2

Genetic Testing

This chapter considers the relevance of genetic variation to human health and how genetic testing can be used to assess that variation. A more detailed description of genetic variation is available in Appendix A, and the technical terms used here are defined in the Glossary. Developments in genetic testing methods and the categories of genetic tests that are available for clinical use are also discussed. Finally, the committee discusses the ethical, legal, and social implications of genetic testing.

GENOMIC FEATURES AND GENETIC VARIATION

Except for germ cells (oocytes and spermatozoa), all cells in the human body have two haploid genomes, one inherited from the mother and one from the father. Each haploid genome has about 3 billion base pairs distributed among 23 chromosomes (autosomes 1–22 and the sex chromosome, X or Y). The genome contains about 20,000 protein-coding genes that are transcribed into a primary RNA transcript. Most RNA transcripts code for proteins, and these undergo splicing to remove intronic sequences and link the exonic sequences colinearly to form a mature mRNA molecule. The mRNA moves from the nucleus to the cytoplasm, where it serves as a template to direct translation of the protein encoded by the corresponding gene. Exons are the segments of a protein-coding gene that are retained in the mature mRNA. They encode the portion of the mRNA that is translated into protein and the non–protein-coding 5’ and 3’ untranslated ends of the mRNA. Typically, exons are about 100–200 base pairs long, and there are about 220,000 of them in the genome. The aggregated exons, referred to as the exome, makes up about 1.5% of our total genome sequence. The remainder of the genome is made up of highly repetitive sequences that have little or no recognized function (about 45%), introns (about 35%), and a large number of functional non–protein-coding elements, many of which are regulatory elements that influence the expression of nearby protein-coding genes (about 10%)¹ (ENCODE Project Consortium, 2012; Kellis et al., 2014).

The so-called reference human genome sequence describes the DNA sequence across the 3,000 Mb of a haploid genome assembled as a composite sequence, which was derived by sequencing a relatively small number of healthy people. Annotated versions of the human

___________________

¹ Available at: https://www.encodeproject.org (accessed January 31, 2016).

reference sequence can be viewed at any of several public genome browsers.² Every genome differs from the human reference genome by about 0.1% or 3–4 million single-nucleotide variants (SNVs). Some SNVs are common: the frequency of their least common variant at a particular position (the minor allele frequency, MAF), ranging from 1 to 49%; these common SNVs are often referred to as single-nucleotide polymorphisms (SNPs). Other SNVs are rare (MAF <1%). More than 150 million SNVs have been observed in sequences of thousands of people from around the world, but SNVs not previously observed continue to be found. For example, a recent report describing the whole-genome sequence (WGS) of about 10,000 healthy people found that each of us has an average of 8,579 SNVs not previously observed (Telenti et al., 2016). The density of SNVs throughout the genome is variable but averages about 55/1,000 base pairs (Telenti et al., 2016). Nearly all SNVs are binary, with only three possible genotypes (homozygous for one allele, heterozygous, or homozygous for the other allele). In addition to SNVs, each of our genomes has structural variation, consisting of insertion–deletion variants, which differ in length by 1–100 base pairs, and copy number variants (CNVs), which are deletions or duplications of sequences of 50 base pairs to 2–3 Mb and in terms of aggregate base pairs contribute about as much variation to the human genome as do SNVs (Zhang et al., 2009; Carvalho and Lupski, 2016). Thus, there is substantial variation in the human genome, and, aside from identical twins, each person has a unique genome sequence.

The functional consequences of the extensive variation can be considered in three classes. The first, and by far the largest, contains variants that have no known functional consequence and are often referred to as neutral variants. A second, much smaller, class contains variants that are highly detrimental and have been shown to cause genetic disease. These disease variants number around a few hundred per person; most are recessive and produce a phenotype only when both copies of the affected gene have disease variants. The third class contains variants that confer some risk of a complex trait or disease, but the phenotype occurs only in individuals with a combination of several risk variants and/or who are also are exposed to environmental risk experiences; these variants contribute risk for the common complex traits of adult life (such as hypertension, coronary artery disease, and neuropsychiatric disease). The goal of genetic testing is to sort through the large number of variants to identify the few (genotype) that either cause a specific monogenic disease (Biesecker and Green, 2014) or contribute risk for a complex trait (Lautenbach et al., 2013). At the current level of understanding, the functional significance of many observed variants is uncertain; such variants are referred to as variants of uncertain significance (VUSs).

Once a genetic test has been performed, interpretation of identified genetic variation can be challenging owing to the extent and variable consequences of the data. Considerations used for this purpose include the type of variant (such as SNV and CNV); the evolutionary conservation of the altered sequence; the frequency of the variant in an appropriate control population; segregation of the variant compared with segregation of the phenotype in the family; the predicted functional consequences of the variant (such as loss of function because of nonsense, missense, or splice site mutations); and comparisons with model organisms that have mutations of similar consequence in the orthologous gene (the same gene in a different species).

___________________

² Examples are the National Center for Biotechnology Information Reference Sequence Database, available at: www.ncbi.nlm.nih.gov/refseq (accessed January 31, 2016); the University of California, Santa Cruz, Genome Browser, available at: genome.ucsc.edu (accessed January 31, 2016); and ENSEMBL, available at: www.ensembl.org (accessed January 31, 2016).

Data on many of those considerations are available in publicly accessible databases. For example, evolutionary conservation is provided at the University of California, Santa Cruz, Genome Browser,³ and consequences of similar variants in the orthologous gene in a model organism can be found at model organism-specific databases, such as MGI⁴ for mouse models. Such databases as the Exome Variant Server⁵ and the Genome Aggregation Database⁶ provide the frequency of human variants in the whole-exome sequence (WES) from more than 6,000 and more than 100,000 people, respectively. ClinVar⁷ provides annotated predictions of the functional consequences of variants: pathogenic, likely pathogenic, uncertain, likely benign, or benign (Landrum et al., 2016). GeneMatcher⁸ and Matchmaker Exchange⁹ enable connections around the world between people who have an interest in a given candidate disease gene (Philippakis et al., 2015; Sobreira et al., 2015). Many tools are available to assist in those analyses (e.g., Kircher et al., 2014), and clinical laboratories use these tools in slightly different ways to make judgments about the pathogenicity of variants (McLaughlin et al., 2014). Additional analyses of functional testing of identified variants might be required to confirm pathogenicity, as might newer approaches, including transcriptomic analyses of cells that express the relevant genes (Cummings et al., 2016). The importance of comparing results in a particular person with those in people of similar ancestry was emphasized in a recent publication (Manrai et al., 2016). The committee expects the challenge posed by analysis and confirmation of causation will be reduced as more disease genes and associated pathologic variants are identified.

A final consideration regarding the interpretation of the results of genetic testing, especially for tests that survey the whole exome or the whole genome, is that VUSs on which current data are inadequate to allow interpretation of medical significance will be identified. A typical WES result, depending on the variant frequency filters and inheritance pattern, might contain hundreds or thousands of such variants. In addition, broad-based tests, such as WES or WGS, might identify known pathologic variants for conditions other than those for which the tests were performed. For example, a young child undergoing WES for evaluation of intellectual disability might be found to have a known pathologic variant in BRCA1, and thus the child, and the parent from whom the child inherited the variant, would be at high risk for breast and ovarian cancer in adulthood. Incidental findings,¹⁰ such as those in the 59 actionable disease genes defined by the American College of Medical Genetics and Genomics (ACMG), are identified in 1–2% of people tested (Green et al., 2013; Amendola et al., 2015; Jurgens et al., 2015; Kalia et al., 2016; Natarajan et al., 2016). Thus, in genetic tests that survey a large fraction of the genome, VUSs and incidental findings can be expected, and such outcomes probably should be discussed with a person to be tested at the time of consenting to the test (Biesecker and Green, 2014).

___________________

³ Available at: http://genome.ucsc.edu (accessed January 31, 2016).

⁴ Available at: http://www.informatics.jax.org (accessed January 31, 2016).

⁵ Available at: http://evs.gs.washington.edu/EVS (accessed January 31, 2016).

⁶ Available at: http://gnomad.broadinstitute.org (accessed January 31, 2016).

⁷ Available at: https://www.ncbi.nlm.nih.gov/clinvar (accessed January 31, 2016).

⁸ Available at: https://genematcher.org (accessed January 31, 2016).

⁹ Available at: http://www.matchmakerexchange.org (accessed January 31, 2016).

¹⁰ Incidental findings refer to the “secondary findings unrelated to the indication for ordering the sequencing but of medical value for patient care” (Green et al., 2013).

GENETIC VARIATION AND HEALTH OUTCOMES

Genetic variation contributes to health outcomes across a broad continuum. At one end of the continuum are highly penetrant monogenic or Mendelian diseases, in which variants in one or both copies of a single pair of genes are sufficient to cause a clinical condition or phenotype. The responsible gene can be on an autosome (one of chromosomes 1–22) or on one of the sex chromosomes (X or Y). Monogenic disorders are inherited either as autosomal dominant disorders, in which a single defective copy of the responsible gene is sufficient to cause a phenotype, or as autosomal recessive disorders, in which both copies of the gene must be defective to produce a phenotype. Some monogenic disorders are caused by variants in genes on a sex chromosome; they usually cause X-linked disorders, but there are a few Y-linked disorders, which are manifested only in males. Typically, X-linked disorders are more severe in males, who have only a single X chromosome and therefore only a single copy of the responsible gene¹¹; females, who have two X chromosomes, are often only mildly affected if they are heterozygous for the responsible X-linked gene. (See Appendix A for a more detailed account of Mendelian inheritance patterns.) There are thousands of Mendelian disorders; most present in the pediatric age range and are individually rare, but in the aggregate they account for 5–10% of hospitalized children (Jimenez-Sanchez et al., 2001; Antonarakis and Beckmann, 2006; Chong et al., 2015). Examples of monogenic disorders¹² are sickle-cell disease (OMIM, 2016a), cystic fibrosis (OMIM, 2016b), and Huntington disease (OMIM, 2016c).

At the other end of the continuum are multifactorial disorders that result from a combination of risk variants in several genes and accumulated environmental experiences. The individual risk variants are by themselves not sufficient to cause a phenotype. The etiology of these common disorders, most of which have onset in adulthood, is often highly heterogeneous, and two people who have the same diagnosis, such as hypertension, might have different but overlapping combinations of genetic risk variants and different but overlapping environmental experiences. Other examples of these complex traits are asthma, neuropsychiatric disorders, and coronary artery disease. Between the extremes of the monogenic disorders and complex traits are many health outcomes that are influenced to a greater or lesser degree by genetic variation.

Chromosomal disorders, another category of genetic diseases, are caused by abnormalities of chromosomal structure or number. Human chromosomes have hundreds or even thousands of genes distributed along their length. Chromosomal abnormalities typically involve large numbers of genes that in the case of trisomy are present as three copies instead of the usual two and in the case of deletions are present in the affected segment as only a single copy. Some genes require two copies to have normal function; they are referred to as dosage-sensitive genes and are thought to produce the phenotypic abnormalities associated with chromosomal disorders. Trisomy 21, or Down syndrome, is an exemplar of this category of genetic disease.

___________________

¹¹ Males with such a disorder are referred to as being hemizygous for the responsible gene.

¹² Online Mendelian Inheritance in Man (OMIM). See omim.org (accessed January 31, 2016) for concise descriptions of the phenotypes and genes responsible for monogenic disorders. OMIM entries are designated by unique numeric identifiers, MIM numbers, which will be provided for the examples used in this chapter so that the interested reader can easily obtain additional information.

ADDITIONAL COMPLEXITY OF MONOGENIC DISORDERS

Genetic Heterogeneity

The monogenic disorders make up a disproportionally large fraction of the clinical scenarios that lead to genetic testing. Although often considered simple or straightforward, these disorders have features that complicate selection of the appropriate test and interpretation of results. First is the phenomenon of genetic heterogeneity, of which there are two types: allelic heterogeneity, in which any of several variants of a particular disease gene is capable of causing the disease, and locus heterogeneity, in which variants of different genes (at different positions in the genome) cause the same or similar phenotype. For example, clinically indistinguishable forms of retinitis pigmentosa (OMIM, 2016d) can be caused by any of several variants in more than 50 genes; retinitis pigmentosa exhibits both allelic and locus heterogeneity (McGee et al., 1997; Rivolta et al., 2002).

Genotype–Phenotype Relationships—Penetrance and Variable Expressivity

A second type of complexity of monogenic disorders concerns the relationship between the genetic variant and the phenotype. The presence of some pathologic variants almost always produce disease; for example, the missense mutation Gly380Arg in the FGFR3 gene (OMIM, 2016e) always produces an autosomal dominant dwarfing condition known as achondroplasia (OMIM, 2015). Geneticists refer to that perfect concordance of genotype with phenotype as complete penetrance. The presence of other variants does not always produce the phenotype; this phenomenon is referred to as incomplete penetrance. For example, women who are heterozygous for a known breast-cancer allele of BRCA1 (OMIM, 2016f) have about an 80% lifetime risk of breast cancer (King et al., 2003). Thus, for the roughly 20% of the women who carry the variant, it is nonpenetrant. The biologic basis of incomplete penetrance is not well understood (Cooper et al., 2013), and this lack of knowledge complicates the counseling for people without symptoms of a disease who undergo genetic testing and are found to have the variant.

Another complication in relating genotype to phenotypic severity is the observation that different people who have the same pathologenic variant in a known disease gene might exhibit extreme variation in the severity of the phenotype. That is referred to as variable expressivity and is exemplified by an autosomal dominant disorder known as neurofibromatosis, type 1 (NF1; OMIM, 2016g), which is characterized by hyperpigmented macules (café au lait spots), benign growths (neurofibromas), and an increased risk of some malignancies, all caused by mutations in the NF1 gene (OMIM, 2016h). Patients who have NF1 have a highly variable phenotype, so even those within the same family who have the same pathogenic variant have considerable differences in the severity of their phenotype, ranging from a few to thousands of neurofibromas and café au lait spots (Nussbaum et al., 2016).

GENETIC TESTS

The National Institutes of Health’s National Human Genome Research Institute (Task Force on Genetic Testing, 1997) defines genetic tests as

A genetic test might be ordered by any of several health care providers, including a primary care physician, a medical specialist, a medical geneticist, an advance practice nurse, or a genetic counselor. In this section, the committee outlines testing methods, reviews the role of family history information in decision making about genetic testing, describes the types of genetics tests, and briefly considers the ethical, legal, and social implications of genetic testing. As noted, the committee focuses on germline DNA-based tests following publicly held conversations with DoD and its internal discussion and deliberations regarding the interpretation of the statement of task.

Testing Methods

Various genetic testing methods are available, and selection of the appropriate method depends on the clinical scenario and knowledge of genetic test results in family members. All the methods describe the constitutional or inherited DNA sequence. Therefore, unlike results of many testing methods used in clinical medicine, results of genetic tests can be useful for the lifetime of the subject and can be referred to repeatedly.

The choice of a testing method or specific test depends on the clinical scenario. A targeted test focuses on a specific disease variant already identified in an affected family member. A test targeted to a specific gene might be indicated if the patient’s presentation is highly suggestive of a disorder known to be caused by mutations in a known disease gene, and testing of several genes (often referred to as a gene panel and tailored for a specific phenotype) might be indicated if the patient’s condition is one known to have locus heterogeneity. In other scenarios, more general testing designed to survey the entire exome (a WES) or the whole genome (a WGS) might be appropriate. The types of tests are reviewed below as described by The Jackson Laboratory (Schott et al., 2015).

Single-Variant Tests

This type of test focuses on determining the genotype of a single base pair of interest. It is useful in two scenarios. In the first, a specific, disease-producing SNV is known to be segregating in a family as determined by previous studies on an affected relative. An example would be testing at-risk family members for a causative SNV in BRCA1 that has been identified in a relative who presented with breast and/or ovarian cancer. The second scenario is screening for a specific disease-causing mutation in a person whose phenotype is consistent with the disease. An example would be screening for the sickle-cell mutation (HBB-E6V) in a person who has clinical features of sickle-cell disease (OMIM, 2016a). The test is highly specific and relatively inexpensive. This testing approach is not appropriate, however, when the causative variants responsible for a particular disorder exhibit allelic or locus heterogeneity. In some

situations, it makes sense to test for multiple variants to address a particular clinical question, for example, in testing several variants in different genes that are all associated with drug response.

Single-Gene Tests

Single-gene tests survey the sequence of a particular gene to detect disease-producing SNVs, CNVs, or other abnormalities. The patient’s sequence is compared with that of the reference sequence, with consideration of the possibility of neutral, population-specific variants that should be distinguished from true pathologic variants. This type of test is often appropriate when the patient’s phenotype suggests that a diagnosis is known to be caused by variants in a specific gene, such as CFTR in a patient who has features of cystic fibrosis (OMIM, 2016i). Such tests can use Sanger sequencing¹³ or, increasingly, next-generation sequencing (NGS) methods; with appropriate controls, both are highly accurate. Because the entire gene is sequenced, allelic heterogeneity is not a problem except for phenotypes with locus heterogeneity, where this method must be modified to sequence all possible genes that could be responsible for the phenotype in question.

Multigene Panels

Multigene panels allow testing of many genes in a single procedure, and they are particularly appropriate for disorders that are known to have locus heterogeneity. The functional segments (exons, splice sites, and so on) of the genes of interest are isolated (“captured”) by hybridization-based methods and then sequenced, typically with NGS methods. The patient’s sequence is compared with a reference sequence to detect SNVs, CNVs, and some chromosomal rearrangements. This method is faster and usually less expensive than testing for each gene known to be responsible for the phenotype in question. An example might be panel testing for genes associated with hypertrophic cardiomyopathy (OMIM, 2016j), which is known to be associated with more than 60 genes.

Whole-Exome Sequencing

As described above, the exome makes up only about 1.5% of the genome. However, the variants that cause Mendelian diseases appear to be disproportionally concentrated in the exome. Current estimates suggest that about 85% of the variants responsible for Mendelian disease are in exons or in the immediately flanking intronic splice sites (Chong et al., 2015). Thus, sequencing of this segment of the genome should detect a large fraction of variants responsible for Mendelian diseases without the added expense of WGS. WES uses capture kits or reagents that are based on hybridization methods to isolate the exome and discard the remainder of the genome. The exome is then sequenced with next-generation methods. Currently, there are only suggested criteria for the quality and quantity of sequencing necessary for WES that will be used for clinical purposes, and ACMG has provided a useful set of recommendations (Rehm et al., 2013). The hybridization step in the capture process is not perfectly efficient, and most capture reagents isolate only 85–90% of the exonic sequence. The “depth” of NGS is usually described numerically and needs to be sufficient to detect heterozygosity at a particular base pair reliably. For example, more than 95% of the targeted sequence at a depth of 20 reads/base would be

___________________

¹³ Sanger sequencing, developed by Frederick Sanger and colleagues in 1977, was the most widely used sequencing method for almost 40 years. More recently, Sanger sequencing has been supplanted by next-generation sequencing methods.

considered minimal for clinical sequencing. Larger amounts of targeted sequence and greater read depths incur greater cost, and laboratories might choose to minimize the expense. It is important to evaluate those and related characteristics when choosing a sequencing laboratory.

Whole-Genome Sequencing

WGS is appealing because it captures virtually all the information and because, like the results of WES, its results are of lifelong value. Unlike WES, WGS does not require a hybridization-based capture step to prepare the sample. For that reason, the number of sequencing reads is proportional to the input DNA and is not influenced by the characteristics of the capture step. Thus, WGS performs better in identifying CNVs than WES. Moreover, exons with some sequence characteristics are not captured well and therefore are not covered by WES but are covered by WGS. A small but important number of SNVs responsible for Mendelian disease are in regions of the genome (such as promoters, other regulatory elements, and deep intronic SNVs that alter splicing) that are not captured for WES but are covered in WGS. Finally, in contrast with the causative SNVs responsible for Mendelian disorders, the vast majority of risk variants that contribute to common complex traits are in non–protein-coding segments of the genome and therefore will be detected only with WGS (Maurano et al., 2012; Andersson et al., 2014). WGS, however, is more expensive, and the large data files are difficult to work with and store.

Other Genomewide Tests

Whole-Genome SNP Array

This test types a dense set (typically 500,000 to more than 2 million) of SNVs or SNPs distributed throughout the genome. The results provide information on both the genotype and the number of copies (strength of signal) of each genotyped SNV. For example, if there are four genotypes (GGG, GGA, GAA, and AAA), the region of the genome tagged by the SNV is probably present in triplicate rather than the normal diploid, and the signal intensity would be increased by about 50%. This method provides information about CNVs throughout the genome. The resolution depends on the density of the SNPs and on the quality of the genotyping but is about 25–200 kB.

Array Comparative Genomic Hybridization

In this test, patient and reference single-stranded genomic DNA molecules are labeled with different fluorescent dyes and applied to a microarray of immobilized small DNA fragments (oligonucleotide probes) to screen for such chromosomal abnormalities as aneuploidy, deletion, duplication, and amplification (Theisen, 2008). The relative fluorescence of the two samples is calculated with a digital imaging system to determine which samples hybridized to the DNA probes. With a dense set of oligonucleotides designed to hybridize, for example, to every exon in the genome, this method can, in principle, detect CNVs that involve deletions of single exons (100–200 base pairs).

Karyotype

This classic method depends on stimulating circulating lymphocytes (B cells) to undergo mitosis in a short-term culture (2–3 days). The chromosomes condense and become visible under a microscope. The chromosomes are stained before being visualized, and each can be identified

by its size, shape, and pattern of bands along its length as detected by the staining. The test detects abnormalities of chromosomal structure and number. Its resolution depends on the degree of chromosomal contraction and the microscopic image and is about 3–5 Mb—lower than that of the molecular methods above.

Family History

The collection and evaluation of family history are important adjuncts to the use of genetic tests. Family history might serve purposes unrelated to genetics, such as providing information about social relationships and important life events, but its primary function in clinical care is to identify people who have an increased likelihood of particular diseases. Family history might identify people who are at relatively high risk for a single-gene condition that is present in their family (such as Lynch syndrome, hereditary breast ovarian cancer syndrome, hemochromatosis, and familial hypercholesterolemia) or at moderate risk for a common complex disorder (such as heart disease or diabetes). A positive family history might indicate that a person is a candidate for diagnostic genetic testing, as in the recommendation to offer BRCA1/2 testing to women who have a family history of some combinations of breast and ovarian cancer (Siu et al., 2016). Family history can also guide preventive-care recommendations, such as when to begin lipid therapy (Jacobson et al., 2014), colorectal cancer screening (Rex et al., 2009), or diabetes screening (CDC Diabetes Cost-Effectiveness Study Group, 1998). This practice depends on the observation that intrafamily variability is less than interfamily variability in many strongly genetic conditions.

When people present with clinical signs or symptoms that suggest a genetic disorder, family history is an important part of evaluation. A detailed three-generation pedigree is used to determine whether other members of the family are similarly affected or have conditions associated with the patient’s condition. Assessment of family history can be crucial in narrowing the differential diagnosis and identifying the most informative tests for further evaluation and might be helpful in interpreting test results.

Family history is also collected before preconception testing or prenatally in obstetric practice to identify the presence of genetic disorders for which carrier screening or other reproductive genetic testing could be offered or to predict potential problems in the fetus or newborn.

Clinicians always need to be mindful of the limitations of family history. People might be unaware of relevant family history or misreport it. Family history might be absent because of premature deaths from other causes or because the family structure limits the manifestation of genetic risk, as in a lack of evidence of hereditary breast ovarian cancer syndrome in a family that has few females. Family history also changes as new diagnoses are made, so it might be less revealing in young adulthood than in later life (Ziogas et al., 2011). For all those reasons, it is useful to update family history during the course of longitudinal care and, when appropriate, to obtain ancillary information from other family members.

Uses of Genetic Tests

The committee organized its thinking around the intended clinical application of a genetic test that is being considered. Applications included diagnostic genetic testing, predictive genetic testing, and reproductive genetic testing. An additional consideration is whether a test is

performed in response to a specific clinical indication or as part of a screening process. Some tests can be used both for screening and also in the context of a clinical problem.

Diagnostic Genetic Testing

Diagnostic genetic testing is used to identify or rule out a specific genetic condition. Genetic testing is often used to confirm a diagnosis when a particular condition is suspected on the basis of physical signs and symptoms. For example, patients with a clinical diagnosis of retinitis pigmentosa (OMIM, 2016d) can have a sequence-based test to identify the variant, in which of the more than 50 genes, that is actually responsible for the patient’s condition. That information not only confirms the diagnosis and individualizes the prognosis, but it can also be useful to relatives and be relevant to treatment.

As noted above, family history might play an important role in identifying appropriate diagnostic tests. For example, genetic testing for inherited cancer syndromes might be performed on the basis of family history suggesting inherited risk, as well as for early onset of cancer. The results of a diagnostic test can influence patients’ and clinicians’ choices about health care and the management of the disorder. Diagnostic testing is not available for all genetic conditions, but the number of tests available has expanded rapidly over the last decade and is likely to continue to increase.

Predictive Genetic Testing

Predictive genetic testing identifies gene variants that increase a person’s risk of developing heritable disorders, such as some types of cancer, before signs or symptoms appear. In some cases, a diagnostic test performed on an affected person yields a result that can be used to recommend a predictive test for relatives. For example, if a woman who has breast cancer is found to have a BRCA1 variant, indicating hereditary breast ovarian cancer syndrome, her relatives can be offered the option of being tested to determine whether they also carry the risk variant for the cancer syndrome. Similarly, when Huntington disease is first diagnosed, first-degree relatives can be tested to determine whether they have inherited the pathogenic variant for the condition. The results of predictive testing can inform preventive care and decisions about lifestyle choices and medical care. The options available after predictive testing vary with the condition. In the case of hereditary breast ovarian cancer, people who have a positive test result can be offered specific screening and prophylactic surgery options to reduce their cancer risk.

Newborn Screening

Predictive tests can be used for population screening, notably in the case of newborn screening (NBS). NBS involves testing infants a few days after birth to screen for evidence of treatable diseases that are known to cause problems in health and development. All states test infants for a large number of inborn errors of metabolism, including phenylketonuria (a genetic disorder that causes intellectual disability), which can be treated with appropriate changes in diet. Most NBS tests use tandem mass spectroscopy to detect biochemical abnormalities that suggest a specific disorder or group of disorders. Testing thresholds are set to optimize identification of affected infants. As a result, NBS tests often have a relatively high rate of false-positive results, and confirmatory testing is used. Current research is exploring the potential for DNA-based tests to complement existing NBS methods to identify additional genetic disorders or to replace

metabolic measures for some genetic disorders (Berg and Powell, 2015; Friedman, 2015; Almannai et al., 2016; Berg et al., 2017).

The potential for comparable genetic screening of adults to identify conditions that are amenable to prevention has been proposed (Evans et al., 2013). Some experts have suggested targeted adult genetic-screening programs, such as screening of young adult women for hereditary breast ovarian cancer (King et al., 2014). However, no adult genetic-screening programs have entered routine practice.

Pharmacogenetic Testing

Pharmacogenetic testing is an important example of predictive genetic testing. It provides information about individual variation in pharmacodynamics (effects on drug receptors) and pharmacokinetics (uptake, distribution, and metabolism) and makes it possible to identify patients who are at increased risk for adverse effects or who are likely to be nonresponders. Pharmacogenetic testing has the potential to help health care providers to tailor therapies by identifying a dose or drug that might work best for a patient, to prevent adverse drug reactions, or to select the people who are likely to respond to a given drug. The evidence base on pharmacogenetics is limited (Wang et al., 2014), but a few pharmacogenetic tests have entered clinical practice, and more are likely to be developed. One form of pharmacogenetic testing is termed companion diagnostics. The tests developed in conjunction with drug development to identify candidates that are likely to respond to a drug (FDA, 2014). Like other types of predictive testing, pharmacogenetic testing can be used for a specific indication (that is, when a particular drug treatment is planned) or as a routine screening test for patients who are receiving care in a particular health care setting; for example, St. Jude Children’s Research Hospital provides a panel of pharmacogenetic testing to pediatric patients in the PG4KDS study, using the comprehensive microarray Drug Metabolizing Enzymes and Transporters (DMET™; Haga and Moaddeb, 2014).

Reproductive Genetic Testing

Reproductive genetic testing offers the opportunity to identify people who are at increased risk for having a child who is affected with a genetic disease or to identify an affected embryo or fetus.

Carrier Genetic Testing

Carrier (heterozygote) genetic testing is used to identify people who are at risk for having a child who has a genetic disease. Most carrier tests identify people who are heterozygous for—who “carry” one variant copy and one normal copy of genes associated with—disorders that are transmitted as autosomal recessive traits (which require both copies of the gene to be pathogenic for the disease to be expressed). Carriers typically show no signs of the disease, but they have the ability to pass on the variant gene to their children; if the other parent is heterozygous for the same disease, the child might inherit defective forms of the same gene from each parent and, as a result, have the genetic disorder. Some carrier tests identify gene variants for X-linked recessive disorders, that is, variants of genes found on the X chromosome that cause disease only in the absence of a normal variant. Thus, a woman who has the variant on one of her two X chromosomes and a normal copy on the other is a heterozygote and usually does not show signs of disease; however, a son that inherits as his only X chromosome the one that has the disease-

causing variant will be affected. Carrier testing is offered either on the basis of family history indicating the presence of a specific inherited disease or as a screening test. Current guidelines recommend that a small number of carrier tests be offered routinely in obstetric practice and that additional carrier tests be offered to prospective parents in particular ethnic groups that have a higher risk of specific inherited diseases (Edwards et al., 2015). If carrier testing identifies a risk for a child who will have a genetic disorder, prospective parents can accept the risk, choose to have prenatal testing with the option of considering pregnancy termination of an affected fetus, or pursue other reproductive options, including preimplantation genetic testing.

Prenatal Genetic Testing

Prenatal genetic testing is used to detect abnormalities in the genes or chromosomes of a fetus. Current guidelines recommend that all pregnant women be offered maternal serum-screening tests to identify pregnancies at increased risk for a trisomy disorder (such as Down syndrome) and for neural tube defect. If the test result is positive, a confirmatory test can be performed. Prenatal testing might also be offered if the parents are known to be at risk for having a child that will have a specific genetic disorder. Some parents choose to terminate a pregnancy if prenatal testing reveals an affected fetus. Others use the information to prepare for the child’s birth and care.

Noninvasive prenatal testing is a new type of screening for genetic abnormalities in a developing fetus that is based on the fact that fetal cells and cell-free DNA circulate in the maternal blood. DNA is extracted from a maternal blood sample and is sequenced by using NGS to screen for the increased chance of specific chromosomal problems, such as Down syndrome, in the fetus. It can also provide information about fetal sex and rhesus (Rh) blood type.

Preimplantation Genetic Testing

Preimplantation genetic testing, or preimplantation genetic diagnosis, is a specialized technique that can reduce the risk of having a child that has a particular genetic or chromosomal disorder. It is used to detect genetic changes in embryos that were created with assisted reproductive techniques, such as in vitro fertilization. To perform preimplantation testing, a small number of cells are taken from the embryo and tested for the presence of pathogenic variants associated with the conditions that are of concern. Only embryos that do not have those changes are transferred into the uterus to initiate a pregnancy.

Clinical Considerations in Using Genetic Tests

Clinicians are often faced with choosing which genetic tests are most appropriate for a particular patient in a particular clinical context. In each scenario, the principle most commonly used is to select a test with high clinical sensitivity, maximal specificity, and minimal cost. Searching for a laboratory that provides testing for a particular gene is greatly enabled by resources such as OMIM and GeneReviews (Pagon, 2006; OMIM, 2017). In some scenarios, WES or WGS might be used clinically but with the intent of analyzing only a single gene or a small number of genes rather than the full dataset; problems related to VUSs and incidental findings can be minimized by offering the test as a targeted analysis of only particular genes in the exome or genome and omitting the remainder. That strategy and its consequences should be clearly explained to the patient to be tested before the test is performed.

A clinical choice that often arises is whether to use a multigene panel or more comprehensive tests, such as WES or WGS. Available multigene panels include about 5–100 genes for a specific condition; other panels, typically of hundreds of genes, are available for genetically heterogeneous conditions, such as intellectual disability or epilepsy (Helbig et al., 2008; Gécz et al., 2009; Rehm, 2013). The more unusual and more specific the constellation of clinical features, the more likely it is to be able to develop a focused differential diagnosis and identify an appropriate gene panel. For clinical presentations, where a genetic etiology is suspected, but a specific set of candidate genes is difficult to identify, WES and WGS might be more appropriate (Biesecker and Green, 2014).

Ethical, Legal, and Social Implications of Genetic Testing

Genetic testing has an array of personal and societal implications that require consideration in defining appropriate testing practice. Those implications are related to shared genetic risk among family members, the use of genetic testing in reproductive decision making, and the potential for genetic information to generate stigma or discrimination. In addition, the rapid development of genomic technology and the many remaining uncertainties about the health implications of genetic risk raise ethical considerations related to health care disparities, clinical data sharing, and the scope of result reporting from genome-scale testing.

Shared Genetic Risk

When a person has a diagnosis of a genetic abnormality, the diagnosis might have implications for family members. The particular family members most at risk of having inherited the condition and the magnitude of the risk depend on the mode of inheritance and the nature of the genetic disease. In the cases of hereditary breast ovarian cancer, Lynch syndrome, and Huntington disease, inheritance follows an autosomal dominant pattern; that is, people of both sexes have a 50% chance of inheriting the condition from the affected parent. However, the nature of the condition is a factor to be considered by family members. For example, males who inherit the hereditary breast ovarian cancer syndrome have a moderately increased risk of breast and prostatic cancer, but most will not develop cancer; they can, however, pass the risk on to their daughters. In the case of Huntington disease, the symptoms typically occur in the fourth decade or later, so children of an affected person might not know that they are at risk until they are adults. For conditions that are inherited as autosomal recessive disorders, one inherits two mutated genes, one from each parent. Such disorders are usually passed on by two parents who are carriers for a specific condition; each of whom has one mutated copy of the gene and one normal copy of the gene for the condition. Two carriers have a 25% chance of having an unaffected child with two normal genes, a 50% chance of having an unaffected child who also is a carrier, and a 25% chance of having an affected child with two recessive genes (Mayo Clinic, n.d.). Thus, for example after a diagnosis of cystic fibrosis is made, it is important to test siblings to identify whether they are likely to become symptomatic or be carriers.

For X-linked recessive disorders, family implications are more complex. If the mother is a carrier, all of her sons have a 50% chance of being affected, and all of her daughters have a 50% chance of being carriers. However, the rate of de novo mutation is high for some X-linked recessive conditions; that is, the condition has occurred as a result of a new mutation in the affected person rather than as a result of inheriting the condition from a carrier.

Genetic testing can help to distinguish inherited from de novo mutations. The distinction can have implications for family members. Parents of an affected child might feel guilt for passing on a genetic condition, and family members who did not inherit a condition that is present in the family might experience survivor guilt. There are also questions about the obligation to share information among family members. Geneticists and genetic counselors typically help patients to identify family members who would most benefit from genetic testing after a genetic diagnosis has been made in a family.

Reproductive Decision Making

Prenatal testing can reveal the presence of a fetus that has a genetic condition. If the condition is serious, some parents might consider termination an option, and others might not. Because decisions about whether to continue a pregnancy are based on personal values, genetics professionals use a “nondirective” counseling approach (Vanstone et al., 2012). Such counseling seeks to provide sufficient information about test results and the condition inherited by the fetus so that parents can make the decision most appropriate for them and their families; counseling is intended to provide support but not to recommend a particular course of action.

As genomic technology develops, health care systems need to consider whether limits should be placed on the genetic testing to be offered prenatally. Genome-scale sequencing and noninvasive prenatal testing, in which fetal test results are derived early in pregnancy from testing of fetal DNA present in maternal serum, make it increasingly possible to generate a wide array of genetic information about the fetus. A joint statement of the American and European Societies of Human Genetics notes the importance of attention to a broad range of concerns in prenatal testing innovation, including technical issues, counseling practices, education, and accountability to all stakeholders; the statement recommends that the scope of prenatal testing be limited to serious congenital and childhood disorders for which valid testing and effective service delivery are available (Dondorp et al., 2015).

Stigma and Discrimination

One of the concerns in the use of genetic testing is that it might result in stigma or discrimination. An early example was manifested in efforts to screen for “sickle trait”—the carrier state for sickle-cell disease—in the 1970s. Broad-based community screening programs were often accompanied by misunderstanding about the health implications of the carrier state, and this led to employment and insurance discrimination (Wailoo, 1999). In one instance, mandatory screening of blacks by an employer led to discriminatory decisions concerning job placement, which resulted in a court decision against the employer.¹⁴ As genetic testing options increased, the potential for discrimination against people who had genetic susceptibilities became a major policy concern, and this resulted in several state antidiscrimination laws and ultimately the passage of the federal Genetic Information Nondiscrimination Act (GINA) in 2008. GINA blocks health insurers and employers from using genetic information in health-coverage or employment decisions. The Patient Protection and Affordable Care Act (ACA) of 2010 added further protection by barring the use of pre-existing conditions to deny coverage or to raise premiums. Nevertheless, private insurers might use genetic information for decisions related to other types of coverage, such as life insurance, mortgage insurance, long-term care insurance, and long-term disability insurance. Despite the protections afforded by GINA and the ACA,

___________________

¹⁴Norman Bloodsaw v. Lawrence Berkely Laboratories, 135 F.2d 1260(9th Cir. 1998).

potential insurance discrimination is of great concern to patients who are considering genetic testing and to people who are participating in clinical research studies that involve genetic testing (Green et al., 2015).

A related concern is that a genetic label might have adverse social consequences. For example, some research has suggested that the perception of a mental illness having a genetic cause results in social distancing, not only for the person who is affected but for family members. As research on behavioral genetics proceeds, the possibility emerges that some people might be labeled as more prone to problematic behaviors, such as substance-use disorders, although genetic associations with such outcomes are likely to be highly probabilistic (Turner, 1995).

Nonetheless, there is considerable movement toward “de-exceptionalizing” genetic medicine and considerable enthusiasm for the idea that genetics will soon be integrated into the day-to-day practice of medicine (Green et al., 2016).

Potential for Health Care Disparities

Genetic research has been conducted disproportionally in European populations (Popejoy and Fullerton, 2016). As a result, there is much less knowledge about genetic variation and its implications for health in non-European populations. As a practical consequence, members of non-European populations who undergo DNA sequencing have a higher likelihood of receiving results that are classified as variants of uncertain significance or an otherwise ambiguous finding (Lek et al., 2016). In a recent study of genetic testing for inherited cardiomyopathy, researchers found that Americans of African descent were more likely than Americans of European descent to have a false-positive finding that erroneously suggested the presence of genetic risk (Manrai et al., 2016). Similarly, the genes and gene variants relevant to warfarin metabolism vary among populations (Schellerman et al., 2008; Yang et al., 2010; Suarez-Kurtz and Botton, 2013), and information on some populations is still limited (Fohner et al., 2015). Those findings point to a health care disparity in genetic testing in which tests are generally more effective and clinically useful for people of European descent. More knowledge of the relevant genetic variations will be needed to provide genetic tests that are of equal efficacy in all populations.

Clinical Data Sharing

Sequencing technologies and other approaches to genome-wide testing often identify novel and rare gene variants of uncertain clinical significance. Those findings generate uncertainty and limit the clinical utility of testing. Resolution of that uncertainty is an important challenge for clinical genetics. The capture of information about associations between genetic variation and clinical outcome that are identified in longitudinally studied populations and in the course of clinical care is needed (Angrist and Jamal, 2015; Natarajan et al., 2016), but efforts to do so require careful attention to scientific accuracy and protection of patient privacy (McEwen et al., 2013). ClinVar,¹⁵ a public archive of information about associations between genotype and phenotype (Harrison et al., 2016; Landrum et al., 2016), offers a partial solution. In that archive, laboratory submissions of gene variants, with clinical interpretations and any supporting evidence, are collated, with linkage to other relevant databases, for example, variant frequency data from the dbGAP data base.¹⁶

___________________

¹⁵ Available at: https://www.ncbi.nlm.nih.gov/clinvar (accessed January 31, 2016).

¹⁶ Available at: https://www.ncbi.nlm.nih.gov/gap (accessed January 31, 2016).

Scope of Return of Results from Genome-Wide Testing

One of the consequences of whole exome, whole genome, and other genome-wide tests is the potential to produce large amounts of information that are not directly relevant to the clinical question for which testing is done (Biesecker and Green, 2014; Johansen Taber et al., 2014). Thus, for example, testing done to evaluate a cardiomyopathy could also generate secondary information about inherited cancer susceptibilities, inherited arrhythmias, and a wide array of other genetic conditions, depending on what analysis of DNA sequence is undertaken. ACMG has proposed that a discrete set of 59 genes be interrogated when genome-wide sequencing is done (Green et al., 2013; Kalia et al., 2016); those genes were chosen based on their potential to generate findings that are medically actionable. That approach has generated discussion and controversy (e.g., Burke et al., 2013; McGuire et al., 2013; Wolf et al., 2013; Anderson et al. 2015; Wilfond et al., 2015). Key questions include: the scope of informed consent; the degree of choice patients should have about receipt of secondary findings; whether findings about adult-onset conditions should be reported in testing of children; and whether analysis of secondary gene targets should occur if patients have declined results.