Spectrum of movement disorders and neurotransmitter abnormalities in paediatric POLG disease

Objectives To describe the spectrum of movement disorders and cerebrospinal fluid (CSF) neurotransmitter profiles in paediatric patients with POLG disease. Methods We identified children with genetically confirmed POLG disease, in whom CSF neurotransmitter analysis had been undertaken. Clinical data were collected retrospectively. CSF neurotransmitter levels were compared to both standardised age-related reference ranges and to non-POLG patients presenting with status epilepticus. Results Forty-one patients with POLG disease were identified. Almost 50% of the patients had documented evidence of a movement disorder, including non-epileptic myoclonus, choreoathetosis and ataxia. CSF neurotransmitter analysis was undertaken in 15 cases and abnormalities were seen in the majority (87%) of cases tested. In many patients, distinctive patterns were evident, including raised neopterin, homovanillic acid and 5-hydroxyindoleacetic acid levels. Conclusions Children with POLG mutations can manifest with a wide spectrum of abnormal movements, which are often prominent features of the clinical syndrome. Underlying pathophysiology is probably multifactorial, and aberrant monoamine metabolism is likely to play a role. Electronic supplementary material The online version of this article (10.1007/s10545-018-0227-7) contains supplementary material, which is available to authorized users.


Introduction
Mitochondrial DNA (mtDNA) depletion syndromes (MDDS) are caused by defects in mtDNA maintenance due to mutations in nuclear genes which affect either mitochondrial deoxyribonucleoside triphosphate supply or components of the mtDNA replication machinery (Rahman and Poulton 2009). DNA polymerase γ (pol γ) is essential for mtDNA replication and repair. Loss-offunction mutations of POLG, encoding the catalytic subunit of pol γ, result in MDDS with evidence of reduced mtDNA content or abnormal mtDNA (multiple mtDNA deletions or point mutations) in affected tissues (Cohen and Naviaux 2010).
POLG-related disease is clinically heterogeneous. In infancy and early childhood, Alpers syndrome (also referred to as Alpers-Huttenlocher syndrome) is the most frequent clinical presentation (Cohen and Naviaux 2010). However, there is a broad phenotypic spectrum, ranging from infantile severe encephalopathy and liver failure to later-onset external ophthalmoplegia, ataxia, myopathy and axonal sensorimotor neuropathy. Epilepsy is a major feature in most cases (Cohen and Naviaux 2010). Movement disorders are commonly described (Morten et al. 2007;Cohen and Naviaux 2010), with parkinsonism most commonly reported in adult patients (Martikainen et al. 2016). In this study, we aimed to describe the clinical spectrum of movement disorders and cerebrospinal fluid (CSF) neurotransmitter profiles in children with POLG mutations.

Patient ascertainment
Paediatric patients (16 years or younger) with confirmed biallelic POLG mutations were retrospectively identified from the Oxford Rare Mitochondrial Disease Service for Adults and Children database, established in 2006. All cases identified between 2006 and 2013 were included in the study. Prior to genetic confirmation, some patients had CSF neurotransmitter analysis as part of routine diagnostic investigation. These patients were identified from the UK CSF Neurotransmitter Service database. Clinical information was ascertained from (i) standardised proformas completed for diagnostic CSF and genetic testing and (ii) patient hospital records, where available (see supplementary data).
For comparative analysis, CSF neurotransmitter profiles of non-POLG patients admitted to a single paediatric intensive care unit (PICU) from August 1999 to November 2011 were reviewed. All patients who had neurotransmitter analysis secondary to non-POLG-related status epilepticus were included in the study.
POLG mutational analysis POLG gene sequencing was performed as previously described ).

CSF metabolite analysis
CSF was collected by lumbar puncture using standardised protocols and neurotransmitters were measured by highperformance liquid chromatography, as previously described (Hyland et al. 1993;Aylett et al. 2013).

Case ascertainment (supplementary data)
In total, 41 paediatric patients with POLG mutations were identified. Twenty of these patients had a documented nonepileptic movement disorder (Tables 1 and 2) and were further studied. The clinical details of eight patients have been published previously (Morten et al. 2007;McCoy et al. 2011;Allen et al. 2014;Rajakulendran et al. 2016;Hikmat et al. 2017).

Genetics
All 20 patients with a movement disorder had biallelic POLG mutations. Of these, 18/20 harboured homozygous/compound heterozygous missense mutations and two cases were compound heterozygotes for missense and nonsense mutations (Table 1).

Age at clinical presentation
The age at neurological presentation ranged from 8 months to 16 years, with 17/20 patients presenting before 24 months of age (median age 13 months).

Clinical features at presentation
Information regarding early clinical features was available for all 20 patients. Encephalopathy and/or status epilepticus was the most common mode of presentation (17/20 cases). Where CSF neurotransmitter analysis had also been performed, 11/15 patients presented either with status epilepticus or epilepsia partialis continua (EPC), preceded by an intercurrent infection in 2/15 cases. The remaining 4/15 patients (D1, D3, D14 and D15) presented initially with a movement disorder, although all eventually developed status epilepticus/EPC in the ensuing weeks or months. Data regarding administered antiepileptic drugs (AEDs) were limited or absent in most cases (Table 1).

Magnetic resonance brain imaging
Many patients had structural abnormalities on brain magnetic resonance imaging (MRI), with bilateral symmetrical thalamic changes evident in 5/14 (Table 1).

CSF analysis
Lumbar puncture was undertaken in 15/20 cases. For most of these patients, CSF neurotransmitter analysis was performed soon (0-4 weeks) after initial neurological presentation. No patient had been administered levodopa prior to CSF sampling. Thirteen of these 15 patients had CSF neurotransmitter abnormalities (Tables 1 and 2). Raised homovanillic acid ( H VA ) w a s s e e n i n 7 / 1 5 a n d a b n o r m a l 5hydroxyindoleacetic acid (5-HIAA) in 8/15 cases (7/15 had high 5-HIAA, 1/15 low 5-HIAA). In fact, 6/15 cases had abnormalities of both HVA and 5-HIAA. Of note, none of the patients were on dopaminergic therapy (including inotropic support) at the time of CSF sampling. Pterin profiles were also frequently abnormal with high neopterin levels in 7/14 patients. 5-Methyltetrahydrofolate levels (5-MTHF), measured in 14 patients, were low in 2/14 cases. 3-Omethyldopa (3-OMD) levels were mildly elevated in 4/8 cases, but not as high as those seen in aromatic L-amino acid decarboxylase (AADC) deficiency (Table 2). Finally, CSF protein and lactate levels were also frequently elevated, where information was available (Table 2); CSF white cell counts were only available in 2/15 patients (D5 and D7) and normal for both cases (data not shown).
In order to determine whether the observed CSF neurotransmitter profiles in POLG patients were disease-specific, we undertook comparative analysis with non-POLG patients who had a similar disease presentation. We identified 1754 paediatric CSF neurotransmitter profiles undertaken between 1999 and 2011 in a single centre. Sixty of 1754 patients underwent CSF analysis during admission to the PICU, of which 15 were for investigation of status epilepticus (Table 2, patients P1-P15). None of these 15 cases were diagnosed with mutations in POLG, although POLG mutations were clinically suspected and subsequently excluded in P6, P7 and P15. A definitive diagnosis was achieved in 6/15 patients (P8-P13). Three of 15 patients (P13-P15) had a suspected or proven mitochondrial disorder, with CSF showing high neopterin levels in 2/3. Additionally, 3/15 patients (P1, P4 and P8) had a suspected or proven central nervous system (CNS) infection, with elevated neopterin in all three cases. Overall, CSF neopterin was elevated in 6/11 cases, where data were available. Two of 15 patients had a raised CSF HVA, one of whom was on dopaminergic therapy, whilst 4/15 had low HVA levels. 5-HIAA levels were abnormal in 5/14 cases (low in 4/14, high in 1/14). CSF 5-MTHF levels, undertaken in 9/15 patients, were low in one patient (P8) ( Table 2). Age-specific (Hyland et al. 1993) CSF HVA and 5-HIAA levels were significantly higher in POLG patients when compared to non-POLG patients (p = 0.001 and p = 0.01, respectively), whereas neopterin levels were similarly elevated in both cohorts (p = 0.68) (Fig. 1).

Discussion
We report the movement disorder semiology and neurotransmitter profiles in children with biallelic POLG mutations. POLG disease has previously been associated with a wide range of movement disorders. In adults and adolescents, ataxia, dystonia, chorea and myoclonus have been described but, overall, parkinsonism seems to be the most commonly encountered motor phenotype (Hinnell et al. 2012;Martikainen et al. 2016). In childhood, choreoathetosis, myoclonus and parkinsonian features have been reported (Morten et al. 2007;Cohen and Naviaux 2010). In our cohort, hyperkinetic motor phenotypes were documented in 20/41 cases, most commonly non-epileptic subcortical myoclonus and choreoathetosis. Ataxia was also We observe that, where CSF neurotransmitter analysis was undertaken, the majority of POLG mutation-positive patients had evidence of abnormal CSF pterin and/or monoamine metabolites. Of these, many (11/15) had an initial presentation of status epilepticus and the majority (12/15) had neurotransmitter analysis performed during a period of increased seizure burden, often whilst in the PICU. Notably, children who presented with a movement disorder in the absence of seizures (patients D1, D3 and D14) had fewer neurotransmitter abnormalities than the POLG status epilepticus group (Table 2). CSF HVA and/or 5-HIAA elevation was evident in 8/15 POLG patients. In fact, CSF monoamine levels were significantly higher in our POLG cohort when compared to those with non-POLG status epilepticus (Fig. 1, Table 2). Similar patterns of HVA and 5-HIAA elevation have been reported previously in a patient with POLG disease (Hasselmann et al. 2010). Importantly, normal HVA:5-HIAA ratios of 1.6-3.4 (normal range 1.0-4.0) (Ng et al. 2015) in all POLG patients discriminate these profiles from other primary neurotransmitter disorders, such as dopamine transporter deficiency syndrome (DTDS), where the HVA:5-HIAA ratios are commonly above 5 (Ng et al. 2015). High levels of HVA and 5-HIAA have also been reported in patients with mtDNA deletions (Pineda et al. 2006). Other mitochondrial diseases are, however, more commonly associated with low HVA and 5-HIAA levels (García-Cazorla et al. 2007;Garcia-Cazorla et al. 2008a), although not as low as in primary neurotransmitter disorders (such as tyrosine hydroxylase or aromatic L-amino acid decarboxylase deficiency), where much lower CSF levels are usually reported (Ng et al. 2015).
Overall, 7/12 POLG patients presenting acutely with seizures or intercurrent infections had high neopterin levels, with levels up to 12 times above the upper limit of the normal reference range (Hyland et al. 1993). Similar neopterin elevation was seen in 6/11 cases of the non-POLG status epilepticus cohort. BH2 and BH4 were also frequently raised in both cohorts, often in tandem with high neopterin levels. High neopterin levels are considered a biochemical marker of inflammation within the CNS and frequently encountered in conditions associated with an exaggerated or aberrant immune response, such as CNS infections, multiple sclerosis and Aicardi-Goutières syndrome (Dale et al. 2009). In keeping with CSF inflammation, CSF protein and/or lactate levels were also high in 9/15 cases, as per previous reports (Cohen and Naviaux 2010). Similar high neopterin levels have previously been reported in a case of POLG disease (Hasselmann et al. 2010). The underlying basis of raised pterin levels in POLG patients is currently unclear,  No definitive diagnosis was achieved for P1-P7, P14 and P15. A mitochondrial disorder was confirmed in P13 and suspected in P14 and P15. Abnormal results are depicted in bold. c values >10% above upper limit of the normal reference range. d >10% below the lower limit of the normal reference range. Reference ranges for protein and lactate measurements are provided by the analysing laboratory but caution in their interpretation is warranted, as studies have indicated that higher age-specific upper limits could also be within the normal range (Leen et al 2012 but it may be related to an immune-mediated response associated with intercurrent infection, frequent seizures at the time of CSF sampling or the underlying disease itself. Two of 14 patients had low CSF 5-MTHF levels, being moderately reduced in one patient (D6) and more markedly reduced in another (D13). Cerebral folate deficiency is reported in several types of mitochondrial disease (Pineda et al. 2006;Garcia-Cazorla et al. 2008b), including POLG mutations (Hasselmann et al. 2010;Rajakulendran et al. 2016), ranging from mild deficiency to more severe forms that can mimic primary folate disorders, such as those due FOLR1 mutations (Cario et al. 2009). The mechanisms underpinning cerebral folate deficiency might include choroid plexus dysfunction, inefficient ATP-dependent transport of folate from blood into the CSF, oxidative stress (Aylett et al. 2013;Rahman 2015) or the presence of blockingtype folate receptor autoantibodies (Hasselmann et al. 2010). Folinic acid treatment sometimes leads to clinical and radiological improvement (Pineda et al. 2006), suggesting a putative link between low CSF 5-MTHF levels and observed phenotypes in these patients (Rahman 2015).
Overall, there seems to be no CSF biomarker that is universally abnormal in POLG patients, at least at disease onset, when CSF is most likely to be obtained; even CSF protein and lactate levels were normal in a few cases (Table 2). However, our results suggest that CSF neurotransmitter analysis might be a helpful tool to herald the possibility of POLG disease in affected patients.
Our study has a number of limitations. Given the retrospective nature of our work, patients were identified as having POLG mutations as part of clinical care and not in the context of a genetic epidemiology study, which may lead to selection bias. However, case identification took place in a nationally commissioned centre performing POLG diagnostic testing; hence, our results are likely to be representative of the paediatric POLG mutation-positive population. Additionally, there was no standardised approach to motor phenotype characterisation while, in some cases, there was insufficient data regarding concurrent AEDs administered, CSF biochemistry, movement disorder semiology and distribution. Furthermore, it is unclear whether the absence of movement disorders in 21/ 41 patients is a true representation or due to under-recognition and/or under-reporting. Regarding CSF biomarkers, we have not examined the neurotransmitter profiles in POLG patients who do not manifest abnormal involuntary movements, and, thus, more studies in this area are warranted. Finally, it is conceivable that whole genome sequencing analysis could help to elucidate the role of additional genetic factors contributing to phenotypic variability in our patient cohort. Overall, despite the above caveats, our findings certainly highlight that POLG disease can be associated with both movement disorders and aberrant CSF neurotransmitter profiles.
The pathophysiology of movement disorders in POLG disease is likely multifactorial. Firstly, previous studies have shown progressive striatonigral degeneration in POLG patients, especially with increasing age (Tzoulis et al. 2016). The early stages of this neurodegenerative process may lead to the abnormal motor phenotypes seen in our cohort. Additionally, the energy-depleted state of POLG disease could render the brain susceptible to acute focal injury triggered by epileptic seizures. The high neopterin levels documented in both POLG patients and controls suggest an acute process common to both groups that may potentially be linked to seizures. However, the high Fig. 1 Cerebrospinal fluid (CSF) neurotransmitter abnormalities in the POLG and non-POLG cohorts. Age-specific homovanillic acid (HVA), 5-hydroxyindoleacetic acid (5-HIAA) and neopterin z-scores in patients with POLG disease (red dots) and non-POLG-related status epilepticus (blue squares) were calculated according to age-related reference ranges (Hyland et al. 1993). Patients on dopaminergic therapy at the time of CSF sample acquisition (patient P8, Table 2) were excluded from this analysis.
The mean values are depicted as horizontal black lines. POLG HVA zscore mean = 1.99 ± 0.56, non-POLG HVA z-score mean = − 0.82 ± 0.46, p = 0.001; POLG 5-HIAA z-score mean = 2.45 ± 0.66, non-POLG 5-HIAA z-score mean = 0.01 ± 0.58, p = 0.01; POLG neopterin z-score mean = 8.71 ± 4.47, non-POLG neopterin z-score mean = 11.23 ± 3.75, p = 0.68. z-Score p-values were calculated using the unpaired t-test. *** = statistically significant (p = 0.001), ** = statistically significant (p = 0.01), ns = not statistically significant (p = 0.68). # = Values from patient P6, who presented with drug-resistant status epilepticus at 5 months of life. Lumbar puncture was performed at 8 months, during an intensive care unit (ICU) admission to manage seizures. POLG mutations and mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS) caused by the common mitochondrial DNA (mtDNA) mutation m.3243A>G were genetically excluded HVA and HIAA levels indicate specific involvement of dopaminergic and serotoninergic systems in the POLG patients but not the controls, and this may underpin the movement abnormalities. Further studies are now warranted in order to investigate whether these high levels are attributed to either increased production of serotonin and dopamine or accelerated monoamine degradation. The raised 3-OMD levels seen in some patients may be indicative of increased L-dopa synthesis. It is also clear that substantia nigra dopaminergic neurons are more vulnerable to defects of mtDNA maintenance than other mtDNA abnormalities (Tzoulis et al. 2016). Therefore, processes other than simple energy depletion or complex 1 deficiency probably underlie their susceptibility. For instance, substantia nigra dopaminergic neurons are specifically vulnerable to defects in mitophagy (a type of mitochondrial quality control) (Narendra et al. 2010), with genetic defects in POLG and Parkin, a key mitophagy protein, exerting synergistic effects in these cells (Pickrell et al. 2015).
In conclusion, hyperkinetic movement disorders are frequently encountered in children with POLG mutations, and may even be the presenting neurological feature, preceding the onset of seizures. Analysis of further cases may allow us to determine the diagnostic utility and biological relevance of observed CSF profiles (raised neopterin/HVA/5-HIAA/3-OMD) in a larger cohort of POLG patients. The mechanisms underpinning movement disorders in POLG disease are not fully understood; however, our report indicates that aberrant dopamine and serotonin metabolism may play a role.