Allopurinol and the risk of ventricular arrhythmias in the elderly: a study using US Medicare data

Recent studies have shown that hyperuricemia and gout, a condition with hyperuricemia associated with joint inflammation and/or renal manifestations, are associated with a higher risk of coronary artery disease (CAD), acute cardiovascular events including myocardial infarction (MI) and stroke, and cardiovascular mortality [1‐8]. Emerging data suggest that gout and hyperuricemia may also be associated with cardiac arrhythmias such as atrial fibrillation [9‐11].

Ventricular arrhythmias (VA) occur  commonly after an acute MI, but are also seen in patients with other cardiac conditions such as valvular or congenital heart disease, cardiomyopathy, hypertension, and other heart diseases [12]. The prevalence of VA (including ventricular tachycardia) in older men and women ranged from 15% to 16% in CAD, 8–9% in hypertension, valvular disease, or cardiomyopathy, and 2–3% in those without cardiac disease [13]; VA is also associated with new coronary events [13]. Therefore, treatment guidelines for VA (and ventricular fibrillation (VF)) emphasize several approaches to reduce associated morbidity and mortality, including immediate treatment using automated external defibrillators in hospitals and the community settings, the use of anti-arrhythmic drugs, and the use of implantable cardioverter-defibrillator, ablation, and revascularization surgery [14, 15].

A recent analysis of the Aspirin Myocardial Infarction Trial that examined mortality rates following daily aspirin administration over 3 years in individuals with documented MI, showed that gout treatments may have had beneficial effects [4]. Compared to MI patients without gout, only MI patients with untreated gout had higher all-cause mortality and CHD mortality, while the risk was not increased in patients with gout treated with gout medications (allopurinol, colchicine, or probenecid) [4]. An observational study reported that allopurinol use was associated with a reduction of the risk of incident atrial fibrillation in the elderly [16]. Together, these data raise an important question: Can allopurinol use reduce the risk of VA?

Animal study data suggest that allopurinol may prevent VA. In both rat and guinea pig models of ischemia-reperfusion injury, allopurinol treatment reduced the incidence of VA, VF, and possibly mortality [17‐19]. This efficacy correlated with significant attenuation of reperfusion-induced transmural conduction delay [18]. Proposed mechanisms for allopurinol’s anti-arrhythmic action include (1) an anti-oxidant action [20], via endothelial nitric oxide synthase reduction [21] that likely underlies the noted improvement of endothelial function [22‐25]; (2) an anti-ischemic action [26] and associated reduction in blood pressure [27, 28]; and (3) the reduction of left ventricular mass [29, 30] and pressure overload [31]. Given this basic science discovery, our recent finding of a beneficial effect of allopurinol use on the risk of atrial fibrillation [16], and findings of mortality reduction with gout treatment after MI in the Aspirin Myocardial Infarction Trial [4], we hypothesized that allopurinol use will be associated with a reduction in the risk of VA.

To our knowledge, there are no published human studies addressing this question. Therefore, we aimed to assess, whether (1) allopurinol use was associated with a lower risk/hazard of VA; and (2) allopurinol use duration was associated with a lower risk/hazard of VA. In an exploratory analysis, we assessed whether the risk reduction in VA with allopurinol use varied by previous MI, and other cardiac conditions that are risk factors for VA, and explored the association of allopurinol with the risk of VF.

In this study of a nationally representative sample of older Americans, we made several novel observations. We found that allopurinol use was independently associated with a lower hazard of incident VA. We also found that, compared to allopurinol non-use, allopurinol use durations greater than 6 months were associated with significant reduction in the hazard of incident VA. Associations of allopurinol use and use duration with hazard reduction of VA were confirmed in patients with gout, and for patient subgroups who were  not on anti-arrhythmic or cardio-protective drugs. The hazard reduction of VA with allopurinol differed by the history of previous MI and the presence of other diseases, which are known risk factors for VA. These novel study findings deserve further elaboration and discussion.

Recent experimental evidence using animal models of ischemia-reperfusion injury showed that treatment with allopurinol reduced the incidence of VA and possibly mortality [17‐19]. To our knowledge, this is the first study in humans to show that, compared to non-use, allopurinol use was associated with a 18% reduction (clinically relevant) in the hazard of incident VA and longer durations of allopurinol use of 181 days to 2 years and more than 2 years were associated with hazard reductions of 24% and 28%, respectively (Table 3). In the absence of prior studies in humans, no comparisons could be done. Animal studies support allopurinol’s anti-arrhythmic action in ischemia-reperfusion injury models [17‐19], including a placebo-controlled study of allopurinol [17]. The mechanism of hazard reduction of VA with allopurinol is unknown. We speculate that it may be related to the significant attenuation of reperfusion-induced transmural conduction delay, as noted with allopurinol use in a guinea pig ischemia-reperfusion injury model [18]. Other proposed mechanisms may be related to other noted beneficial effects of allopurinol, including associated anti-oxidant action [20, 21], an anti-ischemic action [26], blood pressure reduction [27, 28], improvement of endothelial function [22‐25], and reduction of left ventricular mass [29, 30]. Some of these processes may be sequential and many are likely on the causative pathway of VA.

VA often occur in patients with CAD and cardiac dysfunction, including heart failure [13]. The generation of reactive oxygen species in these disorders can contribute to induction of arrhythmias, via multiple mechanisms, including the alteration of cardiac ionic channels [38] and cardiac cell death associated ventricular dysfunction [39]. Oxidative-stress mediated tissue injury during ischemia and reperfusion may be associated with both ischemia and reperfusion-induced arrhythmias [17]. Xanthine oxidase has been implicated in cardiovascular disease and xanthine oxidase inhibition for its treatment [24, 40‐42].

Allopurinol inhibits xanthine oxidase activity, which in turn inhibits superoxide radical production and reduces oxidative stress [43]. Allopurinol’s anti-oxidant effect likely leads to an improvement in endothelial function and possibly has an anti-ischemic effect, which might prevent left ventricular hypertrophy. In particular, both the endothelial function improvement [22‐25], and the anti-ischemic effects [26] associated with the use of allopurinol may be the key mechanisms related to this anti-arrhythmic effect.

Our observation may have practical implications, although they need further confirmation before a widespread implementation. Allopurinol is a well-known effective, affordable treatment for gout, the most common inflammatory arthritis in adults, affecting 5% of adult Americans [44]. The emerging evidence of cardioprotective action of allopurinol provides a greater urgency for optimal use of allopurinol in all patients with gout (except rare instance). This observation of anti-arrhythmic action may lead to a preference of allopurinol over other urate-lowering agents in gout and related conditions. It is possible that there is an even greater advantage of allopurinol use in patients with gout and a concomitant well-known pro-arrhythmic condition. Our findings indicate that there may be a potential role for allopurinol as a xanthine-oxidase inhibitor beyond the joint, including a protective role in preventing VAs in patients with myocardial ischemia and damage.

The estimates for allopurinol use and use duration with VA were robust and were unaltered in multiple sensitivity analyses. This observation of an anti-arrhythmic effect is similar to the recent observations of potentially cardioprotective effects of allopurinol in the elderly [16, 45, 46]. Longer duration of allopurinol use was associated with more VA hazard reduction, representing a dose effect with a magnitude similar to that noted with the associated reduction of myocardial infarction, stroke, or atrial fibrillation [16, 45, 46].

An interesting observation was that the beneficial effect of allopurinol use was similar for subgroup analyses for most diseases that are risk factors for VA, including valvular heart disease, congenital heart disease, heart failure, renal failure, dialysis, cardiopmyopathy, and hyperkalemia. Minor differences in statistical significance seemed related to a smaller sample size for patients with each VA risk factor. One interesting observation was that, although beneficial for both, allopurinol’s beneficial effect was greater in magnitude for patients without CAD versus with CAD, 38% versus 13% hazard reduction (model 1). Allopurinol use was associated with statistically significant hazard reduction for VA (18%) in patients without previous MI, but reduction was not statistically significant in those with previous MI (Model 5). This is not surprising since ischemia-reperfusion injury is one of the best-described pathophysiologic associations between MI and VA [17‐19, 47‐50]. Irreversible structural damage may already have occurred in some patients with CAD/MI, making allopurinol not as effective in patients with CAD/MI. This finding may have important implications, if confirmed in other studies. Additionally, a smaller sample size for patients with MI may have made this analysis underpowered.

We noted an association of older age and Black race with a higher hazard of VA, confirming a similar previous finding [51‐53]. We also noted that older age was no longer significantly associated with the risk of VA once the model was adjusted for VA risk factors, such as CAD, congenital heart disease, renal failure, and others. This indicated that age-VA associations were not due to chronological age but rather reflected the higher risk imparted due to specific diseases that are VA risk factors, more common in the elderly.

Study findings must be interpreted considering their strengths and limitations. An observational study design makes our findings susceptible to confounding bias. We tried to reduce confounding bias by including several important patient characteristics, common cardiac medications and adjusting for risk factor conditions for VA. We used ICD-9 diagnostic codes from Medicare claims for the assessment of VA, which makes results liable to misclassification bias; this likely biased our results towards null. However, similar approaches have been shown to have high positive predictive values in a systematic review of claims-based definitions of ventricular arrhythmias [36]. Sensitivity analysis using the VA code algorithm with the highest positive predictive value (92–100%) reproduced the same result as our main analysis [35]. We realized a priori that there would be insufficient number of VF events, making this analysis underpowered and therefore exploratory in nature, as shown. Limited sources and expected few data for febuxostat (regulatory approval in 2009; another xanthine oxidase inhibitor) prevented us from performing additional comparative effectiveness studies comparing febuxostat to allopurinol.

The main strength of this study lies in the potential for generalization to the elderly US population and to all allopurinol new-users, regardless of the underlying diagnosis. Our study had a large sample size and adequate number of outcome events. Another strength is the incident (or new) user design, which reduces bias by avoiding adjustment for characteristics that may be in the causal pathway and allows capture of both early and late events [54], which is important given the study objectives. We controlled for several potential confounders (risk factors for VA; cardiac medications) to reduce bias and conducted multiple sensitivity analyses, which confirmed the robustness of our findings.

In conclusion, we found a significant association between incident allopurinol use and a lower hazard of VA in the elderly. This result was more pronounced for longer allopurinol use durations. Results for patients with an underlying diagnosis of gout were similar to the entire sample of allopurinol users. We also found some differences in VA risk reduction with allopurinol in patients with and without CAD and previous MI. Mechanisms of reduction of ventricular arrhythmias with allopurinol need to be examined in future studies. Future studies should also examine the underlying mechanisms for why the VA hazard reduction with allopurinol varies by CAD and previous MI.