Risk factors for potential drug-drug interactions of statins in patients with acute coronary syndrome

The aim of our study was to assess risk factors for potential drug-drug interactions (pDDIs) of statins across different phases of treatment of acute coronary syndrome (ACS) patients: from the point of first medical contact to the coronary angiography (first phase), after coronary angiography to the last day of hospitalization (second phase) and at discharge from hospital (third phase). This was a post hoc analysis of the data collected during the retrospective observational cohort study conducted at the Clinic for Cardiology of the Clinical Centre Kragujevac, Serbia. Patients prescribed statins were identified from the original study population: 156, 240 and 236 patients for the first, second and third phases, respectively. At least one statin pDDI was present in 113 (72.4%), 161 (67.1%) and 139 (58.9%) patients in the first, second and third phases, respectively. Heart failure, arrhythmias after ACS, CRP, triglycerides, length of hospitalization, number of prescribed drugs, antiarrhythmic drugs, and clopidogrel seem to increase the risk of statin pDDIs in at least one treatment phase. Physicians should be vigilant to the possibility of statin pDDIs in ACS patients who have factors that may increase their rate.


INTRODUCTION
Statins or 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors are a class of lipid-lowering drugs (Causevic-Ramosevac, Semiz, 2013). Early start of statin treatment has become a guideline-directed standard of care after acute coronary syndrome (ACS) which consists of a broad spectrum of manifestations including STsegment elevation myocardial infarction (STEMI), non-ST-segment elevation myocardial infarction (NSTEMI) and unstable angina pectoris (Nordmann et al., 2016). Current guidelines recommend starting high-intensity statin therapy as early as possible unless contraindicated and maintaining the therapy long-term (Roffi et al., 2016;Ibanez et al., 2018). However, statins have a high potential for drug-drug interactions (DDIs), which may put ACS patients at high risk of adverse events such as hepatotoxicity, myopathy and rhabdomyolysis (Causevic-Ramosevac, Semiz, 2013;Kellick et al., 2014;Bellosta, Corsini, 2018). DDI is defined as a clinically significant change in the exposure and/or response to a drug caused by co-administration of another drug which may lead to precipitation of an adverse event or alteration of its therapeutic effects (Scheife et al., 2015). Potential DDI (pDDI) refers to a co-prescription or co-administration of two drugs known to interact (Scheife et al., 2015).
Depending on the study design, population and setting from 4.6% to 36.6% of statin-treated patients may be exposed to statin pDDIs (Egger et al., 2007;Rätz Bravo et al., 2005;Gavronski et al., 2015;Samardzic, Benkovic, Vrca, 2017;Morival et al., 2018). Up to now, only a handful of studies assessed risk factors for statin pDDIs mostly in outpatients (Egger et al., 2007;Rätz Bravo et al., 2005;Gavronski et al., 2015). In a study conducted in Switzerland, risk factors for statin pDDIs in ambulatory statin-treated dyslipidaemic patients were number of drugs, cardiac failure, arrhythmias and being a patient from the French-speaking part of Switzerland (Rätz Bravo et al., 2005), while analysis of the data from this study according to different age groups have shown that male sex, number of pharmaceutical preparations prescribed and psychiatric disorders were risk factors in patients ≤54 years, whereas in ≥55-year-old patients risk factors were number of pharmaceutical preparations or pharmacologically active substances prescribed and a diagnosis of arrhythmia or heart failure (Egger et al., 2007). In an Estonian nationwide register-based study there was a positive correlation between pDDIs and the number of concomitant medicines and age, while logistic regression analysis showed higher odds of exposure in 61-70 age group, men and simvastatin users (Gavronski et al., 2015).
To the best of our knowledge, statin pDDIs and risk factors for their occurrence have not yet been evaluated in ACS patients. Therefore, the aim of our study was to assess statin pDDIs and risk factors for statin pDDIs across different phases of treatment of ACS patients.

MATERIAL AND METHODS
This was a post hoc analysis of the data collected during the retrospective observational cohort clinical study conducted at the Clinic for Cardiology of the Clinical Centre Kragujevac, a public tertiary care hospital in Kragujevac, Serbia with the aim to evaluate pDDIs across different phases of treatment of ACS patients: from the point of first medical contact to the coronary angiography (first phase), after coronary angiography to the last day of hospitalization (second phase) and at discharge from hospital (third phase) (Pejčić, Janković, Davidović, 2019). Evaluation of pDDIs across different phases of treatment was performed because of the differences in therapy before and after coronary angiography, as well as in the therapy at discharge (Pejčić, Janković, Davidović, 2019). The study is described in detail elsewhere and the Ethics Committee of the Clinical Centre Kragujevac had approved the study before its initiation (Pejčić, Janković, Davidović, 2019). Briefly, pharmacotherapy data regarding every day of the patients' treatment were collected along with demographic and clinical data (e.g. laboratory parameters at baseline, risk factors for coronary heart disease, comorbidities, Charlson comorbidity index, complications during hospitalization, reperfusion therapy) for all consecutive patients admitted to the Clinic for Cardiology between January 01, 2017 to December 31, 2017 who were older than 18 years, had diagnosis of ACS manifested as unstable angina pectoris or acute myocardial infarction with or without ST-segment elevation, who underwent coronary angiography and received at least two drugs during hospitalization longer than 24 hours (Pejčić, Janković, Davidović, 2019). Charlson et al. (1987) comorbidity index (CCI) is one of the most widely used indexes of comorbidity and it consists of 19 selected conditions that are weighted and summed to an index. The original study population consisted of 245 patients in which screening for pDDIs was performed in each of the previously mentioned phases of the treatment based on the drugs prescribed to them in that phase. For this post hoc analysis, we identified patients who were prescribed statin therapy: 156, 240 and 236 patients for the first, second and third phases, respectively.
Potential DDI was defined in accordance with the Consensus Recommendations for Systematic Evaluation of Drug-Drug Interaction Evidence for Clinical Decision Support as "co-prescription or co-administration of two drugs known to interact" (Scheife et al., 2015). The presence of statin pDDIs was determined by Micromedex® interaction checker (available at: http://www.micromedexsolutions.com/ home/dispatch). Micromedex® classifies pDDIs as Contraindicated, Major, Moderate and Minor. Potential DDIs classified as Contraindicated (the drugs are contraindicated for concurrent use), Major (the interaction may be life-threatening and/or require medical intervention to minimize or prevent serious adverse effects) and Moderate (the interaction may result in exacerbation of the patient's condition and/or require an alteration in therapy) were included in the analysis. Potential DDIs classified as Minor were not included in the analysis, considering that they would have limited clinical effects and generally would not require a major alteration in therapy.
All analyses were performed using the Statistical Program for Social Sciences (SPSS version 18). The data were analysed by descriptive statistics. Measures of central tendency (mean) and measures of dispersion (standard deviation and range) were used for continuous variables. Categorical variables were presented as frequencies (%). The influence of potential risk factors on the number of statin pDDIs per patient during each treatment phase was evaluated by multiple linear regression using backward elimination with a probability of F to remove a predictor of ≤ 0.1. In this approach all potential predictor variables are added into the model, then are eliminated one by one, with the variable that has the largest probability of F (i.e., p value) removed until all variables have a p value equal to or less than 0.1. Dichotomous categorical variables were coded with 0 and 1 (0 indicated the absence of a qualitative attribute, while 1 indicated presence, except for gender where 0 indicated female gender, and 1 male gender). The statistical validity of the regression was checked by analysis of variance (F value) and percentage of the outcome (number of pDDIs per patient) variability explained (R 2 ). The influence of potential risk factors on the outcome was assessed by their B coefficients in the regression equation, including 95% confidence intervals (CIs). A p value < 0.05 was considered statistically significant.

RESULTS
Characteristics of the study population of ACS patients who received statin therapy are shown in Table   I, while pharmacotherapy and reperfusion data are shown in Table II. All patients received contrast media during coronarography in the first treatment phase. The most frequently used statin in all treatment phases was atorvastatin (see Table II for frequencies).

List of variables entered at the beginning of the analysis (variables included in the last step of the model are marked with †):
First phase: age, gender †, Charlson Comorbidity Index, heart failure †, renal failure, hyperlipidaemia, diabetes, CRP †, triglycerides, number of physicians who prescribed drugs to single patient, number of prescribed drugs †, antiarrhythmic drugs †, clopidogrel †, antibiotics, calcium channel blockers; Second phase: age, gender, length of hospitalization †, Charlson Comorbidity Index, dementia, infection during hospitalization, heart failure, arrhythmias after ACS †, asthma, renal failure, hyperlipidaemia, diabetes, triglycerides †, number of physicians who prescribed drugs to single patient, number of prescribed drugs, antiarrhythmic drugs †, antibiotics, clopidogrel †; Third phase: age †, gender, length of hospitalization, Charlson Comorbidity Index, dementia, arrhythmias after ACS, asthma, heart failure, renal failure, hyperlipidaemia, diabetes, triglycerides †, number of prescribed drugs, antiarrhythmic drugs †, antibiotics, clopidogrel †.

DISCUSSION
The majority of ACS patients (over 58.9%) in our study were exposed to at least one statin pDDI across all treatment phases. In at least one phase heart failure, arrhythmias after ACS, CRP, triglycerides, length of hospitalization, number of prescribed drugs, antiarrhythmic drugs and clopidogrel seem to increase the risk of pDDIs in ACS patients.
Percentage of patients exposed to statin pDDIs in previous studies ranged from 4.6% in Estonian nationwide study in subjects aged 50 years or older (Gavronski et al., 2015) to 36.6% in subjects in Croatian community pharmacy (Samardzic, Benkovic, Vrca, 2017) which is lower than the percentage in our study probably due to differences in study populations, settings, considered types of pDDIs and information sources used to define pDDIs. Our study was conducted in ACS patients hospitalized in a tertiary care hospital which usually have more severe clinical conditions, so prevalence and pattern of pDDIs observed in our study may differ from ambulatory patients or patients hospitalized in other settings (Morival et al., 2018).
The number of prescribed drugs significantly increased the risk of statin pDDIs in line with previous studies (Egger et al., 2007;Rätz Bravo et al., 2005;Gavronski et al., 2015). ACS patients who are prescribed a higher number of drugs in general have a higher risk to be exposed to pDDIs (Pejčić, Janković, Davidović, 2019). In addition, length of hospitalization was observed to be a significant risk factor for statin pDDIs in our study, as well as for pDDIs in general in ACS (de Lima, de Godoy, 2017) and cardiovascular patients (Kovačević et al., 2017). It seems that longer hospitalization may increase the risk of getting a drug that may interact with statins, particularly in the second phase of treatment.
Increased triglyceride levels were associated with the increased number of statin pDDIs in our study. Considering that statin-fibrate combination therapy may have to be used to treat patients with ACS complicated by elevated triglycerides (Li et al., 2013), this combination therapy probably contributed to this finding. Combined statin and fibrate therapy may lead to rhabdomyolysis and according to the recommendations of the American Heart Association fenofibrate is the preferred fibrate to use in combination with statins because of a reduced incidence of DDIs compared with statin-gemfibrozil combination therapy (Wiggins et al., 2016). However, even a combination with fenofibrate was associated with the occurrence of rhabdomyolysis in ACS patients, so caution is certainly needed even when fenofibrate is prescribed (Jozic et al., 2014). CRP, a prognostic marker of recurrent nonfatal myocardial infarction or cardiac death in NSTEMI, and marker which may reflect the extent of myocardial injury in STEMI (del Val Martin, Sanmartín Fernández, Zamorano Gómez, 2015), was also associated with increased risk of statin pDDIs in our study.
Arrhythmias and heart failure increased the risk of statin pDDIs in our study. This finding is in line with previous studies conducted in ambulatory statin-treated dyslipidaemic patients (Egger et al., 2007;Rätz Bravo et al., 2005). These diagnoses may act as surrogate markers for use of drugs known to interact with statins: digoxin and antiarrhythmic drugs, especially amiodarone (Egger et al., 2007;Rätz Bravo et al., 2005). Certain statins (e.g. simvastatin, lovastatin and atorvastatin) may increase the plasma digoxin concentration due to inhibition of P-glycoprotein, which can be concerning considering its narrow therapeutic range (Rätz Bravo et al., 2005). Amiodarone is a cytochrome P450 (CYP) 3A4 and CYP2C9 inhibitor which may increase levels of statins metabolized via these enzymes and risk of adverse effects (Egger et al., 2007).
Clopidogrel is a prodrug that requires conversion via the CYP system (dominantly by CYP3A4, CYP3A5 and CYP2C19) (Todorović et al., 2016;Thotakura et al., 2018). Drugs interfering with or being co-metabolized via those enzymes, like statins metabolized via CYP3A4, might decrease its antiplatelet effect and increase the risk of thrombotic events (Todorović et al., 2016;Thotakura et al., 2018). Lau et al. (2003) first reported results of an ex vivo study showing that atorvastatin, but not pravastatin, attenuated clopidogrel antiplatelet effects in a dose-dependent manner. In the following years, many conflicting reports regarding this pDDI were published (Steinhubl, Akers, 2006). Although some studies reported that CYP3A4 metabolized statins decrease clopidogrel antiplatelet effects, the majority of them, however, found no influence of statin therapy on platelet function which highlighted the limitations in translating ex vivo platelet testing results to clinical practice and cardiovascular outcomes (Steinhubl, Akers, 2006). The majority of studies involving ACS patients didn't show reduced clinical efficacy of clopidogrel when it was used concomitantly with statins (Ojeifo et al., 2013;Mukherjee et al., 2005). However, some studies suggest that patients with high on-treatment platelet reactivity (HPR) during concurrent use of clopidogrel and atorvastatin may benefit from switching to a non-CYP3A4-metabolized statin which can significantly decrease platelet reactivity and the prevalence of HPR (Park et al., 2012). Also, hepatotoxicity may occur when clopidogrel is used concomitantly with atorvastatin probably due to increased stress on the CYP3A4 enzyme (Thotakura et al., 2018).
When initiating therapy with a statin, both the existing treatment regimen and the risk profile of the patient should be carefully evaluated. The overall clinical impact of a DDI can range from mild to life-threatening, so not all DDIs require a modification in therapy (Wiggins et al., 2016). The variability of the clinical significance of a DDI depends on medication-specific and patient-specific factors, so it is important to evaluate the risk-benefit ratio of the combination therapy individually for each patient (Wiggins et al., 2016).
Our work has some limitations that need to be considered. First, the study was unicentric, which could have introduced bias of local and national quality of medical education in the results. In addition, the study was retrospective, so we could only analyse potential statin DDIs, while their clinical outcomes could not be followed. Although the use of an interaction checker is helpful in identifying pDDIs, this also has its limitations: it is not possible to control for factors influencing the relevance of a pDDI (e.g. dosage, time of administration, beginning and duration of treatment, underlying diseases) (Egger et al., 2007). Actual DDIs, i.e. actual clinical manifestations of pDDIs can be difficult to identify. Some adverse events associated with statin pDDIs e.g. musculoskeletal adverse events occur rarely or may have delayed onset ranging from occurring within 12 months or many years after starting statin therapy (Akimoto et al., 2018) and many factors (such as concomitant disease or genetic predisposition) could make the causality assessment of the interaction difficult (Magro, Moretti, Leone, 2012). Despite these limitations, our findings may serve as a useful input to understand the extent of the problem and taking measures to improve the management of statin DDIs in ACS patients.
In conclusion, the majority of ACS patients in our study were exposed to at least one statin pDDI across all treatment phases. Heart failure, arrhythmias after ACS, CRP, triglycerides, length of hospitalization, number of prescribed drugs, antiarrhythmic drugs, and clopidogrel seem to increase the risk of statin pDDIs in ACS patients. Physicians should be vigilant to the possibility of statin pDDIs in patients who have factors that may increase their rate.