Discussion

Currently, the known genes underlying the pathogenesis of paroxysmal dyskinesias are PRRT2 (phenotype MIM 128200, locus MIM 614386), PNKD (phenotype MIM 118800, locus MIM 609023), SLC2A1/GLUT-1 (phenotype MIM 601042, locus MIM 138140), SCN8A (phenotype MIM 602066, locus MIM 600702), KCNMA1 (phenotype MIM 609446, locus MIM 600150), ECHS1 (phenotype MIM 616277, locus MIM 602292), PDHA1 (phenotype MIM 312170, locus MIM 300502), PDHX (phenotype MIM 245349, locus MIM 608769), DLAT (phenotype MIM 245348, locus MIM 608770), ADCY5 (phenotype MIM 606703, locus MIM 600293), CACNA1A (phenotype MIM 108500 and 141500, locus MIM 601011), GCH1 (phenotype MIM 128230, locus MIM 600225), and SLC16A2 (phenotype MIM 300523, locus MIM 300095).^1,2,3 In our cases, the ATP1A3 variant produces a phenotype similar to PNKD, particularly the wide range of duration and frequency, the lack of sudden movement or exercise as precipitant, and the presence of aggravating factors such as stress and fatigue.^1,2 However, in contrast to PNKD, our patients also presented with interictal mild dystonia, speech arrest, and intellectual disability. Nevertheless, interictal neurological deficits and a wide clinical overlap have been described among subjects with paroxysmal dyskinesias.^1,2 Although a de novo mutation is suspected in our cases, the paternal grandfather who started with late-onset dystonic tremor, tested negative for ATP1A3. The reason for this remains unknown but germinal mosaicism has been recently reported as a mechanism of inheritance in ATP1A3.⁴ Nevertheless, there is still a possibility that the reported variant lacks pathogenicity among these subjects, mainly explained by its absence in their grandfather and responsiveness to the aforementioned treatment.

The large variety of phenotypes due to mutations in the ATP1A3 coupled with clinical features shared with other diseases presents a diagnostic challenge. ATP1A3-related disorders, such as AHC, RDP, and CAPOS syndrome, are just some of the overlapping entities that share common features: abrupt-onset, asymmetric anatomical distribution, and association with triggering factors. However, recent descriptions of diseases with different mutations at the ATP1A3 have shown some similarities to those already known: fever-induced paroxysmal weakness and encephalopathy,⁵ epilepsy with early death and failure to thrive,⁶ and relapsing encephalopathy with cerebellar ataxia,⁷ among others. Other defects associated with less-specific phenotypes are seizures, dystonia, ataxia, and psychiatric conditions.⁸

Pathophysiological mechanisms involved in the development of ATP1A3-related disorders are largely unknown. ATP1A3 encodes the α3 catalytic subunit of the Na⁺/K⁺ ATPase pump, and it is selectively expressed in neurons in the central nervous system. In those expressing α3 mutant alleles, reduced affinity of the Na⁺/K⁺ ATPase for intracellular Na⁺ leads to elevated intracellular Na⁺ concentrations, which in turn can result in an increased influx of Ca²⁺ ions in the cell, via the Na⁺/Ca²⁺ exchange system, with toxic effects and possible liberation of excitatory amino acids. Moreover, since the neuronal uptake of some neurotransmitters relies on a correct Na⁺ gradient, its reduction, due to increased intracellular Na⁺, could functionally impair dopamine uptake, thus leading to dystonia and parkinsonism without degeneration of the nigrostriatal pathway.⁹

A clinical condition similar to PNKD has been described as part of the spectrum of ATP1A3 mutations.¹⁰ Pittock et al.¹¹ described an Irish family with RDP and one of the members (case II-9) showed “intermittent episodes of difficulty speaking, spasms of his arm and hand and unsteadiness in association with stress and anxiety.” This case was catalogued by the authors as an atypical presentation of RDP. Three more subjects depicted mild dystonia in this family but were classified as part of the features of RDP.¹¹ Neither levodopa nor pallidotomy resulted in improvement of the symptoms featured in this family. Rosewich et al.¹² subsequently described 24 subjects with AHC, 23 of whom had dystonia as part of their clinical presentation, in most cases with onset during the first year of life. Choreoathetosis, dysarthria, hypotonia, intellectual disability, and ataxia were also observed in these subjects. Roubergue et al.¹³ described six family members and three had paroxysmal exercise-induced dystonia. Lastly, in a review by Sweeney et al.¹⁴ in 2015, “dystonic spells” are described as part of the clinical criteria for AHC but no mention of paroxysmal dyskinesias is made.

Treatment remains symptomatic. Interestingly, case 2 responded to levodopa while case 1 had an excellent response to GPi DBS. Neither levodopa nor GPi DBS have been previously reported to successfully treat subjects with ATP1A3 mutations.^15,16 Nevertheless, resolution of paroxysmal episodes in our case is in line with the reports of significant and sustained improvement in subjects with PNKD after undergoing bilateral GPi DBS.¹⁷ The reason for improvement through levodopa being limited to one of the two cases remains unclear; however, levodopa has been shown to improve different entities where dystonia coexists: Parkinson’s disease dystonic features, myoclonus dystonia, and dopa-responsive dystonia, among others.^18–20 A milder phenotype in one of the cases could be the reason for responsiveness to levodopa. It seems that a reduced activity at the nigrostriatal dopaminergic system would be responsible for the emergence of dystonia as the causative mechanism.

In conclusion, this is the first description of an ATP1A3 variant causing paroxysmal worsening of dystonia resembling PNKD and status dystonicus in more severe episodes.²¹ Our cases suggest that ATP1A3 mutations should be considered in early onset generalized dystonia especially when combined with intellectual disability and paroxysmal movement disorders.