Amiodarone-induced thyroid dysfunction in Taiwan: a retrospective cohort study

Explainable Machine Learning Techniques To Predict Amiodarone-Induced Thyroid Dysfunction Risk: Multicenter, Retrospective Study With External Validation

BACKGROUND

Machine learning offers new solutions for predicting life-threatening, unpredictable amiodarone-induced thyroid dysfunction. Traditional regression approaches for adverse-effect prediction without time-series consideration of features have yielded suboptimal predictions. Machine learning algorithms with multiple data sets at different time points may generate better performance in predicting adverse effects.

OBJECTIVE

We aimed to develop and validate machine learning models for forecasting individualized amiodarone-induced thyroid dysfunction risk and to optimize a machine learning-based risk stratification scheme with a resampling method and readjustment of the clinically derived decision thresholds.

METHODS

This study developed machine learning models using multicenter, delinked electronic health records. It included patients receiving amiodarone from January 2013 to December 2017. The training set was composed of data from Taipei Medical University Hospital and Wan Fang Hospital, while data from Taipei Medical University Shuang Ho Hospital were used as the external test set. The study collected stationary features at baseline and dynamic features at the first, second, third, sixth, ninth, 12th, 15th, 18th, and 21st months after amiodarone initiation. We used 16 machine learning models, including extreme gradient boosting, adaptive boosting, k-nearest neighbor, and logistic regression models, along with an original resampling method and 3 other resampling methods, including oversampling with the borderline-synthesized minority oversampling technique, undersampling-edited nearest neighbor, and over- and undersampling hybrid methods. The model performance was compared based on accuracy; Precision, recall, F₁-score, geometric mean, area under the curve of the receiver operating characteristic curve (AUROC), and the area under the precision-recall curve (AUPRC). Feature importance was determined by the best model. The decision threshold was readjusted to identify the best cutoff value and a Kaplan-Meier survival analysis was performed.

RESULTS

The training set contained 4075 patients from Taipei Medical University Hospital and Wan Fang Hospital, of whom 583 (14.3%) developed amiodarone-induced thyroid dysfunction, while the external test set included 2422 patients from Taipei Medical University Shuang Ho Hospital, of whom 275 (11.4%) developed amiodarone-induced thyroid dysfunction. The extreme gradient boosting oversampling machine learning model demonstrated the best predictive outcomes among all 16 models. The accuracy; Precision, recall, F₁-score, G-mean, AUPRC, and AUROC were 0.923, 0.632, 0.756, 0.688, 0.845, 0.751, and 0.934, respectively. After readjusting the cutoff, the best value was 0.627, and the F₁-score reached 0.699. The best threshold was able to classify 286 of 2422 patients (11.8%) as high-risk subjects, among which 275 were true-positive patients in the testing set. A shorter treatment duration; higher levels of thyroid-stimulating hormone and high-density lipoprotein cholesterol; and lower levels of free thyroxin, alkaline phosphatase, and low-density lipoprotein were the most important features.

CONCLUSIONS

Machine learning models combined with resampling methods can predict amiodarone-induced thyroid dysfunction and serve as a support tool for individualized risk prediction and clinical decision support.

Prevalence of amiodarone-induced hypothyroidism; A systematic review and meta-analysis: Amiodarone-induced hypothyroidism epidemiology

Amiodarone is a common anti-arrhythmic agent mostly used to treat and prevent different kinds of arrhythmia with several considerable side effects, most commonly on the thyroid gland. We aimed to assess the frequency of hypothyroidism among chronic amiodarone users. PubMed/Medline, Web of Science, and Scopus databases were screened in the title and abstract sections with no time limitation. Relevant published records reported amiodarone-induced hypothyroidism (AIH) among patients with normal thyroid function at baseline were recruited with further analysis according to gender and study locations. We found 29 records on 14143 individuals. Total population age ranged from 18 to 92 years (males: 58.2% (8158 out of 13999)). The AIH prevalence was found to be 14% (95% confidence interval (CI): 12-17%). Further gender stratified showed an insignificant higher AIH frequency in females versus males (17%, 95% CI: 13-22% vs. 14%, 95% CI: 11-19% P= 0.304, respectively). Despite no significant difference in AIH prevalence according to different continents, African subjects had marginally lower AIH frequency compared to Asian (7%, 95% CI: 4-13% vs. 15%, 95% CI: 12-19%, P= 0.012) and South American persons (7%, 95% CI: 4-13% vs. 54%, 95% CI: 9-93%, P= 0.038). This review suggests the occurrence of AIH is quite considerable regardless of gender and area of residence, and several periodic thyroid assessment strategies should be developed for earlier recognition and therapeutic interventions in clinical settings.

Amiodarone-Induced Thyrotoxicosis - Literature Review & Clinical Update

Amiodarone is widely used in treating atrial and ventricular arrhythmias; however, due to its high iodine concentration, the chronic use of the drug can induce thyroid disorders. Amiodarone-induced thyrotoxicosis (AIT) can decompensate and exacerbate underlying cardiac abnormalities, leading to increased morbidity and mortality, especially in patients with left ventricular ejection fraction <30%. AIT cases are classified into two subtypes that guide therapeutic management. The risks and benefits of maintaining the amiodarone must be evaluated individually, and the therapeutic decision should be taken jointly by cardiologists and endocrinologists. Type 1 AIT treatment is similar to that of spontaneous hyperthyroidism, using antithyroid drugs (methimazole and propylthiouracil) at high doses. Type 1 AIT is more complicated since it has proportionally higher recurrences or even non-remission, and definitive treatment is recommended (total thyroidectomy or radioiodine). Type 2 AIT is generally self-limited, yet due to the high mortality associated with thyrotoxicosis in cardiac patients, the treatment should be implemented for faster achievement of euthyroidism. Furthermore, in well-defined cases of type 2 AIT, the treatment with corticosteroids is more effective than treatment with antithyroid drugs. In severe cases, regardless of subtype, immediate restoration of euthyroidism through total thyroidectomy should be considered before the patient progresses to excessive clinical deterioration, as delayed surgery indication is associated with increased mortality.

Polypharmacy Is Associated With Amiodarone-Induced Hypothyroidism

Background: Amiodarone is rich in iodine, so in clinical practice amiodarone-induced hypothyroidism (AIH) is a major side effect. This drug is used in patients with arrhythmias, especially atrial fibrillation, the most common sustained arrhythmia. Polypharmacy, which can result in complex drug-drug interactions, occurs in more than 70% of the patients with atrial fibrillation. Therefore, polypharmacy may be involved in the expression of AIH. In this study, we investigated the association between polypharmacy and AIH. Methods: We conducted a retrospective study using data from January 2006 to May 2020 collected from a large, organized database of prescriptions constructed by the Japan Medical Information Research Institute, Inc. (Tokyo, Japan). To investigate the association between number of prescribed drugs with amiodarone and AIH, we divided patients into two groups: polypharmacy (≥ 5 prescribed drugs) and non-polypharmacy (< 5 prescribed drugs). We then performed a sequence symmetry analysis on the two groups: incident thyroxine after incident amiodarone and incident thyroxine before incident amiodarone. Finally, we conducted a case-control study on two further groups: those prescribed thyroxine after incident amiodarone (AIH group; n=555) and those not prescribed thyroxine after incident amiodarone (non-AIH group; n=6,192). Results: Sequence symmetry analysis revealed a significant association between amiodarone and thyroxine in both the polypharmacy and non-polypharmacy groups. The ranges for the adjusted sequence ratio in the two groups were 12.0-16.7 and 7.3-9.0, respectively. The case-control study showed that ≥5 prescribed drugs at the first prescription of amiodarone were found to significantly increase the odds of AIH (odds ratio: 1.48, 95% confidence interval: 1.18-1.84). Conclusion: Polypharmacy was suggested as an independent risk factor for AIH. Careful assessment of the appropriateness of prescription is warranted.

Issues in amiodarone-induced thyrotoxicosis: Update and review of the literature

Amiodarone, a benzofuranic iodine-rich pan-anti-arrhythmic drug, induces amiodarone-induced thyrotoxicosis (AIT) in 7-15% of patients. AIT is a major issue due to its typical severity and resistance to anti-thyroid measures, and to its negative impact on cardiac status. Classically, AIT is either an iodine-induced thyrotoxicosis in patients with abnormal thyroid (type 1), or due to acute thyroiditis in a "healthy" thyroid (type 2). Determination of the type of AIT is a diagnostic dilemma, as characteristics of both types may be present in some patients. As it is the main etiological factor in AIT, it is recommended that amiodarone treatment should be stopped; however, it may be the only anti-arrhythmic option, needing to be either continued or re-introduced to improve cardiovascular survival. Recently, a few studies demonstrated that amiodarone could be continued or re-introduced in patients with history of type-2 AIT. However, in the other patients, it is recommended that amiodarone treatment be interrupted, to improve response to thioamides and to alleviate the risk of AIT recurrence. In such patients, thyroidectomy is recommended once AIT is under control, allowing safe re-introduction of amiodarone.

Adherence to guidelines in monitoring amiodarone-induced thyroid dysfunction

RATIONALE, AIMS AND OBJECTIVES

Baseline thyroid function testing and regular follow-up of thyroid function under amiodarone usage was recommended by guidelines. Little is known about the status of amiodarone monitoring in real-world clinical care in Taiwan. The objective was to determine the rate of thyroid monitoring and to assess the clinical and physicians' characteristics associated with adequate monitoring in a tertiary referral centre for arrhythmia.

METHODS

We reviewed the medical records of patients receiving amiodarone during the period 2008-2009 at Taipei Veterans General Hospital. The rate of baseline and follow-up thyroid function monitoring during amiodarone therapy were calculated. Factors associated with guideline adherence to monitoring were analysed.

RESULTS

Among the 1319 enrolled cases, 36.4% (n = 480) underwent baseline thyroid function testing and 1.1% (n = 15) received measurement of anti-thyroid peroxidase antibody before amiodarone initiation. Regular follow up of thyroid function under amiodarone usage occurred in only 8.6% (n = 114) of cases. Baseline thyroid function was more likely to be present in patients of younger age (P < 0.001), female sex (P = 0.01), and in those who received amiodarone therapy from cardiologists (P < 0.001) or electrophysiologists (P < 0.001) with fewer years of service (P < 0.001). Upon multivariate analysis, only physicians' expertise (cardiologist versus non-cardiologist, OR = 5.67, 95% CI: 2.44-13.16) and years of service (OR = 0.97, 95% CI: 0.95-0.998) were significantly associated with adequate thyroid monitoring.

CONCLUSIONS

The rate of thyroid monitoring with amiodarone therapy had been suboptimal. Strategies to enhance guideline adherence are needed.

2016 American Thyroid Association Guidelines for Diagnosis and Management of Hyperthyroidism and Other Causes of Thyrotoxicosis

BACKGROUND

Thyrotoxicosis has multiple etiologies, manifestations, and potential therapies. Appropriate treatment requires an accurate diagnosis and is influenced by coexisting medical conditions and patient preference. This document describes evidence-based clinical guidelines for the management of thyrotoxicosis that would be useful to generalist and subspecialty physicians and others providing care for patients with this condition.

METHODS

The American Thyroid Association (ATA) previously cosponsored guidelines for the management of thyrotoxicosis that were published in 2011. Considerable new literature has been published since then, and the ATA felt updated evidence-based guidelines were needed. The association assembled a task force of expert clinicians who authored this report. They examined relevant literature using a systematic PubMed search supplemented with additional published materials. An evidence-based medicine approach that incorporated the knowledge and experience of the panel was used to update the 2011 text and recommendations. The strength of the recommendations and the quality of evidence supporting them were rated according to the approach recommended by the Grading of Recommendations, Assessment, Development, and Evaluation Group.

RESULTS

Clinical topics addressed include the initial evaluation and management of thyrotoxicosis; management of Graves' hyperthyroidism using radioactive iodine, antithyroid drugs, or surgery; management of toxic multinodular goiter or toxic adenoma using radioactive iodine or surgery; Graves' disease in children, adolescents, or pregnant patients; subclinical hyperthyroidism; hyperthyroidism in patients with Graves' orbitopathy; and management of other miscellaneous causes of thyrotoxicosis. New paradigms since publication of the 2011 guidelines are presented for the evaluation of the etiology of thyrotoxicosis, the management of Graves' hyperthyroidism with antithyroid drugs, the management of pregnant hyperthyroid patients, and the preparation of patients for thyroid surgery. The sections on less common causes of thyrotoxicosis have been expanded.

CONCLUSIONS

One hundred twenty-four evidence-based recommendations were developed to aid in the care of patients with thyrotoxicosis and to share what the task force believes is current, rational, and optimal medical practice.

Amiodarone induced liver dysfunction in elderly patients with arrhythmia: Clinical characteristics and management

AIM

to explore the clinical characteristics and management of amiodarone induced liver dysfunction in elderly patients with arrhythmia.

METHODS

Two hundred and seventy-eight elderly patients with cardiac arrhythmias treated with amiodarone at Central Hospital of Jinhua and Lishui People's Hospital from January 2013 to December 2015 were enrolled, including 183 male patients and 95 female patients. Their age ranged from 60 to 85 years, with a mean age of 71.52 years ± 7.19 years. Before the medication each patient underwent liver ultrasound and venous blood liver function tests, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, direct bilirubin, indirect bilirubin, albumin, globulin, and-glutamyltransferase, to exclude liver function abnormalities. The liver function of patients was monitored by liver function tests every 2 wk, and by liver ultrasound monthly during the process of medication.

RESULTS

Abnormal liver function was detected in 71 patients, including 48 males and 23 females. Within 2 wk after amiodarone medication, ALT increased from 15.08 U/L ± 1.17 U/L to 102.27 U/L ± 1.72 U/L, and AST increased from 19.83 U/L ± 1.17 U/L to 125.27 U/L ± 1.72 U/L. After 2 wk all other indicators rose more than two times than those before medication. However, these indicators returned to normal after discontinuation of amiodarone for 1 mo. Liver ultrasound also showed that after discontinuation of amiodarone, ultrasonographic changes returned to normal spontaneously only in patients with abnormal liver function. In patients with liver cirrhosis, ultrasonographic changes were irreversible.

CONCLUSION

Amiodarone is a main cause of liver dysfunction, and clinical manifestations of amiodarone induced liver dysfunction show individual differences. After discontinuation of amiodarone, liver injury can recover spontaneously; however, in patients with liver cirrhosis, liver injury is irreversible.

Risk factors for amiodarone-induced thyroid dysfunction in Japan

Background

Amiodarone is associated with a number of significant adverse effects, including elevated transaminase levels, pulmonary fibrosis, arrhythmia, and thyroid dysfunction. Although thyroid dysfunction is considered to be a common and potentially serious adverse effect of amiodarone therapy, the exact pathogenesis remains unknown because of its complex manifestations. Therefore, the prevalence of, and risk factors for, amiodarone-induced thyroid dysfunction in Japanese patients were investigated in the present study.

Methods

A retrospective analysis of patients treated with amiodarone between January 2012 and December 2013 was performed. A total of 317 patients with euthyroidism, or subclinical hyperthyroidism or hypothyroidism, were enrolled in this study.

Results

After being treated with amiodarone, 30 (9.5%) and 60 patients (18.9%) developed amiodarone-induced hyperthyroidism and amiodarone-induced hypothyroidism, respectively. Ten (33.3%) patients with amiodarone-induced hyperthyroidism and 40 (66.6%) with amiodarone-induced hypothyroidism were diagnosed within two years of the initiation of amiodarone therapy. Dilated cardiomyopathy (DCM) [Adjusted odds ratio (OR) 3.30 (95% confidence interval (CI): 1.26–8.90)], and cardiac sarcoidosis [Adjusted OR 6.47 (95% CI: 1.60–25.77)] were identified as predictors of amiodarone-induced hyperthyroidism. The baseline free thyroxine (T4) level [Adjusted OR 0.13 (95% CI: 0.03–0.68)], and thyroid-stimulating hormone (TSH) level [Adjusted OR1.47 (95% CI: 1.26–1.74)] were identified as predictors of amiodarone-induced hypothyroidism.

Conclusion

DCM and cardiac sarcoidosis were identified as risk factors for amiodarone-induced hyperthyroidism. Risk factors for amiodarone-induced hypothyroidism included higher baseline TSH level and lower baseline free T4 level, suggesting that subclinical hypothyroidism may be a potential risk factor for the development of amiodarone-induced hypothyroidism.

Environmental Iodine Content, Female Sex and Age Are Associated with New-Onset Amiodarone-Induced Hypothyroidism: A Systematic Review and Meta-Analysis of Adverse Reactions of Amiodarone on the Thyroid

OBJECTIVES

To investigate the incidence of new-onset amiodarone-induced hypothyroidism (AIH) and the associated risk factors.

METHODS

We performed a systematic search in MEDLINE, Embase, the Cochrane Library and the Chinese database from 1995 to 2015. Studies that investigated amiodarone-related adverse reactions on the thyroid were included. A random-effect model was used for the meta-analysis to investigate the incidence rate of AIH and associated risk factors.

RESULTS

We identified 465 studies, of which data from 9 studies were included, comprising 1,972 patients. The incidence of AIH was 14.0% (95% confidence interval, CI, 8.7-21.7%) as a whole; it was higher in areas with a high than a low iodine content in the environment (20.3 vs. 8.7%, p < 0.001); subgroup analysis showed that AIH occurred in 19.2% (95% CI 10.2-33.1%) of women and 13.3% (95% CI 7.9-21.7%) of men (p < 0.001). Meta-regression analysis indicated a positive correlation with the mean age and percentage of women.

CONCLUSIONS

The occurrence of AIH is a relatively frequent complication of amiodarone, and older women are more likely to develop AIH, especially in areas with a high iodine content in the environment, and restriction of total exposure to iodine might decrease the incidence of AIH.

Prevalence of amiodarone-induced thyrotoxicosis and associated risk factors in Japanese patients

Amiodarone is a widely used agent for life-threatening arrhythmias. Although amiodarone-induced thyrotoxicosis (AIT) is a major adverse effect that can cause recurrence of arrhythmias and exacerbation of heart failure, risk factors for AIT among amiodarone-treated Japanese patients have not been elucidated. Here, we investigated the prevalence and predictive factors for AIT. The study subjects were 225 patients treated with amiodarone between 2008 and 2012, who were euthyroid before amiodarone therapy. All patients with AIT were diagnosed by measurement of thyroid hormones and ultrasonography. Among the 225 subjects, 13 patients (5.8%) developed AIT and all the patients were classified as Type 2 AIT. Baseline features of patients with AIT were not different from those who did not develop AIT, except for age (AIT, 55.1 ± 13.8, non-AIT, 68.1 ± 12.0 years, P < 0.001). Multivariate analyses using the Cox proportional hazard model identified age as the sole determinant of AIT (hazard ratio: 0.927, 95% confidence interval: 0.891-0.964). Receiver operating characteristic curve analysis identified age of 63.5 years as the cutoff value for AIT with sensitivity of 70.3% and specificity of 69.2%. In summary, this study showed that the prevalence of AIT is 5.8% in Japanese patients treated with amiodarone and that young age is a risk factor for AIT.