The Frequency of Reversible Parkinsonism and Cognitive Decline Associated with Valproate Treatment: A Study of 364 Patients with Different Types of Epilepsy


Address correspondence and reprint requests to Dr. D. Sokić at Institute of Neurology, CCS, Dr. Subotica 6, 11000 Belgrade, Serbia. E-mail: dsokic@sezampro.yu


Summary: Purpose: We report the frequency of parkinsonism and cognitive decline (P/CD) in patients treated with valproate (VPA) after 1 year of treatment and at least 1 year of follow-up.

Methods: Three hundred sixty-four patients with various epileptic syndromes and seizure types were treated with VPA mono- or polytherapy for more than 1 year.

Results: We found five cases of P/CD (1.37%; 95% CI, 0.18–2.56%). Among 140 patients with different adverse effects (AEs) of VPA, P/CD were among the rarest in frequency but significant in terms of drug discontinuation (five of 17).

Conclusions: Early identification of this type of AE and discontinuation of the drug led to complete recovery in affected patients.

Valproate (VPA) is considered to have a good safety profile. Sporadic reports and a few case series or clinical studies of reversible syndrome of parkinsonism and cognitive decline (P/CD) have been made in patients treated with VPA (Zaret and Cohen, 1986; Armon et al., 1996; Easterford et al., 2004; Masmoudi et al, 2006). The prevalence of this type of treatment complication has been estimated at 6% in a smaller group of consecutively recruited patients (Easterford et al., 2004). We report the frequency of P/CD in patients treated with VPA after 1 year of treatment and with ≥1 year of follow-up.



In total, 364 patients (55.2% women) aged between 17 and 73 years were treated with VPA mono- (37.6%) or polytherapy (VPA as comedication with enzyme-inducing anticonvulsants, 26.3%; VPA with lamotrigine, 22.5%; and VPA with other anticonvulsants, 13.4%) for >1 year were prospectively included in this study (Table 1). Of these, 110 patients had generalized and 254 had partial epileptic syndromes. Patients were recruited from the Epilepsy Department of the Institute of Neurology, Clinical Center of Serbia, during a 9-year period. Compliance was tested at least once in all patients (VPA plasma level in the range of 50–100 mg/L). In total, 429 patients treated with VPA for <1 year (in most cases, <3 months) were not included in the study. CT/MRI studies were performed in all patients. Other exclusion criteria included progressive neurologic conditions, exposure to drugs that produced extrapyramidal side effects 2 years before commencement of the study, neuroradiologic findings suggesting small blood vessel disease of the brain, and a positive heredo-degenerative history in first-degree relatives (16 patients). All patients signed informed consents, approved by the Ethical Committee.

Table 1. Demographic data (all patients and patients with adverse effects)
 PercentageMedian age (yr)Duration of epilepsy (yr)Average dose of VPADuration of therapy (yr)
Total no. of patients (n = 364)10025.08.81,352 mg3.3
(95% CI, 24.0–27.0)(95% CI, 6.7–11.1)(95% CI, 1,286.2–1,418.9)(95% CI, 2.9–3.6)  
Patients with38.425.09.91,507 mg3.2
  adverse effects(95% CI, 33.5–43.3)(95% CI, 24.0–27.0) (95% CI, 8.57–11.39)(95% CI 1,408.1–1,627.7)(95% CI, 2.80–3.74)
Mann–Whitney test p = 0.74p = 0.21p = 0.14p = 0.44

Patient evaluation

Patients were followed in 6- to 12-month scheduled visits (mean period of follow-up was 4.2 ± 3.2 years). Mini Mental State Examination (MMSE) and neurologic examination were performed on each visit. All adverse effects (AEs) were registered. Parkinsonism was defined as an associated syndrome with the presence of at least three of these: bradykinesia, resting tremor, rigidity/cogwheel, and impairment of posture. Cognitive decline was indicated by an MMSE score of ≤24. Patients with parkinsonism were further assessed by A.J.R. and D.S. by using the Unified Parkinson Disease Rating Scale (UPDRS), parts I through III. Video footage of all patients with P/CD, routine laboratory tests, thyroid hormone screening, EEG, antiepileptic drug (AED) plasma levels, and brain CT/MRIs were taken at the beginning and ≥3 months after discontinuation of VPA treatment.


We identified five patients with parkinsonism (1.37%; 95% CI, 0.18–2.56%) that developed during 2.5–10 months of VPA therapy (Table 2). In all except two patients (patients 4 and 5), all four signs determining parkinsonism were present. Mild to moderate cognitive impairment was registered in three patients with parkinsonism (patients 1–3), including psychomotor slowness, apathy, and impairment in memory and previously gained knowledge recall. Laboratory tests, vitamin B12, thyroid status, and VPA plasma level were within reference limits.

Table 2. Demographics and clinical features of patients with reversible P/CD
Patients/sex1/female2/female3/female4/female5/maleAverage (95% CI)
  1. VPA, valproate; LTG, lamotrigine; CBZ, carbamazepine; T, tremor; R, rigidity; B, bradykinesia; IP, impairment of posture; UPDRS, Unified Parkinson Disease Rating Scale; MMSE, Mini Mental State Examination; GTCS, generalized tonic–clonic seizure; CPS, complex partial seizure; SBA, slow background activity; AS, atonic seizure; AAS, atypical absence seizure.

Age575764655559.9 (53.9–65.2)
Epilepsy duration (yr)30170.589.3 (5.6–24.2)
VPA treatment duration/daily dose/comedication (dose)4 yr/1,250 mg1.5 yr/1,500 mg/1.5 yr/500 mg/0.5 yr/1,000 mg0.5 yr/1,250 mg1.5 yr (0.31–3.31)/
 LTG (200 mg) CBZ (800 mg)  1,100 mg(629.2–1570.7)
VPA treatmentMonotherapyPolytherapyPolytherapyMonotherapyMonotherapy 
Symptoms duration (mo)10862.54 
Clinical featuresT/R/B/IPT/R/B/IPT/R/B/IPT/R/BT/R/B 
UPDRS I, II, III (total)7, 19, 48 (74)6, 8, 32 (46)4, 9, 35 (48)2, 3, 19 (24)1, 3, 16 (20) 
Brain CTSubcortical atrophyNormalCortical atrophyNormalCortical atrophy 

In all patients with P/CD, VPA treatment was discontinued and replaced with another AED (carbamazepine, lamotrigine, or topiramate). Improvements on quantitative scale measures (UPDRS/MMSE) were observed in all patients between 9 and 24 weeks after discontinuation, most obvious in patient 1 (39/27), patient 2 (20/30), and patient 3 (18/30), and less noticeably in slightly affected patient 4 (17/28) and patient 5 (15/30). We discerned no significant changes on brain CT/MRI findings of the affected patients during follow-up. Slow background EEG activity, identified in two patients, resolved after discontinuation of VPA. Clinical improvement in patients no longer using VPA was confirmed by a specialist in movement disorders through video analysis performed in blinded fashion retrospectively.

AEs occurred in 140 patients (31.6% of patients initially treated with VPA and in 41.4% treated with add-on VPA) (Table 1); 75% of patients experienced AE after >8 months of VPA therapy. We found no statistically significant difference between groups with and without AEs when comparing demographic parameters (Mann–Whitney test for age, duration of epilepsy and duration of VPA treatment).

The most frequent AE was tremor (20%); these patients had no symptoms of P/CD. Other registered AEs were hair loss (8.7%), weight gain (8.4%), gastrointestinal symptoms (6.5%), ataxia (5%), and menstrual disorders (3.7%). Thrombocytopenia and elevation of liver enzymes each occurred in four cases. Treatment needed to be terminated in a total of 17 (4.6%) cases. With the exception of patients with reversible P/CD, treatment was discontinued in six patients because of various and simultaneous AEs; in three because of isolated ataxia, in two because of severe tremor, and in one because of toxic hepatitis.


A previous study found a 10-fold increase in the incidence of parkinsonism in people taking VPA (Easterford et al., 2004) when compared with community-based studies of the prevalence and incidence of Parkinson's disease (PD) and parkinsonism in the population of London (Schrag and Quinn, 2000). Results in our study would support the lower level of 95% CI (2%) in the aforementioned published data (Easterford et al., 2004), and probably represent a more accurate estimation of the frequency of this kind of AE in chronic VPA therapy because of a larger sample. However, even in our data, a 95% CI range leaves room for a more precise determination of the rate of occurrence of P/CD with prolonged VPA use. Epilepsy clinic settings may further contaminate results on account of the larger number of pharmacoresistant epilepsies. Again, the high prevalence of P/CD reported in other studies seems unlikely and presumably may represent bias because these studies were not designed to address a particular prevalence issue (Armon et al., 1996).

Strict exclusion criteria diminish the likelihood of the overall possibility of parkinsonism or dementia owing to reasons other than VPA toxicity. We have been impressed with the association of postural instability and higher UPDRS scores with more-significant cognitive impairment in our patients. Dementia is a well-recognized clinical feature in advanced stages of PD (Dubois and Pillon, 1997), and follows a mostly subcortical pattern, as seen in our cognitively impaired patients. Furthermore, postural reflex impairment is also an attribute of PD progression (Martilla and Rinne, 1977), and with dementia, suggests damage in multiple neurotransmitters systems in CNS. The absence of a levodopa response and normal radioligand neuroimaging in earlier studies (Easterford et al., 2004) further supports a hypothesis of probable dopaminergic neuron sparing and widespread damage of nondopaminergic systems not equally sensitive to the toxic nature of VPA. This damage—with vaguely defined pathophysiologic mechanisms—is reversible within a few weeks or months of VPA discontinuation (Onofrj et al., 1998).

We have not, however, observed a total regression of symptoms and signs in all our patients during follow-up. This may be due to insufficient duration of follow-up or could point to certain irreversible processes that may begin and go clinically unnoticed at any time during VPA treatment. It could be argued that factors other than VPA may contribute to the initiation and maintenance of neuronal damage.

Chronic toxicity, despite therapeutic plasma ranges of VPA, is well known (Perucca, 2000). Consequently, our finding of the late appearance of AE was not surprising. Because of the limited number of patients with P/CD in our group, it is hard to estimate the contribution of add-on therapy to this particular AE.

Although several younger patients with P/CD caused by VPA treatment were described (Alvarez-Gomez et al., 1993), in our study, we observed a tendency toward P/CD in older patients as a result of prolonged VPA treatment. Although a smaller number of affected patients (five) would render statistical analysis invalid when compared with the entire sample, this may suggest a greater vulnerability in elderly patients and would support a more cautious administration of the drug to this population. Aberrant metabolism of the drug in elderly patients, production of toxic metabolites, or deranged distribution of the drug may be possible explanations.

Improvement is most obvious in most severely affected patients, but complete recovery is likely in early-recognized patients with lower UPDRS scores. P/CD remains one of rarest yet most severe AEs during VPA treatment (Alvarez-Gomez et al., 1993) and is probably fully reversible. However, it can be subtle, and thus escape detection unless the clinician is aware of it. P/CD was the reason for drug discontinuation in almost a third of our patients who had AEs due to VPA, necessitating discontinuation of the drug.