AbstractNext-generation sequencing (NGS) will soon be used for clinically heterogeneous, inherited disorders and the increasing number of disease-causing genes reported. Diagnostic laboratories therefore need to decide which NGS methods they are going to invest in and how to implement them. We discuss here the challenges and opportunities of using targeted resequencing (TRS) panels for diagnosing monogenetic disorders. Of the different NGS approaches available, TRS panels offer the opportunity to sequence and analyze a limited set of predetermined target genes. At present, TRS panels offer better base-pair coverage, running times, costs and dataset handling than other NGS applications such as whole genome sequencing and whole exome sequencing. However, working with TRS panels also poses new challenges in variant interpretation, data handling and bioinformatic analyses. To optimize the analyses, TRS panel testing should be performed by bioinformaticians, clinicians and laboratory staff in close collaboration. KeywordsIn recent years, next-generation sequencing (NGS) has revolutionized molecular genetic research. Different platforms for massive parallel sequencing have been developed and applied in research settings for whole genome sequencing (WGS), whole exome sequencing (WES), RNA sequencing and some specific applications such as miRNA sequencing. Despite this revolution on the research side, in routine clinical diagnostics, the use of Sanger sequencing to detect mutations is still the gold standard in most diagnostic DNA laboratories. However, given its potential applications, it is likely that NGS will soon be implemented in clinical diagnostics on a large scale. The technical possibilities of NGS are advancing rapidly, and the numbers of newly annotated disease genes are increasing in parallel, leading to an ever-growing number of requests to add new disease genes to diagnostic packages. This growing list of disease-causing genes has revealed that many disease phenotypes are in fact quite heterogeneous and many different genes can give rise to the same or comparable clinical phenotypes. It is often no longer possible to pinpoint the most likely disease gene in a specific patient based on only their clinical features, despite careful clinical phenotyping. Examples of these heterogeneous disease categories are cardiomyopathy, movement disorders, deafness and epilepsy; for example, in epilepsy, more than 250 genes are tested in one targeted resequencing (TRS) panel [1]. Using Sanger sequencing to determine such large numbers of genes in a diagnostic setting is time consuming and costly and, in most cases, an almost impossible task. In contrast, they can be easily tested in a single run using NGS and can thus provide a genetic diagnosis in heterogeneous disease categories. However, implementing NGS does not just require a simple change in laboratory routines. Diagnostic laboratories first need to decide which of the many available NGS methods and machines they should invest in and how to adjust their practical routines, while all the staffs need to learn how to apply these new possibilities. Similar to what has happened in molecular genetic research, the handling of large datasets and use of bioinformatics programs for their analysis need to become part of the daily routine and incorporated into existing quality control systems. There are now professional standards and guidelines available to guide diagnostic laboratories in how to implement these NGS techniques, and these also offer important information for their quality control systems [2,3]. Here, we discuss the requirements, challenges and opportunities of implementing NGS technology in routine clinical DNA diagnostics. We will focus on the use of gene panels based on TRS and restrict ourselves to its use in the diagnosis of monogenetic Mendelian disorders. NGS methodsToday, different NGS techniques can be used for diagnostic purposes. These include sequencing the patient’s entire genome (WGS) or the exons of every protein-coding gene (WES) or targeting specific (known) disease-causing genes (TRS). Whole genome sequencingWGS can be used to determine the complete nucleotide sequence of an entire DNA sample, including all the coding and non-coding sequences. The major advantage of this approach is that the complete genomic information will be available and the actual sequencing is relatively fast, as only limited time is required for sample preparation. This is why it has been successfully applied for the rapid diagnostic sequencing of critically ill newborns in intensive care units [4]. Other exciting WES applications not related to monogenetic disorders include, for instance, the detection of chromosomal aberrations in circulating cancer cells [5]. One disadvantage of sequencing the whole genome is that it yields a large amount of data that requires extensive filtering to provide meaningful data with regard to (known) disease-causing mutations. In addition, there may be problems regarding unsolicited findings, that is, the identification of potential or definite pathogenic mutations in a patient that are unrelated to the disease for which the genetic diagnosis was originally requested. These can be encountered in both WGS and WES and raise ethical issues that are discussed later. Whole exome sequencingIn contrast to WGS, which basically analyzes every single nucleotide of the genome, WES focuses only on coding nucleotides. This allows the simultaneous analysis of the coding regions of annotated disease genes as well as of potentially disease-causing genes. To perform WES, all the coding exons in a DNA sample need to be enriched. A major advantage of WES is that all the potential disease-causing genes are included in the analysis: both known disease genes and genes not yet related to a disease. In particular, WES can reveal new disease genes or mutations that have not yet been associated with certain clinical phenotypes. The advantage of WES over WGS is that considerably higher read depths can be reached at relatively lower cost. However, some major concerns for WES are its incomplete representation, in most cases because exons were simply not included in the manufacturer’s capture design, and low coverage of base-pair reads in certain exons, most often related to insufficient capturing of GC-rich sequences. In the past, incomplete representation and low coverage have led to clinically relevant mutations being missed when WES was used in clinical diagnostics and these still represent serious quality issues compared to traditional Sanger sequencing [6,7]. The advantage of being able to test large numbers of genes by WES is thus counterbalanced by its incomplete sequence depth; this issue must be carefully considered if WES is offered in diagnostic testing. However, the use of WES in clinical diagnostics is certainly justified for those diseases that do not yet have a well-defined gene panel because the knowledge about their genetic basis is still too limited. WES should also be considered in genetic diagnostics for rare monogenetic diseases that cannot justify the investment required for the design and production of a disease-specific gene panel capturing kit. Although WGS or other gene panel-based analyses could also be used, WES is probably the best technique for identifying de novo mutations in a parent–patient trio approach for heterogeneous disorders with very large numbers of putative genes, such as mental retardation [8]. Finally, WES has not only been successfully applied in the diagnostics of rare monogenetic disorders but also in clinical oncology. Examples of the use of WES are the genotyping of multiple genetic variants in tumor tissue [9], obtained in a short period of time [10] and the use of WES to guide targeted therapy [11]. Like WGS, unsolicited findings are also an issue when WES is used in clinical diagnostics for monogenetic disorders. These ethical aspects of unsolicited findings will be discussed later. Targeted resequencingThe third, widely used, NGS strategy is to enrich only the coding regions of genes of interest for a specific disease or diagnostic category. TRS diagnostic panels are used both for diagnosing hereditary monogenetic disorders as well as for assessing cancer risk. There are some major advantages to restricting the mutation analysis to a limited set of genes: targeted enrichment and subsequent resequencing provides a superior quality of representation and a much higher read depth than WGS or WES. In addition, as the focus is on known disease genes, laboratory staff faces fewer challenges in analyzing the datasets and interpreting variants that contributes to significantly shorter turn-around times for test results [12]. Moreover, TRS minimizes the problem of unsolicited findings and thereby increases the willingness of patients and their families to participate in NGS diagnostic testing [13]. Depending on the platform and the enrichment strategy used, several hundred target genes can be analyzed for multiple patients in a single run. Thus, not only can large numbers of genes be tested but also large numbers of patient samples can be analyzed within a relatively short processing time. Costs of NGS diagnosticsThere is limited data available on all the costs and benefits related to NGS diagnostics. However, the direct costs of WGS have fallen considerably in recent years [14] and it is likely that WGS costs will drop further, thereby facilitating the use of WGS in routine diagnostics. However, high-capacity NGS sequencers are needed for WGS to obtain high-quality results and this, together with the costs of reagens, means that the costs of WGS are considerably higher at present than for the other two NGS applications. Considerably higher read depths can be reached by WES at relatively lower cost compared to WGS, but TRS results are produced at a much lower cost than for either WES or WGS. Using TRS, diagnostic panels of up to several hundred target genes can be analyzed for multiple patients in a single run. Thus, not only can large numbers of genes be tested but also large numbers of patient samples can be analyzed in a relatively short processing time. In assessing the costs of sequencing, only the direct costs are usually taken into account [14], and these have continued to drop since 2007. However, the actual costs involved in interpreting test results, reporting variants and maintaining variant databases have not changed. These costs relate to the actual number of variants observed in each analysis and favor the use TRS panels [15]. NGS diagnostics can, however, guide clinicians in the use of targeted therapy and help avoid ineffective and potentially harmful treatments, thereby saving costs and improving individualized patient care [15]. Surprisingly little has been published as to whether NGS testing really improves patient care. Randomized clinical trials are urgently needed to answer these important questions, while the cost–effectiveness of NGS testing compared to current standards of care must be investigated in randomized trials [16]. Enrichment strategies in TRSIn all three NGS applications, the sequence quality largely depends on the minimum coverage at every base-pair position and the optimization of enrichment procedures of the genes of interest (target sequences) in a diagnostic panel is therefore of major importance. Various enrichment methods are available and include solid phase-based microarrays, micro-droplet-based PCR (Rain Dance Technologies, Lexington, MA, USA), amplicon-based (Fluidigm, San Francisco, CA, USA) and solution phase-based hybridization methods such as Agilent’s Sure Select Targeted strategies (Santa Clara, CA, USA) and Illumina’s TruSeq Custom strategies (Illumina, San Diego, CA, USA). Each of these different approaches has advantages and limitations. Enrichment strategies that are PCR or amplicon based have fast and easy workflows for the multiplexed PCRs and can also be used on fixed tissue samples. However, amplicons may be missed due to polymorphisms located in primer-binding sites and artifacts may be introduced during the amplification procedure [17,18]. Moreover, structural variations may be difficult or even impossible to detect. In hybridization-based assays, the enrichment is usually good, the assays are cost–efficient, and structural variations with an exonic breakpoint can be detected [19]. An important disadvantage of these hybridization-based enrichment strategies is that they often reduce the coverage of the GC-rich regions in the target genes [17]. In contrast to WGS and WES, the targeted fragments or amplicons obtained with the different TRS enrichment strategies can be analyzed on bench-top instruments, because relatively less material needs to be sequenced. Such bench-top machines include the Ion Torrent PGM (Life Technologies Ltd., Paisley, UK), the 454 GS Roche Junior (Roche Applied Science, Indianapolis, IN, USA) and the Illumina MiSeq (Illumina, San Diego, CA, USA) [20]. Diagnostic performance of TRS panelsRequirements for diagnostic TRSCoverage of almost every nucleotide of interest is of major importance for the application of NGS technology in clinical diagnostics and sequence depth is therefore an important quality parameter in NGS applications. Meynert et al. [21] demonstrated that, at a mean on-target read depth of 20×, which is commonly used in WES studies for diagnosing rare disorders for instance, one would miss 5–15% of the heterozygous and 1–4% of the homozygous single nucleotide variants in the targeted regions. However, to use TRS for clinical diagnostics, high-quality data is essential, which means that not the mean on-target depth, but ideally all nucleotides are seen at a minimal read depth of 20×–40× [22]. Currently, this requirement can only be achieved at reasonable cost using gene panel-based TRS methods. In our recent experience with a targeted panel for cardiomyopathy, the mean coverage per target was approximately 250× [12] and more than 99% of each nucleotide was covered at least 30×. When we looked at targets with insufficient coverage in more detail, we found the poor cover was usually caused by only a few nucleotides within the target and was mostly due to a high GC content. Diagnostic yieldWeiss et al. [22] proposed the use of ‘diagnostic yield’ as a criterion for the performance in NGS-based testing, as well as the usual test characteristics of sensitivity and specificity. They defined diagnostic yield as the number of patients who receive molecular confirmation of a given clinical diagnosis and considered this to be an important output parameter. In other words, diagnostic yield is the likelihood that a test will provide the information needed to establish a genetic diagnosis and laboratories could use this as a criterion to decide whether they are justified in switching from Sanger sequencing to NGS-based methods. Such a switch should result in a diagnostic yield that at least matches that from sequential gene testing using Sanger sequencing [22,23]. In our own experience of using a TRS panel with 55 cardiomyopathy genes, the percentage of patients in whom disease-causing mutations were found improved greatly, from 15% to about 50% in over 250 patients tested [Pósafalvi et al., Manuscript in Preparation], showing how patients with cardiomyopathy can benefit from introducing NGS-based methods in clinical diagnostics. Comparably, Pugh et al. [24] reported that the diagnostic yield increased when more genes were added to a TRS diagnostic panel for cardiomyopathy. Their yield increased to 37% when they increased their panel size from 5 to 46 genes. Using diagnostic yield as a criterion also means that the genes included in a panel should be carefully selected, that is, only genes that have sufficient data to prove their involvement in the respective disease should be included. In addition, the diagnostic yield does not necessarily increase by simply including an increasing number of genes in a diagnostic panel. The clinical genetic diagnostic laboratories in the Netherlands proposed they should define and maintain ‘core disease gene lists’ for which the relevant genes are selected by a team of medical and genetic experts [22]. This list should then at the least be included in an NGS-based diagnostic test to achieve maximum mutation detection. And logically this implies that high-quality sequencing of the respective genes in the test is ensured. Of course, including additional genes in gene panels is optional as long as complete coverage and a high quality for the core disease gene list are guaranteed. Sensitivity & specificityResults reported for the sensitivity and specificity of validated TRS panels compared to Sanger sequencing are very encouraging. In our previous studies using our cardiomyopathy panel with 55 genes, we were able to reach 100% sensitivity (95% confidence 97.76–100%) and a specificity of nearly 100% (0.00315% false-positive rate) [12]. Likewise, Gowrisankar et al. [25] reported almost similar findings for their TRS panel, although their panel contained only 19 cardiomyopathy genes. This level of performance has also been reported for other heterogeneous disorders, such as deafness and cancer risk estimation, with panels for BRCA and hereditary colon carcinoma [26–28]. Together these data support the idea that, when properly designed, TRS gene panels will outperform traditional Sanger sequencing [12]. Specific attention should be paid to the detection of small insertions and deletions (indels). To enable this, the respective sequences should be sufficiently covered and the analyzed reads should be of appropriate length to diminish the risks of misalignment and improve diagnostic accuracy, as shown by Voelkerding et al. [29]. In our own experience, using base-pair reads of 150 bp enabled us to detect insertions of up to 6 base-pairs and deletions of up to 18 base-pairs [12]. In contrast, shorter reads are more likely to align to highly homologous sequences such as pseudogenes and those present in gene families and domain analogs. This misalignment can interfere with sequence analyses and have a negative influence on the diagnostic specifications [30]. The use of longer read lengths (>1 kb) will significantly improve the ability to detect long indels and structural variants as was recently demonstrated [31]. In the current transition phase to TRS panels, many laboratories are still confirming the presence of variants identified with NGS by performing Sanger sequencing, as this is still the gold standard in DNA diagnostics. Zhang et al. [32] stressed that this is necessary for two reasons: first, to remove incorrect calls due to experimental errors, and second, to confirm the genetic diagnosis using a different technique. However, the TRS panels now being developed will likely include more and more genes and confirming the identified mutations could become a burden that might affect a laboratory’s efficiency and turnaround times. Moreover, the reason for confirming the presence of an identified mutation was not only meant to confirm the genetic diagnosis, but also – and more importantly – to exclude the possibility of a putative sample swap. We suggest that it is no longer necessary to confirm all the positive results with Sanger sequencing as long as a minimal coverage of 30× per nucleotide is reached [12] and a standard check for putative sample swaps is included in the diagnostic procedure. In addition, possible variants identified in targets that are poorly covered can either be excluded from the final test report or Sanger sequencing of them can be performed in parallel. Data analysis & storageA major difference between traditional Sanger sequencing and all the NGS modalities lies in the various aspects of data analysis and variant interpretation. Typically, very large numbers of variants need to be filtered out of the raw data. After defining variant thresholds, further elimination of benign variants is usually done using web-based databases, such as 1000 genomes, exome variant server, dbSNP and variants lists generated and managed locally. The remaining variants need to be weighted as benign, variant of uncertain clinical significance or (likely) pathogenic using web-based browsers with prediction programs, such as SIFT and PolyPhen, often combined with custom-made software tools and extended with literature searches. At present, this process of data analysis and data interpretation is considered one of the most complex and challenging aspects of working with NGS diagnostic tests [20]. Data obtained from diagnostic procedures need to be stored for a long period of time as data storage is part of the laboratory’s accreditation requirements. The size of the data files generated by TRS diagnostic panels is considerably less than those obtained through WES and WGS, making it possible to store not only the variant files but also the FASTQ files for longer periods [22]. Unfortunately, there is no uniformity in file formats or the terms and conditions for storage. When files are stored, the pipeline version used and the quality control parameters should also be included. Quality managementSeveral national and international recommendations have been published on how NGS should be incorporated into quality management systems [2]. Gargis et al. [33] reported general guidelines for assuring the quality of NGS in clinical laboratory practice. These guidelines were formulated by a national working group from the US CDC. In their study, they addressed four topics that should be accounted for: proper test validation, quality control procedures to assure and maintain accurate test results, independent assessment of test performance through proficiency testing or alternative approaches and use of accurate reference materials. However, these recommendations are not actually much different from non-NGS quality control guidelines in the lab. What is quite different and essential in NGS approaches are the procedures for data analysis and variant interpretation, which were also discussed by Weiss et al. [22]. These procedures, software and machinery should also be incorporated into quality control management systems certified by the International Standardization Organization. Particular attention should be paid to bioinformatics procedures, because many labs perform NGS sequence analysis using ‘in-house’ pipelines. According to the guidelines proposed by Weiss and colleagues, the use of such pipelines is acceptable as long as the data flow and processing are transparent and there is strict regulation enforced of versioning and updating once the pipeline has been validated for diagnostic use. They suggested that: specific changes to software should be documented (i.e., the changes made from one version to the next); there should be a system for tracking issues and problems (e.g., a ticket system) and samples should be traceable and analyses should be completely reproducible at a later time point. This automatically implies that the settings of instrument control software, analysis settings in the pipeline and the different versions of analysis programs and databases are retraceable, and that redundant versions of the pipeline are properly archived [22]. Challenges in introducing TRS in the clinical diagnostics labDeveloping new techniques in a research setting for implementation in clinical DNA diagnostics is nothing new, but the magnitude and speed at which NGS technology is advancing are unique. At present, the technical advances cause a paradox in NGS applications in clinical diagnostics: the technical possibilities to retrieve information from DNA have greatly increased, but our ability to understand the effects of the genetic variants identified on disease, or disease predisposition, according to Ellard et al. [34], is still in its infancy. The interpretation and classification of variants is therefore one of the major challenges in applying all the NGS modalities, including TRS, in routine DNA diagnostics and this should not be underestimated, either in terms of manpower investments in laboratory staff and bioinformaticians or the computer systems and data storage required. Thus, nowadays, the laboratory staff involved in NGS diagnostics not only has to keep up with the continuously improving technical possibilities, but also with the ever-changing medical and molecular information related to the disorders they are analyzing. Whereas laboratory staff in the past were always in charge of the full diagnostic process, from receiving the sample to sending the test results to the clinician, now there is a new situation in which the handling and storage of large datasets and the daily retrieval of information on variants from a variety of web-based servers and bioinformatics pipelines have become part of the laboratory infrastructure and routine. Moreover, given the power of TRS and the large numbers of genes that can be included in diagnostic panels, this may mean that laboratory staff will start working with genes of which they have limited, or sometimes even no, experience. This new situation automatically implies that dedicated bioinformaticians have become a crucial part of the diagnostic workflow. However, the colleagues working on bioinformatics pipelines and data storage often come from a research environment and they need to become familiar with the strict requirements of quality control and daily management in a DNA diagnostic laboratory. They also need to be trained in writing standard operating procedures. It is therefore important to ensure that the pipelines are constructed to support generic data analyses and that versions of pipelines, once they have been validated, are not modified unless it is to improve the diagnostic workflow or their performance. These new versions must then be validated again. This rapidly changing laboratory environment and work routine require staff to have different skills and a different focus than a few years ago. Surprisingly, little has been published on the human aspects of the NGS revolution, despite the fact that adopting NGS methods not only means investing in machines but also in the skills of laboratory staff. Apart from the challenges associated with interpreting variants and the central role of bioinformatics in analyzing TRS data, there are also some potential pitfalls in the widespread use of NGS diagnostics. One of the areas where conventional sequencing is still superior to TRS is in detecting repeating nucleotides and their disease-causing expansions. Frequently, the number of repeats is associated with the age of onset and/or with disease severity. There is a large group of genetic disorders for which expanded nucleotide repeats are an underlying cause and it includes Huntington’s disease, Fragile-X syndrome and the autosomal dominant spinocerebellar ataxias. These nucleotide repeats or triplets are often filtered out using standard variant-calling algorithms in NGS applications. However, as elegantly shown by Trujillano et al. [35], dedicated software tools can be developed to detect and report the number of these specific repeat sequences. Using TG repeats in the CFTR gene, as an example, they demonstrated how software tools could be used to report the repeat expansions using TRS. Custom-made software tools are also necessary to tackle the problem of detecting large structural alterations and deletions or duplications of one or more exons. Even with the use of longer reads, these large structural alterations are still difficult to detect using TRS. Most genetic laboratories are currently testing algorithms to detect (large) deletions or duplications in NGS datasets. It has been shown that such software tools, for instance, to normalize read counts, can be used to detect structural large alterations, thereby eliminating the need for additional comparative genome hybridization arrays or multiplex ligation-dependent probe amplification-based assays [36,37]. Another interesting challenge is related to the number of genes that are being analyzed using TRS panels and whether this will increase the possibility of detecting multiple mutations in multiple genes. As an example, Gl?ckle et al. [38] reported using a panel of retinal dystrophy genes (n = 105) with which they examined 170 patients: they found some patients had multiple disease-causing variants and discussed that these could be acting as disease modifiers. Such findings add to the complexity of interpreting variants found in NGS diagnostics and these cases warrant careful clinical follow-up to determine whether multiple deleterious variants are actually disease modifiers or not. Such data should be shared in collaborative networks, and although the importance of this is generally acknowledged, actual sharing of this kind of information is less straightforward than it seems. This also raises the point of how these findings should be reported to the treating physicians and their patients. In our own data on 250 patients with cardiomyopathy, more than one likely pathogenic mutation was seen in over 10% of cases [Pósafalvi et al., Manuscript in Preparation], creating challenges in interpreting our sequence results similar to those reported by Gl?ckle et al. [38]. In another study by Yang et al. [39], exome sequencing was used for clinical diagnostics and surprisingly two separate genetic diseases were diagnosed in four out of 250 patients (1.6%). These four patients proved to have a second, non-overlapping, genetic diagnosis that was felt was important to report because of its clinical relevance. This once again highlights the issue of unsolicited findings that can be encountered using NGS applications, in particular in WES and WGS, and emphasizes why patients and families should be carefully counseled before testing. A new paradigm for DNA diagnostics: opportunities for laboratory staff & cliniciansDespite the fact that the rapidly increasing technical possibilities contain many challenges for the laboratory staff, the new NGS diagnostics also offer many opportunities. It has already been shown that the diagnostic yield increases for selected diagnostic categories when TRS is implemented in routine diagnostics and a widespread use of NGS will likely increase the diagnostic yield for many patients suffering from rare genetic disorders. On the other hand, as large numbers of TRS panels or other NGS modalities become available in clinical diagnostics, clinicians will need to redefine their diagnostic strategies when requesting NGS tests. They should first, on the basis of clinical parameters, try to confirm a high a priori risk that the patient is indeed suffering from a genetic disorder. NGS methods should preferably not be used if the a priori risk is rather low. This is important, because when numerous genes are tested, the chances are high that variants might be interpreted as disease causing in genes that are in fact unlikely to explain the clinical phenotype. Performing reliable NGS testing thus depends on having comprehensive clinical information on a patient, as this will guarantee an optimum workflow, safe variant interpretation, short turn-around time and fewer risks of over-interpreting any variants found. As clearly stated by Hennekam and Biesecker [40], when working with NGS methods, we should appreciate that NGS and computers will not magically make diagnoses in all, or even perhaps in most patients. Diagnoses can only be made in close collaboration between laboratory staff and clinicians and this is the area where many new opportunities will arise in the care of patients with rare genetic disorders. Dedicated teams of laboratory staff, bioinformaticians and clinicians should design, implement and use TRS panels [41]. Together, these dedicated teams can progress to even more innovative diagnostic modalities, such as combined disease and pharmacogenetic test panels, for truly personalized medicine applications or preconception testing for future parents. Ethical issues in NGS testingWith the introduction of NGS in clinical diagnostics, the handling of unsolicited findings and other ethical issues will need to be addressed. The issue of unsolicited findings is very relevant in WGS and WES and Green et al. [42] has published recommendations on this subject. It was recently reviewed that the majority of parents of children tested by WES want significant unsolicited findings to be disclosed, but a minority does not want to know these results [43]. This again warrants the need for a close collaboration between laboratory staff, clinicians and those offering careful pre-test genetic counseling. Unsolicited findings are less of an issue in TRS diagnostic panels because only a preselected list of genes is being tested and the situations in which, for instance, a heterozygous state attributes a certain risk can be discussed in advance with patients. The risk of unsolicited findings in WGS and WES might result in less acceptance of these NGS procedures by patients and families, and fewer of them may give their consent. Neveling et al. [13] observed that 10% of the families who received pre-test counseling did not consent to the DNA test because of the risk of unsolicited findings, in particular in families with children and young adults. Such an observation also supports the use of TRS diagnostic panels, which, in our experience, have not posed any problems for patients and/or their parents. However, there are certain ethical issues related to TRS diagnostic panels. For instance, how should panel results be analyzed and reported in relation to the clinician’s original diagnostic question? If only one specific mutation analysis has been requested, it is usually better not to use a diagnostic panel, but what should be done if a limited number of gene tests from a larger panel have been requested? Ideally, only these test results should be disclosed. If these test results are negative, the remaining information from the diagnostic panel can be disclosed if requested by the clinician. Of course, the patient’s consent needs to be obtained in advance for the various scenarios. Finally, the rapidly expanding list of disease-associated genes means targeted diagnostic panels need to be updated regularly. Not only the addition of new disease genes but also new information on variant and how best to weigh them might be valid reasons to recontact or retest patients. This subject has been debated in the past, but the widespread introduction of NGS diagnostics is forcing clinicians and geneticists to reconsider the issues relating to the recontact and retesting of (old) patients [44]. A decade ago, a duty to recontact was considered by some healthcare lawyers as undesirable [45], but others working in medical genetics considered it ethically desirable, although impossible to carry out in practice [46]. This situation has not changed much today and Pyeritz [44] considers clinicians should accept an explicit duty to recontact with caution – and should discuss their position when they counsel patients about genetic testing. These issues clearly need to be worked out, while, in the meantime, laboratories will need to give regular updates on the content of their TRS diagnostic panels and changing views on variant interpretation to their requesting physicians. The rapid implementation of NGS diagnostics will, of course, lead to the implementation of protocols on how to handle the various ethical, legal and societal aspects that accompany unsolicited findings and other information that becomes available about patients’ genomes. Expert opinionThe advantages of NGS are clear and NGS technology will soon be commonplace in DNA diagnostics. As we have discussed above, TRS panels are faster, less expensive per test sample and allow easier data handling than WGS and WES approaches. TRS panels can replace traditional Sanger sequencing for the majority of indications and will allow the detection of multiple genes with high sensitivity and specificity. With the datasets and numbers of variants that need to be interpreted in TRS being relatively small, this approach will permit large numbers of tests to be analyzed. By the time a large number of TRS panels are being used, this will offer the flexibility to add new genes to the panel or to remove genes that do not contribute to the diagnostic yield. Although there are many evident advantages in using NGS technology, each laboratory that starts with NGS will have to weigh carefully the advantages and disadvantages of the different NGS methods and machines available. This is because there are differences between the available NGS platforms in terms of costs per sample, laboratory procedures and ease of use. The magnitude of these issues was nicely demonstrated by Li et al., who tested a small TRS panel for inherited arrhythmia syndromes on two different NGS platforms [47]. Laboratory performance and results with short turnaround times is important for all diagnostic tests, because the sooner a molecular diagnosis becomes available, the better clinicians can make well-informed decisions about potential diagnostic and therapeutic options for their patients. The same is true for TRS panels, but, in addition, short turnaround and reporting times will also be good reasons for clinicians to participate in NGS testing and perhaps even join NGS diagnostic teams. Physicians will soon be able to experience the benefits and opportunities of using TRS diagnostic panels and other NGS approaches in their clinic. NGS has revolutionized molecular genetic research and will soon revolutionize clinical DNA diagnostics as well. At present, of all the NGS modalities available, TRS has the best characteristics with respect to identifying mutations in known disease genes and the results can be obtained at relatively low cost and with short turnaround times. TRS panels enable a large number of genes on multiple patient samples to be tested in a single run. As for all diagnostic tests, TRS procedures should be well embedded in quality management systems and this is particularly important for the handling and storage of datasets and bioinformatics pipelines. The interpretation and classification of variants is one of the major challenges in using TRS diagnostic panels and in the other NGS modalities and will likely remain so in the coming years. For this reason, we recommend that the development and implementation of TRS panel testing in DNA diagnostic laboratories should be done by laboratory staff, bioinformaticians and clinicians in close collaboration. Five-year viewIn the next 5 years, there will be a transition from using only traditional Sanger sequencing to NGS diagnostics. In the majority of DNA diagnostic laboratories, one or more of the NGS modalities will be in use, but it is difficult to predict which modality will be most preferred. It is tempting to think that WGS will be the dominant diagnostic approach, because all genomic information can be disclosed with just one technique. The costs of WGS will certainly drop further, making more applications possible in routine clinical diagnostics. However, the costs of TRS diagnostic panels and WES will also drop and given that panels are fast and efficient, they will have a firm place in routine diagnostics. The machines on which diagnostic panels will be analyzed will become considerably smaller and some people think that in 5 years’ time hand-held sequencers will be widely used for bedside testing, for instance to genotype pathogens and for pharmacogenomics [48]. For the clinically and genetically heterogeneous diseases such as mental retardation, which have many (unidentified) disease-causing genes, WGS will likely be the primary diagnostic approach within the next few years. Traditional Sanger sequencing will be reserved for those situations in which testing of known mutations in a single gene can be requested and that is sufficient to reach a molecular diagnosis, for instance, in family screening. With the widespread use of NGS diagnostics, data sharing will become a prominent focus in the coming years, in combination with automated variant interpretation and classification. Sharing data need not be restricted to genomic data, but can include the sharing of phenotypic datasets with detailed phenotypic information. These will be equally important for diagnostic laboratories. Only by sharing the combined genotypic and phenotypic data will we be able to determine the effects of multiple variants, for instance, in the light of a patient’s symptoms. Standardization of file formats and storage will become a prerequisite for the widespread exchange of data. To enhance variant interpretation and classification, several research tools have been launched recently and are looking very promising [49], but their use in clinical diagnostics first needs to be validated. In the coming years, more insight into the variants in the non-coding regions of genes will become available and more deep intronic variants will be identified as disease-causing mutations. The same is true for promoter and enhancer mutations. If these non-coding variants are known to be associated with disease, the TRS panel design can be adjusted to detect these variants [50,51], but if this is not the case, it will be difficult to address these problems with regular techniques other than WGS. However, another promising application to gain insight into (non-coding) DNA variants will be the incorporation of RNA sequencing results into diagnostics procedures, thereby demonstrating the effects of variants on transcription and splicing [52]. Once we have tackled how to retrieve, interpret and handle information from the coding and intronic genomic regions, the next step in NGS diagnostics will be to incorporate the wealth of information that is present from the remaining non-coding parts of our genome [53]. As clearly shown by the results of the ENCODE project, the ‘non-coding’ regions and ‘non-coding’ RNAs have a significant influence on gene functions and gene regulation. Biostatisticians and sophisticated bioinformatics tools will open up this information for us, but like the current introduction of NGS into the diagnostic setting, laboratory staff and clinicians will together need to learn to translate this knowledge into diagnostic tests that can be applied in routine diagnostics. Finally, what we have discussed with regard to the future of NGS diagnostics might be somewhat biased towards the opportunities that will arise with advances in technology. However, the extent to which all these tools will be implemented will largely depend on how they can be embedded within the current healthcare systems, given the budgets available for innovation in molecular diagnostics and the ways of paying for diagnostic tests and data analysis. A detailed insight into the true costs of implementing NGS and its widespread use in diagnostics, and into the benefits of NGS to patients and society is urgently required. AcknowledgementsWe thank Jackie Senior for editing the manuscript. Financial & competing interests disclosureThe authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the production of this manuscript. Key issues
References
Papers of special note have been highlighted as: · of interest ·· of considerable interest |
||||||||||||||||||||||||||||||||||||||||||
|
|
来自： zhuqiaoxiaoxue > 《重测序跟疾病文章》
