Prenatal exome sequencing for fetuses with structural abnormalities: the next step

Fetal anomalies complicate up to 3% of pregnancies and are being diagnosed by ultrasound prenatally with increasing sensitivity and specificity1. The overall prognosis and long-term outcome is very variable, being dependent upon: i) the type of fetal anomaly, ii) whether the anomaly is isolated or there are additional abnormalities suggestive of a known malformation syndrome and iii) the underlying genetic etiology. For many years, certain congenital abnormalities (e.g. omphalocele and complex cardiac disease) have been associated strongly with chromosomal aneuploidy and parents with affected pregnancies have been offered fetal karyotyping, as a chromosomal abnormality significantly worsens the overall prognosis2. Other congenital abnormalities may suggest a single-gene disorder, but often it is very difficult to make a definitive molecular diagnosis unless it has been detected previously in the family. Therefore, having a detailed family history and being aware of consanguinity can be extremely helpful in the assessment. Numerous genetic mechanisms account for the development of congenital structural abnormalities, of which the most common are aneuploidy, deletions and duplications of segments of DNA, known collectively as copy number variants (CNVs), and single nucleotide variants (SNVs) in protein-coding sequences that damage the encoded protein3-7. Knowing the cause of a congenital structural anomaly can aid clinicians in making a more accurate diagnosis and provides information about prognosis and recurrence risks for parents. Despite the importance of diagnosis, currently, only a minority of children affected by congenital and developmental diseases receive a genetic diagnosis8. Following the detection of one or more fetal congenital anomalies, the current practice is to offer the parents an invasive procedure, either chorionic villus sampling (CVS), amniocentesis or fetal-blood sampling, followed by genetic investigations. These include quantitative fluorescent polymerase chain reaction (QF-PCR) or fluorescent in-situ hybridization (FISH), both of which are rapid prenatal tests for the common trisomies, i.e. trisomies 13, 18 and 21, and sex-chromosome aneuploidy. These investigations are run in parallel with conventional cytogenetic techniques, which involve culturing the fetal cells and subsequently isolating and treating the chromosomes to make them visible for microscopic evaluation. This enables the detection of aneuploidy, relatively large deletions or duplications and other structural rearrangements, such as balanced and unbalanced translocations. Such classical cytogenetic G-banding (conventional karyotyping) carries the disadvantages of potential cell-culture failure and clonal selection and has a diagnostic resolution of 5–10 Mb9. One megabase (Mb) is a unit of length of DNA and is equivalent to 1 000 000 base pairs; 1 kilobase (kb) is equivalent to 1000 base pairs. With advances in cytogenetic technology, in addition to excluding aneuploidy, 'targeted' karyotyping using FISH can also identify rare syndromes. A relatively common example of this is 22q11 microdeletion syndrome, which should be excluded in fetuses with complex heart disease10. There are many examples of subtelomeric-microdeletion syndromes which present in babies with congenital malformations, but their prenatal diagnosis is only possible if specifically requested for particular syndromes or if the diagnosis has been made previously in the family11. Chromosomal microarray analysis (CMA, also known as array comparative genomic hybridization (aCGH) or CGH microarrays) uses up-to-date methods to detect much smaller chromosomal imbalances, usually with a resolution of 10–400 kb. Replacing conventional karyotyping with prenatal CMA improves by 4–6% the detection rate of chromosomal abnormalities, due to the detection of submicroscopic chromosomal rearrangements (pathogenic CNVs), in babies with congenital malformations4, 12. Indeed, the American College of Obstetrics and Gynecology has recommended that prenatal CMA should be the diagnostic test of choice to investigate a prenatally identified congenital anomaly, and could also be used in structurally normal fetuses undergoing invasive prenatal testing13. This is not presently the case in the UK, although routine use of CMA, rather than conventional karyotyping, is being implemented increasingly in many regions in England when a prenatal structural anomaly is identified. National guidance in this area is currently being formulated. Consensus needs to be reached quickly, as a specialized service, CQUIN (Commissioning for Quality and Innovation), has been issued by the Women and Children Programme of Care, 2014–15, entitled 'Maximising use of aCGH as primary ante-natal test (replacing karyotyping)'. There are certain concerns with using CMA prenatally, so informed consent is essential. It could identify an imbalance in the chromosomes unrelated to the ultrasound findings, but which may have implications for the future health of the baby and, potentially, for other family members: for example, a microdeletion that includes a 'cancer gene'. Rarely, CMA may detect chromosomal changes known as 'variants of unknown significance' (VOUS), genetic changes whose association with disease risk is currently unknown. There is ongoing debate about whether either of these findings should be reported4, 14. The resolution at which genetic variants can be analyzed and interpreted in prenatal diagnosis is increasing at a remarkable pace; the highest resolution would be a mutation at a single base pair and is achievable using sequencing techniques. DNA sequencing is the process of determining the precise order of nucleotides within a DNA molecule. Sanger sequencing, which identifies alterations in the order of bases within a single contiguous segment of DNA, is the classical method of DNA sequencing and has been used widely for approximately 25 years. More recently, next-generation sequencing (NGS) has revolutionized genetic testing, as it enables many genes to be sequenced together in a single test. Compared with traditional Sanger sequencing methods, NGS has dramatically faster processing and therefore a quicker turnaround of results, reducing overall cost and making the technique more feasible for use in the prenatal setting. Using NGS, the whole genome or exome can be sequenced; however, exome sequencing is often favored over whole-genome sequencing as it targets only exons, regions of the genome that code for proteins. Exons represent only 1–2% of the entire genome but contain up to 85% of mutations known to cause genetic disorders15, making this an efficient method of sequencing in terms of cost, time and computational resources. For exome sequencing, genomic DNA is extracted from samples (amniotic fluid, CVS or fetal blood are often used prenatally), and fragmented into many millions of small sections of DNA. A genomic library is constructed from these fragments. The library is then enriched for exonic sequences, using targeted hybridization probes. Finally, the DNA is sequenced. Many different technologies or 'platforms' are available for this, including the commonly used HiSeq™ (Illumina, San Diego, CA, USA), which uses reversible dye terminator technology (Figure 1). Briefly, to analyze the data, the high-quality patient DNA fragments, or 'reads', are typically aligned to the reference human genome, and differences between the reads and the reference genome are determined. To identify mutations that cause rare disease, known rare mutations are prioritized, particularly those predicted to result in loss of protein function. Many tools exist to accomplish these tasks, and the exact workflow depends on the samples, the expected mode of inheritance and the nature of the phenotype, amongst other factors. Interpreting the relevance of genetic variance in the prenatal setting, however, is difficult, given the limited phenotype seen on ultrasound examination (e.g. it is not possible to ascertain developmental delay). The bioinformatics are complicated and accurate interpretation is reliant on genotype–phenotype correlations using consistent phenotypic description, such as the Human Phenotype Ontology (HPO), as well as inputting of follow-up data. Databases such as DECIPHER (Wellcome Trust Sanger Institute, Cambridge, UK, http://decipher.sanger.ac.uk/) are comprised mostly of information from the postnatal setting but are likely to be adapted imminently to display data from prenatal clinical–genomic analysis; the more data that are entered, the more accurate will become the reporting of prenatal exome sequencing. The first report on the use of NGS to identify the genetic cause of a rare disease was made in 2010, whereby a genetic etiology was confirmed in a case of Miller Syndrome (postaxial acrofacial dysostosis: OMIM 263750)16. In the pediatric literature, the use of NGS, to identify genetic variants throughout the genome, is being introduced in the evaluation of children with rare malformation syndromes and those with unexplained neurodevelopmental morbidity. A recent prospective study, using whole exome sequencing to investigate children with neurodevelopmental morbidity and other phenotypic abnormalities present across childhood and adolescence, revealed a significant proportion of previously undiagnosed genetic syndromes17. Indeed, 86 mutated alleles that were highly likely to be causative were identified in 62 of the 250 patients, achieving a 25% (95% CI, 20–31%) molecular diagnostic rate. In the UK, the Deciphering Developmental Delay (DDD) project has been awarded a grant from the Health Innovation Challenge Fund (a partnership between the UK Department of Health and the Wellcome Trust). The project aims to understand better the genetic etiology of developmental delay in children, and catalyze the use of NGS as a diagnostic tool for these children within the National Health Service (NHS). The project is a UK-wide multicenter collaboration involving over 95% of consultant clinical geneticists, with all sequencing being performed at the Wellcome Trust Sanger Institute in Cambridge, UK. Importantly, the phenotypic information is recorded carefully and defined using the HPO criteria18. The aim is to perform exome sequencing for 12 000 trios (family structures of children and their parents). Any uncertain results are assessed by an expert review panel. Final results are disseminated to local clinical geneticists who then feed back the data to the family. Within the context of the research project, a turnaround time for these results has not been given to families. The use of NGS (whole-genome or exome sequencing) for fetuses with congenital anomalies has been described in only a small number of cases. The first study, published in 2012, was a single case study, in which whole-genome 'jumping library' sequencing of amniocytes identified prenatally an apparently balanced de-novo translocation that disrupted the CHD7 gene, causing CHARGE syndrome in a fetus19. In the study of Yang et al.17, although the majority of patients were children, the cohort also included four fetuses which had been terminated. In one fetus with a phenotype consistent with Cornelia de Lange syndrome, a causative de novo splicing variant in the NIPBL gene was noted. Finally, Filges et al.20 used exome sequencing to identify a novel, autosomal recessive, lethal ciliopathy-like condition. They sequenced two parents, their affected daughter and postmortem samples from two affected fetuses, and found compound heterozygous mutations in the KIF14 gene in both fetuses. These mutations were deemed highly likely to be the cause of the condition. In a recent collaboration between The Birmingham Women's Foundation Trust, the University of Birmingham, and the Wellcome Trust Sanger Institute, Cambridge, we performed exome sequencing on 30 prenatal and neonatal samples, in an attempt to identify what additional information exome sequencing might provide over other techniques, including high-resolution CMA, when a structural abnormality is seen at prenatal ultrasound examination. To our knowledge, this is the largest published cohort study to date relating to exome sequencing in fetuses with structural abnormalities7. To summarize, the 30 euploid fetuses had a diverse range of structural abnormalities detected by prenatal ultrasound examination. Exome sequencing was performed using the HiSeq (Illumina) platform in a trio design. We sought both de novo and inherited (recessive or X-linked) rare coding variants (SNVs, insertions or deletions, known as indels, and CNVs). These variants were classified through a systematic manual process as highly likely to be causal, possibly causal or causality unknown. All CNVs were validated using a CMA, as exome sequencing is not always an ideal method for their detection. Variants that were highly likely to be causal, due to identical or similar variants having been shown previously to cause a similar phenotype in humans, were identified in three of the 30 fetuses, giving a diagnostic yield of 10%. The first, in a male fetus with features consistent with a lethal skeletal dysplasia, was a de novo missense variant that is highly likely to be pathogenic, in the gene FGFR3. This mutation is one of the most common in fetuses and infants with thanatophoric dysplasia Type I6, and its presence confirms this diagnosis. A second de novo missense mutation was identified in the COL2A1 gene of a male fetus with increased (>3.5 mm) nuchal translucency, tricuspid valve regurgitation, abnormal legs and feet, an extended posture and bilateral talipes equinovarus anomaly. COL2A1 mutations can cause a wide range of skeletal and ocular conditions, referred to as the Type-II collagenopathies, sometimes with cardiac conditions21, 22. Although this mutation has not been reported previously, its position within an important region of the gene suggests that it is highly likely to be causative. Finally, we found a de novo 21 kb deletion in a female fetus with ventriculomegaly and agenesis of the corpus callosum. This CNV removed most of the gene OFD1. Mutations in this gene cause orofaciodigital syndrome Type I, which is associated in 40% of cases with central nervous system anomalies, including absence of the corpus callosum23. Being de novo, all three variants had a low recurrence risk, allowing us to reassure the families regarding further reproductive decisions. Only one of these three variants (the 21 kb deletion) could have been detected by high-resolution CMA. Variants that were possibly causal but required further confirmatory genetic and functional studies were found in a further five (17%) cases; these variants were not reported back to families. This pilot study therefore highlighted the increased rate of detection of prenatal genetic abnormalities by exome sequencing compared with karyotyping and CMA alone4, 12. Additionally, it revealed the potential diagnostic strengths of exome sequencing in the prenatal setting. However, a much larger cohort is required to answer and address some of the issues that would arise regarding the clinical use of this tool in prenatal genetic diagnosis. New methods for genetic testing, such as exome sequencing, can create ethical dilemmas. There is ongoing debate about whether incidental findings and VOUS should be fed back to patients. The American College of Medical Genetics and Genomics have produced guidelines for clinical reporting in the postnatal setting, recommending that patients should be informed if variants are seen in any of 57 specific disease genes, regardless of the referral indication. However, these guidelines excluded fetuses24. Subsequently, a revised document was produced, stating that patients should be allowed to opt out of receiving this information. In the UK, the DDD research study is not returning incidental findings to patients25. Discussion about the best mode of practice in this area is ongoing. Of course, providing information prenatally on incidental findings may have profound consequences, due to the possible option of terminating the pregnancy14. Also, incidental prenatal findings have the potential to give life-changing information to families. In some cases it could be argued that such information would be life-saving; for example, mutations in the BRCA1 gene confer an 80% risk of developing cancer at some point in life26. Yet it could also provide families with potentially unwanted information regarding their own and their extended family's health. The reporting of VOUS to families in the prenatal setting has been discussed in the context of VOUS found by CMA testing; some women found the information to be 'toxic', causing considerable amounts of anxiety in pregnancy and beyond, into their baby's childhood14. There is a real risk of labeling an unborn child as having a genetic disorder when, in fact, they have a benign genetic variant. We await national guidance, which will become available in the next few months. Meanwhile, our current practice in the clinical setting is not to report when there is uncertainty about whether a particular genetic change is significant. We also await the result of research in this area. It is vital that any ongoing studies not only answer questions about the feasibility of prenatal exome sequencing, in terms of DNA quality from prenatal samples and the turnaround time of results, but also address ethical dilemmas created by the testing. The PAGE project will be a UK-wide collaboration between the Wellcome Trust Sanger Institute in Cambridge, in which the exome sequencing will take place, and fetal medicine centers throughout the UK. The two co-ordinating fetal medicine centers will be Birmingham Women's Foundation Trust and University College Hospital Foundation Trust, London. In total, 1000 fetal exome sequences will be analyzed in a trio design. In addition to this, whole-genome sequencing will be implemented in a smaller cohort. A novel assay will be designed, and the potential for a non-invasive approach will be explored. Moreover, a rigorous health economic assessment (allowing for the inevitable decrease in genetic sequencing costs) will take place, as well as a review of the social aspects of this testing, by means of qualitative work and an 'ethical review' to aid future guidance in this area. Importantly, we expect to find an increase in the number of pregnancies in which we can reveal to couples the cause of their baby's structural anomaly and inform them regarding future reproductive decisions. It will also highlight difficulties in interpretation in the prenatal setting. For instance, how should a clinician counsel a patient if a genetic variant is identified which confers susceptibility to a condition that cannot be confirmed prenatally? For example, in the case of autism, a variant may be identified that conveys no or low susceptibility to some patients and high susceptibility to others, but it is currently not possible to assess accurately the risk in the prenatal, or indeed the postnatal, setting. The number of known genetic variants likely to cause fetal abnormalities will increase, at least initially, with the use of exome sequencing and, with detailed databases (such as DECIPHER), their clinical significance should become clearer. Results from the PAGE project will be made available to women and their partners only after pregnancy is complete, due to the current timescale for testing. Of course, turnaround time is of particular significance in the prenatal setting due to the biological timeframe of pregnancy. Although the outcome can be used to inform the prognosis in childhood, many women would value receiving the information within a suitable timeframe such that an informed decision can be made as to whether or not to interrupt the pregnancy. The project will be a catalyst, in terms of reducing turnaround times, as the minimum amount/quality of DNA required is explored, new assays are developed and experience in the analysis of results increases, speeding up the reporting process. The PAGE project has opted to feedback to patients only information on variants known to be a cause of the structural abnormality seen on the ultrasound scan. A clinical review panel will assess any ambiguous results, with the emphasis on giving clear information to couples rather than adding uncertainty. Late-onset adult diseases will not be sought actively when interpreting the exome sequencing results and, if found, will not be relayed to the clinician or family. It will be important to examine parental acceptance of this testing and their views on the reporting of results. Exome sequencing in the prenatal setting is inevitable; the challenge will be to incorporate it into prenatal diagnostics in a responsible manner, with an ethical framework. We must support patients at a time when they are faced with making the most complex and difficult decisions, rather than create more uncertainty. The PAGE collaboration team comprised Professor Lyn Chitty, Dr Nicholas Lench, Professor Michael Parker, Ms Jane Fisher and Professor Eamonn Maher.