Frequency, Characteristics, and Risk Factors for Amiodarone Neurotoxicity

Objective  To assess the incidence, severity, and spectrum of neurologic toxic effects associated with administration of amiodarone hydrochloride.

Design  Retrospective medical record analysis of cardiac patients treated with amiodarone between January 1, 1996, and July 31, 2008.

Setting  Residents of Olmsted County, Minnesota, treated at the Mayo Clinic.

Patients  The Mayo Clinic medical records of all adult Olmsted County residents prescribed amiodarone between January 1, 1996, and July 31, 2008, were reviewed and all possible neurologic adverse effects that might be attributable to amiodarone were tabulated.

Main Outcome Measures  Risk factors and clinical characteristics of patients developing neurologic problems were compared with those who did not.

Results  Over the 151 months of analysis, 707 patients were treated with amiodarone. Among these patients, the cumulative incidence of likely amiodarone neurotoxic effects was 2.8%; 1.6% of all amiodarone-treated patients were referred to Neurology for a neurotoxic reaction. Neurologic problems included tremor, gait ataxia, peripheral neuropathy, and cognitive impairment. The primary risk factor for amiodarone neurotoxic effects was duration of treatment, not age, drug dose, sex, or indication for therapy. Where this could be assessed, the adverse effects were usually but not always reversible.

Conclusions  Amiodarone infrequently causes clinically significant neurologic toxic effects. Substantially higher estimates of neurotoxic effects in the early amiodarone era may be related to a much higher daily dose.

Amiodarone hydrochloride is a class III antiarrhythmic drug prescribed commonly for atrial fibrillation and ventricular arrhythmias. It has several clinically important toxic effects, including pneumonitis, hepatotoxicity, thyroid dysfunction, and corneal microdeposits with visual disturbances.^1-3 The initial experience with this drug also suggested substantial potential for neurologic toxic effects, with frequencies in these early reports ranging from 27.5% to as high as 74%^1,2,4-6; the most common of these neurologic adverse effects have been tremor, ataxia, and peripheral neuropathy. However, various other neurologic adverse effects have also been reported, including cognitive impairment/encephalopathy,^4,6,7 parkinsonism,^8,9 optic neuropathy,¹⁰ and myopathy.^6,11 On the other hand, the frequency of neurologic adverse effects in longer-term amiodarone clinical trials has been much lower, at less than 5%.^12-15

Tremor, ataxia, neuropathy, parkinsonism, encephalopathy, and myopathy are frequent problems in the general population of neurologic clinics. Currently, amiodarone is also commonly prescribed and, by chance, may be associated with these neurologic problems. Amiodarone is unlike certain other drugs that can be transiently discontinued on a trial basis. It may be crucial to control a life-threatening cardiac dysrhythmia and, moreover, has a long elimination half-life that may require several months for drug washout.¹⁶ This presents a dilemma to neurologists who encounter amiodarone-treated patients with one of the earlier-mentioned problems. Thus, the likelihood that amiodarone is culpable is crucial to any decision to discontinue amiodarone treatment.

Given the very disparate frequencies of amiodarone neurotoxic effects in the literature, we analyzed this among patients treated at the Mayo Clinic, restricting the analysis to patients living in the community (Olmsted County, Minnesota) to minimize referral bias. Although these community patients may receive medical care from non-Mayo clinicians, the vast majority of cardiac and neurologic care for Olmsted County residents is provided by the Mayo physicians. Thus, we retrospectively investigated the frequency, spectrum, severity, and reversibility of amiodarone neurotoxic effects among Olmsted County residents cared for at the Mayo Clinic.

After receiving Mayo institutional review board approval, we identified all Olmsted County residents treated with amiodarone at the Mayo Clinic (Rochester, Minnesota) between January 1, 1996, and July 31, 2008 (151 months), using the electronic medical records linkage system. Patients were identified via computer search of the electronic medical records to identify the text words “amiodarone,” “Pacerone,” or “Cordarone.” Patients were limited to those with Olmsted County residence (confirmed by zip code) who were older than 18 years.

We limited the primary analysis to the amiodarone-treated patients who were seen in the neurologic clinic (for any reason) after long-term amiodarone therapy was initiated. Mayo in-house neurologic consultations can be obtained with minimal delays and would likely be requested by cardiologists and nonneurologists addressing important potential amiodarone adverse effects. We believed that the neurologic record would permit more reliable judgments about both the character of the problem and the relationship to amiodarone. We also analyzed a subset of all the amiodarone-treated patients to estimate the frequency of amiodarone neurotoxic effects that were not associated with neurologic referral (see later).

Records were reviewed for neurologic problems developing after initiation of long-term amiodarone therapy. The neurologic adverse events were characterized as definite, probable, possible, or doubtful amiodarone toxic effects using the Naranjo algorithm.¹⁷

We also compared the clinical characteristics of the patients with amiodarone neurotoxic effects with all patients prescribed amiodarone over the period of the study who did not develop neurologic problems for sex, treatment indication, age, treatment duration, and dose. We divided the number of cases of amiodarone neurotoxic effects by the total number of individuals taking amiodarone during the same period to estimate the overall cumulative incidence of amiodarone toxic effects in the population.

We also estimated the number of amiodarone neurotoxic effect cases that might have been missed because the patients had not been referred to Neurology. For this, we analyzed the records of a subset of all Olmsted County amiodarone-treated patients, confined to the 2-year period of 2006 and 2007.

We identified 29 amiodarone-treated Olmsted County patients who had been evaluated by Mayo Neurology between January 1, 1996, and July 31, 2008. On review of their records, 7 patients were excluded because the neurologic problem developed prior to the initiation of amiodarone therapy and 2 were excluded because the problem was not neurologic. Of the 20 remaining patients, 7 had problems considered to be biologically implausible as drug effects (3 strokes, 1 meningioma, 1 subdural hematoma, 1 carpal tunnel syndrome, and 1 transient ischemic attack secondary to a documented carotid dissection). Another 2 problems were considered as possible adverse effects but with other, likelier causes (asymmetric neuropathy in the setting of systemic vasculitis in 1 patient and peripheral neuropathy in the setting of carboplatin and paclitaxel therapy in the other patient). The remaining 11 problems were judged to be biologically plausible and likely drug effects: new-onset tremor (4 patients); worsening of preexisting essential tremor (2); peripheral neuropathy (2); tremor with gait ataxia (1); gait ataxia plus mild peripheral neuropathy (1); and cognitive impairment (1). In 2 cases, these clinical problems were noticed incidentally by neurologists when assessing the patients for other problems. Of these 11 patients (8 male), 9 were prescribed amiodarone for atrial arrhythmias and 2 for ventricular arrhythmias. The mean age of the affected patients at the time they were seen in Neurology was 74.1 years (range, 66-84 years) and the mean age when starting amiodarone therapy was 71.5 years (range, 64-83 years), the mean daily dose of amiodarone hydrochloride was 209 mg (range, 100-400 mg), and the average length of time receiving treatment was 31.6 months (range, 2 weeks to 84 months). In 7 cases, administration of the drug was stopped or reduced, and of these, all but 1 patient improved. Using the Naranjo Adverse Drug Reaction probability algorithm, which estimates the likely causality of an adverse drug reaction, of the 11 biologically plausible drug effects, 7 were considered probable and 4 were considered possible. None were considered either definite or doubtful. A summary of the clinical characteristics of the 11 patients is provided in Table 1.

We compared these patients with all 707 Olmsted County adults who received amiodarone therapy between January 1, 1996, and July 31, 2008, documented by our record review. As shown in Table 2, the mean age, sex, type of arrhythmia, and mean daily dose were similar, comparing the 696 patients who did not develop neurotoxic effects with those 11 patients who did. However, those with neurotoxic effects took amiodarone for significantly longer (mean, 31.6 months vs 17.2 months).

We also analyzed a subset of the 707 amiodarone-treated patients to determine how many cases of amiodarone neurotoxic effects might have been missed by restricting the primary analysis to patients referred to Neurology. This subset consisted of amiodarone-treated patients seen within a 2-year time frame, January 1, 2006, to December 31, 2007. This included 230 amiodarone-treated adults, of whom 10 had been referred to Neurology and are included in the primary analysis. Of the remaining 220 patients, there was no mention of any form of neurologic problems in 179. Of the remaining 41 patients, cerebrovascular pathology due either to the underlying arrhythmia or anticoagulation treatment formed the bulk of the neurologic impairments. Thirteen had strokes either before or after amiodarone initiation, 4 had transient ischemic attacks, 1 had an anoxic brain injury secondary to a cardiopulmonary arrest secondary to ventricular tachycardia, 2 had subdural hematomas, and 1 had a traumatic subarachnoid hemorrhage following a motor vehicle accident. Several patients had preexisting neurologic diseases: 3 cases of Parkinson disease; 1, epilepsy; 1, chronic headache; 1, essential tremor; and 1, cognitive impairment, all prior to starting amiodarone therapy. Several other patients had problems that were unlikely to be related to amiodarone treatment: 2 had peripheral neuropathy attributed to chronic diabetes mellitus; 1 had polyneuropathy attributed to pathologically proven amyloidosis; and 1 had an ulnar neuropathy. The clinical details of the remaining 9 patients are shown in Table 3. The average age of these patients at the time of the neurologic complaint was 71.2 years (range, 48-80 years), the average dose of amiodarone hydrochloride was 244.4 mg (range, 100-400 mg), and the average length of time receiving treatment was 28.8 months (range, 1-105 months). Of these 9 patients, 3 were given diagnoses other than amiodarone neurotoxic effects and their problems resolved with appropriate treatment (patients 1, 4, and 5); 2 others had plausible explanations other than amiodarone toxic effects but did not improve or worsened (patients 3 and 7). One appeared to be a secondary adverse effect due to liver failure caused by amiodarone use (patient 8). Only 3 patients (patients 2, 6, and 9) appeared to have had genuinely possible amiodarone adverse effects, and in 2 of these patients, the neurologic problems were nonprogressive despite continuation of amiodarone therapy (patients 2 and 6).

Thus, the proportion of patients prescribed amiodarone over this 2-year period who developed possible amiodarone neurotoxic effects but were not referred to Mayo Neurology was 3/220 = 1.4%. By extrapolation, 1.4% of the entire cohort of 678 patients yields an approximate estimate of 9 patients with neurotoxic effects not seen by Mayo Neurology. Combining the 11 cases of amiodarone neurotoxic effects confirmed with neurologic evaluations plus these 9 other cases provides an overall upper-limit estimate of 20/707 = 2.8%.

The early published experience with amiodarone suggested that neurotoxic effects were frequent, with several series reporting substantial neurologic problems in more than a third of amiodarone-treated patients.^1,2,4,6 Our experience suggests that with current prescribing habits, the frequency of such problems is but a tiny fraction of these early reported values. Thus, only 1.6% of amiodarone-treated patients developed neurotoxic effects sufficient to lead to neurologic consultation. Allowing for patients who may not have been referred, the total frequency among amiodarone-treated patients over the 151-month time frame was only 2.8%.

This is crucial information for neurologists in the clinic, given the common use of amiodarone in current practice and the broad spectrum of neurologic problems that can be provoked by amiodarone (tremor, gait ataxia, cognitive impairment, peripheral neuropathy, and, rarely, parkinsonism or myopathy). Thus, amiodarone enters into the differential diagnosis of these conditions, but usually not at the top of the list.

When amiodarone cannot be excluded as the cause for a neurologic disorder and culpability seems likely, treatment can be problematic. Amiodarone is frequently used as rescue therapy when other measures for life-threatening cardiac dysrhythmias have failed,¹⁸ and there may be few or no alternatives. Amiodarone is very lipophilic and has an extremely long elimination half-life, on the order of 6 months¹⁶; thus, confirmation of a reversible adverse effect requires months of observation of the patient not taking the drug. It is assumed that these neurologic adverse effects are indeed reversible, but because drug discontinuation and prolonged observation is often impossible, there may be exceptions.¹¹

The spectrum of neurologic adverse effects we identified was relatively limited and included tremor, peripheral neuropathy, gait ataxia, and cognitive impairment; these are consistent with previous studies.^1,2,4-6 Also consistent with previous studies, we observed tremor to be the most common manifestation of neurotoxic effects. Tremor, of course, can be secondary to amiodarone-induced hyperthyroidism and this should be excluded in clinical practice. Other movement disorders, including parkinsonism, myoclonus, and various dyskinesias, as well as optic neuropathy and myopathy, have also been described^6-11 but were not observed in our study.

The marked disparity in the frequency of neurotoxic effects in the current era compared with the decade when amiodarone was first introduced likely relates to dosing differences. Maintenance doses of 600 mg daily were often used in the 1980s,^1,2 whereas currently, the recommended maintenance dose after the initial loading phase is 200 mg daily,¹⁶ consistent with the maintenance doses used in this series (Table 2). Table 4 illustrates the striking relationship between the long-term amiodarone maintenance dose and the frequency of neurotoxic effects among published series (Table 4). Thus, although the daily amiodarone dose did not surface in our study as a risk factor, this may be because very high doses were infrequently maintained. Other investigators have concluded that amiodarone toxic effects are dose related.^4,18-20

The main risk factor for amiodarone neurotoxic effects in our study was length of time receiving therapy. This is in agreement with a meta-analysis of amiodarone toxic effects in 1997 that concluded that exposure to long-term (at least 12 months) amiodarone therapy doubled the odds of neurologic adverse effects compared with placebo (despite low doses [150-330 mg/d]).¹⁵ Age was not a risk factor in our study, in agreement with a previous report,⁴ and neither was sex or indication for amiodarone therapy.

With the amiodarone dosing strategy of the current era, the risk of neurologic toxic effects is small; however, it may be suspected among patients with otherwise unexplained tremor, gait ataxia, peripheral neuropathy, or cognitive impairment.

Correspondence: J. Eric Ahlskog, MD, PhD, Department of Neurology, Mayo College of Medicine, 200 First St SW, Rochester, MN 55905 (eahlskog@mayo.edu).

Accepted for Publication: February 13, 2009.

Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data analysis. Study concept and design: Orr and Ahlskog. Acquisition of data: Orr. Analysis and interpretation of data: Orr and Ahlskog. Drafting of the manuscript: Orr and Ahlskog. Critical revision of the manuscript for important intellectual content: Orr and Ahlskog.

Financial Disclosure: None reported.

Additional Contributions: Barbara Abbott helped with data retrieval, Katie Millard assisted with the Tables, and Jennifer St. Sauver, PhD, provided advice on epidemiological analysis.