Types of Genetic Variation

Genetic variation comes in many forms, and single nucleotide variants (SNVs), small insertions/deletions (indels), exonic deletions, and trinucleotide repeats have all demonstrated roles in the pathogenesis of lung disease ( Fig. 3.2 ). SNVs are single nucleotide changes in the genome and can occur in coding and noncoding regions. SNVs in coding regions can have relatively little effect on protein function (e.g., no change in amino acid sequence [synonymous] or conservative amino acid substitution), can result in a truncated or nonfunctional protein (e.g., nonsense, frameshift SNVs), or can have intermediate effects (nonconservative missense, splice site, and in-frame indels). SNVs that are present in greater than 1% of the population (minor allele frequency > 0.01) are referred to as single nucleotide polymorphisms (SNPs), and these “common” variants have enabled research methodologies, including genome-wide association studies (GWAS). SNVs present in less than 1% of the population are termed “rare” variants, and, although most are still of modest effect size, they are more likely to have larger effect sizes than SNPs. The term “mutation,” often used interchangeably with SNVs and SNPs, has a more negative connotation and may even imply a “disease-causing change,” whereas the term “polymorphism” has a more neutral connotation and indicates a more common variant. Indels may affect one or more nucleotides and can be “in-frame,” meaning they occur in multiples of three nucleotides and add or delete amino acids in that region without disrupting the remainder of the amino acid sequence, or they can cause a “frameshift,” meaning they disrupt the reading frame and all subsequent amino acid sequence. In-frame indels generally result in less disruption to the reading frame, but, as evidenced by the delF508 mutation in CFTR, even the deletion of three bases and a single amino acid can be deleterious. Indels that result in addition or deletion of bases that shift the reading frame are frequently pathogenic.

Types of single nucleotide variation. A representative DNA sequence of nucleotide triplets comprising a codon and the corresponding encoded amino acid (blue box) is shown at the top. A single nucleotide substitution (red) may have no effect on amino acid sequence (synonymous), may change the amino acid sequence (nonsynonymous), or may encode a signal (TAG) to stop transcription prematurely, which typically results in an unstable mRNA and absence of detectable protein. An insertion (or deletion) of a single nucleotide changes the triplet reading frame and changes the subsequent amino acid sequence.

Although the vast majority of monogenic lung disease is caused by SNVs, there are important examples of less common genetic variation causing lung disease. For example, the absence of dozens or hundreds of base pairs encompassing one or more exons (so-called exonic deletions) can cause serious disease. For the pulmonologist the prime examples are Duchenne and Becker muscular dystrophies. Methods for assaying SNVs including Sanger sequencing will not reliably detect exon skipping (see Table 3.3 later in this chapter), and methods such as multiplex ligation-dependent probe amplification (MLPA) must be used when there is high clinical suspicion for these disorders. Disorders such as myotonic dystrophy result from expansion of three nucleotide repeat (so-called trinucleotide repeat) regions in coding regions that result in unstable and abnormal proteins. Methods for assaying SNVs will not reliably detect trinucleotide repeats, and methods such as polymerase chain reaction (PCR) must be used when there is high clinical suspicion for these disorders.

In addition to single gene variation, copy number variation (deletions, duplications) may occur at the whole chromosome level (e.g., trisomy 13, 18, 21), where pulmonary effects are largely of secondary importance as compared with other organ dysfunction, but also at the microscopic or submicroscopic level, affecting only a portion of a single chromosome or as a translocation involving two or more chromosomes. Autosomal dominant disorders may be caused by deletions, autosomal recessive conditions may be unmasked by deletion of the “normal” allele, and translocations with breakpoints within a given gene may cause similar effects.

Technologies to Identify Genetic Variation

The Human Genome Project, with initial sequencing completed in 2001, has led to the development of multiple technologies of relevance to pediatric lung disease from both clinical and research perspectives. These technologies are being increasingly integrated into clinical medicine, often blurring the lines between clinical diagnostic evaluation and research. Thus it is important to understand both the technologies in current clinical use, as well as those being used on a research basis, to interpret an emerging body of literature and to anticipate the maturation of such technologies for clinical application (see Table 3.2 ).

Historically, chromosomal or cytogenetic variation was assayed using the classical karyotype or fluorescent in situ hybridization (FISH); however, array-based technologies have become more prevalent in both research and clinical care. Chromosomal microarray (CMA) consists of hundreds of thousands to millions of DNA probes (as oligonucleotides with or without SNPs) spotted onto a solid surface. DNA (or RNA) is hybridized to these probes, and hybridization is measured by a fluorescent reporter. CMA can assess for copy number variation (microdeletions, microduplications, and unbalanced translocations) and (when SNPs are used) regions of excessive homozygosity, suggesting increased risk for recessive or imprinting disorders. Although often used as a first-line tool for genetic analysis, CMA will not detect small changes in the sequence of single genes including point mutations, tiny duplications or deletions within a single gene (e.g., between array probes), or balanced chromosomal rearrangements (translocations, inversions). Microarray can also be used to assess RNA expression as a research tool (e.g., comparing RNA transcripts for an SNV predicted to alter exon-intron splicing).

The International HapMap Project demonstrated that SNPs at different positions in the genome are frequently inherited together, a phenomenon known as linkage disequilibrium. Although millions of SNPs were present in the human genome, one could infer, through linkage disequilibrium, many more variants by sequencing only a subset of these “tagSNPs.” The use of tagSNPs facilitated the rise of GWAS using arrays that contain up to 2 million previously identified variants.

The Human Genome Project also fostered significant advancement in sequencing technologies. Classically, sequencing of single genes was done by Sanger sequencing. Highly reliable and able to generate reads of greater than 500 base pairs, Sanger sequencing is still used to validate limited numbers of selected variants generated by high-throughput sequencing technologies; however, it does not scale well to sequence multiple genes. “Next-generation” sequencing technologies allow rapid sequencing of hundreds of genes or even whole exomes or genomes simultaneously by combining microarray technology and parallel sequencing of short reads (50–400 base pairs) with computational techniques to align the fragments to a reference sequence.

Whole exome sequencing (WES) targets the exonic (protein coding) regions of all approximately 20,000 genes in the genome simultaneously and has become a widely used clinical diagnostic tool to identify rare sequence variants in patients with a phenotype suspected to be due to disruption of a single gene. However, exons comprise only approximately 1% of the genome. Whole genome sequencing (WGS) has therefore emerged as a diagnostic tool and offers several advantages over exome sequencing, including detection of structural variation and variation in nonexonic regulatory, intronic, and intergenic regions. In silico programs to interpret structural and noncoding variation are emerging but lag behind those used for coding variation; currently, genome sequencing is not widely available and is significantly more expensive than exome sequencing. Clinical exome and genome sequencing is estimated to identify a causative variant in approximately 25% of trios (affected child plus both parents), but this depends largely on patient selection based on phenotype, family history, and suspected inheritance pattern. Clinical exome and genome sequencing can lead to the identification of previously identified variants or suspected pathogenic variant(s) in a gene known to be associated with human disease. Alternatively, exome or genome sequencing may identify suspected pathogenic variants in a gene not previously associated with a human phenotype, in which case additional clinical or laboratory investigation may be needed to establish pathogenicity. Positive results from clinical exome or genome sequencing are highly accurate, but false-negative results can occur, depending on the quality of data from a specific genomic region. In addition, the clinician must consider how well the putative variant(s) explains the patient’s phenotype in terms of clinical assessment and interpretation of existing medical literature. In some individuals, more than one candidate gene may be identified or, alternatively, no candidate variants may be identified. Reasons for negative findings include lack of coverage of a genomic region, limitations of variant prediction algorithms, polygenic inheritance, epigenetic mechanisms, or lack of an underlying genetic etiology. Reanalysis of sequence data may be of use as variant prediction algorithms mature, additional patients with similar phenotypes are reported, and the functions of additional genes are characterized.

Interpretation of Genetic Variation

Because each individual has millions of genetic variants, the vast majority of which are functionally insignificant, the principal challenges of sequencing are interpretation, as opposed to sequence generation. For previously described SNVs many databases exist to assist in interpretation or classification, examples of which are listed in Table 3.2 . Some programs, such as ANNOVAR, efficiently combine the results of multiple in silico prediction algorithms. However, it should be noted that such databases are not infallible and variant classification is very dynamic (i.e., a variant that might be classified as “benign” or “of unknown significance” might be reclassified with accumulation of additional information). ClinVar provides the most current cataloging of these variants and also allows for conflicting interpretations of the same variant( http://www-ncbi-nlm-nih-gov.easyaccess2.lib.cuhk.edu.hk/clinvar/ ). Furthermore, because sequencing has the potential of identifying novel variants, both the clinician and the researcher must have a basic understanding of in silico prediction methods and their limitations. Wu and Jiang provide an overview of available tools for this purpose and caution that, although specific algorithms may outperform others under specific conditions, best results are typically obtained by using multiple algorithms and integrating knowledge over all available domains. In addition to in silico prediction algorithms, functional studies and identification of variants in other diseased (or nondiseased) individuals can also aid in variant interpretation. Given the limitations of in silico prediction, there is increasing demand for biologic confirmation of in silico predictions. Although a detailed exploration of the various methods of developing model systems is beyond the scope of this chapter, special attention should be paid to the emerging technology of CRISPR/Cas9, which permits targeted genome editing and is being applied to existing model organisms to test functionality suggested by in silico prediction tools.

If one takes a genomic approach and obtains WES/WGS, many (most) of the genes sequenced and variants that are identified will not have anything to do with the clinical phenotype. Variants in such genes are termed secondary findings and are an area of significant controversy. As the availability of genomic testing spreads and nongeneticist clinicians are able to order such testing, they must be aware of the potential for secondary findings and provide adequate counseling on this possibility. The American College of Medical Genetics has a position statement on secondary findings that can guide clinicians.