Valproic acid (VPA) is a first-generation antiepileptic drug (AED). Its effect is directly related to its plasma levels (current reference range: 50-100 μg/mL). Regarding distribution, VPA exhibits saturable plasma protein binding even in the therapeutic range, which may explain the lack of correlation between dose and plasma concentrations1,2. The biotransformation of VPA mainly occurs in the liver and is excreted in urine as glucuronide conjugate. This metabolic pathway is reversible by acyl-peptide hydrolase (APEH), which induces the deconjugation reaction. Carbapenem antibiotics inhibit APEH, promoting the glucuronidation of VPA and its renal clearance. Thus, concomitant treatment with a carbapenem and VPA leads to a rapid decrease in plasma VPA levels and compromises its antiepileptic effectiveness. In fact, previous studies have reported reductions in its plasma concentrations of more than 80%2-7. This interaction is classified by UpTo-Date as risk D (i.e. it is generally recommended that therapy modification be considered)8.

Despite theoretical knowledge of this interaction, there are few published studies analysing its relevance in routine clinical practice7,9.10. However, these studies have not addressed factors inherent to antiepileptic treatment that, in practice, could influence seizure control. These factors include the number of AEDs prescribed for each patient, potential alterations in liver function, or the form of administration of VPA.

Hospital pharmacists are highly competent in the correct management of pharmacotherapy and thus have a relevant professional role in the review of inpatient treatment. Within the pharmaceutical care of inpatients, the review of interactions is included as one of the actions to guarantee the pharmacotherapeutic safety of patients11.

The main objective of this study was to assess the relevance of this interaction in routine clinical practice by quantifying differences in the frequency of seizures experienced by epileptic patients under treatment with VPA when exposed to any carbapenem marketed in Spain. The secondary objectives were to determine which variables are associated with a higher incidence of epileptic seizures and to analyse the impact of pharmaceutical intervention in the prevention of this interaction.

A retrospective observational study conducted in a Group 5 hospital graded according to the Spanish Ministry of Health 2008 DRG classification12. The study was approved by the ethics committee of the hospital in which the study was conducted.

We assessed all patients admitted between January 2016 and July 2020 who had received concomitant treatment with VPA and a carbapenem marketed in Spain (i.e. imipenem/cilastatin, meropenem, or ertapenem). Inclusion criteria were as follows: 18 years or older, a diagnosis of epilepsy with chronic treatment with VPA (i.e. for at least 3 months prior to admission), and the absence of infections or neoplasms involving the central nervous system.

We recorded concomitant pharmacotherapeutic treatment for each patient during the VPA-carbapenem interaction period, and the presence or absence of other interactions that decreased VPA plasma concentrations. These were determined using Lexi-Interact®, with special attention placed on the detection of interactions widely described in the literature2,8.

In order to differentiate interaction-free seizures and interaction-based seizures, two periods were defined: the year prior to admission and the admission period itself, during which period patients were prescribed concomitant treatment with VPA and a carbapenem.

Hospital and primary care electronic medical records were used to determine the pharmacological treatment prescribed and the frequency of seizures during the admission period under study and during the previous year.

If concomitant treatment was detected, pharmaceutical intervention was conducted in writing via the electronic prescription and by telephone to the prescribing physician. Information was provided on the available evidence on the interaction, pharmacokinetic monitoring of VPA was recommended, and, if feasible, changeover from the antibiotic to a safer alternative was proposed10,13. In our centre, the pharmacokinetic monitoring of VPA levels is measured as total plasma concentrations.

Data were recorded and analysed in a pseudoanonymised data collection notebook. Regarding the descriptive analysis, quantitative variables are expressed as medians (interquartile range) and qualitative variables are expressed frequency distributions (%).

The main study objective was assessed by calculating the seizure incidence rates (IR) during the year prior to admission and during the VPA-carbapenem interaction period. The seizure incidence rates were used to calculate the incidence rate ratios (IRR) for the total sample with their respective 95% confidence intervals (95% CI) and compared by calculating the absolute risk difference14. The number needed to harm was also calculated15.

Associations between different pharmacotherapeutic variables and seizure rates were also assessed by calculating the IRs and the IRRs with their respective 95%CIs. To this end, patients were grouped by the carbapenem prescribed and the number of antiepileptic drugs administered. We calculated the IRRs for each subgroup and compared the IRRs associated with the same variable with each other. Cases in which the confidence intervals did not overlap were considered to be statistically significantly different.

We assessed the acceptance and impact of the pharmaceutical intervention by recording the actions taken by the prescribers after they received information and recommendations from the pharmacists. Episodes were distributed according to whether VPA pharmacokinetic monitoring had been initiated or not. For each group, we counted the number of episodes where carbapenem or VPA had been discontinued and the number of episodes in which both prescriptions remained unchanged.

We included 37 episodes in 31 patients (see Table 1 for the main data). None of our patients were treated with imipenem/cilastatin.

It is noteworthy that during the search for interactions other than VPA-carbapenem, we also found one episode involving VPA-darunavir and VPA-ritonavir interactions that may have potentially decreased the plasma concentrations of the antiepileptic. Both could be implicated in a potential reduction of antiepileptic drug levels and are classified by UpToDate as risk C: thus, monitoring is recommended. However, the patient did not experience an increase in the seizure rate during this period.

In 13 episodes (35.1%), pharmacokinetic monitoring of VPA was performed during the drug-combination period. During all episodes, sub-therapeutic levels of VPA were found with a mean total plasma drug concentration of 15.5 ± 12.1 μg/mL.

During the VPA-carbapenem interaction period, the seizure incident rate was 8.56% (8.56 seizures per 100 patient-days), whereas in the previous year, without interaction, it was 3.28%. Thus, the IRR was 2.60 (95% CI: 1.61-4.21), the absolute risk difference was 5.28%, and the number needed to harm was 19 combinations.

In the subgroup analysis, no statistically significant differences were found between patients receiving meropenem and those receiving ertapenem. In contrast, a statistically significant increase in the IRR was found for patients treated with antiepileptic polytherapy (PAE) versus patients treated with VPA as antiepileptic monotherapy (MAE) (Table 2).

The Pharmaceutical Care department for inpatients detected 24 episodes (64.9%) and conducted interventions in all of them. These pharmaceutical interventions prompted clinical decisions that led to the discontinuation of 13 combinations (Figure 1).

Figure 1.

Actions taken by prescribing physicians after pharmaceutical intervention (n = 24). VPA: valproic acid.

(0,06MB).

This study obtained pharmacokinetic results that are in line with those described in the previous literature. We found that the interaction between VPA and meropenem or ertapenem is clinically relevant and potentially severe. These results were obtained not only because the study patients exposed to meropenem or ertapenem had sub-therapeutic plasma concentrations of VPA, but also because there was a significant increase in the seizure episode rate compared to that during the period in which VPA was prescribed without associated carbapenem therapy. Moreover, strategies based on increasing the dose of VPA have been shown to be ineffective in restoring plasma VPA levels7,10,16.

The relevance of the interaction is such that studies are currently underway on the clinical applicability of the use of carbapenems in VPA poisoning17,18. However, more information is needed on their efficacy and safety in this setting.

We found a marked difference in seizure rates between the antiepileptic monotherapy and antiepileptic polytherapy groups. This finding may have been obtained because patients requiring more than one AED have a history of poorer seizure control and their condition is more resistant to pharmacological treatment1,19. This factor is also a limitation since we did not estimate what proportion of the increase in seizures could be related to it. Given that antiepileptic polytherapy is associated with increased seizure rates, we recommend that patients receiving this type of treatment should be closely monitored.

Due to the significant predominance of the use of immediate-release oral forms in the sample, no analysis was performed of the different forms of VPA administration in these patients. It would be of interest to include this variable in future studies, especially to understand the impact of the interaction in patients receiving continuous infusion VPA.

This study was conducted in a single centre with a small sample, which might explain the lack of significant differences by carbapenem administered. Previous studies have found this type of difference and shown that the reduction in VPA levels was less in patients receiving imipenem than in patients receiving meropenem or ertapenem7.

This study is also limited by its retrospective design: for example, the potential omission of seizure episodes prior to admission that may not have been recorded in the clinical history.

This study has demonstrated the relevance of this interaction and its pharmacokinetic impact. Based on these results, we recommend that the use of meropenem or ertapenem in epileptic patients receiving VPA therapy should be avoided unless there is no viable alternative. Furthermore, given the available evidence on interactions with imipenem, we extend this recommendation to the carbapenem group marketed in Spain7,9,16. Previous studies have proposed the use of levofloxacin or piperacillin/tazobactam as an alternative to the use of carbapenems10, although antibiotherapy should always be individualised. If patients need carbapenem therapy, it has been suggested that another antiepileptic is used instead of VPA9,16.

In addition, when the concomitant use of these drugs is detected, it is well established that pharmaceutical intervention should be a priority because it contributes to the discontinuation of the VPA-meropenem or VPAertapenem combination in a considerable proportion of cases.

To maximise the impact of the intervention, we recommend the implementation of early warning systems to assist pharmaceutical validation to avoid inaction in unnoticed cases. However, the optimal approach to this interaction is to avoid the concomitant prescription of potentially interacting drugs, rather than conducting interventions when this situation is detected during pharmaceutical validation20,21. To this end, pharmacists should be integrated in health care teams.

No funding.

No conflict of interests.