Clinical Experience of Neurological Mitochondrial Diseases in Children and Adults: A Single-Center Study

It is still a challange to diagnose a primary mitochondrial disease (MD) in a clinical setting. In general, MDs are characterized as a dysfunction of oxidative phosphorylation (OXPHOS) and/or enzymes involved in the mitochondrial role in the oxidation of sugars, amino acids and fats to water and carbon dioxide. Primary MD is a genetic and/or a biochemical defect of OXPHOS. Mitochondrial diseases are the most commonly inherited metabolic diseases in neurology, both in children and in adults. The minimum prevalence rate for mitochondrial DNA (mtDNA), mutations was one in 5000 (20 per 100,000) and for nuclear mutations was 2.9 per 100,000 adults in Northeast England [1]. Childhood MD prevalence has been estimated at 3.0–6.2 per 100,000 with a point prevalence in Sweden of mitochondrial encephalomyopathies in children of 4.8 per 100,000 [2]. The clinical, biochemical and genetic heterogeneity is shaping the right diagnostic approach, one which is changing rapidly with next-generation sequencing (NGS) methods [3].

Mitochondria is now recognized as able to perform multiple essential functions involved in cellular homeostasis and cellular energetics [4]. Oxidative phosphorylation is a very sophisticated mitochondrial mechanism being able to produce chemical energy in a form of adenosine triphosphate (ATP). Glucose, glutamin and branched chain amino acids are the main micronutrients that are the substrates of the chain of reactions, where a pool of acetyl coenzyme A (acetyl-CoA) and Krebs cycle intermediates in a mitochondrial matrix is filled by three key-enzymes: pyruvate dehydrogenase complex, alpha-ketoglutarate dehydrogenase complex and branched chain alpha-keto acid dehydrogenase complex with a thiamine, flavin adenine dinucleotide (FAD) and Mg²⁺ as essential cofactors in all three enzymatic complexes. The Krebs cycle produces reduced equivalents in a form of NADH₂ and FADH2, which push the respiratory chain enzymes to create a proton gradient that promotes ATP synthesis. Mitopathogenic mechanisms are recognized in a number of medical disciplines [5].

A muscle biopsy has been the gold standard used to set a diagnosis of primary MD in the pre genomic era. Nevertheless, the procedure has been invasive, costly, and usually requires general anesthesia in children. In order to evaluate the probability of underlying MD according to scoring of the clinical picture, Nijmegen MD Criteria (MDC) was developed [6,7].

Recent reanalysis of the MDC revealed that they are still useful in the clinical diagnosis of MD, both in children and adults, interpreting whole-exome results, as well as in deciding whether muscle biopsy needs to be performed [8]. In the last few years, NGS methods of leucocyte-derived DNA with yield of 40.0% have reversed the diagnostic approach from a “function to gene” to a “gene to function” approach and decreased the need for muscle biopsy [9]. Nevertheless, a muscle biopsy remains the method of choice for enabling diagnosis in selected clinical cases, where a confirmatory diagnostics is required.

In this paper, we describe the clinical characteristics and the results of imaging and laboratory investigations [muscle histology, histochemistry, electron microscopy (EM), respiratory chain enzymes (RCEs), pyruvate dehydrogenase complex (PDHc) activities, and/or molecular genetics] in a cohort of Slovenian neurological mitochondrial patients, of whom 26 are children and 36 are adults. Use of MDC, new NGS methods and repositioning of a muscle biopsy in the genomic era is emphasized. The study outlines the clinical experience of neurological mitochondrial diseases in Slovenia, a European country of 2 million people. It is important to improve early recognition and noninvasive diagnostics of MDs due to its devastating character and current unsatisfactory treatment options.

Clinical data (i.e., clinical, metabolic, electrophysiological, muscle biopsy and brain imaging data) of the children and adult patients with MDs have been summarized and analyzed. Data have been collected retrospectively analyzing the medical records of patients from 2000–2018, performing neurological examination and taking medical histories of all patients by the first author.

The main clinical charasteristics of the patients in the childrens’ cohort is listed in Table 1, and those of patients in an adult cohort is listed in Table 2. The MDC were applied to all children and adults in the study. Clinical comparison between both cohorts of patients (i.e., children and adults), was also performed and the data are presented in Table 3.

In a cohort of children, RCEs and/or PDHc deficiencies were found in all subjects. Twelve children had a PDHc deficiency. Molecular genetic analysis of the PDHA and PDX genes did not reveal a mutation in any child. Five children had Leigh syndrome, 10 children had RCE deficiencies, and four children had combined PDHc and RCE deficiencies. Diagnostic evaluation in these children is presented in Table 4, and results of molecular genetic testing are available in Table 5.

In a cohort of adults the main clinical presentation of MD was chronic progressive external ophthalmoplegioa (CPEO) or CPEO+, where a clinical diagnosis with musculus levator palpebrae or musculus orbicularis oculi biopsy and EM in 16 patients. Classical mitochondrial syndromes (CPEO, CPEO plus, MELAS, MERRF, MNGIE) were found in 22 patients (Table 2). Twelve adults were found to carry a diagnosis of definite MD, and 15 adults were found to carry a probable MD. A muscle biopsy with histochemical analysis has been performed in 30 patients, but RCEs were measured in only 14 patients. Diagnostic evaluation of these adult patients is presented in Table 6, and results of molecular genetic testing are available in Table 5.

Twenty children were scored according to MDC as a definite MD, five children as a probable MD and one child as a possible MD. The mean clinical score (I) was 3.7 (range 2–4), with a maximum score of 4 [including multi-system involvement taken alone, as part of the clinical score, had a mean score of 1.0 (range 0–2)]. The metabolic and brain imaging score (II) was 3.0 (range 0–4), and the histology score (III) was 1.8 (range 0–4). Before muscle biopsy, the mean MDC score (I+II) was 6.6 (±1.4 SD), and the final score (I+II+III) was 8.4 (±1.8 SD). Twenty children (77.0%) scored a total (I+II+III) of ≥8, a result that is comparable with the diagnosis of a definite MD.

Twelve adults were scored according to MDC as a definite MD, 15 adults as a probable, and eight adults as a possible MD. The mean clinical score (I) was 3.5 (range 2–4), with a maximum score of 4 [including multisystem involvement, which taken alone, had a mean score of 1.4 (range 0–3)]. The metabolic and brain imaging score (II) was 0.4 (range 0–3), and the histology score (III) was 2.6 (range 0–4). Before muscle biopsy, the mean MDC score (I+II) was 3.8 (±1.2), and the final score (I+II+III) was 6.5 (± 2.0). Twelve adults (35.0%) scored a total (I+II+III) of ≥8, a result that is comparable with the diagnosis of a definite MD. If we also include in this group of patients the patients with CPEO who were not genetically confirmed, the number of definite MD patients would be increased to 22 (63.0%).

Of 24 children evaluated with brain MRI imaging, five children (20%) had MRI signs of Leigh syndrome. Morphological brain MRI changes, which are not classified in MDC, but might be in accordance with the diagnosis of an MD, have been found in the following: eight (33.0%) children with delayed myelination, seven (29.0%) children with cerebral atrophy, two (8.0%) children with cerebellum changes, and in six (25.0%) children in whom neuronal migration disorders were found. In summary, brain MRI morphological changes, which can be an imaging clue for the diagnosis of MD, were found in 17/24 (71.0%) children.

Variability of results of the MRS spectrum was large in children with MD. No statistically significant difference at relaxation time TE 35 ms was found in N-acetyl aspartate, myoinositol and cholin ratios between basal ganglia and white matter between children with MD and children in a control group. Lactate peak was found in one child with Leigh syndrome and complex I plus IV deficiency at 1.33 ppm. However, statistically significant metabolic differences were found between basal ganglia and white matter in a healthy control group, leading us to conclude that the MRS spectrum was well performed.

Follow-up of this group of children with brain MRI has been important to show the “normalization” of myelination in these children after several years of active disease, to evaluate the progression of morphological changes in Leigh syndrome, and revealed new neuronal migration disorders previously not found on brain MRI images (Table 7). Figure 1 demonstrates the progression of Leigh syndrome in an 18-year-old girl with a pathogenic homozygous mutation c.626C>T (p.Ser209Leu) on the MTFMT gene (OMIM #614947).

Figure 1

We concluded this phase of the study with a comparison of the clinical characteristics of MD in children and adults (Table 3). Mitochondrial diseases in children have revealed a chronic course with metabolic decompensation with increased serum lactate and often negative genetic analysis. However, MDs in adults have also presented a slow but chronic progression over the years and this has been more often related to mitochondrial genome rearrangements with normal serum lactate. Nuclear mitochondrial gene mutations can have a different clinical presentation depending on whether patients are children or adults. Figure 2 demonstrates brain MRI changes in an adult with a pathogenic homozygous mutation p.Ser282fs on the SURF1 gene, a pathogenic variant that usually presents in childhood as Leigh syndrome (OMIM #256000).

Figure 2

Finally, we analyzed positive results of molecular genetic analysis methods for different groups of patients scored by Nijmegen MDC (Table 8). Positive molecular genetic results were in the great majority (17/19) in accordance with definite diagnosis of MD. Results also showed that NGS methods are beneficial to discriminate molecular mimicry and determining a diagnosis in cases of possible and probable groups of patients with MD (Tables 5 and 8). Despite a limited number of patients with mtDNA clinical syndromes, buccal swabs for mtDNA isolation are a promising noninvasive option instead of muscle biopsies (Table 8).

Slovenia is a small European high-income country with 2.1 million inhabitants and an estimated prevalence of MD of three per 100,000 inhabitants, a finding that is in accordance with other population studies of MD prevalence [2]. Determining a diagnosis of MD is a complex algorithm due to its extremely heterogeneous character, genetic diversity and complex laboratory diagnostic methods. With the majority of our mitochondrial patients we still have to use a laborious approach with muscle biopsies in the end. Obviously, in the genomic era, the new diagnostic algorithms will improve the diagnostic approach, and influence the development of new standards for early diagnosis and possible treatment [3,9,20].

Our study has mostly been done on unspecific encephalomyopathies in children and neuromuscular patients in adults. Thirty-one percent of cases were defined by CES in both our cohorts. Clinical-exome sequencing has been used when a pathology in the mitochondrial genome was not suspected or was excluded. In almost all of the children in the study, a diagnosis of MD was first suspected, based on the results of metabolic deviations in plasma, urine or CSF or abnormal brain MRI. In adults, an MD diagnosis has still been defined by muscle histology findings. Twenty of 30 adult patients had abnormal mitochondria on EM and 16/30 adult patients presented ragged-red fibers. Therefore, clinical knowledge, metabolic investigations, brain imaging and muscle biopsies have still been valuable in our clinical scenario until 2018. Promising potential of whole-exome sequencing (WES) as a first diagnostic option in finding mutations in genes related to mitochondrial dys-functions will probably reverse the classical diagnostic approach [21,22].

Our clinical and diagnostic patient data were put together according to MDC scoring. It has previously been demonstrated that a high MDC scores increases the likelihood of finding a genetic diagnosis by WES [23,24]. Witters et al. [8] found that the combined metabolic findings and imaging sub scores were higher in mitochondrial patients who were diagnosed when Sanger sequencing was used as a genetic analysis tool, and where a clinician-directed genetic analysis to a target gene was based on a clinical phenotype. The MDC scores are still found very useful in the clinical diagnosis of MDs. The MDC scores are helpful in interpreting exome-sequencing results or deciding on the need to perform a muscle biopsy [8]. The importance of MDC scoring and recognizing the patient's clinical phenotype is in accordance with what we found in our clinical data. Moreover, in our cohort, MDC score was used to conduct muscle biopsy not molecular genetic studies. Twenty of 26 children and 22/36 adults in our cohort reached a score of a definite MD. In all children, a diagnosis was confirmed by RCEs and PDHc activities measurements and when a high MDC score was reached before a muscle biopsy. Molecular genetic diagnosis was confirmed in 5/26 children. In adults a diagnosis of MD was confirmed in six patients with measurements of RCE activities, and reached a confirmation in 11 patients by molecular genetic analysis. Sixteen of 36 patients had CPEO syndrome, which is still challenging to diagnose at the genetic level [25]. The MDC score was found to be a useful tool for determining whether a muscle biopsy was used in both cohorts of patients, as well as for affirming the certainty of a diagnosis. In future, the MDC score might be also useful in using WES results to reach a diagnosis of a definite MD before performing a muscle biopsy, especially in children.

Next-generation sequencing methods are becomig the first diagnostic tool in the genomic era. Whole-genome sequencing (WGS) of blood DNA is also going to contribute as the first-line method for diagnosing a primary MD [3]. Whole-genome sequencing has overtaken WES as the preferred NGS method for DNA sequencing [26,27,28]. Whole-genome sequencing is better for detecting copy number variations (CNVs), single nucleotide variants, insertions/deletions, and in principle, microRNA (miRNA) variants. However, even with WES, results are not going to be so straightforward. Four scenarios of molecular results are possible for diagnoses derived from the WES approach [9]. At the very least, a skin biopsy and fibro-blast-related studies will be needed to evaluate variances of unknown significance (VUS) or new candidate genes. For now, the results of CES in our cohort of patients, gave us a promising yield (positive in 7/22 or 31.0% cases) compared to a yield of 39.0–55.0% in centers abroad [23,29]. Next-generation sequencing methods can also discriminate molecular mimicry or confirm a molecular diagnosis, even in cases of probable or possible groups of MD (Table 8). Buccal swabs for mtDNA isolation are showing us a potential as a noninvasive approach to diagnose mtDNA pathogenic variations, but we had a limited number of suspected mtDNA syndrome cases (Table 8).

Large mitochondrial centers are searching for a genotype-phenotype correlation in neuroimaging of MDs [30,31,32]. Mitochondrial diseases are not only commonly described as multi-systemic diseases, but also demonstrate neuroradiological changes in several functional systems of the central nervous system. Pathognomonic MRI changes for MDs can be found in the cerebral cortex, white matter, basal ganglia, cerebellum and brainstem [31]. The same is true in neuroimaging of MDs, as the meaning of pleitropy is known in genetics. For example, it is true that one pathognomonic MRI characteristic can be a result of many genes related to mitochondrial dysfunction. A genetic cause of Leigh sydnrome can also be found in 75 genes [33]. Nevertheless, through our study, we can demonstrate that in 71.0% of children with MD we have found brain MRI changes that can be a clue to a diagnosis of MD. Therefore, we can suggest that MRI is still the diagnostic tool of choice in selected cases that serves to provide at least one piece of the puzzle, particularly if we have a genetic result of VUS or negative results by WES. Magnetic resonance imaging also demonstrates that MD has the dynamic nature of a chronic disease (Table 7) (Figure 1). Some of the brain MRI characteristics present as a degree of cortical atrophy, dysmyelination and/or basal ganglia hyperintensities can change over a period of several years of follow-up (Table 7) (Figure 1).

One of the diagnostic problems in MDs is diagnostic mimicry. Clinical presentation of MD can present as a metabolic, neurodegenerative, neuromuscular or neuroin-flammatory diseases [34,35]. When we suspect a primary MD, another molecular cause can be found and mitochondrial dysfunction is only secondary [36]. This is demonstrated by the example of two clinical cases with pathogenic variants in a gene for a congenital myasthenic syndrome and Allan-Herndon-Dudley syndrome of triiodo-thyronine resistance that have been found by CES. In these cases, even muscle morphology and OXPHOS measurements of a skeletal muscle specimen and brain MRI have been misleading. We emphasize how important WES is in the differentiation of rare neurogenetic diseases mimicking as an MD.

In conclusion, MDs have remained a very heterogeneous group of neurometabolic diseases on a clinical, biochemical and genetic level of diagnostics. The goal of being noninvasive and cost-efficient is guiding us to choose a new way of the “gene-to-function” approach, instead of the old way as the “function-to-gene” approach. Solving the diagnostic puzzle(s) in either way is still a challenging task. For clinical purposes, the MDC has been a valuable tool in the diagnosis of MDs in children and adults. Obviously, the NGS methods are likely to contribute to end many diagnostic odysseys.