The use of targeted genomic capture and massively parallel sequencing in diagnosis of Chinese Leukoencephalopathies

Leukoencephalopathies are diseases with high clinical heterogeneity. In clinical work, it’s difficult for doctors to make a definite etiological diagnosis. Here, we designed a custom probe library which contains the known pathogenic genes reported to be associated with Leukoencephalopathies, and performed targeted gene capture and massively parallel sequencing (MPS) among 49 Chinese patients who has white matter damage as the main imaging changes, and made the validation by Sanger sequencing for the probands’ parents. As result, a total of 40.8% (20/49) of the patients identified pathogenic mutations, including four associated with metachromatic leukodystrophy, three associated with vanishing white matter leukoencephalopathy, three associated with mitochondrial complex I deficiency, one associated with Globoid cell leukodystrophy (or Krabbe diseases), three associated with megalencephalic leukoencephalopathy with subcortical cysts, two associated with Pelizaeus-Merzbacher disease, two associated with X-linked adrenoleukodystrophy, one associated with Zellweger syndrome and one associated with Alexander disease. Targeted capture and MPS enables to identify mutations of all classes causing leukoencephalopathy. Our study combines targeted capture and MPS technology with clinical and genetic diagnosis and highlights its usefulness for rapid and comprehensive genetic testing in the clinical setting. This method will also expand our knowledge of the genetic and clinical spectra of leukoencephalopathy.

Leukoencephalopathies are disorders that primarily affect the white matter of the central nervous system (CNS). It contains acquired leukoencephalopathy 1-3 (leukoencephalopathy induced by ischemia, hypoxia, intoxication, infection, traumatic brain injuries, etc.), genetic leukoencephalopathy 4-6 (such as metachromatic leukodystrophy, globoid cell leukodystrophy, X-linked adrenoleukodystrophy, etc.) In addition, it also contains some mitochondrial diseases, cerebral cortical degenerative disorders, and so on. Clinically, after considering clinical history, symptoms and brain MRI features, doctors may be able to give a diagnosis for acquired leukoencephalopathies. However, leukoencephalopathy is a disease with high clinical heterogeneity and may involve in multiple genes, it is difficult even for experienced neurologists to make definite diagnosis [7][8][9] . Therefore, we are in urgent need of finding an efficient, economical, and practical method for diagnosing leukoencephalopathies.

Results
Demographic and Clinical characteristics of the total 49 patients. We summarized the clinical characteristics of the total 49 patients enrolled in this study and found 39 are male and 10 are female. The age at onset of symptoms varied from 20 days to 7 years and the average onset age was almost 1.2 years. The main neurologic complaint of these patients include developmental delay/regression (27/49, 55.1%), epilepsy (15/49, 30.6%), weakness (7/49, 14.3%), ataxia (5/49, 10.3%) and dystonia (5/49, 10.3%). The severity of the disease course is reflected in the developmental milestones achieved. Two patients have suspected familial clustering. One has been diagnosed as adrenoleukodystrophy by gene testing. His mother's elder brother had the same clinic feature and MRI findings, and died at his age of 12. The other one has been diagnosed as mitochondrial complex I deficiency, and his elder sister has a similar brain MRI changes without significant neurological disease manifestation. The wide spectrum of MRI findings was noted in the study. Abnormality in periventricular, subcortical white matter and cerebellar hemisphere were common. Lumbar puncture and CSF analysis were performed in 19 patients. None of them had a positive result.
Twenty patients were identified pathogenic mutations in this study, and their demographic and clinical characteristics were shown in Table 2. However, more than half (29/49, 59%) of patients in our study did not reach the diagnosis.

Discussion
With the widespread use of imaging examinations in nervous system diseases, finding the pathogeny of cerebral white matter lesions becomes an important clinical clue for neurologists. Because of the strong heterogeneity of hereditary leukoencephalopathy, it is difficult even for experienced doctors to make a definitive diagnosis, and a multistep process is often needed 7,21 . Currently, routine clinical diagnostic tests for leukodystrophy often consist of screening for genes on the basis of ethnic origin, MRI features, family history, personal history and findings from physical examinations 22 . In China, the problem seems more serious, with the lack of a referral system, many patients and their families wasted valuable time, finances, and medical resources seeing various doctors and getting repeat examinations in search of a correct diagnosis. Some patients who could even be cured missed the opportunity for effective treatment. However, due to the high cost of Sanger sequencing for the long list of candidate genes, more effective genetic screening methods are needed.
In recent years, targeted capture and MPS technologies have been widely used in clinical practice and have got satisfactory results 15,[23][24][25][26][27][28] . To this end, we designed the gene panel contains 118 genes which are reported to be associated with leukoencephalopathies, not only contains genes associated with genetic leukoencephalopathy, but also mitochondrial disease, cerebral cortical degenerative disorders, etc. associated genes. Then we designed the probe library and performed this study to assess the utility and effectiveness of targeted capture and MPS in diagnosing leukoencephalopathy patients.
In our study, 40.8% (20/49) of the patients detected pathogenic mutations, which is higher than that of other commercially available chips. In our department, the positive rate of a mitochondrial disease chip is only 9.5%, and that of a metabolic disease chip is 16% (data not shown). These differences may be explained by the variety of pathogenic mutations and lack of a specific clinical phenotype associated with these disorders. Moreover, the results achieved using the leukoencephalopathy probe library may be explained by the distinctive brain MRI patterns that characterize leukoencephalopathy that was seen in most of the patients, providing a guide in the diagnostic process. In addition, patients had been thoroughly examined before the screening for leukoencephalopathy-associated genes, and other secondary causes were already excluded.
Among the result, one patient (case 19) was diagnosed with Zellweger syndrome with PEX6 gene compound heterozygous mutations, PEX6 gene mutation is reported to be associated with Peroxisome biogenesis disorder 4A/B 29,30 . The patient in our study was a 5.9-year-old girl exhibiting mental and motor retardation for 5 years, and with deterioration for 3 months (Clinical features and auxiliary examinations are included in Table 2). Brain MRI showed symmetrically increased signal intensity in T2-weighted images with gadolinium enhancement in the posterior limbs of the internal capsules (Fig. 2). There was no diffuse restriction or gadolinium enhancement in the periventricular area and deep white matter, similar to the features of X-ALD 31,32 . However, this is a female and   the ABCD1 gene in this patient exhibited a normal sequence and gene dosage. Given the diagnostic uncertainty, targeted capture and MPS were performed. Molecular testing identified PEX6 gene compound heterozygous mutations, supporting the Zellweger spectrum disorder diagnosis in this patient. Genetic analysis showed that the two mutation sites were respectively inherited from the parents. The result showed us the effectiveness of this targeted capture and MPS method for the diagnosis of leukoencephalopathies. The targeted capture and MPS method can not only diagnose genetic leukoencephalopathies, but also can make the diagnosis of mitochondrial diseases with white matter abnormal as the primary imaging changes. In our study, three of our patients had pathogenic gene mutations associated with mitochondrial complex I deficiency.
Our team was the first to report leukoencephalopathy associated with mitochondrial complex I deficiency due to a novel mutation in the NDUFAF1 gene (c.278A > G; c.247G > A) 33 . Mitochondrial complex I deficiency is the most frequent cause of respiratory chain defects in childhood, which accounts for various clinical presentations 34,35 . As the report, mutations have been described in 28 of these, including the 7 mitochondrial genes and 21 nuclear genes. Brain lesions caused by mitochondrial complex 1 deficiency are usually located in the brainstem, periaqueductal gray matter, the thalamus, etc. While, diffuse supratentorial leukoencephalopathy involving the deep lobar white matter may also occur in patients with mitochondrial complex 1 deficiency, especially in patients with nuclear DNA (nDNA) mutations. Some patients were available with abnormal white matter containing cysts in FLAIR sequences, and other patients may have notably hyperintense on T2 and very hypointense on T1 weighted images, suggesting cysts 33,36 . Therefore, containing pathogenic genes associated with mitochondrial diseases can promote the diagnosis of patients with leukoencephalopathies.
The patients in our study came from six provinces in central-south China. Therefore, the results may represent the specific disease incidence in this region. When clinicians encounter children with prominent cerebral white matter lesions that can't be explained by a certain disease, application of leukoencephalopathy probe library gene screening may be useful. Targeted capture and MPS can detect multiple candidate genes at the same time in a fast, cost-effective way, and can facilitate clinical diagnosis. Moreover, by reaching a definitive diagnosis for children with leukoencephalopathy, we can better judge the prognosis for patients and provide genetic counseling.
In summary, our data demonstrate that the use of targeted capture and MPS technology coupled with NGS has great promise as a tool for screening leukoencephalopathy-related genes for diagnostic purposes in patients. At the same time, genetic testing results combined with detailed clinical phenotypes help us expand our knowledge of the clinical spectra of each type of leukoencephalopathy. This method enables clinicians to identify leukoencephalopathy even the clinical performance is not typical. Moreover, the entire process of targeted capture, sequencing, analysis, and parental analysis was rapid (requiring only 10 days for up to 12 patients).
While targeted genomic capture and MPS technology also has its limitation, it can only identify the known pathogenic mutations. With the development of gene testing technology, a lot more pathogenic mutations will be detected, so the panel should be renewed with the latest findings, and the patients with negative results of genetic testing can be re-tested using the newest panel. With the fast development of NGS sequencing, the price will be more accessible, we can choose whole exome sequencing (WES) if the targeted analysis is unrevealing, or we can directly choose the WES technology. WES will be the inevitable trend, but under the condition of most countries and before this come true, our panel with cheap price, fast testing speed and strong pertinence, still have the irreplaceable advantage. Thus, we expect that this method can serve as an inspirational starting point. This technology will enable us to conduct straightforward, comprehensive screening for more known leukoencephalopathy-related genes, and to expand and redefine the genetic and clinical spectra of leukoencephalopathies.   Genetic testing. Genomic DNA was isolated from peripheral blood leukocytes (Promega, Beijing).
Target-fragments are capture by SureDesign target enrichment kit (Agilent, Santa Clara, CA) and high throughput sequencing by HiSeq2500 sequencer (Illumina Inc, San Diego, CA) were conducted in house. Overall, 49 samples were sequenced pre lane and the mean depth is 583X.
Bioinformatic Pipeline. For the quality control, the Cutadapt and FastQC were used to remove 3′ -/5′ -adapters andlow-quality reads, respectively. The clean reads were mapped to the reference human genome using the BWA (Burrows-Wheeler Aligner) program with at most two mismatches. The alignment files (bam) were generated with SAM tools and the reads of low mapping quality (< Q30) were filtered out. Clonal duplicated reads that may be derived from PCR artifacts were removed using Picard Tools by default parameters. Short read alignment and annotation visualization were performed using the IGV (Integrative Genomics Viewer). The percentage of alignment of the clean read to the exome regions was obtained using our custom Perl scripts on the base of alignment files. SNVs and indels were detected by GATK (Genome Analysis ToolKit). Comprehensive annotation of all of the detected SNVs and indels were annotated by ANNOVAR, including function implication (gene region, functional effect, mRNA GenBank accession number, amino acid change, cytoband, etc.) and allele frequency in 1000 Genomes, ExAc. Damaging missense mutations were predicted by SIFT, PolyPhen-2 and MutationTaster.