Author + information
- aDivision of Cardiology, Department of Medicine, University of Iowa Hospitals and Clinics, Iowa City, IowaDivision of Cardiology, Department of Medicine, University of Iowa Hospitals and Clinics, Iowa City, Iowa
- bUniversity of Tennessee Health Science Center, Memphis, TennesseeUniversity of Tennessee Health Science Center, Memphis, Tennessee
- ↵∗Address for correspondence:
Dr. Brian Olshansky, Division of Cardiology, Department of Medicine, University of Iowa Hospitals, 200 Hawkins Drive, Iowa City, Iowa 52242.
Observations connecting electrolyte anomalies to the genesis of cardiac arrhythmias date to the 1880s, when A.G. Mayer noted changes in jellyfish contractions by varying solutions in electrolyte baths in which they were immersed. Development of the string galvanometer and subsequent advancements in electrocardiogram (ECG) recordings led to observations in humans that automaticity, cardiac conduction and repolarization depend on serum electrolyte concentrations. Today, electrolyte abnormalities, including hypokalemia, hyperkalemia, hypocalcemia, hypermagnesemia, and others, have well-known effects on ECG recordings and can cause serious rhythm disturbances, and death. However, no large, systematic study relates electrolyte concentrations to specific ECG intervals in the general population, until now.
In this issue of the Journal, Noordam et al. (1) report a robust meta-analysis evaluating associations between serum electrolyte concentrations and ECG intervals. In this consensus-building, multicentered, global assessment of 153,014 individuals from 38 study groups of 33 cohorts in 5 distinct ancestries, RR, QT, QRS, JT, and PR intervals were associated with variations in Ca+2, Mg+2 K+, and Na+ concentrations stratified by quintiles and adjusted for confounders including hypertension (drug user vs. nonuser), sex, and ancestry. Upon examining 20 distinct ECG-electrolyte associations, the strongest link was “clinically relevant” (>5 ms) QT prolongation at lowest Ca+2 concentrations (2% of the population). QT prolongation also occurred with higher Mg+2 concentrations. Apparent K+ concentration effects were driven primarily by antihypertensive drugs. Na+ concentration was associated with small changes in QRS duration.
For the first time, “big data” from large, diverse populations associated electrolyte concentrations to ECG intervals. Notwithstanding numerous intriguing relationships, on close inspection, these data raise more questions than provide answers.
Although some results are anticipatory (e.g., QT lengthening with lower Ca+2), how electrolyte concentrations affect ECG intervals remains uncertain. Associations do not prove causality and even if a causal relationship exists, what is it? Electrolyte concentrations can modulate multiple cardiac channels but each channel type may have discrepant, or even competing, physiological effects on depolarization and repolarization.
Consider that increasing Mg+2 concentrations were associated with increasing QT intervals. Should we worry? Mg+2 concentrations were not associated with changes in PR intervals as may be expected. Why is this? Intriguingly, changes in K+ (after data adjustment) were not associated with ECG abnormalities. Observed differences in ECG intervals between men and women were not explained by electrolyte concentrations alone but, then, what are the explanations? Do putative associations provide new insights? We are not so sure.
The extremely large size of this study is undoubtedly a strength but there are inescapable weaknesses. With big data like these, statistically significant changes can be detected with exceedingly small differences in observed ECG intervals or electrolyte concentrations but small differences in QT interval may be devoid of any clinical or physiological meaning. This crucial point pertains to overinterpreting results when exceedingly small differences in very large populations appear significant. Big data do not necessarily provide big answers to critical, challenging, pathophysiological conundrums.
At extremes, the relationship of ECG intervals to K+ concentrations is nonlinear. This may be true for other electrolytes as well, especially at extremes, but what about in the general population? A beta coefficient, suggesting degrees of association, may help understand what ranges are important, or critical, but differences may become manifest only at outlying values. Moreover, these data cannot help extrapolate relationships of varying electrolyte levels to ECG intervals for any given individual.
The authors attempt to adjust for confounding variables but do not ascertain synergistic, or counteractive, effects of one electrolyte with another. Any level at which relevant interactions occur remains uncertain. Higher Mg+2 and lower Ca+2 levels each prolong the QT but what is the combined effect? Is all QT prolongation problematic? Mg+2 may increase QT intervals but prevent torsades de pointes. Why? Big data alone are not big enough to answer these questions.
It remains unclear what to do with these data. Small changes have not been shown to have an adverse or beneficial effect. Any putative relationship of intervals to outcomes is extrapolated from other data and, in so doing, provide a dangerous precedent. What do these data tell us about populations at large? What interventions make a difference? Answers to these questions are, by no means, obvious.
No data presented here implicate relationships between lower Ca+2 concentrations, for example, and adverse outcomes. Perhaps, a specific genetic abnormality exists in which lower Ca+2 concentrations are associated with longer QT intervals and higher risk for ventricular arrhythmias. Perhaps not. We caution overinterpretation of data that implicate slight lengthening in QT intervals as “clinically relevant.”
Slight QT prolongation, by itself, may have no specific meaning or prognostic value. Considering the Food and Drug Administration 5-ms threshold for regulatory concern following a “thorough QT/QTc study” in healthy volunteers testing drugs that may lengthen the QT interval has no specific applicability here because no evidence implicates longer QT intervals related to elevated Mg+2 levels alone, for example, with outcomes or arrhythmias.
The ECG is, after all, a rather crude measure. Repolarization, manifest by longer QT/JT intervals alone may have little to do with 3-dimensional directionality of repolarization and potential complexities of arrhythmogenesis. Interval measurement has nothing to do directly with atrial repolarization, automaticity or conduction disturbances. Data relating electrolytes to ambient arrhythmias may be more valuable.
This study does not help discern which patient with a specific electrolyte concentration–QT interval relationship would be a good candidate for a drug that affects the QT interval. Generally, drugs that lengthen QT intervals and that can cause torsade de pointes affect the IKr (HERG) channel. However, while higher Mg+2 concentrations were associated with longer QT intervals, Mg+2 may prevent, rather than cause, torsade de pointes by blocking the HERG channel. While antihypertensive medications in this study may influence the QT interval, without knowing specifics about the drugs involved, clinically relevant inferences cannot be drawn about any risk these drugs may cause.
It will be important to gain deeper insight into mechanistic relationships between fluctuations within relatively normal electrolyte concentrations producing minor changes in ECG intervals as they relate to important measurable outcomes. Understanding such associations would further enhance our understanding of how to integrate this information into patient care and whether to consider specific drug restrictions in those with slightly prolonged QT intervals, for example.
Moving forward, there may be specific information more telling on the ECG that cannot be determined by interval measurement alone. As artificial intelligence is strategically applied to ECGs, it may be possible to detect the imprint of electrolyte concentrations and determine whether effects are good or bad. For example, Mg+2 may lengthen the QT interval but transform T waves with a specific morphologic configuration. The ECG may be rich with undetectable information to date. Artificial intelligence might help distinguish interrelationships between genomics, drugs, electrolytes, gender and outcomes.
Data from Noordam et al. (1) provide an exciting glimpse into relationships between electrolyte concentrations and ECG intervals in the general population. However, big data do not necessarily provide big insights leaving much to unravel about clinical outcomes, risk assessment, mechanisms, causal relationships, and interventions. We congratulate the authors for providing this novel interplay between electrolytes and ECG intervals suggesting proposed dynamic relationships, but caution against overinterpretation of this information for potential clinical applications.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Olshansky has served as a speaker for Lundbeck; has served as a consultant for Boehringer Ingelheim and Respironics; has served on the Data Safety Monitoring Board for Amarin; and has received honoraria from Lundbeck, Boehringer Ingelheim, Respironics, and Amarin. Dr. Bhattacharya has reported that he has no relationships relevant to the contents of this paper to disclose.
- 2019 American College of Cardiology Foundation
- Noordam R.,
- Young W.J.,
- Salman R.,
- et al.