Abstract
Objective. Corrected QT (QTc) interval predicts all-cause and cardiovascular mortality and may contribute to the increased mortality risk in rheumatoid arthritis (RA). Animal experiments have shown that proinflammatory cytokines [tumor necrosis factor (TNF)-α and interleukin 1 (IL-1)] can prolong cardiomyocyte action potential. We sought to determine whether elevations in circulating inflammatory cytokines were independently associated with QTc prolongation in patients with RA.
Methods. One hundred twelve patients [median age 62 (interquartile range 17) yrs; 80 women (71%)] from a well-characterized RA cohort underwent baseline 12-lead electrocardiograms for QT interval measurement and contemporary blood sampling to assess concentrations of inflammatory markers including C-reactive protein (CRP), TNF-α, and interleukins (IL-1α, IL-1β, IL-6, IL-10). QTc was calculated using the Bazett (QTBAZ = QT ÷ √RR) and Framingham Heart Study (QTFHS = QT + 0.154 × [1 – RR]) heart rate correction formulas.
Results. Inflammatory cytokines (TNF-α, IL-1β, IL-6, IL-10) were positively correlated with QTBAZ (Spearman rank correlation coefficient rho = 0.199, 0.210, 0.222, 0.333; all p < 0.05). In multivariable regression analysis, these associations were all confounded by age except IL-10, where higher tertile groups were independently and positively associated with QTBAZ (β = 0.202, p = 0.023) and QTFHS (β = 0.223, p = 0.009) when compared to the lower tertile. CRP (per unit increase) was independently associated with QTBAZ (β = 0.278, p = 0.001), but not QTFHS.
Conclusion. To our knowledge, ours is the first study demonstrating a contemporary link between inflammatory cytokines and QT interval in humans. Our results suggest that a lower inflammatory burden may protect against QTc prolongation in patients with RA. However, further studies are required to confirm the effects of pro- and antiinflammatory cytokines on QTc interval.
The QT interval represents the time from onset of ventricular depolarization (beginning of the Q wave) to completion of repolarization (end of the T wave). Corrected QT (QTc) prolongation is a weak independent predictor of all-cause and cardiovascular mortality in the general population1, but a more significant one in patients with rheumatoid arthritis (RA)2. Compared to the general population, the risk of sudden cardiac death is almost doubled in RA3 and the case fatality rate (death rate following an acute cardiovascular event) is also increased4,5,6. This increased risk may in part be explained by prolonged QTc interval, a likely surrogate of sudden arrhythmic death2,7.
Given the variability of the QT interval in relation to heart rate, several formulae are used to yield a corrected measure (QTc). The Bazett formula (QTBAZ = QT ÷ √RR), one of the earliest and most commonly used heart rate correction formulae, has been criticized because of an overcorrection and undercorrection of QT interval at high and low heart rates, respectively8. To overcome this problem, a linear regression model using data from the Framingham Heart Study (QTFHS = QT + 0.154 × [1 – RR]) was developed9.
Prior studies (including 1 in patients with RA) have suggested that QTc prolongation (QTBAZ) may be driven by increased inflammatory burden2,10,11,12,13,14. However, few studies have assessed the direct effects of inflammatory cytokines on QT interval. In 1 animal experiment, tumor necrosis factor (TNF)-α prolonged the cardiomyocyte action potential by inhibiting the hERG (the human ether-à-go-go–related) gene (KCNH2) that codes for a protein known as Kv11.1, the α subunit of a potassium ion channel11. In another animal study, interleukin 1 (IL-1) was found to prolong cardiomyocyte action potential through its effects on calcium channels15. In addition, systemic inflammation may lead to QTc prolongation by accelerating the development of structural heart disease7, a major risk factor for QTc prolongation that is more prevalent in patients with RA16. Acceleration of atherosclerosis17, as well as induction of myocardial injury and remodeling18,19,20, have been suggested as potential mechanisms by which systemic inflammation may directly promote the development of structural heart disease7. Alterations in the autonomic nervous system (ANS) function may also result in QTc prolongation7,13. Reduced cardiac parasympathetic activity21 and increased sympathetic activity22, a pattern of ANS dysfunction prevalent in patients with RA23,24,25,26,27, have been shown to contribute to QTc prolongation. The relationship between inflammation and ANS remains to be fully elucidated and is a focus of research28,29,30,31. Evidence from animal experiments demonstrates that IL-1β32, IL-633, and TNF-α34 cause sympathoexcitation, which potentially may prolong QT interval. Collectively, these studies in animal models suggest several potential mechanisms by which proinflammatory cytokines may lead to prolongation of the QT interval, but it remains unclear whether such pathways are involved in humans.
Our study aimed to test the hypothesis that serum proinflammatory and antiinflammatory cytokine concentrations were correlated with contemporary QTc interval in patients with RA. We also sought to determine whether these associations were affected by the QT correction method used (QTBAZ vs QTFHS).
MATERIALS AND METHODS
One hundred forty-six subjects from a well-characterized cohort of patients with RA35,36 were assessed with 12 lead electrocardiograms (ECG) for QTc interval and contemporary blood samples that were analyzed for biochemical markers of inflammation. Disease Activity Score at 28 joints (DAS28)37 was used to assess their current disease activity. RA diagnosis was based on the revised 1987 American College of Rheumatology classification criteria38. Patients were recruited consecutively at the outpatient clinics of the Dudley Group National Health Service Foundation Trust, UK, from July 2004 to October 2006. Our study was approved by the local ethics committee and all subjects gave written informed consent.
QT interval (time between the beginning of a Q wave and the end of a T wave) was manually measured by 2 experts and corrected for heart rate using the Bazett formula (methods described in detail elsewhere2). Additionally, QT was corrected using the Framingham Heart Study formula (QTFHS = QT + 0.154 × [1 – RR])9. Serum was collected on the same day and frozen immediately at −70°C until analyzed. Concentrations of IL-6, TNF-α, IL-1β, and IL-10 were analyzed by flow cytometry using FlowCytomix kits (Bender MedSystems GmbH). Briefly, FlowCytomix kits allowed the analyses of multiple targets in 1 sample. Uniquely color-coded polystyrene beads were coated with monoclonal antibodies that targeted the required cytokine, following which they were allowed to react with the sample. A biotinylated detection antibody specific for a different epitope on the cytokine was then added. Finally, after the addition of a streptavidin-phycoerythrin complex, the emitted fluorescent signals were detected using flow cytometry. C-reactive protein (CRP; Vitros 5.1-FS, Ortho Clinical Diagnostics) and erythrocyte sedimentation rate (StaRRsed Compact, Vitech Scientific Ltd.) were also analyzed. History of ischemic heart disease was defined as having any of the following: angina diagnosed by a physician or elicited by the use of the Rose questionnaire39, myocardial infarction, angioplasty, or coronary artery bypass grafting or/and ischemic heart failure. The Minnesota Code Classification System was used to code for the presence of ST abnormality on the ECG40.
Statistical analysis was performed using SPSS software, version 19 (IBM SPSS Inc.). Continuous variables were tested for normality using the Kolmogorov-Smirnov test. Correlations between QTc (QTBAZ, QTFHS) interval and continuous variables were assessed using Pearson correlation for parametric and Spearman correlation coefficient (rho) for nonparametric data. Associations between QTc (QTBAZ, QTFHS) interval and dichotomous categorical variables were assessed using a 2-tailed, independent Student t test. In addition, cytokines and CRP were divided into tertiles. Differences in QTc interval between tertiles was assessed using a 1-way ANOVA with Tukey posthoc test. Stepwise multivariable linear regression analysis was performed to determine independent associations between QTc (QTBAZ and QTFHS separately) interval and other pertinent measured variables (variables considered clinically significant or those with a univariate association with QTc demonstrating a p value < 0.1 were included in the model). Cytokine tertile groups were entered into the regression models as continuous variables (calculating a β per tertile increase). To minimize the risk of collinearity, variables that are known to interact were entered into separate models (e.g., β-blocker, ischemic heart disease, cytokines). Values were expressed as mean ± SD for parametric data, median [interquartile range (IQR)] for nonparametric data, and frequency (percentages) for categorical variables. A p value of less than 0.05 was considered statistically significant throughout.
RESULTS
Of the 146 patients studied, 34 (23%) were excluded because of the presence of bundle branch block precluding QT interval assessment. Analysis was performed on the remaining 112 patients. Median age (IQR) was 62.4 years (51.5–68.6) with 80 females (71%; Table 1). Other baseline patient characteristics are summarized in Table 1.
Baseline characteristics for patients with RA. Values are mean ± SD for parametric data, median (IQR) for nonparametric data, or n (%) for categorical data.
Heart rate correction
Mean QTc interval was 423.2 ± 22.5 ms (QTBAZ) and 412.5 ± 18.0 ms (QTFHS). QTBAZ was positively correlated with heart rate while QTFHS was not (Table 2).
Correlations between QT, QTc (QTBAZ, QTFHS) heart rate, age, CRP, ESR, and cytokines. Values are correlation coefficient (Spearman rank rho or Pearson product ra), p, and n.
Demographic and disease-related variables
QTc interval (QTBAZ and QTFHS) was weakly positively correlated with age and was higher in women compared to men (QTBAZ = 12.5 ms and QTFHS = 7.8 ms higher; Table 3). Patients with a history of ischemic heart disease had a higher QTc interval (QTBAZ = 16.8 ms and QTFHS = 20.4 ms higher). Hydroxychloroquine (HCQ) use was associated with a higher QTc interval (QTBAZ = 12.9 ms, QTFHS = 9.8 ms higher). There was a trend for β blockade to lower QTBAZ, but not QTFHS. Aside from HCQ, no other medications, including biologic agents (i.e., TNF-α inhibitors) and those with a known association with risk of Torsades de Pointes (thiazide diuretics, amitriptyline, and other antidepressants; www.crediblemeds.org), were associated with QTc interval.
Significant associations between QTc and selected variables. Associations between QTc interval and dichotomous categorical variables were tested using 2-tailed, unpaired Student t test. Multiple factors (found in Table 1) were assessed; however, only those with p < 0.1 were included in this table. Values are mean ± SD unless otherwise specified.
Patients with hypercholesterolemia had a higher QTFHS, while there were trends for higher QTFHS in patients with coexistent osteoarthritis, those receiving proton pump inhibition, and in seropositive patients (anticyclic citrullinated peptide, rheumatoid factor). There were no significant associations between QTc interval and ECG abnormalities (including left ventricular hypertrophy), other comorbidities (including hypertension), medications, and RA-related characteristics (DAS28; Table 3).
Inflammatory markers
Correlation analysis showed that CRP, TNF-α, IL-6, IL-1β, and IL-10 were all significantly associated with QTBAZ (weakly positive) while IL-1α showed only a trend (Table 2). However, when QTFHS was used, only IL-1β and IL-10 were significantly associated (weakly positive) while there was a trend with TNF-α and IL-1α. There was no significant association between QTFHS and CRP or IL-6.
QTBAZ and QTFHS were significantly lower in the first tertile for IL-10 compared to the second and third; this was attenuated after multivariable adjustment for age, sex, presence of ischemic heart disease or hypercholesterolemia, HCQ, and β-blocker use (Table 4). QTFHS was significantly lower in the first tertile for TNF-α compared to the third tertile; however, this was insignificant after adjustment for age and other variables (Table 4). QTBAZ tended to be lower in the first tertile for CRP compared to the third tertile. There was no significant association between IL-6, IL-1α, and IL-1β tertile groups and QTc interval.
Associations between QTc and inflammatory markers according to tertiles before and after multivariable adjustments.
Multivariable linear regression
Stepwise linear regression analysis showed that age, sex, and CRP were independently associated with QTBAZ interval, accounting for 22% of the variation (Table 5). The associations (and trends) found between the cytokines and QTBAZ/QTFHS in univariate analysis were all confounded by age with the exception of IL-10 (Table 5). IL-10 tertile group (when entered into the model as a continuous variable) was independently associated with QTBAZ and QTFHS intervals independent of other factors (including age, sex, β-blocker or HCQ use, and presence of ischemic heart disease).
Stepwise multivariable linear regression with QTc interval as dependent variable. To minimize risk of collinearity, these were entered into separate models: β blocker, heart disease, and hypercholesterolemia, as well as anti-CCP positivity and RF positivity. Standardized β and unstandardized (B) coefficients are shown.
DISCUSSION
To our knowledge, our study is the first to report a correlation between QTc interval and contemporary serum concentrations of proinflammatory and antiinflammatory cytokines in humans. CRP and IL-10 demonstrated a positive correlation with QTc, suggesting that a higher inflammatory burden associates independently with longer duration of QTc interval. Given that QTc prolongation increases the risk of potentially life-threatening arrhythmias (e.g., Torsades de Pointes), our results are in keeping with a possible mechanistic link between increased inflammatory burden in RA and cardiovascular mortality. Cardiovascular case fatality in RA is reported to decline in parallel with better control of disease activity41, which supports the hypothesis that a lower inflammatory burden may protect against QTc prolongation and hence reduce the risk of sudden cardiac death.
Of the cytokines analyzed, IL-10 was independently associated with QTc interval prolongation. IL-10, the major antiinflammatory cytokine, is released by monocytes, macrophages, and B cells in response to inflammatory stimuli, and is known to inhibit the production and release of proinflammatory cytokines (including TNF-α, IL-1β, and IL-6)42,43,44,45. In addition, IL-10 reduces the surface expression of TNF receptors and promotes the release of TNF receptors into the circulation46,47. There are a number of possible explanations for the positive correlation between IL-10 concentration and QTc prolongation found in our study. The surge of IL-10 production is known to follow that of proinflammatory cytokines45 and hence, increased levels of IL-10, as seen in RA48, may represent increased inflammation. Conversely, undetectable levels of IL-10 may indicate lower circulating concentrations of TNF-α and IL-1, and may result in a lower QTc interval; given that both TNF-α11 and IL-115 were shown to prolong cardiomyocyte action potential duration in animal studies. This is supported by our finding that IL-10 was strongly correlated with other proinflammatory cytokines, including TNF-α (rho = 0.775, p < 0.001), IL-1α (rho = 0.853, p < 0.001), and IL-1β (rho = 0.902, p < 0.001).
Additionally, given that the release of IL-10 is stimulated by adrenaline45 and that sympathoexcitation may be associated with QTc prolongation22, lower levels of IL-10 may reflect lower resting sympathetic nerve activity. Finally, the direct effects of IL-10 on cardiomyocyte action potential are currently unknown and further studies are warranted to explore the mechanisms by which IL-10 and other inflammatory cytokines may affect QTc interval.
In our study, we found a significant and independent association between CRP and QTBAZ, but no association between CRP and QTFHS. We are the first to report such a discrepancy between QTBAZ and QTFHS; to our knowledge, prior studies showing associations between QTc interval and CRP all used the Bazett heart rate correction formula. In 1 study of 101 patients with inflammatory arthritis, elevated concentrations of high-sensitivity CRP were associated with higher QTBAZ, but also a higher resting heart rate13. One possible explanation is that heart rate may be confounding the relationship between QTc interval and CRP; once heart rate is adequately corrected with the stricter formula (QTFHS), this relationship disappears. This is further supported by our finding that CRP was correlated with heart rate. Interactions between the immune and ANS could also explain the associations we found between QTc interval and inflammation. Both the sympathetic and parasympathetic nervous system have been shown to contribute to QTc interval21,22. Left cardiac sympathetic denervation shortens QTc interval in patients with the congenital long QT syndrome22, while cholinergic stimulation with pyridostigmine shortens QTc interval in patients with coronary artery disease21. In patients with RA, increased sympathetic and reduced parasympathetic activities have been identified24,25,26,27 and could contribute to QTc interval prolongation. Evidence suggests a bidirectional relationship between the immune system and ANS. For example, IL-6 has been shown to increase sympathetic activity in rats33, while the ANS has been implicated in regulation of the immune response through cholinergic nerve fibers, known as the cholinergic inflammatory reflex31. Indeed, the principal vagal neurotransmitter acetylcholine inhibits the release of cytokines TNF-α, IL-1β, and IL-6, but not IL-1049, and direct electrical stimulation of the vagus nerve in rats during endotoxemia inhibited synthesis of TNF-α49.
The weak correlation between CRP and IL-6 found in our study and reported in prior studies50,51 is interesting given that IL-6 stimulates hepatocytes to produce CRP. This finding may be because of hysteresis (i.e., the delayed effect of IL-6 on CRP production may not be reflected in concurrent blood sampling), or perhaps the pleiotropic effects of IL-6 (i.e., antiinflammatory), or altered activity in RA. One of the major limitations of our present study is the cross-sectional nature of the associations between inflammatory cytokines and QTc interval. With a lack of followup data, these results cannot prove causality and therefore should be viewed as hypothesis-generating. Another limitation is the one-off blood sampling that does not allow for cytokine fluctuations over time to be measured. Finally, the variables included in our regression models explained about 21%–24% of the variation in QTc interval, suggesting the presence of other confounders not being explained. These may include autonomic dysfunction, electrolyte abnormalities, structural heart disease, and the presence of genetic polymorphisms causing congenital long QT syndrome. Studies are needed to further explore the relationship between inflammatory cytokines and QTc prolongation. It is likely that concurrent assessments of ANS function and QTc interval with multiple methods of heart rate correction can help provide greater understanding of the precise mechanisms by which inflammatory cytokines contribute to QTc prolongation in RA.
Our study, to our knowledge, is the first to report a contemporary association between higher serum proinflammatory and antiinflammatory cytokine concentrations and QTc prolongation in patients with RA. Undetectable concentrations of IL-10 were independently associated with shorter QTc interval while proinflammatory cytokines (TNF-α, IL-1β) were positively associated with prolonged QTc interval — although this association was attenuated when adjusting for age. Further studies are needed to confirm the effect of circulating inflammatory cytokine levels on QTc interval and to assess whether this effect is contemporary or cumulative.
Footnotes
-
Supported by the Dudley Group of Hospitals R and D Directorate cardiovascular program grant. The study was in part funded by a research grant from Arthritis Research UK (grant number: 196633).
- Accepted for publication November 10, 2014.
