Review Paper

Systematic Overview of Drug Interactions With Antidepressant Medications

Carmine Nieuwstraten, RPh, BScPhm¹, N Renee Labiris, PhD², Anne Holbrook, MD, PharmD, MSc, FRCPC³

Objective:Antidepressants are commonly used drugs with potential for numerous drug interactions. This study aims to systematically review the literature on drug interactions with antidepressants. Method: We searched MEDLINE (1966 to November 2003) and EMBASE (1980 to 2003), using the heading drug interactions combined with individual antidepressant names. We restricted searches to English-language articles and human studies. We screened drug interaction texts and review articles for relevant studies. We included articles reporting original human data on drug interactions with antidepressants commonly used in North America. Articles were independently evaluated by 2 reviewers on clinical effect, clinical significance, and quality of evidence. Discrepancies were resolved by consensus. Results: There were 904 eligible interactions, involving 9509 patients, for a total of 598 summary interactions. Of these, 439 (73%) demonstrated an interaction, 148 (25%) had no effect, and 11 (2%) had conflicting evidence. For 510 interactions (85%), the quality of evidence was poor. It was fair for 67 (11%) interactions and good for 10 (2%) interactions. There were no interactions with excellent quality of evidence. There were 145 (24%) interactions of major clinical significance. These were predominantly hypertensive emergencies and serotonin syndrome. Most interacting drugs had central nervous system (CNS) activity. As expected, monoamine oxidase inhibitors (MAOIs) appear to be the most problematic family in terms of potential for serious drug interactions. Conclusion: Drug interactions with antidepressants are an important cause for concern, but this concern is based primarily on poor evidence. We recommend caution when combining antidepressants with other CNS drugs, particularly when coadministering MAOIs with other substances. (Can J Psychiatry 2006;51:300–316) Click here for author affiliations.  Click here for information on funding and support.  Clinical Implications Physicians should be aware of proposed drug interactions with antidepressant medications, but they should also recognize that the supporting literature is generally of poor quality. Other CNS-active drugs are most commonly reported to interact with antidepressants. MAOIs seem to be associated with major drug interactions more often than other groups of antidepressants. Limitations The literature is of such poor quality that it is difficult to make any high-quality recommendations about definite interactions or lack thereof. Since drug interactions are not currently targeted outcomes in clinical trials and are frequently unrecognized by clinicians, drug interactions are likely underreported. Adverse events attributed to medications are often inadequately assessed for other potential causes. The search was limited to English language reports and did not include all psychiatric resources potentially containing drug interaction reports. Key Words: antidepressant medications, drug interactions, systematic review, case reports, causation, bias, quality of evidence. Résumé : Aperçu systématique des interactions des médicaments avec les antidépresseurs

Depression is the second most common reason for visiting an office-based physician in Canada (1). Antidepressant drugs are commonly prescribed. Data from IMS Health for 2002 noted that antidepressants were the third ranked therapeutic class in global pharmaceutical sales, totaling $17.1 billion and growing at a rate of 5% annually (1). In Canada in 2004, more than 79 million prescriptions for antidepressant drugs were filled at a cost of more than $2.8 billion (1). In addition to treating depressive disorders, antidepressant drugs are increasingly used to treat other illnesses, including anxiety disorders, chronic pain, and eating disorders (2). In many cases, antidepressant drugs are used concurrently with other medications, thus exposing patients to potential harm owing to adverse drug interactions (3).

Drug interactions may be synergistic, antagonistic, or idiosyncratic. Interactions may be further classified as pharmacokinetic or pharmacodynamic. Many of the available antidepressant drugs are inhibitors and (or) substrates of the CYP 450 enzyme system and therefore have potential for pharmacokinetic interactions (3). The potential for pharmacodynamic interactions is also concerning given the range of physiologic systems affected by antidepressant drugs (for example, the genitourinary, cardiovascular, and endocrine systems) (2).

Of the many published reviews of drug interactions involving antidepressants (4–13), none offer a systematic review of the literature, formal evaluation of the quality of evidence, and summary of findings. Many reviews mix data from small trials with individual case reports, data from animal studies with human studies, and clinical with nonclinical outcomes. We previously published a systematic drug interaction evaluation methodology in which we evaluated the quality and clinical relevance of reports of drug interactions involving antidepressants commonly used in North America (14–16).

Methods

Literature Review

We consulted the relevant compendia to compile a list of antidepressant drugs to evaluate for potential drug interactions. On the evaluation, we included 24 antidepressant drugs and one herbal remedy, St John’s wort (Table 1). Although nefazodone was removed from the Canadian market prior to the completion of this review, we included the results as nefazodone remains available in the US and other countries. We searched MEDLINE from 1966 to November 2003, using the MeSH heading drug interactions in combination with the MeSH heading or textword for each individual drug name. The searches were restricted to articles in English and to those involving human studies. We also searched EMBASE from 1980 to November 2003, using the headings drug interactions combined with each individual drug name. We screened bibliographies of 60 review articles for other relevant studies. We checked the indexes of 2 drug-interaction textbooks for all interactions cited for each of the antidepressant drugs, and we retrieved any additional references (17,18). We did not evaluate articles involving animal or in vitro studies, nor did we review articles or interactions involving drugs unavailable for, or commonly used in, current clinical practice.

Drug Interaction Evaluation

All articles were evaluated independently by 2 reviewers. We resolved discrepancies by consensus. In cases where discrepancies were unresolvable, a third reviewer was consulted. Studies that reported positive interactions, such as enhancement of therapeutic response without precipitation of adverse events, were excluded, as were drug overdoses and unintentional drug ingestions. We excluded studies involving interactions with intravenously administered tyramine. Those involving orally administered tyramine were included as those studies may have potential relevance for adverse interactions between food containing tyramine and drugs such as MAOIs.

We evaluated interactions in 3 domains: clinical effect (outcome of interaction), clinical significance (clinical impact), and quality of evidence (rating from study design and sample size).

Clinical Effect

Outcome of Interaction. We classified interaction outcomes as augmentation, inhibition, no effect, or conflicting evidence. To define an interaction as augmentation, the drug effect had to be greater than the sum of the individual drug’s effects. To define an interaction as inhibition, the drug effect had to be less in the presence of another drug or substance than the expected effect of that drug alone. Outcomes in which drugs neither augmented nor inhibited each other’s effects were classified as no effect. Conflicting evidence was used when a particular study or mix of studies did not report a uniform clinical effect but found that certain patients exhibited drug augmentation while others exhibited drug inhibition or no effect.

Clinical Compared With Nonclinical Effect. If there was a clinically evident effect, such as symptomatology or laboratory evidence of a clinical effect (that is, INR), we classified an augmentation or an inhibition outcome as clinical. If there were statistically significant alterations in the pharmacokinetic parameters without evidence of clinical effect, an augmentation or inhibition outcome was classified as nonclinical.

Clinical Significance

From the severity of the outcome, we rated the clinical significance of an interaction as major, moderate, or minor. We classified a clinical outcome causing death or considered life-threatening as major (for example, conditions urgently requiring IV medications to lower blood pressure, intubation, IV fluids or electrolytes, or the application of cooling blankets). We classified a clinical outcome that required medical attention, caused the patient great discomfort, or interfered with a patient’s ability to function in daily activities as moderate. A clinical outcome that was not likely to affect a patient’s usual activities and could be tolerated without seeking medical attention (that is, symptoms causing mild discomfort, such as headache or slight nausea) was classified as minor. Interaction outcomes that were rated nonclinical were classified as having minor clinical significance.

Quality of Evidence

From the methodologic quality of the study and the sample size supporting the interaction (total number of patients in the report), we rated the quality of evidence for each interaction report as excellent, good, fair, or poor. We rated an RCT with more than 100 patients as excellent quality. We rated an RCT with 21 to 100 patients good quality. We considered an RCT with 10 to 20 patients fair quality and an RCT with fewer than 10 patients poor quality. Nonrandomized trials, such as cohort studies, case–control studies, case series, case reports, and pharmacokinetic studies, were all classified as poor quality.

Drug Interaction Summaries

We used a hierarchy of rating for situations in which the conclusions of individual reports differed: 1) when there were both patient-based and healthy volunteer-based studies, the conclusions of the studies involving patients took priority regardless of quality of evidence; 2) when the studies involved the same subject population, we used the interaction conclusion from the article with the highest quality of evidence; 3) for articles with the same quality of evidence and studying the same subject population, we used the conclusion of the one reporting a higher clinical significance; and 4) we gave a conflicting evidence rating to articles of the same quality of evidence and the same subject population that reported different interactions of the same clinical significance.

The summary criteria were inapplicable to 8 drug interactions. In these cases, an overall rating was devised by consensus in consultation with a third reviewer.

SS and NMS

All reports of SS were evaluated separately according to Sternbach’s criteria (19). Interactions were rated as SS only if all criteria were met. If criteria were possibly but not conclusively met, we rated the interaction unclear. If one or more criteria were not met, the interaction was rated no effect. Interactions were also evaluated for NMS according to DSM-IV criteria (20). We rated NMS in the same way as we rated SS.

Results

Twenty-five antidepressant drugs were eligible for the analysis (Table 1). A total of 2541 citations were retrieved and 1478 articles were evaluated (the remainder were excluded for reasons noted in the methods section), resulting in a total of 904 reported interactions involving 9509 patients or volunteers. These were collated into 598 summary interaction reports. An interaction was suggested in 439 cases (73%), no effect in 148 cases (25%), and conflicting evidence in 11cases (2%). Of the 439 interactions, 389 (89%) augmented a drug’s clinical effect, whereas 50 interactions (11%) inhibited a drug’s effect. We classified 92 interactions (21%) as nonclinical. SSRIs and tricyclic antidepressants were involved in 32% and 33% of the interactions, respectively.

We classified 145 interactions (33%) as having major clinical significance, 128 (29%) as having moderate clinical significance, and 166 (38%) as having minor clinical significance. Table 1 summarizes the clinical significance and quality of evidence for each antidepressant drug’s interactions. Tables 2 and 3 summarize the interactions of major clinical significance, excluding SS and NMS. All the major interactions were derived from individual case reports or case series, with only 16 (18%) of the 88 interactions derived from more than one report. Fifteen (10%) of the major interactions involved antidepressant drug combinations. An additional 29 summary interactions (20%) involved nonantidepressant drugs with CNS activity. These included olanzapine, clozapine, chlorpromazine, haloperidol, thioridazine, trifluoperazine, fluphenazine, pimozide, benztropine, lithium, methylphenidate, clonazepam, nitrazepam, ethanol, cyproheptadine, and dextroamphetamine. Other medications involved in major adverse drug interaction combinations included fluconazole, estrogen, prednisone, dextromethorphan, ritonavir, propofol, clarithromycin, erythromycin, pentazocine, interferon, indapamide, altretamine, levodopa, presumed tyramine in foods, disulfiram, atracurium, droperidol, enflurane, isoproterenol, clonidine, propranolol, nifedipine, tinzaparin, warfarin, amiodarone, phenylephrine, simvastatin, irinotecan, chlorpropamide, cyclosporine, propoxyphene, meperidine, selegiline, and theophylline. Phenelzine and tranylcypromine were the most commonly implicated antidepressant drugs in major clinical adverse interactions.

Outcomes of the major interaction reports included death, organ failure, hypertensive emergencies, syncope, SS, delirium, psychosis, suicide attempts, transplant rejection, seizures, rhabdomyolysis, subarachnoid hemorrhage, and related adverse bleeding (Tables 2 and 3).