Prevalence and type of potential pharmacokinetic drug- drug interactions in old aged psychiatric patients

Purpose: The number of prescribed drugs increases with age, and resulting polypharmacy bears the risk of drug-drug interactions (DDI). We assessed the prevalence and type of cytochrome P450 (CYP) associated pharmacokinetic DDI that were considered as clinically relevant in elderly psychiatric patients under conditions of every day pharmacotherapy. Methods: For retrospective analysis a large therapeutic drug monitoring (TDM) data base was used. Patients included were elderly inand outpatients, aged ≥ 65 years with a psychiatric disorder for whom serum level measurements of a psychotropic drug had been requested. Medication data were examined for co-prescription of clinically relevant CYP inhibitors or inducers and drugs that are substrates of these enzymes (victim drugs). Results: Data from 1243 patients (65.3% female) with a mean (± standard deviation) age of 73 ± 6 years were available for analysis. Mean number of prescribed drugs was 5.3 ± 3.4 per patient. Moderate (n=189, 73.8%) or strong (n=67, 26.2%) CYP inhibitors or inducers were prescribed in 18.9% of all patients. Most frequently prescribed CYP inhibitors were duloxetine (CYP2D6 inhibitor, n=73, 31.7%), melperone (CYP2D6 inhibitor, n=63, 27.4%) and omeprazole (CYP2C19 inhibitor, n=20, 8.7%). Carbamazepine was the most commonly prescribed CYP inducer (primarily CYP3A4 inducer, n=25, 96.2%). Taken inhibitory, inducing and substrate properties of combined drugs together, CYP-mediated DDI considered as clinically relevant could be detected in 133 (10.7%) of included patients. Conclusions: CYP-mediated DDI are common among elderly psychiatric patients. To avoid potentially supraor sub-therapeutic concentrations of victim drugs, physicians should be more aware of these preventable interactions and adjust the dose. Correspondence to: Gudrun Hefner, Institute of Clinical Chemistry and Laboratory Medicine, University Medical Center, Langenbeckstrasse 1, 55131 Mainz, Germany, Tel: +49 (0) 6131 17-7092; Fax: +49 (0) 6131 17-6627; E-mail: gudrun.hefner@unimedizin-mainz.de


Introduction
The potential for drug-drug interactions (DDI) is increasing exponentially [1] with increasing drug intake. Especially in the elderly, polypharmacy [2] increases the risk of DDI [3] and therefore the probability of adverse drug reactions (ADRs) [4,5].
DDI are classified into two main groups. Pharmacodynamic interactions act by potentiation or antagonism at the target structure. Pharmacokinetic DDI occur when a drug alters the absorption, distribution, metabolism, or clearance of another, the victim drug [6].
Several enzymatic systems are involved in pharmacokinetic DDI, most prominent is the cytochrome P450 (CYP) system [7]. Inhibition or induction of CYP metabolism is recognized as most important mechanisms of pharmacokinetic DDI, although DDI related to involvement of CYP isoenzymes constitute only a small proportion of ADRs [8]. Pharmacokinetic DDI are important because they are often predictable and therefore avoidable [6].
Pharmacokinetic DDI may result in serious toxicological serum concentrations or in sub-therapeutic concentration and loss of action of a victim drug if the dosage is not adjusted [9]. The seriousness of a DDI depends on the serum concentration of the victim drug and its therapeutic index [10]. To avoid pharmacokinetic DDI, substrate, inhibitor, and inducer properties for the CYP isoenzymes of all combined drugs must be considered [11].
A growing appreciation for CYP-related DDI resulted in an identification of multiple drugs as CYP substrates, inhibitors or inducers [9,12]. Thereby the magnitude of potential drug interactions was often neglected. This has lead to alerts that are clinically unimportant and do not require changes of medications or dose (overalerting) [13]. Therefore, the CYP-inhibitory or -inducing potential of combined drugs must be considered. The Food and Drug Administration (FDA) categorized inhibitors or inducers as strong, moderate or weak based on pharmacokinetic studies. This was published in an updated version on a FDA website for drug development and drug interactions [14]. The last update was in July 2011.
Moderate and strong inhibitors can be regarded as clinically relevant, since for moderate inhibitors, a more than 50% decrease of the clearance is defined which leads to an at least twofold increase of drug concentrations in blood.
Most pharmacokinetic DDI of psychotropic drugs occur at the metabolic level involving the CYP enzyme system [15]. Psychotropic DDI give rise to highly variable serum-concentrations of the respective victim drug [9,10,16], leading to ADRs [17,18]. Thereby, elderly patients are particularly sensitive to ADRs, especially because of age-related changes in pharmacokinetics and pharmacodynamics [19,20], a higher prevalence of comorbidity [21] and a high rate of drug consumption [2] which increases the risk of drug interactions [3]. For risk minimization due to application of potentially harmful drug combinations, physicians should amongst others consider drug history and prescribe appropriate drugs with a minimum of critical combinations [22] to improve safety and effectiveness of psychopharmacological treatment.
Until recently, studies on the prevalence and relevance of DDI regarding the gradual impact on the metabolism of a victim drug in a naturalistic clinical setting are lacking. This study examined the prevalence of combinations of clinically relevant CYP enzyme inhibitors or inducers, considered as "strong" or "moderate" based on a pharmacokinetic classification [14], with drugs that are preferred substrates for these enzymes (victim drugs) in elderly psychiatric patients. It was hypothesized for old aged psychiatric patients that the number of CYP associated DDI increases with increasing number of prescribed drugs. Moreover, it was investigated which CYP enzymes and drugs are primarily affected by pharmacokinetic DDI in old aged psychiatric patients.

Study design
The present retrospective analysis used information from a large therapeutic drug monitoring (TDM) database. It originated from 2999 TDM requests raised in the TDM laboratories of the Departments of Psychiatry and Psychotherapy at the Universities of Mainz and Regensburg (Germany) between 1998 and 2014. These laboratories receive blood samples for TDM of psychoactive drugs from multiple psychiatric hospitals in Germany. Overall, a data set from 922 TDM requests were obtained from Regensburg [23] and data from 2077 TDM requests were obtained from the University Medical Center of Mainz. All included patients were aged at least 65 years and had a psychiatric diagnosis according to International classification of Mental and Behavioural Disorders, 10th Revision (ICD-10). Databases included person-specific information given on the TDM request forms [9]. Retrospective analysis of clinical data for this study was in accordance with regulations given by the local ethics committee of the Ärztekammer Rheinland-Pfalz.

Patients
Age, gender, psychiatric diagnosis and medication were recorded for analysis. Applied medications at the day of blood-withdrawal were considered. The total number of prescribed medications was calculated, excluding dietary supplements, as vitamins and mineral nutrients. Each patient was included for analysis only once. When multiple serum level measurements were requested for the same patient, only the first request was considered. Thus, a higher prevalence rate of pharmacokinetic DDI was expected at the first request, as the reported TDM result includes dose and medication recommendation (DDI check) by the laboratory. Patients were excluded when data about medication or age were insufficient or when the patient identification number was not available.

Drug combinations
Medication data were examined for co-prescription of CYP inhibitors or inducers and corresponding substrates (victim drugs). CYP inhibitors were defined "strong" or "moderate" in accordance with a classification proposed by the FDA [14]. CYP inhibitors identified as "strong" (>5-fold increase in area under the plasma concentration-time curve (AUC) or >80% decrease in clearance) or "moderate" (>2-fold but <5-fold increase in the AUC or 50-80% decrease in clearance) and CYP inducers defined as "strong" (≥ 80% decrease in AUC) or "moderate" (50-80% decrease in AUC) were considered for analysis. In addition to this information and based on reported literature, melperone [24] and levomepromazine [25] were designated as moderate CYP2D6 inhibitors and perazine as a strong CYP1A2 inhibitor [26][27][28].
CYP substrate properties were considered to define victim drugs in accordance with the Consensus Guidelines for TDM in Psychiatry of the Arbeitsgemeinschaft für Neuropsychopharmakologie und Pharmakopsychiatrie (AGNP) [9] who defined victim drugs for distinct CYP enzymes. For non-psychotropic victim drugs, CYP substrate properties were identified by the CYP450 Drug Interaction Table of Indiana University [29].

Statistical analysis
Descriptive statistics of patients' data were presented as mean values ± standard deviations (mean±SD). For comparisons between gender and patients with and without a detected DDI, the Mann-Whitney U test was used, as the variables were not normally distributed. To evaluate a difference between two nominal variables (gender and DDI), the chi square test for independence was used. Statistical analysis was carried out using IBM ® SPSS ® Statistics version 21.0 (IBM GmbH, Ehningen, Germany). A 95% confidence interval (CI) was applied and a p-value <0.05 was considered statistically significant.
Carbamazepine was the most frequently prescribed CYP inducer (Table 1)      *CYP inducers or inhibitors were classified as "strong" or "moderate" as specified by the U.S. Food and Drug Administration [14].
DDI Drug-Drug Interaction John´s wort was prescribed once (n=1, 3.8%). Carbamazepine is graded as a strong CYP3A inducer and moderate CYP2C9 inducer, St. John´s wort as a strong CYP3A inducer. Combinations with victim drugs were detected for 56% of the patients who took carbamazepine ( Table 1).

Discussion
This retrospective, naturalistic analysis of data obtained from elderly patients ≥ 65 years revealed that the prevalence of CYPmediated pharmacokinetic DDI is considerably high in a psychiatric setting (10.7%). In particular, 19% of elderly psychiatric patients received a CYP inhibitor or inducer.
A strength of this study was that, to avoid over-alerting [30], only clinically relevant CYP-mediated DDI were considered for analysis. Clinically relevant meant that only "strong" or "moderate" CYP inhibitors or inducers [14] were considered. In this study, a high proportion of ingested CYP inhibitors or inducers were listed as strong (26.2%) or as moderate (73.8%) by the FDA [14]. We chose three sources of CYP inhibitor, inducer and substrate properties of a drug [9,14,29] because they were comprehensive and amongst others based on clinical data. It is concluded that, with this selection-method, an overestimating of the prevalence-rate of CYP-mediated DDI could be avoided.
The comparison of the results in this study with results of other studies [17,[31][32][33] is complicated by the fact that these studies did not consider of the intensity of CYP inhibition or induction. Some studies even used clinical decision support systems to detect CYP-mediated DDI. These systems may help to avoid critical drug combinations, but they tend to produce a large number of irrelevant alerts, in turn leading to over-alerting and causing alert fatigue [13,30]. Finally, different patients´ inclusion criteria and study-settings complicated comparison *CYP substrate properties were considered as specified by the AGNP (Arbeitsgemeinschaft für Neuropsychopharmakologie und Pharmakopsychiatrie) expert group for therapeutic drug monitoring in psychiatry [9] and by experts of the Indiana University [29].  In this study, CYP inhibitors were prescribed to 17.1% of the elderly patients, most frequently duloxetine and melperone. Some studies [24,34] identified melperone as clinically relevant CYP2D6 inhibitor, however, the drug was often neglected in other studies [17,[31][32][33]. As melperone is not approved in the United States, this may explain why the drug is not in the FDA interaction table [14]. It should, however, be considered as a moderate CYP2D6 inhibitor in lists of CYP inhibitors, as melperone is a frequently used drug in Europe.
It is also remarkable that in particular fluoxetine was not prescribed in any of the elderly patients. Furthermore, paroxetine was only prescribed in 8 patients. This may be due to the long and well known CYP2D6-inhibitory properties of these drugs [35]. Therefore, prescription could have been avoided by the physician in these elderly patients. On the other hand, the CYP inhibitory potential of bupropion, duloxetine or melperone and for internistic drugs like omeprazole may be underestimated. In particular, the CYP inhibitory potential of melperone seems likely to be less well known, as a higher rate of DDI (68.3%) could be detected in those patients who took melperone, compared with other CYP inhibitors (40-50%).
As the number of detected CYP inducers was much lower, compared with CYP inhibitors [14], the lower prevalence rate in prescribing inducers (2.1% vs. 17.2%) was expected. The long-established moodstabilizer carbamazepine [36] was the most frequently prescribed CYP inducer in this study. As DDIs were detected in 56% of the patients who took carbamazepine, physicians should consider the CYP inducing potential of carbamazepine, and, if necessary, switch to an alternative mood-stabilizer without CYP inducing potential.
CYP-enzyme inhibitors and inducers were prescribed in this study in 235 (18.9%) of our patients. The number of pharmacokinetic DDI was notably smaller, i.e. 10.7%, since DDI require combinations with victim drugs.
In our sample, 12.6% of the patients were under monotherapy and therefore no CYP-mediated DDI was possible. We included those patients to obtain the overall, real-life prevalence rate of CYP-mediated DDI in the psychiatric setting.
Consistently, Preskorn et al. [35] examined a prevalence rate of CYP-mediated DDI of 11% in veterans affairs who had a prescription for an antidepressant in the preceeding year and a current prescription for at least one systemically active drug. Secondly, Adson et al. [32] examined DDI in 14% of psychiatric patients (17-84 years) seen on a consultative psychiatry service. Contradictory, Guo et al. [17] analyzed from claims data for adult patients with a diagnosis of schizophrenia a prevalence rate of 23% for CYP-mediated DDI. The high prevalence rate could be explained by the fact that other methods of CYP inhibitor, inducer and substrate grouping and selection were chosen. E.g. sertraline and haloperidol were defined as CYP2D6 inhibitor in the study by Guo et al. [17], but not in the present study.
In this study, the most frequently CYP-isoform that was affected was CYP2D6, in accordance with previous studies [17,18]. As some psychotropic drugs, e.g. duloxetine, fluoxetine, paroxetine, melperone and bupropion inhibit CYP2D6 [9,24,35], the major role of CYP2D6 was expectable for the psychiatric setting.
Referring to the previous result, the most frequently detected victim drugs were the CYP2D6 substrates venlafaxine and risperidone.
In accordance, Guo et al. [17] detected that patients with diagnosis of schizophrenia who took the antipsychotic risperidone were most likely to be exposed to an interaction, compared with other psychotropic drugs.
In accordance with a previous study [31], our patients with a detected DDI received significantly more drugs (p<0.001) than patients without a detected DDI (7.9 ± 3.8 vs. 5.0 ± 3.3). In this study, 5.6% of the patients who received 2-4 drugs, 14.1% of the patients who received 5-9 drugs and 24.8% who received ≥ 10 drugs developed a CYP-mediated pharmacokinetic DDI. Doan et al. [33] detected in community hospitalized patients ≥ 65 years receiving five or more drugs a much higher probability of at least one CYP-mediated DDI, with 50% for persons taking 5-9 drugs and 81% taking 10-14 drugs. Amongst others, this could be explained by the fact that the medication profiles of these patients were analyzed with a multidrug cytochromespecific software program that had another selection of CYP inhibitors, inducers and substrates.
A significant gender-difference concerning CYP-mediated DDI was detected in male patients (13.5%), compared with female patients (9.2%). Other studies investigated a higher risk for female patients [8,17], but the mean number of drugs in the gender groups was not considered. After adjusting for the number of drugs dispensed, Doan et al. [33] found no gender differences in the risk of potential CYPmediated DDI. Since the mean number of administered drugs in the present study was similar in females (5.3 ± 3.4) and males (5.3 ± 3.5), the increased risk for a CYP-mediated DDI may be simply explained by the fact that different kind of drug combinations were preferably prescribed for females compared with males.
It has to be mentioned that this study only considered pharmacokinetic DDI related to the CYP450 system. Other pharmacokinetic DDI were not considered for this analysis. As an example pharmacokinetic DDI between lithium and non-steroidal anti-inflammatory drugs [5,37] and pharmacokinetic DDI related to the p-glycoprotein transporter or including the uridine 5-diphosphate glucuronosyltransferases were not considered. Furthermore, as we only analyzed pharmacokinetic interactions between two drugs, smoking status was not considered. Therefore, a much greater overall prevalence rate must be considered for pharmacokinetic DDI in the psychiatric setting.
In addition, it has to be mentioned that some augmentation strategies, for example combination therapy with fluvoxamine and clozapine, are intentional therapeutic DDI [38]. In this study, two patients received a combination therapy with fluvoxamine and clozapine. Furthermore, potentially hazardous drug combinations can be used beneficially if appropriate precautions, e.g. dose adjustment, are occasioned [34]. TDM can help to adapt the serum concentration of a victim drug within the therapeutic reference range [9]. This study has not considered the doses, serum levels and the therapeutic indices of the victim drugs and therefore dose-dependent toxicity. Thus, the clinical relevance of the CYP-mediated DDI remains questionable.
Concerning the clinical relevance of our findings, a considerable number of drug-combinations were identified as high risk combinations. This was suggested, since the combination of these drugs reportedly leads to clinical relevant increase or decrease of drug serum concentration of the victim drug due to the proven high inhibitory or inducing potential of combined drugs [14]. Serum concentrations of victim drugs would have been most informative to evaluate the magnitude of DDI effects. Blood level measurements, however, were done only for some psychotropic drugs. For most drugs, however, such data were not available, especially for non-psychiatric medications. Therefore, we referred to data in the literature and included Table  4 showing clinically relevant DDIs reported by other investigators [24,34,[39][40][41][42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58]. Since drug combinations of Table 4 were similar to the combinations observed by us, we assumed clinical relevance for our findings.
This study had several limitations that need to be considered. Since included patients were selected from a TDM database, a selection bias may be assumed. Patients may not be representative for the ''average'' patient in a standard psychiatric setting. However, according to the AGNP guidelines for TDM in psychiatry [9], age ≥ 65 years is an indication to use TDM. Moreover, a number of indications such as combination treatment with a drug known for its interaction potential or pharmacokinetic comorbidities are specific indications to use TDM. In accordance with this guideline recommendation TDM is highly recommended for elderly psychiatric patients in the participating centers. Patients of this study are therefore regarded as representative for elderly psychiatrically ill patients. This view is supported by the finding that the mean number of drugs of the selected patient sample and the gender distribution was similar to that of a previous study on geriatric psychiatric patients [59].
In addition, this study relies on retrospective data which has the limitation that information on medication may be assumed less reliable than in case of a prospective study. On the other hand, prospective studies may not reflect every day practice of psychopharmacotherapy, since supervision of patients and treating physicians may lead to reduction of the number of combined drugs.

Conclusions
From our findings it may be concluded that the prevalence of CYP-mediated pharmacokinetic drug interactions is high in old aged psychiatric patients. This confirmed our expectation that elderly patients frequently receive strong or moderate CYP inhibitors and to a lesser extent CYP inducers that may give rise to drug interactions when interfering drugs are combined with victim drugs.

Clinical implications
The results of our study provide implications for the practice of health care especially for old aged patients. CYP related pharmacokinetic drug interactions are frequent in this group. Therefore, especially elderly patients require individualized pharmacotherapy with a comprehensive geriatric assessment, limiting the number of drugs prescribed for an elderly patient as few as is necessary and reviewing the use of drugs regularly [6].
Since CYP mediated drug interactions may lead to severe ADRs [34] and since most of these adverse reactions are considered as preventable medication errors, physicians have to be aware about this risk when prescribing combinations of drugs. Thus, efficacy and safety of psychopharmacotherapy could be improved [60]. CYP inhibitor and inducer properties of drugs must be an element of medication management when more than a single drug is given to a patient. Computer programs can be helpful to control of medications for potential DDIs [30]. Available programs, however, are often incomplete, because of the multiple possible drug combinations [30]. Moreover, computerized alerts are often not meaningful [13]. Regarding CYP inhibiting properties, the physician should pay special attention to CYP2D6 inhibitors. Overall, whenever a drug like melperone, duloxetine or carbamazepine with strong or moderate CYP inhibiting or inducing properties (Table 1) is identified in the medication, the treating physician must check if victim drugs are also present ( Table 2). Such combinations should be avoided for safety reasons. If an alternative, non-interacting drug is not available and the critical combination needed for therapy, monitoring of drug concentrations can be helpful to adapt the dosage of the victim drug [9]. If therapeutic drug monitoring guided dose adaptation is not available for the victim drug, careful clinical supervision is mandatory. Using such strategies for drug management and increased awareness of clinically relevant drug interactions will improve the efficacy and safety of pharmacotherapy.

Disclosure of potential conflicts of interest
Christoph Hiemke has received speaker's or consultancy fees from Pfizer Deutschland GmbH, Janssen-Cilag GmbH and Servier Deutschland GmbH, outside the submitted work. He reports no conflict of interest with this publication. Ekkehard Haen reports personal fees from Pfizer Deutschland GmbH, Servier Deutschland GmbH, Novartis Pharma GmbH, and Johnson&Johnson GmbH, outside the submitted work. He reports no conflict of interest with this publication. Gudrun Hefner has received speaker's fee from Servier Deutschland GmbH. She reports no conflict of interest with this publication. All other

CYP inducer/ inhibitor + distinct victim drug
Influence on drug serum concentration Potential clinical consequence Carbamazepine +Quetiapine Decrease of quetiapine serum concentration by 80 to 86% [39][40][41][42] Loss of efficacy +Trazodone Decrease of trazodone serum concentration by 24% and its active metabolite by 40% [43,44]   authors declare no conflicts of interest as well. The research study did not receive funds or support from any source.

Statement of human rights
For this type of study formal consent is not required.