Valproate-associated Movement Disorder: A Literature Review

: Valproate (VPA) was first synthesized in 1882, but it was only in the early 1960s that its anticonvulsant properties were discovered. The aim of this literature review is to evaluate the clinical epidemiological profile, pathological mechanisms, and management of VPA-associated movement disorder (MD). Relevant reports in six databases were identified and assessed by two reviewers without language restriction. A total of 138 reports containing 362 cases of subjects who developed a MD secondary to VPA were reported. The MD identified were parkinsonism (PKN) (252), myoclonus (MCL) (54), dystonia (DTN) (17), dyskinesia (DKN) (16), stutters (4), tics (3), akathisia (AKT) (1). In the not clearly defined group, 15 extrapyramidal symptoms, 3 AKT, 2 DTN, 1 rigidity, 1 unstable gait were assessed. The mean and median age was 55.8 (SD: 16.58) and 61 years (range: 4–87 years). The most common VPA-indication was epilepsy, and 51.36% were males. The mean and median time from the VPA start to the MD onset was 32.75 (SD: 30.05) and 21.15 months (range: 1 day – 20 years). The mean and median time from the VPA withdrawal until the MD recovery was 2.89 (SD: 2.79) and 3 months (1 day – 12 months). The most common management was drug withdrawal. A complete recovery was obtained in 80.61%. VPA-associated MD was extensively reported in the literature. PKN was the most well-described. Future studies need to clearly report the clinical history of the patient, considering the full investigation of other adverse events during their entire life.


Introduction
Valproate (VPA), and its pharmacological forms such as valproic acid, sodium valproate, and valproate semisodium are anticonvulsants (Figure 1). In 1882, Beverly Burton synthesized VPA for the first time; this compound was used for almost eighty years as an inert solvent in laboratories (Lempérière, 2001). Pierre Eymard, in the early 1960s, during animal studies to develop a new antiepileptic drug, noted that the substances dissolved in VPA had apparently better anticonvulsant properties (Henry, 2003). After this observation, many clinical studies showed the efficacy and safety of VPA for the management of focal seizures (Brugger et al., 2016). In 1967, it was approved as an antiepileptic drug in France (Henry, 2003). Only in 1983, the Food and Drug Administration approved this medication for the treatment of epilepsy (Lempérière, 2001). The first study assessing the efficacy of VPA in bipolar disorder was done by Lambert et al. at the end of the 1960s in France, soon after the approval for epilepsy, which showed good results, but for many years these data were believed to be incidental, due to the small number of subjects studied (Henry, 2003). About ten years later, German clinical trials followed by North American studies supported the hypothesis of Lambert et al. In 1995, VPA was approved as monotherapy during manic episodes by the FDA (Lempérière, 2001).
The mechanism of action of VPA is not completely understood ( Figure 2) (Lempérière, 2001;Löscher, 2002;Bowden, 2003;Henry, 2003;Brugger et al., 2016). Its main interactions are related to the voltage-gated sodium channels blockage and increased brain levels of gamma-aminobutyric acid (GABA) (Löscher, 2002). The increased concentration of this neurotransmitter is believed to occur due to indirect inhibition of the GABA's reuptake and degradative enzymes. Also, it is worth mentioning that this GABAergic mechanism probably explains the anticonvulsant and antimanic properties of this drug (Bowden, 2003). Other pathways that VPA is related include the Kv7.2, AKAP5, and histone deacetylase (Löscher, 2002). It is hypothesized that the inhibition of the histone deacetylase may have neuroprotective effects due to the increased uncoiling of DNA promoting more transcriptional activity of chromatin structures (Bowden, 2003).
The adverse effects of this medication that affect more than ten percent of users are nausea, vomiting, headache, coagulation disorders (Rissardo et al., 2019), alopecia, asthenia, somnolence, amblyopia, diarrhea, dizziness, dyspepsia, nystagmus, and tinnitus (Bowden, 2003). In the label of VPA, there is a black box warning about hepatotoxicity in susceptible individuals (those with mitochondrial diseases), teratogenicity, and pancreatitis (Löscher, 2002). Other common side effects secondary to VPA are movement disorders (MD) such as tremor and ataxia, which can significantly impact the quality of life of an important percentage of the VPA users. Moreover, these abnormal movements are challenging to diagnose and manage in the clinical practice, because the majority of affected individuals have a pre-existing psychiatric or neurologic comorbidity.
In the literature, there are few reviews about VPA and MD that were not focused solely on tremors. To be more specific, we found two reviews about VPA-induced parkinsonism (PKN). Mahmoud and Tampi published a study about this topic in 2011; they objectively selected elderly patients, and a total of thirteen case reports were analysed. In 2016, Brugger et al. searched on four databases for papers in English about VPA and PKN, a total of 116 patients were evaluated; their purpose was to discuss the possible hypotheses for these adverse effects. In this context, the aim of the present literature review is to evaluate the clinic-epidemiological profile, pathological mechanisms, and management of VPA-associated MD.  of all papers found in the initial search. Disagreements between the authors were resolved through discussion. Cases where the cause of MD was already known and either motor symptoms did not worsen or were not related to VPA were excluded. Also, cases that were not accessible by electronic methods, even after a formal request to the authors (by e-mail) were excluded. Reports that had more than one factor contributing to the MD were evaluated by the probability of occurrence of the event based on the Naranjo algorithm.

Data extraction
For VPA a total of 6,443 papers were found; 5,279 were irrelevant and 1,026 were unrelated to the complication, duplicate, inaccessible electronically, or provided insufficient data ( Figure 3). Data abstraction was performed. When provided, we extracted author, department, year of publication, country of origin, number of patients affected, VPA indication including off-label uses, time from first VPA-dose until MD onset, time from VPA withdrawal or management to symptoms improvement, patient's status at the last follow-up, and important findings of clinical history and management. The majority of the reports did not provide specific

147)
VPA-associated MD Prague Medical Report / Vol. 122 (2021) No. 3, p. 140-180 information about the times of MD onset and recovery. Data were extracted by two independent authors, double-checked to ensure matching, and organized by whether or not the MD was a side effect of VPA use.

Statistical analysis
Categorical variables were represented as proportions; continuous variables were represented as mean, standard deviation (SD), median, and range.

Definitions
The clinical characteristics and definitions of the MDs such as parkinsonism, tics, dyskinesia, dystonia, myoclonus, restless legs syndrome, akathisia, tremor, chorea, ataxia, and ballism were obtained from the reference Jankovic and Tolosa (2007). The clinical diagnosis for the psychiatric conditions was obtained from the Diagnostic and Statistical Manual of Mental Disorders (DSM-5 ® ) (American Psychiatric Association, 2013). The Naranjo algorithm was used for determining the likelihood of whether an adverse drug reaction was actually due to the drug rather than the result of other factors (Naranjo et al., 1981). In the cases where the non-English literature was beyond the authors' proficiency (English, Portuguese, Spanish, Italian, French, and German) and the English abstract did not provide enough data, such as Japanese, Korean, Chinese, Russian, and Dutch, Google Translate service was used (De Vries et al., 2018).

Results
For the years 1975 to 2019, a total of 138 reports containing 362 cases, from thirtythree countries, of individuals who developed a movement disorder (MD) secondary to valproate (VPA) were reported (Table 2). Figure 4 shows the number of reports     palsy rating scale at 12 months was significantly higher in the VPA than in the placebo group but was similar between the two groups at 24 months. According to the authors' this strongly suggests poor tolerability due to side effects than permanent neurological damage.
Wang et al.
The summary data about VPA-associated MD is provided in Table 3. Herein, we will describe the general data of all clearly defined cases.
The mean and median time from starting VPA use to the MD onset was 32.75 (SD: 30.05) and 21.15 months (MD onset time range: 1 day -20 years), respectively. About 75% of the individual had abnormal movement within 50 months of the VPA treatment. The mean and median time from the VPA withdrawal until the MD recovery was 2.89 (SD: 2.79) and 3 months (MD recovery time range: 1 day -12 months), respectively. In the subgroup of subjects that had improvement of the symptoms, the complete recovery was achieved within 9 months of the drug withdrawal in almost all cases (99%). Figure 5 shows a comparison between the percentage of patients who developed a MD since the beginning of the treatment and the percentage of patients recovering after drug withdrawal when outliers were removed. The most common management was drug withdrawal. Other options were the VPA-dose reduction, replacement of the drug probably interacting with VPA, and the prescription of other drugs after the VPA discontinuation, such as levodopa, benztropine, benzodiazepines, biperiden, haloperidol, and diphenhydramine. In addition, the replacement of VPA in tablet form for the same dosage in sprinkles presentation was sufficient to improve the symptoms in one case. A complete recovery was observed in 80.61% of the patients (158/196).

Discussion
General VPA-associated MD was widely reported in the literature. We believe that the availability, costs, and some historical factors of VPA probably had contributed to this. VPA is among the safest and most effective medicines needed in a health system, as attested by the World Health Organization's List of Essential Medicines, and it is marketed in the majority of countries. Also, VPA was the 126 th most prescribed medication in the USA with almost six million prescriptions in 2017 (ClinCalc, 2020). Furthermore, the well-known description of flunarizine and cinnarizine developing PKN in 1984, promoted the awareness of the drug-induced MD resulting in an increasing number of reports about all abnormal movements secondary to medications including those associated with VPA (Teive et al., 2004b).
Based on the data available in Table 2, we can hypothetically illustrate a case. A middle-aged European male with poorly controlled epilepsy resorts to his neurologist. VPA 250 mg with a gradual increase until five-six tablets a day was prescribed. Within three years, the patient started complaining of stiffness, rigidity, and resting tremor; neurological examination revealed bradykinesia, and a diagnosis of PKN secondary to VPA was done. VPA was established and lamotrigine or carbamazepine was started. In the follow-up after three months, the patient had a full recovery and was able to walk without assistance and the tremor disappeared.
The majority of the incidences of abnormal movements associated with VPA are not well described in the literature. Table 3 is a summary of the percentages of some abnormal movements secondary to VPA (Anthony, 1977;Bowdle et al., 1979;Friis et al., 1983;Zaccara et al., 1984;van der Zwan, 1989;Armon et al., 1996;Nouzeilles et al., 1999;Easterford et al., 2004;Ristić et al., 2006;Jamora et al., 2007;Zadikoff et al., 2007;Lance and;Leclair-Visonneau et al., 2016;Makhlouf et al., 2018;Baizabal-Carvallo and Alonso-Juarez, 2021); the data was extracted from the clinical trials and population-based studies that provide sufficient data for Table 2. The incidences of VPA-associated abnormal movements extensively vary throughout the literature. For example, VPA-induced PKN was observed from 1.37 to 75% of the individuals.
Herein, we would like to discuss some of the MDs in subtopics to allow a better comprehension of the data.

167)
VPA-associated MD Prague Medical Report / Vol. 122 (2021) No. 3, p. 140-180 Parkinsonism (PKN) History In 1979, Lautin et al. reported the first case of VPA-induced PKN. They described a middle-aged male who was prescribed VPA 1,000 mg for schizophrenia; four days later, the patient complained of PKN symptoms. Benztropine and trihexyphenidyl were started, but the symptoms did not alleviate. Only when VPA was withdrawn the patient had a full recovery. Also, the individual had a previous history of PKN secondary to metoclopramide and haloperidol. Therefore, we believe that this has contributed significantly to the literature because a similar presentation of VPA and antidopaminergic drugs in the same individual suggested a common neuronal pathway associated with extrapyramidal symptoms of these two drug classes. It is worth mentioning that in the same year Nutt et al. (1979) published the cases of four individuals with Parkinson's disease with worsening gait and resting tremors that were using VPA.

Epidemiology
The incidence of PKN following VPA use found in the literature was 1.37, 1. 60, 2.27, 5.04, 6.00, 10.16, 10.71, 73.33, and 75.00% (Table 4). The majority of the individuals reported were males, the mean age was 60.87 years, the mean VPAdose was 1,315 mg. The time since starting VPA until MD onset, and the time until resolution after VPA discontinuation were 3.38 years and 3.41 months, respectively. When we compare the present study with the Brugger et al. (2016); the main differences encountered are that the present work has assessed a greater number of patients (252 vs. 116), of which the majority was male (52.86% vs. 41.4%); interestingly, the findings of Brugger et al. (2016) for mean age (63.5 years), VPA main indication (epilepsy) and median VPA dose (1,250 mg) were almost identical to those already described in this revision.

Presentation and clinical diagnosis
The presentation in the majority of the cases was a symmetric akinetic-rigid syndrome, with predominant postural/action over the resting tremor. Sometimes signs and symptoms of cognitive impairment were observed. The severity of the clinical presentation ranged from mild to severe with loss of physical independence. Some patients had pre-existing diseases other than the indication for VPA prescription such as Parkinson's disease, multiple system atrophy, progressive supranuclear palsy, Huntington's disease, systemic lupus erythematosus, and some brain damage. Interestingly, multiple system atrophy-like and progressive supranuclear palsy-like syndromes were reported as the presenting symptoms. A clear distinction between VPA-induced MD and idiopathic Parkinson's disease based only on clinical criteria is challenging in clinical practice. Therefore, we proposed some clinical tools to help with the diagnosis of this syndrome (Table 5).

Pathophysiological mechanism
In the literature, we found five possible pathophysiological mechanisms to explains the VPA-induced PKN ( Figure 6) (Brugger et al., 2016). First, the VPA can increase the concentration of GABA (Löscher, 2002), which inhibits the globus pallidus connections with the thalamus, decreasing activity of the direct pathway. Second, another effect of VPA is the inhibition of the histone deacetylase that may increase the expression of some genes and decrease others such as those involved in the synaptic transmission (Löscher, 2002), which was already suggested in cell studies. Third, most affected individuals were elderly, so they may already have an imbalance of dopaminergic and cholinergic activity, and when VPA is used, a decrease in dopamine occurs, favoring the indirect pathway (Sawle et al., 1990). Fourth, VPA can, in a normal concentration of neurotransmitters (balanced state), affect mitochondrial enzymes causing cellular energy deficiency, what increases the likelihood of an oxidative stress and consequently a neurodegenerative process, especially in the dopaminergic system (Löscher, 2002); a supporting fact for this theory is that individuals with particular mitochondrial lesions are more susceptible to the development of side effects related to VPA (Henry, 2003). Fifth, the diagnosis of Parkinson's disease in a significant percentage of the patients cannot be ruled out; as a result, perhaps the use of VPA was only by chance present in these individuals, who may develop uncorrelated Parkinson's disease.

Management
The most commonly reported management was the VPA discontinuation, adopted in more than seventy-five percent of the individuals. In some cases, dopamine precursors were attempted to manage and a partial improvement of the PKN symptoms was achieved; in the follow-up, levodopa showed to be effective and reduced the recovery time, but apparently these cases were after diagnosed with Parkinson's disease. One individual received bromocriptine, but no details were provided regarding treatment response. The VPA-induced PKN had the secondworst prognosis, full recovery was obtained in 77.02% of the subjects; about 10% of subjects only had partial improvement of the symptoms, with permanence of at least one symptom even after the last follow-up.

Myoclonus (MCL)
MCL was the first VPA-associated MD identified and was the second most commonly reported in the literature. The incidence of MCL related to VPA use found in the literature was 1.66-5.26% (Table 4). MCL-individuals were approximately twenty years younger than those affected by PKN, also, the VPA dose was lower, and MD onset and recovery happened sooner than in general data. The majority of the subjects involved were female (3:2). The presentation was asterixis and multifocal MCL. The MCL source was cortical and subcortical. It is worth mentioning that an important percentage of the cases only describe the neurological examination, giving the diagnosis without providing the findings of the electrodiagnostic studies. The management was drug withdrawal or the reduction of VPA-dose. A feature reported in an important percentage of the patients, and possibly related to the mechanism of VPA-induced MCL, is the high serum concentrations of ammonia, with no sign of liver failure described in the eight individuals assessed. Therefore, some authors believe that the explanation for MCL in the group of hyperammonemic individuals is the decrease of inhibitory neurotransmitters caused by ammonia, turning the individuals more susceptible to the development of MCL (Campostrini et al., 1983;Gastaut and Mege, 1985). On the other hand, the possibility that the ammonia levels found in these cases may be incidental cannot be excluded, as clinical trials with VPA already reported higher levels of this compound in individuals without any complaint (Löscher, 2002;Bowden, 2003). Also, this hypothesis can support the idea that perhaps VPA action on the central nervous system may lead to the development of MCL. Moreover, we hypothesized that the mechanism behind VPA-induced MCL is probably related to VPA interaction with serotonin. In rat models, VPA caused both increase and decrease in serotonin concentration, depending on the site of action (Baf et al., 1994).

Dystonia (DTN)
In the DTN group, the data obtained for doses and times until onset and recovery from MD are comparable to general data on drug-induced DTN found in literature. Dick and Saunders (1980) probably described the first case of VPA-associated DTN. They reported the case of an individual with cervical and axial DTN that they attempted to treat with VPA, and the DTN-symptoms worsened.
The presentation in descending order of frequency was axial, cervical, oromandibular, blepharospasm, status dystonicus, and spasmodic dysphonia. We included the spasmodic dysphonia in the DTN group, but some authors believe that this disorder is a different entity, which goes beyond the aim of this review (Oh et al., 2004). In the same way, dropped head syndrome commonly reported with anticonvulsants may or may not be related to DTN (Werner et al., 2006). Possible interactions with clozapine, risperidone, quetiapine, and butamirate citrate were described.
One of the possible assumptions to explain the VPA-induced DTN is based on the GABAergic neurotransmission (Löscher, 2002). We believe that due to increased GABA levels by VPA the direct and indirect pathways that go to the thalamus might be interrupted. But the indirect pathway subactivity could probably predominate, and this disruption can increase the thalamocortical drive and eventually lead to DTN . Another hypothesis related to dopaminergic activity in a mechanism similar to that proposed for the VPA-induced PKN can also be assumed (Brugger et al., 2016). Friis et al. (1983) reported in 53.33% of the individuals receiving VPA the development of DKN (Table 4). The presentation was orofacial, choreiform, hemichoreiform, and choreoathetotic. The association with another MD was observed with axial DTN and multifocal MCL. The DKN, stutter, and tics had the best prognosis with 100% recovery after the management.

Dyskinesia (DKN)
The effects of VPA in the dopaminergic system probably explain the VPA-induced DKN. One fact that can support this hypothesis is the long time from starting VPA treatment until the MD onset, which was after 2.21 years. It is believed that due to the dopamine blockage, antipsychotics trigger inflammatory processes and the release of reactive oxygen species causing abnormal adaptations of the striatal organization, and ultimately leading to overactivation of the direct pathway (Lepping et al., 2011).
The most frequent management was VPA withdrawal. Another option was the VPA-dose reduction in those individuals that a possible interaction related to protein intake or other medications was assumed. Moreover, Lancman et al. (1994) reported in one subject the substitution from VPA tablet to sprinkles improving the symptoms. A possible relation with the higher VPA plasma concentration and the DKN occurrence can be proposed in this case, since the sprinkles have their peak within four hours and other VPA formulation in one hour, especially the syrup (Cloyd et al., 1992).

Stutter, tic, and akathisia (AKT)
In our analysis, we included stutter because of the possible differential diagnosis of DTN, MCL, and even DKN due to poor description of the neurological examination. It was observed only in young adult males with bipolar disorder. The MD's times of onset and recovery were the shortest, and the VPA-dose was the lowest reported in relation to the general data. These features can support the assumption of possible DTN diagnosis. Also, the prognosis was excellent, with 100% recovery. The most effective treatment was the drug withdrawal. Mukherjee et al. (2015) attempted the VPA reintroduction, which caused the reappearance of the symptoms. Thus, in the VPA-induced stuttering, we believe that the rechallenge of VPA should not be done.
Tics were observed in three individuals and corresponded to less than one percent of VPA-induced MD reports. In the data extraction of the Zadikoff et al. (2007) study, the percentage of individuals developing tics with VPA use was 1.69%. The patients presented with motor, motor and phonic, or only excessive eye blinking tics. Two cases reported had possible drug interactions, so a clear association can only be suspected. The drugs interacting with VPA were ziprasidone and lamotrigine; VPA can increase the levels of ziprasidone/lamotrigine by decreasing their metabolism (Alonso-Navarro et al., 2007;Thome-Souza et al., 2012). The VPA-dose decrease was enough for the achievement of a full recovery in the reports.
The less frequent MD published in the literature in association with VPA was AKT. But, it is noteworthy that this does not represent clinical practice. Friis et al. (1983) assessed 15 individuals using VPA, all of them developed some degree of AKT. Clos (2001) reported the case of a young adult female with bipolar disorder who was prescribed VPA 1,200 mg, and within one month she developed AKT-symptoms, but the diagnosis may be doubtful because the individual was in concomitant use of lithium.

Conclusion
In sum, VPA-associated MD was extensively reported in the literature probably due to availability, costs, and some historical factors of VPA. The most frequent and well described MD was PKN. In descending order of frequency, the following MD related to VPA were encountered: PKN > MCL > DTN > DKN > Stutter > Tic > AKT. Further studies are warranted to elucidate the occurrence of these MD associated with VPA and its underlying pathophysiology. Future reports need to clearly describe the clinical history of the patient considering a full investigation of other adverse events during their entire life as well as a long-term follow-up. We believe that the knowledge of VPA-associated MD raises the awareness about MD, and especially those drug-induced, which sometimes are challenging in the clinical practice to diagnose and manage.