et al (22). Left ventricular hypertrophy and right atrial enlargement were defined according to ECG criteria (16). Only the subjects with all serum variables collected within a 24 h period were considered for the present study (n=492). Their charts were screened for pre-existing arrhythmias, chronic intake of proarrhythmogenic drugs (amiodarone, chlorpromazine, chinolones, haloperidol, methadone, sotalol, class III antiarrhythmogenic drugs) or a pre-existing pacemaker (8,12,13,16) – subjects with these conditions were excluded from the study. A total of 77 records were statistically analyzed.

Patients were divided into high (HFer) and low (LFer) ferritin level groups according to the descriptive analysis. The mean serum ferritin level of the entire cohort (576 ng/mL) was used as the cut-off point to separate subjects into the HFer and LFer groups. A similar value was used by other studies to identify patients with an iron overload (1,19,20,23). The Shapiro-Wilk W test and distribution plots were used to examine the normality of the distributions. Results are expressed as mean ± SD for variables that exhibited a normal distribution. These variables were compared using the Student’s t test after the patients were divided into the HFer and LFer groups. For categorical variables, the Pearson’s χ² test was applied. The r²-Pearson momentum and multiple regression with stepwise forward method was calculated for some analyses. The predictors were indexes of iron metabolism (serum iron, TIBC, serum ferritin), and covariables included serum electrolyte levels, although abnormalities in ECG were deemed the outcome variables. A double-sided P value of less than 0.05 was considered to be statistically significant for all tests. Statistical analyses were performed with Statistica version 8.0 (StatSoft Inc, USA).

RESULTS

The demographic variables are presented in Table 1. There were no differences in age, sex and diagnosis admission codes between the HFer and LFer groups. Also, the measured laboratory values were not significantly different between studied groups, except for serum ferritin level and TIBC, which were significantly higher in the HFer group.

TABLE 1.

Patient demographic information, admission codes and laboratory values

Variable	HFer (n=30)	LFer (n=47)
Sex, n
Male	17	21
Female	13	26
Age, years	59.1±16.8	55.7±18.5
Diagnosis-related group admission codes, n
Ethanol intoxication	7	2
Anemia	7	3
Liver dysfunction	10	4
Pneumonia	11	7
Stroke	6	3
Newly diagnosed cardiac arrhythmia	2	3
Psychiatric condition	2	6
Others	2	2
Laboratory values
Serum iron, ng/mL	42.0±39.6	57.2±62.3
Total iron-binding capacity, %	232.1±81.9*	297.4±114.3*
Serum ferritin, ng/mL	1394.2±573.9*	306.3±230.6*
Serum Na+, mg/dL	137.2±4.9	137.9±5.1
Serum K+, mg/dL	4.04±0.59	4.22±0.78
Corrected total serum Ca2+, mg/dL	8.8±0.74	8.9±0.78
Serum Mg2+, mg/dL	1.42±0.21	1.48±0.32

Data presented as mean ± SD unless otherwise stated.

^*P<0.05 was considered statistically significant. LFer Low ferritin; HFer High ferritin

Analyses of ECGs showed that all patients had underlying sinus rhythm. There was no difference in heart rate between the HFer and LFer groups. There was no difference in STT deflection from the isoelectric baseline or QRS width between the studied groups. The PQ interval was slightly longer in the HFer group but the difference was statistically insignificant (Figure 1). Both QT and QTs intervals were significantly longer in subjects with elevated serum ferritin level (Figure 1). However, three subjects in the HFer group (3.95% of all subjects) and one in the LFer group (1.32% of all subjects) had prolongation of the QT interval longer than 450 ms. The levels of electrolytes were not significantly altered in these four subjects (Table 2). No difference in the frequency of left ventricular hypertrophy was seen between HFer versus LFer groups.

Figure 1).

Statistically significantly (X) longer QT and QTs intervals were seen in high ferritin (HFer) versus low ferritin (LFer) study groups. QTc Corrected QT

TABLE 2.

The individual value of serum electrolytes in four patients with prolongation of QTs

	Electrolyte
	Na+	K+	tCa2+	Mg2+
High ferritin group
Patient 1 (QTs=512 ms)	142	4.8	8.9	2.0
Patient 2 (QTs=490 ms)	139	4.5	9.0	1.8
Patient 3 (QTs=473 ms)	139	3.8	9.4	1.4
Low ferritin group
Patient 4 (QTs=467 ms)	137	4.5	8.8	2.0

tCa Total calcium

A correlation analysis between QT and QTs demonstrated a statistically significant weak link with serum iron (r²=–0.29; P=0.05). However, the association between QT/QTs and serum ferritin levels was much stronger (r²=0.49) and robust (P<0.001) (Figure 2). In the subsequent regression analysis, only serum ferritin concentration (β=0.49; P<0.0001) and serum iron levels (β=–0.24; P=0.042) were found to be significant predictors of QT/QTs variance. Serum levels of Na⁺, K⁺, Mg²⁺, Ca²⁺ and TIBC were excluded because they contributed insignificantly to the variance of QT/QTs.

Figure 2).

The correlation between serum ferritin level and QT interval is strong (r²=0.49) and robust (P<0.001)

DISCUSSION

In the present retrospective study, we demonstrated that there is a relatively strong relationship between serum levels of ferritin and QT/QTs. Regression analysis showed that 24% of the variation in QT/QTs is dependent on the serum ferritin level. Three subjects in HFer and one in LFer had sufficient prolongation of QTs to recognize pathological prolongation of QT. Because serum levels of Mg²⁺, Ca²⁺, K⁺ and Na⁺ affect depolarization and propagation of the action potential through the myocardium, we investigated their effects (8,13,14,16). We found that these variables had no effect on the duration of the QT interval. Heart rate, PQ interval, QRS axis and width, STT deflection, and frequency of left ventricular hypertrophy or right atrial enlargement were not significantly different between HFer and LFer groups.

A literature search did not identify any studies that examined the effect of serum indexes of iron metabolism on ECG changes in a population of acutely ill patients. Most studies (3,4,17,19,23) have examined the effect of changes of iron metabolism and have been conducted in chronic iron overload conditions such as hemosiderosis or renal failure. Abnormalities in iron metabolism are common among subjects with acute pathology (2,20). Cardiac arrhythmias are also very prevalent in this population (8,14,15). The underlying mechanisms for cardiac arrhythmias may be due to structural heart abnormalities in siderotic hearts, the generation of free radicals in iron-overloaded hearts or the interference of the function of several ion channels due to abnormalities of intracellular iron metabolism. Of these three etiologies, the latter is more plausible because it has been observed among acutely ill patients without pre-existing siderotic cardiomyopathy. Thus, we hypothesized that there may be a link between aberration of iron metabolism and propagation of action potentials through the heart. This would be an important finding because mortality from cardiac arrhythmias, especially from a prolonged QT interval, is substantial among acutely ill subjects admitted to the intensive care unit (8,12).

Our study population consisted of a diverse group of patients, suggesting that the observed effect of serum ferritin on QTs is fairly universal. Similar findings were observed in dialyzed patients. Wu et al (19) independently showed that serum iron levels affect the dispersion of QT in patients treated with peritoneal dialysis. In this population, serum ferritin levels were routinely in the upper limit of the level considered to be normal in the general population. Similar data were obtained in hemodialyzed patients (23). However, studies of dialyzed subjects are frequently hampered by abrupt changes in electrolyte levels and varying degrees of cardiomyopathy (24). These comorbidities influence the propagation of action potentials through the heart, possibly obscuring the effect of iron metabolism abnormalities (8,16). Heterogeneity in the studied sample is responsible for increased variability of data obscuring the effect of the measured variable. We tried to limit the effect of several covariables by eliminating subjects with pre-existing arrhythmias or those taking medication known to affect depolarization of the heart (12), although some of these medications may have been overlooked. An alternative approach would be to enlarge our sample size and control for the effect of various medical conditions; however, the increase in the number of subjects that would be required would be prohibitive.

Signal conduction from atria to ventricles, and its propagation through the ventricle was not affected by measures of iron metabolism in our study. Some authors (1,3,6) suggest that increased fibrosis secondary to hemosiderosis can be responsible for this effect. Because we studied acutely ill patients, the expected prevalence of hemosiderosis was low (5). Also, our study was underpowered to detect prolongation of PQ in subjects with elevated ferritin levels. Abnormal propagation of the electrical impulse through myocardia is disrupted by ischemia, hypertrophy or pre-existing structural abnormalities of the heart (18,25). These changes can be detected only by echocardiography, a measure that was not included in the present analysis.

The observed correlations were specific to the QT interval. Cardiac arrhythmias in the present study were observed in a similar study (21) that used a very small group of patients. This report suggested that by lowering the ferritin level, some cardiac arrhythmias can be reversed. The experimental data suggest that the propagation of the action potential through the membrane of myocardial muscles is primarily affected by elevated serum iron and ferritin levels by reduction of current through the Na⁺ channels as a whole (9,10). These channels are critical for depolarization, whereas prolongation of QT is predominantly dependent on the K⁺ rectifier current (12,13,15,16). Excessive iron deposition in hearts resulted in alteration of outward K⁺ current but inward rectifier K⁺ current was unchanged (7). The data suggest that the observed prolongation of QT (which correlated with increased serum ferritin level) is an intracellular phenomenon. Recently, new data emerged suggesting that in epithelial cells, there is a concurrent uptake of Na⁺ and iron, the latter being sequestrated as ferritin (26). This process affects the inward K⁺ current as well. It is possible that this ferritin-dependent K⁺ current is involved in the pathology. Alternatively, the function of the rectifier K⁺ channels may be disrupted by free radicals that were possibly generated from ferritin (8,13,15,27). The data also suggest that a link between iron metabolism and QT prolongation depends on intracellular interactions versus abnormalities of extracellular electrolytes. Serum ferritin is a reflection of intracellular stores of iron and this variable was very tightly linked with QT/QTs. In contrast, serum iron levels, which are mostly extracellular, were not as significant.

Our study has several limitations. It is a retrospective chart review, and as such, has several methodological limitations. However, the present study was designed to be a pilot study for a larger project. We enrolled 77 subjects, which is not sufficient to detect the effect of heart abnormalities and left ventricular hyperthrophy on variance of the observed results. The power was significant to detect QT/corrected QT prolongation but it was not enough to see a 10% increase in PQ interval. Expansion of the groups would be beneficial for the statistical validity of the study. The relatively low number of included subjects was due to the study design attempting to limit the influence of several covariables. This severely limited the number of subjects enrolled; ie, the specificity of the study was sacrificed for power. An additional limitation is that not all medications affecting prolongation of QTs were included in our methodology (8,12,13). There is a chance that surveyed medical documentation was incomplete, resulting in possible inclusion of patients taking medications from the exclusion list. We did not present any data with respect to cardiac function. Cardiac ischemia or remodelling affects propagation of the depolarization through myocardium (13). Despite the correlation between serum ferritin and QT/QTs, the majority of subjects had a QT below the pathological duration of 450 ms (15,22). If the QT interval is longer, the risk of sudden cardiac death is increased (13). Because our study design was retrospective, we were not able to survey long-term mortality in the studied cohort.

An elevated serum ferritin level can act as a cofactor in inducing deadly QT prolongation or be a sole culprit. It is possible that in the presence of an elevated ferritin level, other offending factors can be lethal at lower concentrations. Our study was not designed to detect this proposed mechanism because we had limited our entry criteria to a large extent. Serum ferritin level is very dependent on iron intake, underlying pathology, clinical status of the patient, extent of the host’s inflammatory response and other variables (1,2,4). Our study shows a correlation between elevated serum ferritin level and prolongation of QT, but it is unable to answer whether they are related in a causative fashion because of the retrospective nature of the study. It is also speculative as to how acute elevated serum ferritin and iron levels affect ECG characteristics. It is unlikely that among acutely ill patients there is a buildup of iron deposits in the heart. Cardiac biopsy data are necessary for this determination, but such a study would be premature at this step. It has been reported that iron is an important trigger in the production of free radicals. Iron can also interfere with the function of several ion channels, leading to observed ECG abnormalities. It is also possible that aberrant serum indexes of iron turnover are a coexisting result of other processes that lead to ECG changes as well. Again, the retrospective nature of the present study precludes or limits drawing any conclusive steps.

The present data suggest that some indexes of abnormal iron metabolism correlate strongly with QT in populations of acutely ill patients. However, further studies are necessary to detect the effect of acute elevation of ferritin on long-term survival of the patient and its role as cofactor in lethal cardiac arrhythmias.