Peters A, Ehlers M, Blank B, Exler D, Falk C, Kohlmann T, Fruehwald-Schultes B, Wellhoener P, Kerner W, Fehm HL. Excess Triiodothyronine as a Risk Factor of Coronary Events. Arch Intern Med. 2000;160(13):1993-1999. doi:10.1001/archinte.160.13.1993
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2000
Abnormalities in cardiac function, eg, arrhythmias and congestive heart failure, often accompany thyrotoxicosis. A relationship between thyroid hormone excess and the cardiac complications of angina pectoris and myocardial infarction (MI) remains largely speculative.
The results of thyroid function studies on blood samples drawn from a total of 1049 patients (aged 40 years or older) immediately on emergency medical admission were related to frequencies of angina pectoris and myocardial infarction as determined according to current diagnostic algorithms. After 3 years, those patients who had initially presented with angina pectoris or acute MI were observed for subsequent coronary events; of these (n=185), 98% of the subjects (n=181) could be reevaluated.
On hospital admission, the relative rate of angina pectoris and MI was markedly high (odds ratio, 2.6; 95% confidence interval, 1.3-5.2; P=.007) in patients with elevated serum free and total triiodothyronine (T3) levels. An initially elevated free T3 level was a risk factor for subsequent coronary events during the 3-year follow-up (adjusted odds ratio, 4.8; 95% confidence interval, 1.3-17.4; P=.02).
An elevation of serum free T3 levels at hospital admission is associated with a 2.6-fold greater likelihood of the presence of a coronary event. Moreover, an initially elevated T3 level is associated with a 3-fold higher risk of developing a subsequent coronary event during the next 3 years. Excess T3 seemed to be a factor associated with the development and progression of acute myocardial ischemia.
THE CLOSE relationship between the thyrotoxic state and abnormalities in cardiac function, eg, arrhythmias and, in the presence of underlying heart disease, congestive heart failure, is well known to clinicians.1,2 It remains unclear, however, whether the presence of thyrotoxicosis also poses an increased vulnerability to angina pectoris or acute myocardial infarction (AMI).
In the thyrotoxic state, circulatory alterations occur, which may be expected to both increase and decrease the likelihood of developing or progressing myocardial ischemia in any given patient. Among the unfavorable prognostic factors are the increase in cardiac output and contractility.3,4 In addition, the increase in total blood volume imposes a hypervolemic burden on the heart. As in the cardiovascular response to exercise, these changes increase the left ventricular workload, which may lead to myocardial ischemia. Another unfavorable prognostic factor is a reduction in contractile reserve, which is reflected in the failure of exercise to elicit an increase in the ejection fraction. This phenomenon is reversed after antithyroid treatment.5 Furthermore, coronary vasospasms occasionally occur, which may result in angina pectoris or myocardial infarction (MI) even in the absence of underlying cardiac disease.6,7
Conversely, a fall in systemic vascular resistance may be a favorable factor in the setting of either ischemic or hypertensive heart disease.4 Raising serum triiodothyronine (T3) concentrations in patients undergoing coronary artery bypass surgery lowers systemic vascular resistance and increases cardiac output, but does not change outcome or alter the need for standard postoperative therapy.8 Moreover, a slight enhancement of myocardial contractility may be beneficial because it lowers the end diastolic pressure and increases the epicardial-to-endocardial pressure gradient of coronary flow.9
The relative importance of these various factors in regard to cardiac functioning in the thyrotoxic state has not yet been determined, especially as to whether beneficial or deleterious effects have greater impact. Therefore, we have designed this study to test the hypothesis of whether thyroid hormone excess is related to the cardiac complications of angina pectoris and MI.
This study was based on the evaluation of 1049 patients from the population of Lübeck, Germany, and its vicinity. All persons aged 40 years or older presenting to the medical emergency treatment center at the Medical University of Lübeck from January 1 to April 30, 1995, were included, with the exception of those who had received intravenous vasopressive therapy or mechanical respiration, as well as those who had been resuscitated following cardiac arrest prior to admission (21 patients). Characteristics of the 1049 patients on hospital admission are given in Table 1. Heparin had not been administered to any patient before phlebotomy, not even by the referring emergency physician. As the university hospital is 1 of 2 facilities serving the area, emergency admissions are made to this institution on odd-numbered days of the month while the second hospital admits on respective even-numbered days of the month. This special allocation system in Lübeck allowed for drawing an approximately 50% sample of all patients presenting for emergency medical care with a minimum of potential selection bias. Every patient gave written consent. The consent form and the study protocol have been approved by the local ethics commission.
After 3 years, from January 1 to April 30, 1998, we did follow-up evaluations of those patients who had initially presented with angina pectoris or AMI. Of these (n=185), 98% of the subjects (n=181) could be recalled and reevaluated. Baseline characteristics of the 181 patients who completed the follow-up are given in Table 1.
Serum thyroxine (T4) and T3 concentrations were determined from blood samples drawn immediately after arrival in the emergency department and prior to any intravenous therapy, including administration of heparin. Ascertainment of the respective free T3 and T4 (FT3 and FT4, respectively) and total T3 and T4 (TT3 and TT4, respectively) serum concentrations was accomplished using enzyme immunoassays (Boehringer Mannheim Immunodiagnostica, Mannheim, Germany). Total serum T3 and T4 levels were measured for comparative purposes to rule out possible laboratory inaccuracies in determination of the respective free hormone concentrations.10 The reference intervals for our laboratory were FT3, 4.5 to 9.0 pmol/L (3-6 pg/mL); FT4, 10 to 25 pmol/L (0.8-1.9 ng/dL); TT3, 1.2 to 2.7 nmol/L (0.8-1.8 ng/mL); and TT4, 58 to 151 nmol/L (4.5-11.7 µg/dL). These intervals correspond to those established by Gerhardt and Keller,11 who used the same type of assays on a population from nearby Scandinavia. Thyrotropin (TSH) was measured using an immunoluminometric assay (Brahms Diagnostica, Berlin, Germany; reference interval, 0.2-3.5 mU/L). The samples were analyzed either on the day of hospital admission or, after immediate storage at −20°C, within 72 hours following admission.
On hospital admission, angina pectoris, and AMI were diagnosed on the basis of acute symptoms, electrocardiographic (ECG) classification, cardiac enzymes, and patient history related to coronary heart disease. Angina pectoris was diagnosed if acute chest pain at rest had arisen within 12 hours prior to hospital admission and at least 1 of the following 3 criteria for "suspected ischemia"12 was present: (1) angina, based on a standardized questionnaire,13 (2) history of a possible infarction based on the same questionnaire, or (3) positive ECG signs according to the items of the Minnesota code. A diagnosis of definite MI was based on current World Health Organization diagnostic algorithms according to the MONICA study protocol.14
Information on possible interfering variables was collected in a structured interview with each patient, through contact with the general practitioner, from earlier hospital records, or from current laboratory tests and ECGs. We assessed the presence of the classic risk factors (hypercholesterolemia; history of hypertension, diabetes, or smoking), drug effects (prehospitalization medication, contact with iodine-containing contrast agents), and coexisting diseases (atrial fibrillation documented in ECG at admission; history of thyroid or liver disease).
In the cohort study, the occurrence of subsequent coronary events up to 3 years after initial evaluation was assessed by using the current World Health Organization diagnostic algorithm for AMI14: Patients who had initially presented with angina pectoris were observed for the occurrence of MI; patients who had initially presented with MI were observed for the occurrence of coronary death (<28 days after initial evaluation) or recurrent MI (≥28 days after initial evaluation). Information on coronary events and on possible interfering variables was collected in a structured interview with the patient's general practitioner, from hospital records, autopsy findings, in a structured telephone interview with each surviving patient, and through contact with family members of each patient who had died. Coronary revascularisation procedures (percutaneous transluminal coronary angioplasty, stent, bypass) during the 3-year follow-up were assessed; however, in case of a recurrent coronary event, only procedures prior to the respective event were included in the analysis. Investigators collecting the data on events during follow-up were blinded to thyroid status.
The data were evaluated using logistic regression analyses to estimate odds ratios (ORs) and their 95% confidence intervals.15 Odds ratios were calculated to analyze the relationship between the results of the thyroid function tests and the occurrence of coronary events. These ORs were adjusted for age and sex. Additional covariates (hypertension, diabetes, smoking, hypercholesterolemia, liver disease, atrial fibrillation, etc) were conditionally selected in a forward stepwise procedure. Calculations were done using the SPSS statistics program, version 6.0 (SPSS Inc, Chicago, Ill).
Among the 1049 patients examined, elevations of serum FT3 or serum FT4 levels, or suppressed serum TSH levels were found in 60, 99, and 108 patients, respectively (Table 1). Most of these patients had 2 or 3 of these thyroid function alterations corresponding to the strong associations between FT3, FT4, and TSH (Table 2). There were 109 respective diagnoses of angina pectoris and 76 cases of AMI. Using bivariate analysis, we found elevated FT3 levels consistently correlated with both angina pectoris and MI (Table 3). If FT3 level was considered a continuous variable, the correlation with angina pectoris and MI was also significant (regression coefficient, 0.0148; SE, 0.0056; P=.008). The multivariate logistic regression revealed an adjusted OR of 2.6 (95% confidence interval, 1.3-5.2; P=.007) for the occurrence of angina pectoris or MI in the presence of elevated serum FT3 levels (Table 4).
As possible interfering variables, the classic risk factors, eg, history of hypertension, smoking, or elevated low-density lipoprotein cholesterol, were significant at ORs of 1.7, 1.5, and 1.9, respectively. Atrial fibrillation was more frequent in patients with suppressed TSH levels (age-adjusted OR, 2.2; 95% confidence interval, 1.3-3.8; P=.004; no further predictors of atrial fibrillation could be identified using a stepwise forward logistic regression model). Nonetheless, there seemed to be no relationship between atrial fibrillation and the manifestation of coronary events after application of the logistic regression model to our data (Table 4). Finally, the occurrence of documented liver disease also showed no effect on the incidence of coronary events in the logistic regression.
A summary of the substances having potential influence on thyroid function that appeared in the histories of the 60 patients with elevated serum FT3 levels (amiodarone, β-adrenergic antagonists, glucocorticoids, carbamazepine, salicylates, furosemide, radiographic contrast agents) is given presented in Table 5. Contact with iodine-containing contrast agents within 6 months prior to admission was reported by 9 (15%) of the 60 patients. In total, 140 (13%) of the 1049 patients reported having received contrast substances (Table 1).
Assessments of TT3 levels, performed to confirm the validity of the laboratory analyses in the determination of hypertriiodothyroninemia, supported the results of the FT3 studies. In the subgroup of 60 patients with elevated FT3 levels, we also found increases in TT3 concentrations, above the reference range in 48 (80%) of these patients, and within the upper third of the reference range in 12 (20%).
In the cohort study, elevated FT3 levels were confirmed to be strongly related to coronary events during the 3-year follow-up. Using bivariate analysis we found elevated FT3 levels as a risk factor for subsequent coronary events. The crude relative risk was 3.0 (95% CI, 1.3-7.1); the crude OR, as an estimate of the relative risk, was 4.0 (95% CI, 1.2-12.7; P=.01) (Table 6). Because of strong associations between FT3, FT4, and TSH, we used a multivariate model with thyroid function alterations conditionally selected in a forward stepwise procedure (inclusion criterion, P<.10). Of thyroid function alterations, only FT3 was a significant risk factor for subsequent coronary events: FT3, FT4, and TSH with P values of .02, .19, and .16, respectively. After adjustment for other known risk factors and potentially confounding variables, the multivariate logistic regression revealed an adjusted OR of 4.8 (1.3-17.4; P=.02) for the occurrence of subsequent coronary events in patients with initially elevated FT3 levels (Table 7).
Our data demonstrate a strong association between excess T3 and the rate of manifestation of coronary events, ie, angina pectoris or MI. A relatively large number of patients (60) presented with T3 excess (Table 1). Angina pectoris or MI was diagnosed in 1 of 5 presenting patients (Table 1). Logistic regression analysis, as presented in Table 4, revealed an unexpectedly high adjusted OR of 2.6 for the occurrence of angina pectoris or MI in the presence of elevated free T3 levels.
The coastal region of northern Germany, where Lübeck is situated, is an iodine-deficient area.16 As expected, we observed a markedly higher rate of elevated T3 levels in our patients than in hospital admissions in geographic areas with an adequate iodine supply.17 For 3 reasons, iodine deficiency offers a unique opportunity to study the effects of thyroid hormone excess in a patient population. First, thyrotoxicosis, especially with excess T3, occurs more frequently in iodine-deficient areas.18 Second, slight elevations in serum T3 levels, as occurred in few patients in our study, also occur in iodine-deficient areas owing to adaptive changes, including an increased conversion of T4 to T3 in thyroidal and extrathyroidal tissues, and an increased secretion of TSH.19 Rarely, serum TSH levels may be normal in the presence of slightly elevated T3 levels because T3 is not very effective as an inhibitor of TSH secretion.20,21 Third, hypertriiodothyroninemia has been documented to occur as a precursor to thyrotoxicosis in some patients.22 Thus, it is possible that we detected an early stage of evolving thyrotoxicosis in some patients that had led to hospital admission for, eg, cardiac complications.
Possible interfering variables also known to affect the rate of coronary events or levels of T3 have also been taken into consideration in our calculations. The classic risk factors, ie, hypertension, smoking, and hypercholesterolemia, were, as expected, shown to be independent risk factors in the regression model. As previously described by other researchers, we found low serum TSH levels associated with atrial fibrillation23 but atrial fibrillation not associated with coronary events, making an interfering effect of atrial fibrillation unlikely. As another factor, liver disease may fundamentally alter thyroid hormone metabolism,24 but our analysis demonstrated no relationship between liver disease, thyroid hormones, and coronary events. Finally, with the exception of 1 patient taking amiodarone, no clinically relevant pharmacological effects could be established for the patients with T3 excess.25,26 In conclusion, we consider it improbable that interfering effects could have been the basis of the observed relationship between thyroid function and coronary events.
Based on these cross-sectional data, with all due caution and recourse to knowledge outside of this study, we may raise some conjectures as to the nature of the relationship.27 On the one hand, it may be that the T3 excess found in our patients is the precipitating cause of the coronary events, reflecting the direct action of T3 on the heart.28 Indeed this may be the major mechanism operative in the relationship evidenced in our data. On the other hand, coronary events in return might also have affected thyroid function. While we found no evidence in the literature that the acute stress associated with coronary events could have led to T3 excess, there is enough evidence that nonthyroidal diseases, such as coronary events, can lead to a low serum T3 concentration.29- 31 Typical changes in nonthyroidal illness include the decrease in extrathyroidal conversion of T4 to T3 accompanied by a decreased serum TSH level. This type of association, however, failed to explain our data because one would expect a coronary event to be associated with low T3 levels, but this was not what we observed. In spite of the likelihood of these T3-lowering effects, ie, altered thyroid function in nonthyroidal illness, we found T3 level elevations. Thus, the view that the cardiac illness could have led to these elevated T3 levels is not convincingly supported by the literature. More consistent with the literature, however, would seem the view of circulating excess T3 playing an important role in its direct effect on the cardiovascular system.
It was a surprising finding that our results failed to show a clear effect of serum T4 and of TSH on coronary events. These findings may result from 3 mechanisms. First, T4, the primary secretory product of the thyroid, is relatively inactive and is converted to the active hormone T3 by the enzyme thyroxine-5′-deiodinase.32 The actions of thyroid hormone are primarily the result of the interaction of T3 with T3 nuclear receptors that bind to regulatory regions and modify their expression.33 These T3 receptors are present in the nuclei of most tissues, and mediate nearly all the known actions of thyroid hormone.34 Thus, T4 can be considered a prohormone, although it may have some intrinsic activity. Second, when T4 secretion is increased, then more T3 is usually produced in extrathyroidal tissue. It could be that the patients' illnesses resulted in decreased extrathyroidal conversion of T4 to T3. Accordingly, of all hyperthyroxinemic patients (99 patients), most had normal T3 levels (66 patients). As a result, limited sensitivity of the myocardium to T4 and limited extrathyroidal conversion could have accounted for the lack of a statistically significant interaction between T4 and coronary events in our study. Third, TSH secretion is directly inhibited by both T4 and T3. Thyroxine seems to be the major feedback regulator of pituitary TSH secretion in the normal individual, while T3 only in high serum concentrations may inhibit TSH secretion.20,21 The probable reason is that T4 contributes more to the nuclear T3 content of the pituitary than it does in many other tissues because the pituitary has considerable thyroxine-5′-deiodinase activity.19 In fact, serum FT4 was a better predictor of low serum TSH levels in our study population (OR, 6.4) than was free T3 (OR, 4.3) (Table 2); ie, TSH seemed more strongly correlated with T4 (being less active in the myocardium) than with T3 (being active in the myocardium). Therefore, our findings on serum T4 and TSH seem to be consistent with a direct effect of T3 on the heart.
In the cohort study, we could demonstrate that baseline excess T3 was associated with a 3-fold higher risk of the occurrence of subsequent coronary events during the following 3 years. Since follow-up of subjects was almost complete (98%), it is unlikely that selection bias has distorted the results. We carefully assessed the presence of potential confounding factors, such as the classic cardiovascular risk factors and coronary revascularization procedures (percutaneous transluminal coronary angioplasty, stents, bypass). As expected, the risk of subsequent coronary events increased with male sex, history of smoking, hypertension, and diabetes, and decreased after coronary revascularization. The effect of elevated T3 levels on subsequent coronary events was independent of any of these potentially confounding factors. The results from the cohort study provide evidence for the prognostic value of excess T3.
In conclusion, we could demonstrate that an elevated T3 concentration at hospitalization is associated with a 2.6-fold greater likelihood of the presence of a coronary event. Moreover, an initially elevated T3 level is associated with a 3-fold higher risk of developing a subsequent coronary event during the next 3 years. Excess T3 seemed to be a factor associated with the transition from chronic coronary artery disease to AMI.
Accepted for publication March 17, 2000.
We are indebted to H. H. Raspe, MD, PhD, F.S. Keck, MD, and H. Djonlagic, MD, for their invaluable suggestions, and to J. Levejohann, MD, and S. Sömmer for their help in data collection.
Reprints: Achim Peters, MD, Medical Clinic 1, Medical University of Lübeck, Ratzeburger Allee 160, D-23538 Lübeck, Germany.