In postmarketing surveillance, the US Food and Drug Administration has reported the development of lung masses, thyroid cancer, and skin cancer after amiodarone therapy.
In postmarketing surveillance, the US Food and Drug Administration has reported the development of lung masses, thyroid cancer, and skin cancer after amiodarone therapy.
Using the Taiwan National Health Insurance Research database, the authors conducted a population-based cohort study. Patients who were treated with amiodarone between 1997 and 2008 were enrolled. Those with antecedent cancer were excluded. Standardized incidence ratios (SIRs) of cancers were calculated to compare the cancer incidence of the study cohort with that of the general population. A multivariate Cox regression model was used to evaluate the association between cumulative defined daily doses (cDDDs) of amiodarone and cancer occurrence.
The study included 6418 subjects, with a median follow-up of 2.57 years. A total of 280 patients developed cancer. The risk of cancer increased with borderline significance (SIR, 1.12; 95% confidence interval [95% CI], 0.99-1.26 [P = .067]). Male patients had a higher risk (SIR, 1.18; 95% CI, 1.02-1.36 [P = .022]). The total cohort of patients and the male patients with > 180 cDDDs within the first year were found to have SIRs of 1.28 (95% CI, 1.00-1.61; P = .046) and 1.46 (95% CI, 1.11-1.89; P = .008), respectively. After adjustment for age, sex, and comorbidities, the hazards ratio was 1.98 (95% CI, 1.22-3.22; P = .006) for the high tertile of cDDDs compared with the low tertile.
The results of the current study indicate that amiodarone may be associated with an increased risk of incident cancer, especially in males, with a dose-dependent effect. Cancer 2013;119:1699–1705. © 2013 American Cancer Society.
Amiodarone, which was approved by the US Food and Drug Administration for the treatment of arrhythmias in 1985, is an antiarrhythmic drug used for the treatment of tachyarrhythmias. Because amiodarone is fat-soluble and its elimination half-life is very long (53 days),1 large amounts of the drug can accumulate in the soft tissues after long-term treatment. In addition, the structure of amiodarone is very similar to that of thyroxine, and therefore thyroid function abnormalities are common during its administration. Both hypothyroidism and hyperthyroidism can occur after treatment with amiodarone.2 To the best of our knowledge, the most serious reaction caused by amiodarone therapy is pulmonary toxicity, which can result in pulmonary fibrosis. Long-term treatment with amiodarone is also associated with blue-gray hyperpigmentation and photosensitivity of the skin.3 In animal studies, amiodarone has been reported to increase the risk of thyroid tumors in rats.4 In postmarketing surveillance, the Food and Drug Administration reported the development of lung masses, thyroid cancer, and skin cancer after treatment with amiodarone.5 Some case reports indicated that amiodarone might increase the risk of pulmonary masses,6-12 thyroid cancer,13-15 and skin cancer.16-19 In addition, a meta-analysis of randomized trials unexpectedly revealed a borderline significant increase in cancer mortality among patients assigned to the amiodarone group. However, to the best of our knowledge, no large-scale study to date has addressed this issue. Therefore, the objective of the current study was to determine whether the use of amiodarone is associated with an increased risk of cancer.
A retrospective cohort study was conducted from January 1, 1996 to December 31, 2008 using the National Health Insurance Research Database (NHIRD) in Taiwan. Patients who received amiodarone for ≥ 28 days between January 1, 1997 and December 31, 2008 were enrolled. Patients aged < 20 years or those with antecedent malignancies were excluded. Information regarding comorbidities, including diabetes mellitus, cirrhosis, chronic obstructive pulmonary disease, chronic kidney disease, heart failure, hypertension, cardiovascular disease, dysrhythmia, hyperthyroidism, and hypothyroidism, was collected for analysis. Data regarding monthly income levels were collected as a surrogate of socioeconomic status. We also gathered information concerning urbanization levels of the residential area.
The National Health Insurance (NHI) program is a mandatory general health insurance program offering complete medical care coverage to all Taiwanese residents, with a coverage rate of up to 98% since 1995.20 The NHI program covers inpatient, outpatient, and emergency care, as well as traditional Chinese medicine, dental services, and prescription medication. The NHIRD is managed, filed, and publicly released by the National Health Research Institutes (NHRI) in Taiwan. Personal information that could identify an individual patient is encrypted. The Longitudinal Health Insurance Database, a subset of the NHIRD, is a representative database of 1,000,000 patients randomly sampled from the registry of all NHI enrollees by the NHRI. The NHI Catastrophic Illness Registry is a subset of the NHIRD that can provide comprehensive information regarding all patients with severe diseases, including catastrophic illnesses such as cancer, who have been exempted from copayment under the NHI program. The confidentiality of the data abides by the data regulations of the NHI Bureau and the NHRI. Because the NHI data set consists of deidentified secondary data for research purposes, the current study was exempt from full review by the Institutional Review Board.
The endpoint of the current study was cancer occurrence. The Catastrophic Illness Registry was used to identify patients who were diagnosed with cancer. Pathohistological confirmation is required for a diagnosis of cancer to be reported in the Catastrophic Illness Registry. The patients who were treated with amiodarone were followed until the occurrence of cancer, dropout from the NHI program, death, or the end of 2008.
The risk of cancer among the amiodarone-treated cohort was determined with the standardized incidence ratio (SIR), which is defined as the observed number of cancer cases divided by the expected number. The expected number of cancer cases was calculated by multiplying the national incidence rate of cancers according to age (in 5-year intervals), sex, and calendar year by the corresponding stratum-specific person-time accrued in the cohort. The incidence rates of cancers among the general population were acquired from the Taiwan Cancer Registry. The cancer registry in Taiwan was initiated in 1979 and therefore the entire follow-up of the study cohort was under the coverage of the registry. The 95% confidence intervals (95% CIs) for the SIRs were estimated under the assumption that the observed number of cancers followed a Poisson probability distribution. We determined the SIRs for the subgroups according to sex and age group. Because the cancer incidence most likely would be inflated in the first year due to the surveillance bias, a subgroup analysis stratified by the duration of enrollment (0-1 years, 1-5 years, and ≥ 5 years) was performed. To determine whether amiodarone was associated with specific types of malignancies, we also calculated the SIRs for each cancer site. Because there is likely a latent period of at least 2 years between exposure and the development of clinically significant cancer,21 we calculated the cumulative defined daily doses (cDDDs) of amiodarone within the first year and estimated the SIRs 3 years after enrollment for patients with high (> 180) and low (≤ 180) cDDDs to determine whether a dose-effect exists.
We used univariate and multivariate backward conditional Cox proportional hazards models to analyze the correlation between the cDDDs and cancer, and to identify predictors of cancer development among patients receiving amiodarone. Risk factors with a P value < .1 were entered into the multivariate analysis. Only the cDDD within the first year was calculated and the person-time within the first year was excluded to avoid immortal time bias.22 In other words, we redefined time 0 as being 1 year after the initial treatment with amiodarone. Accordingly, we excluded patients with a follow-up of < 1 year from the survival analysis. The cDDD was treated as a time-dependent covariate, with a lag time of 2 years (ie, the exposure had no effect in the first 2 years). We also tried using different models that incorporated > 180 cDDDs versus ≤ 180; > 90 cDDDs versus ≤ 90; or cDDDs of low, intermediate, and high levels (with cutoff points at the 33.3rd and 66.7th percentiles). Kaplan-Meier analysis and the log-rank test were performed to evaluate the time trend and test the difference in cumulative incidence among patients with low, intermediate, and high cDDDs. The person-time and events within the first year were excluded to avoid immortal time bias. Interactions between variables were tested by entering multiplicative interaction terms into the model.
Extraction and computation of data were performed using the Perl programming language (version 5.12.2; Perl Foundation, Walnut, Calif). Microsoft SQL Server 2005 (Microsoft Corporation, Redmond, Wash) was used for data linkage, processing, and sampling. All statistical analyses were performed using IBM SPSS statistical software (version 19.0 for Windows; IBM Corporation, New York, NY). A P value < .05 was considered to be statistically significant.
We identified 6851 patients who were treated with amiodarone. Of these, 433 patients had antecedent malignancies and therefore were excluded; the final study cohort consisted of 6418 patients, 42.8% of whom were female. In total, the cohort was observed for 21,684.4 person-years from 1997 to 2008. The median follow-up was 2.57 years (interquartile range [IQR], 0.78 years-5.25 years). The median age of the patients at the time of diagnosis was 70 years (IQR, 60 years-78 years). The demographic data and comorbidities of the cohort are shown in Table 1.
|No. of patients||6418||3674||2744|
|Person-y at risk||21,684.4||12,488.5||9,195.9|
|Median follow-up (IQR), y||2.57 (0.78-5.25)||2.60 (0.80-5.32)||2.53 (0.77-5.17)|
|Age at diagnosis, y|
|No. of comorbidities (%)|
|Diabetes mellitus||2506 (39.0)||1339 (36.4)||1167 (42.5)|
|Cirrhosis||205 (3.2)||130 (3.5)||75 (2.7)|
|COPD||2822 (44.0)||1702 (46.3)||1120 (40.8)|
|Chronic kidney disease||1627 (25.4)||934 (25.4)||693 (25.3)|
|Heart failure||3041 (47.4)||1664 (45.3)||1377 (50.2)|
|Hypertension||4906 (76.4)||2756 (75.0)||2150 (78.4)|
|Cardiovascular disease||4613 (71.9)||2649 (72.1)||1964 (71.6)|
|Hyperthyroidism||285 (4.4)||113 (3.1)||172 (6.3)|
|Hypothyroidism||102 (1.6)||37 (1.0)||65 (2.4)|
During the observation interval, 280 cancers developed. Compared with the general population, patients who received amiodarone had a borderline significant increase in their overall risk of cancer (SIR, 1.12; 95% CI, 0.99-1.26 [P = .067]).
The risk of all cancers was found to be significantly increased in men (SIR, 1.18; 95% CI, 1.02-1.36 [P = .022]) but not in women (SIR, 0.99; 95% CI, 0.79-1.23). In the subgroup analysis performed according to age (20 years-59 years, 60 years-79 years, and ≥ 80 years), the SIRs were not significantly different from unity, except for those of male patients aged 20 years to 59 years (SIR, 1.67; 95% CI, 1.07-2.48 [P = .025]) and ≥ 80 years (SIR, 1.41; 95% CI, 1.07-1.83 [P = .016]). The incidence of cancer was increased within 1 year of amiodarone therapy (SIR, 1.32; 95% CI, 1.05-1.64 [P = .002]), but not after 1 year (SIR, 1.02; 95% CI, 0.89-1.18). At 3 years after enrollment, the SIRs for patients with > 180 cDDDs increased significantly among the total number of patients and among the male patients, with SIRs of 1.28 (95% CI, 1.00-1.61; P = .046) and 1.46 (95% CI, 1.11-1.89; P = .008), respectively. In contrast, patients with ≤ 180 cDDDs did not appear to have a higher risk of developing cancer compared with the general population. The results of the subgroup analysis are summarized in Table 2.
|Characteristics||Observed||Expected||SIR (95% CI)||Observed||Expected||SIR (95% CI)||Observed||Expected||SIR (95% CI)|
|All cancers||280||250.21||1.12 (0.99-1.26)||198||167.29||1.18 (1.02-1.36)||82||82.91||0.99 (0.79-1.23)|
|20-59||32||22.07||1.45 (0.99-2.05)||24||14.37||1.67 (1.07-2.48)||8||7.70||1.04 (0.45-2.05)|
|60-79||168||164.01||1.02 (0.88-1.19)||118||113.27||1.04 (0.86-1.25)||50||50.74||0.99 (0.73-1.30)|
|≥80||80||64.12||1.25 (0.99-1.55)||56||39.64||1.41 (1.07-1.83)||24||24.48||0.98 (0.63-1.46)|
|Duration of treatment, y|
|0-1||84||58.70||1.43 (1.14-1.77)||55||38.70||1.42 (1.07-1.85)||29||20.00||1.45 (0.97-2.08)|
|≥1||196||191.51||1.02 (0.89-1.18)||143||128.59||1.11 (0.94-1.31)||53||62.91||0.84 (0.63-1.10)|
|1-5||140||136.63||1.02 (0.86-1.21)||100||91.33||1.09 (0.89-1.33)||40||45.30||0.88 (0.63-1.20)|
|≥5||56||54.88||1.02 (0.77-1.33)||43||37.26||1.15 (0.84-1.55)||13||17.62||0.74 (0.39-1.26)|
|cDDDs within first ya|
|≤180||46||51.49||0.89 (0.65-1.19)||37||34.40||1.08 (0.76-1.48)||9||17.09||0.53 (0.24-1.00)|
|>180||73||56.98||1.28 (1.00-1.61)||57||38.99||1.46 (1.11-1.89)||16||17.99||0.89 (0.51-1.44)|
There were no significant differences noted with regard to incidence for any of the specific cancer types, including lung and mediastinum (44 observed cases vs 39.41 expected cases; SIR, 1.12 [95%, CI, 0.81-1.50]), thyroid cancer (1 observed case vs 1.94 expected cases; SIR, 0.51 [95% CI, 0.01-2.87]), or skin cancer (6 observed cases vs 5.99 expected cases; SIR, 1.00 [95% CI, 0.37-2.18]). The SIRs of specific cancer sites are summarized in Table 3. The cancer incidences of each site were not found to be significantly different from those of the general population after the exclusion of person-time and events within the first 1 or 3 years (data not shown).
|Site of Cancer||Observed||Expected||SIR (95% CI)||Observed||Expected||SIR (95% CI)||Observed||Expected||SIR (95% CI)|
|All cancers||280||250.21||1.12 (0.99-1.26)||198||167.29||1.18 (1.02-1.36)||82||82.91||0.99 (0.79-1.23)|
|Head and neck||22||15.59||1.41 (0.88-2.14)||18||13.59||1.32 (0.78-2.09)||4||2.00||2.00 (0.55-5.13)|
|Digestive||124||108.66||1.14 (0.95-1.36)||90||73.75||1.22 (0.98-1.50)||34||34.91||0.97 (0.67-1.36)|
|Esophagus||3||5.00||0.60 (0.12-1.75)||2||4.49||0.45 (0.05-1.61)||1||0.51||1.97 (0.05-10.98)|
|Stomach||17||17.17||0.99 (0.58-1.59)||16||12.34||1.30 (0.74-2.10)||1||4.83||0.21 (0.01-1.15)|
|Colon and rectum||47||41.94||1.12 (0.82-1.49)||33||26.43||1.25 (0.86-1.75)||14||15.51||0.90 (0.49-1.51)|
|Liver and biliary tract||51||39.20||1.30 (0.97-1.71)||34||27.22||1.25 (0.87-1.75)||17||11.98||1.42 (0.83-2.27)|
|Pancreas||6||5.35||1.12 (0.41-2.44)||5||3.26||1.53 (0.50-3.58)||1||2.09||0.48 (0.01-2.66)|
|Lung and mediastinum||44||39.41||1.12 (0.81-1.50)||37||29.05||1.27 (0.90-1.76)||7||10.36||0.68 (0.27-1.39)|
|Bone and soft tissue||1||1.71||0.58 (0.01-3.25)||1||1.16||0.86 (0.02-4.78)||0||0.55||0.00 (0.00-6.70)|
|Skin||6||5.99||1.00 (0.37-2.18)||6||3.34||1.80 (0.66-3.91)||0||2.65||0.00 (0.00-1.39)|
|Breast||9||9.93||0.91 (0.41-1.72)||0||0.20||0.00 (0.00-18.06)||9||9.73||0.93 (0.42-1.76)|
|Genitourinary||47||46.26||1.02 (0.75-1.35)||30||32.35||0.93 (0.63-1.32)||17||13.91||1.22 (0.71-1.96)|
|Cervix||4||5.39||0.74 (0.20-1.90)||—||—||4||5.39||0.74 (0.20-1.90)|
|Uterus||4||1.38||2.90 (0.79-7.43)||—||—||4||1.38||2.90 (0.79-7.43)|
|Ovary||3||1.46||2.05 (0.42-6.00)||—||—||3||1.46||2.05 (0.42-6.00)|
|Prostate||22||19.97||1.10 (0.69-1.67)||22||19.97||1.10 (0.69-1.67)||—||—|
|Bladder||6||10.99||0.55 (0.20-1.19)||4||8.35||0.48 (0.13-1.23)||2||2.64||0.76 (0.09-2.74)|
|Kidney||8||7.07||1.13 (0.49-2.23)||4||4.03||0.99 (0.27-2.54)||4||3.04||1.32 (0.36-3.37)|
|Thyroid||1||1.94||0.51 (0.01-2.87)||0||0.65||0.00 (0.00-5.63)||1||1.29||0.78 (0.02-4.32)|
|Hematologic||14||10.78||1.30 (0.71-2.18)||11||7.16||1.54 (0.77-2.75)||3||3.62||0.83 (0.17-2.42)|
|All others||12||9.92||1.21 (0.62-2.11)||5||6.02||0.83 (0.27-1.94)||7||3.90||1.80 (0.72-3.70)|
On multivariate analysis, age (hazards ratio [HR], 1.04 for being 1 year older; 95% CI, 1.03-1.06 [P < .001]), male sex (HR, 1.90; 95% CI, 1.38-2.62 [P < .001]), cirrhosis (HR, 3.70; 95% CI, 2.10-6.52 [P < .001]), socioeconomic status, and cDDD (HR, 1.001 for 1 additional daily dose; 95% CI, 1.000-1.002 [P = .022]) were found to be significant factors. These results are summarized in Table 4. The interactions between variables were not found to be significant. If the > 180 cDDDs were replaced with ≤ 180 or > 90 cDDDs were replaced with ≤ 90 on the multivariate analysis, the adjusted HRs were 1.46 (95% CI, 1.01-2.11; P = .047) and 1.89 (95% CI, 1.17-3.07; P = .009), respectively. When cDDDs were divided into low, intermediate, and high levels with cutoff points at the 33.3rd percentile (103 cDDDs) and 66.7th percentile (253 cDDDs), the adjusted HRs were 1.70 (95% CI, 1.02-2.84; P = .042) and 1.98 (95% CI, 1.22-3.22; P = .006) for the intermediate and high levels, respectively. The HR estimates of other covariates differed only slightly when different covariates representing cDDDs were applied in the models (data not shown).
|Univariate Analysis||Multivariate Analysisa|
|Variables||HR (95% CI)||P||HR (95% CI)||P|
|Ageb||1.04 (1.03-1.06)||<.001||1.04 (1.03-1.06)||<.001|
|Male sex||1.98 (1.45-2.72)||<.001||1.90 (1.38-2.62)||<.001|
|Diabetes mellitus||1.11 (0.81-1.51)||.524|
|Cirrhosis||3.81 (2.16-6.69)||<.001||3.70 (2.10-6.52)||<.001|
|Chronic kidney disease||1.04 (0.69-1.55)||.856|
|Heart failure||0.96 (0.72-1.30)||.807|
|Cardiovascular disease||0.99 (0.74-1.33)||.961|
|Hyperthyroidism||0.25 (0.06-0.99)||.048||0.33 (0.08-1.35)||.123|
|SES (mo income, NTD)c|
|20,000-39,999||0.51 (0.37-0.70)||<.001||0.63 (0.45-0.87)||.006|
|≥40,000||0.52 (0.34-0.78)||.002||0.62 (0.41-0.94)||.024|
|cDDDse||1.001 (1.000-1.002)||.008||1.001 (1.000-1.002)||.022|
On Kaplan-Meier analysis, which excluded the person-time and events occurring within the first year, the cumulative incidence among patients with low, intermediate, and high cDDDs (≤ 103, 103-253, and > 253) differed significantly (log-rank P = .023) (Fig. 1). The numbers of patients in the low, intermediate, and high cDDDs groups were 1527, 1527, and 1526, respectively, and the numbers of cancer occurrences were 49, 67, and 80, respectively. If only the intermediate-level and high-level cDDD groups were compared, there was no significant difference noted (log-rank P = .545). In addition, the difference between dose levels was not significant if only the person-time and events occurring within the first 2 years were evaluated (log-rank P = .832), and the numbers of cancer occurrences were 26, 26, and 25, respectively.
To the best of our knowledge, the current study is the first large, population-based, cohort study to evaluate the risk of cancer among patients treated with amiodarone. We found that there was a borderline significantly increased risk of cancer among patients who received amiodarone compared with the general population. Patients either of male sex or those with > 180 cDDDs within the first year had a significantly higher risk of developing cancer, and those with both factors had an even greater SIR of 1.46 (P = .008). The dose-effect was also confirmed with the multivariate Cox regression model, which demonstrated an adjusted HR of 1.97 (P = .006) when comparing the incidences in patients with cDDDs in the high and low tertiles.
Our research design, which included an unbiased subject selection, strict amiodarone regimen, and SIR estimations matched with patient age, sex, and calendar year of treatment, supported the validity of these findings. Amiodarone therapy requires complete clinical and electrocardiographic assessments. Because participation in the NHI is mandatory, and all Taiwanese residents can access medical care with low copayments, the follow-up is complete with what to our knowledge is the lowest referral bias reported to date. Pathologic proof of malignancy is necessary to apply for a cancer catastrophic illness certificate. Laboratory and imaging data must also be supplied for the peer review. In addition, a patient with a certificate of catastrophic illness is free from related medical costs, especially hospital expenses. Therefore, data regarding treatment with amiodarone and the diagnosis of cancer in the current study were not only reliable but also exhaustive.
Amiodarone-related adverse effects and toxicity occur in multiple organ systems, especially the thyroid, lung, skin, and liver. Although mutagenicity studies concerning amiodarone were negative, an animal study revealed its potential carcinogenicity.4, 5 In addition, amiodarone has been reported to enhance the effects of radiation on the skin and mucosa,23 and thus possibly accentuate the effects of radiation carcinogenesis. Several case reports have shown evidence of a plausible correlation between amiodarone and malignancies in humans.6–19 Furthermore, a meta-analysis of 15 randomized controlled trials, 4 of which reported cancer deaths, unexpectedly revealed that there were more cancer deaths in patients assigned to treatment with amiodarone compared with patients in the control groups (0.8% [13 of 1609 patients] vs 0.3% [4 of 1597 patients]; P = .0501).24
In previous case reports,13-15, 18, 19 cancers typically developed after regular use of amiodarone for 2 years to 5 years, suggesting that a latency period and high cumulative doses might be necessary for the development of amiodarone-associated malignancies. On the Kaplan-Meier analysis performed in the current study, the overall cancer incidences of patients with low, intermediate, and high cDDDs were found to be significantly different. However, no differences were observed if only the person-time and events occurring within the first 2 years were analyzed. This result was compatible with a latency of 2 years.
In the current study, the dose-effect of amiodarone was confirmed by both external comparison (the SIR estimation) and internal comparison (multivariate Cox regression model). Compared with the general population, the cancer incidence was significantly higher among patients with > 180 cDDDs (SIR, 1.28; P = .046), but not among those with ≤ 180 cDDDs. cDDDs, either as continuous variables or transformed into categorical variables with different cutoff points, were found to be significantly correlated with the development of cancer on multivariate analysis after adjustment for age, sex, and comorbidities. The risk was nearly 2-fold when using a cutoff point of 90 cDDDs (HR, 1.88; P = .010) or when comparing the highest with the lowest tertile (HR, 1.97; P = .006).
In the current study, male patients had an increased risk of cancer, but female patients did not. One possible explanation for this difference is that there is a 37% higher clearance rate of amiodarone in females than in males because of differences in cytochrome P450 3A4 activity and the percentage of body fat.25 In addition, hormonal stimuli might play a role in the gender difference. The increased cancer incidence within 1 year of treatment with amiodarone may be due to a surveillance bias. To provide an initial, thorough evaluation of the etiology of arrhythmias and to monitor the toxicity of amiodarone in follow-up studies, an increased number of medical examinations are performed in patients treated with amiodarone. Thus, the cancer incidence within the first year falsely increases due to early detection.
Amiodarone-induced thyrotoxicosis is a well-known complication among patients who undergo prolonged treatment with amiodarone. The incidence of amiodarone-induced thyrotoxicosis was reported to be 3% to 12%,26-28 and was found to be higher in areas with low iodine intake. Some studies have found an increased risk of thyroid cancer in patients with thyrotoxicosis and goiter.29 Amiodarone is associated with a statistically significant, dose-related increase in the incidence of thyroid tumors (follicular adenomas and/or carcinomas) in rats.5 Several case reports have demonstrated a possible correlation between amiodarone-induced thyrotoxicosis and thyroid cancer.13-15 In the current study, the use of amiodarone was not found to be associated with thyroid cancer (SIR, 0.51; 95% CI, 0.01-2.87), but the case number was too small to draw a definite conclusion. Furthermore, hyperthyroidism and hypothyroidism were not found to be associated with cancer occurrence on our multivariate Cox regression analysis.
Rare cases with amiodarone-induced pulmonary masses, amiodaronomas, have been reported, and the oncogenic effects on the lung were considered.6-12 Furthermore, the long-term administration of amiodarone has been associated with photosensitivity and a blue-gray hyperpigmentation of light-exposed areas of the skin.3, 30 Several cases of multiple basal cell carcinoma were reported after the development of amiodarone-associated skin toxicity.16-19 In the cohort in the current study, the risks of lung cancer and skin cancer were not found to be significantly higher than in the general population, but again the case number was inadequate for detecting a subtle difference.
The current study has several limitations. Several potential risk factors, including obesity, smoking, alcohol use, environmental exposure, and family history of malignancy, were not available in our analyses. Nevertheless, if these factors were meaningful confounders (ie, if these factors were distributed unevenly among patients treated with different dose levels of amiodarone), the difference in cancer incidence would then likely be present from the beginning of follow-up and not after a latency period, as in the current study. We could not directly evaluate the severity of amiodarone-related toxicity (eg, the presence of organ dysfunctions such as pulmonary fibrosis, hyperthyroidism, or hypothyroidism). In our research, the median follow-up was 2.57 years, due to the greater comorbidities and higher mortality rate noted in this patient population. This follow-up might be too short to detect cancer development. Furthermore, to avoid the immortal time bias, we only calculated the cDDDs within the first year. Thus, inevitably, there would be a nondifferential misclassification bias of the exposure variables (cDDDs) (ie, some patients with high cDDDs in the first year might actually take low cDDDs of amiodarone thereafter, and vice versa). However, this is a bias toward the null hypothesis (ie, no association between cDDDs and cancer). Therefore, the actual effect would be larger if the misclassification bias was corrected.
The results of the current study demonstrate a higher cancer risk in patients, especially males, treated with higher cDDDs of amiodarone than in the general population. There is a dose-effect relation between amiodarone and cancer occurrence. It should be noted that the evidence regarding amiodarone as a carcinogen is weak, because the HR is only 1.001 for 1 additional daily dose. Although extensive screenings for occult cancers in patients currently undergoing treatment with amiodarone appears to be impractical, we suggest that cancer events should be routinely reported in future amiodarone trials, and further observational research is necessary.
We thank Miss Chiu-Mei Yeh for her contributions to the revision of the article through further statistical analyses and interpretation of the data.
Supported in part by a grant from Taipei Veterans General Hospital (V101D-001-2).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.