Human Inheritance, Genetic Testing and Counseling

Everything in this book comes together when a family sits down with a clinician and asks what a diagnosis means for them, their children and their relatives. Human genetics applies Mendel, Hardy and Weinberg, Bayes and the molecular biology of mutation to real decisions: whether to test, how to read a result, how to estimate a risk and how to share it. This chapter covers the major patterns of single gene disease, the genetics of common complex disease, the tests now available from before conception to adulthood, the arithmetic of genetic counseling and the ethical questions that come with knowing more about our genomes.

Section 1Single gene disorders in humans

How human inheritance is studied

Human geneticists cannot arrange crosses, so they work backward from families. A pedigree records relationships and phenotypes across generations, and the inheritance pattern is inferred from who is affected. The person through whom a family comes to attention is the proband; the person seeking advice is the consultand. Single gene (Mendelian) disorders are individually rare, but there are thousands of them, and together they affect roughly 1 percent of people. Catalogs of human genes and phenotypes now list more than 7,000 Mendelian conditions with a known molecular basis.

Autosomal dominant disorders

In autosomal dominant (AD) inheritance, one copy of the variant allele is enough to cause disease. The pedigree shows affected individuals in every generation (vertical transmission), males and females affected about equally, and male to male transmission, which rules out X linkage. Each child of an affected heterozygote has a 1/2 chance of inheriting the variant. Most affected people are heterozygous; homozygotes, when they occur, are often more severely affected.

Dominant disorders arise by several molecular mechanisms. Some involve haploinsufficiency, in which half the normal protein is not enough (familial hypercholesterolemia from LDLR variants, with half the normal LDL receptors). Some involve a dominant negative effect, in which an abnormal protein disrupts the normal one, as in Marfan syndrome (FBN1, fibrillin 1) and osteogenesis imperfecta (collagen type I). Others are gain of function, as in achondroplasia, where a specific FGFR3 variant (Gly380Arg) makes a growth plate receptor overactive, and Huntington disease, where an expanded protein acquires toxic properties.

Three features complicate dominant pedigrees:

• Incomplete penetrance: some people with the genotype show no phenotype. Penetrance is the percentage of genotype carriers who are affected. Hereditary retinoblastoma is about 90 percent penetrant; BRCA1 variants confer a high but incomplete lifetime cancer risk. A nonpenetrant carrier produces a "skipped generation."
• Variable expressivity: carriers differ in severity. In neurofibromatosis type 1, one family member may have only a few café au lait spots while another has thousands of neurofibromas. Penetrance is all or none for a person; expressivity is a matter of degree.
• New mutations: about 80 percent of people with achondroplasia have unaffected parents. Many such new variants arise in sperm, and their frequency rises with paternal age. Occasionally a parent carries the variant in part of the germ line only (germline mosaicism), so two children are affected although both parents test negative in blood.

Autosomal recessive disorders

In autosomal recessive (AR) inheritance, disease requires two variant alleles. Affected people usually have unaffected heterozygous (carrier) parents; the pattern is horizontal, appearing in siblings rather than across generations; males and females are equally affected; and consanguinity raises the risk, because related parents are more likely to carry the same rare allele. Two carrier parents have, at each conception, a 1/4 chance of an affected child, 1/2 of a carrier and 1/4 of a noncarrier. Most recessive disorders result from loss of function, typically of enzymes, which explains why carriers (with 50 percent activity) are healthy. Many inborn errors of metabolism follow this pattern.

X linked disorders

Because males are hemizygous for the X chromosome, a single recessive allele on the X is expressed in a male. In X linked recessive inheritance, affected people are mostly males; affected males are related through unaffected carrier females; there is no male to male transmission (fathers pass a Y to sons); and all daughters of an affected male are obligate carriers. A carrier mother has a 1/2 chance of passing the allele to each child, so half her sons are affected and half her daughters are carriers. Carrier females are usually healthy but can show symptoms because of skewed X inactivation. Duchenne muscular dystrophy, hemophilia A and B, red green color blindness and G6PD deficiency are examples. In X linked dominant disorders (such as X linked hypophosphatemic rickets), affected fathers pass the condition to all daughters and no sons, and females are affected about twice as often as males, usually more mildly.

Mitochondrial inheritance

The 16,569 base pair mitochondrial genome encodes 13 proteins of oxidative phosphorylation plus 22 tRNAs and 2 rRNAs. Mitochondria are inherited through the egg, so mitochondrial disorders show maternal inheritance: an affected mother can pass the condition to all her children, while an affected father passes it to none. Cells contain hundreds to thousands of mitochondrial genomes, and a mixture of normal and variant genomes (heteroplasmy) is common. Because the proportion can shift at random during egg formation (the mitochondrial bottleneck) and among tissues, severity varies greatly within a family, and symptoms appear when the variant fraction passes a tissue's threshold. Tissues with high energy demand (brain, muscle, retina, heart) suffer most, as in MELAS (mitochondrial encephalomyopathy, lactic acidosis and strokelike episodes) and Leber hereditary optic neuropathy.

Repeat expansions and anticipation

Several neurological disorders are caused by trinucleotide repeat expansions, which are unstable and tend to grow when transmitted. Longer repeats cause earlier onset and more severe disease, so the disorder appears earlier and worse in successive generations, a phenomenon called anticipation. The parent of origin matters: Huntington disease repeats expand mainly in male meiosis (juvenile cases usually inherit from the father), while fragile X and congenital myotonic dystrophy expansions occur through the mother.

Genomic imprinting and other exceptions

For a few genes, only the copy from one parent is expressed, because the other is silenced by DNA methylation laid down in the parent's germ line. This is genomic imprinting. The classic pair is on chromosome 15q11 to q13. Loss of the paternal contribution (a paternal deletion in about 70 percent of cases, or maternal uniparental disomy, two maternal copies and no paternal copy, in about 25 percent) causes Prader-Willi syndrome, with neonatal hypotonia, later overeating and obesity, and mild intellectual disability. Loss of the maternal contribution, specifically of maternal UBE3A expression in neurons, causes Angelman syndrome, with severe intellectual disability, seizures, absent speech and a happy demeanor. The same deletion produces two different diseases depending on which parent it came from.

Two more terms are frequently confused. Locus heterogeneity means variants in different genes cause the same phenotype (hearing loss from any of more than 100 genes; two deaf parents with recessive deafness in different genes can have all hearing children). Allelic heterogeneity means different variants in the same gene, such as the more than 2,000 known CFTR variants, cause disease. Pleiotropy means one gene affects several apparently unrelated traits, as FBN1 affects the eye, skeleton and aorta.

Section 2Complex and multifactorial disease

From single genes to many

Most common diseases, including coronary artery disease, type 2 diabetes, asthma, schizophrenia, most birth defects and most cancers, run in families but do not follow Mendelian ratios. They are complex or multifactorial traits, shaped by many genetic variants of small effect acting together with environmental factors. Some, like height or blood pressure, are quantitative traits that vary continuously. Others, such as cleft lip or pyloric stenosis, are present or absent, yet still have a polygenic basis. The bridge between these two ideas is the liability threshold model.

The liability threshold model

Imagine that every person has an unseen liability to a disease, the sum of many genetic and environmental contributions. Because it is built from many small additive factors, liability is normally distributed in the population. Only people whose liability exceeds a threshold develop the disease. Relatives of an affected person share some of that person's risk alleles, so their liability distribution is shifted toward the threshold, and a larger fraction of them cross it.

The model explains several characteristic features of multifactorial traits that distinguish them from Mendelian ones:

• Recurrence risk is low but above the population risk, typically 2 to 5 percent for first degree relatives of an affected person, compared with a population risk often well under 1 percent.
• Risk falls off steeply with distance of relationship: much more than the halving seen for a dominant gene. First degree relatives share half their genes, second degree a quarter, and the risk drops toward the population value quickly.
• Risk rises with the number of affected relatives: two affected children suggest the parents carry many risk factors. (For a Mendelian disorder, by contrast, the risk to the next child of carrier parents stays 1/4 no matter how many children are affected.)
• Risk rises with severity: relatives of a person with bilateral cleft lip and palate have a higher risk than relatives of a person with a small unilateral cleft.
• Risk is higher for relatives of the less often affected sex. Pyloric stenosis is about five times more common in boys. A girl must carry more liability to be affected, so her relatives have a higher risk than relatives of an affected boy. This is the Carter effect.

Heritability and twin studies

Heritability is the proportion of the variation in a trait, within a particular population and environment, that is due to genetic differences. Broad sense heritability (H² = V_G/V_P) includes all genetic variance; narrow sense heritability (h² = V_A/V_P) includes only additive variance, the part that predicts resemblance between parents and offspring. Heritability describes variation in a population, not the fraction of an individual's trait due to genes, and it can change when the environment changes. A trait can be highly heritable yet strongly changed by the environment: height is about 80 percent heritable, yet average heights rose substantially over the last century as nutrition improved.

Twin studies estimate genetic influence by comparing monozygotic (MZ) twins, who share essentially all their DNA, with dizygotic (DZ) twins, who share on average half, like ordinary siblings. For an either/or trait, we compare concordance, the proportion of twin pairs in which both twins are affected given that one is. If MZ concordance is much higher than DZ concordance, genes matter; if MZ concordance is well below 100 percent, the environment (or chance) also matters. For quantitative traits, we compare correlations, and Falconer's formula estimates h² = 2(r_MZ - r_DZ).

Finding the genes: genome wide association studies

Linkage analysis, which tracks a marker through large families, works well for Mendelian disease but poorly for common disease, where each variant adds only a little risk. The genome wide association study (GWAS) instead compares the frequency of hundreds of thousands to millions of single nucleotide polymorphisms (SNPs) between many unrelated cases and controls. A SNP associated with disease is either causal or, more often, in linkage disequilibrium with a nearby causal variant. Because so many tests are performed, the conventional threshold for genome wide significance is p < 5 × 10^-8 (roughly 0.05 divided by one million independent tests).

The strength of an association is reported as an odds ratio (OR): the odds of carrying the risk allele among cases divided by the odds among controls. Most GWAS hits for common diseases have odds ratios of only 1.1 to 1.5. Well known exceptions with larger effects include the APOE ε4 allele in late onset Alzheimer disease (roughly 3 fold risk for one copy and 8 to 12 fold for two) and variants in the HLA region in autoimmune disease.

GWAS revealed that common diseases are highly polygenic: hundreds to thousands of loci each contribute a small effect, and most lie in noncoding regulatory DNA rather than in protein coding sequence. Together, identified SNPs often explain less of the heritability than twin studies predict, the problem of missing heritability, attributed to many variants with effects too small to detect, rare variants not captured on SNP arrays and gene by environment interactions. Summing a person's risk alleles, each weighted by its effect size, gives a polygenic risk score (PRS). People in the top few percent of a coronary disease PRS can have a risk comparable to carriers of a monogenic familial hypercholesterolemia variant. However, PRSs predict less accurately in populations that differ in ancestry from the GWAS participants, who have been overwhelmingly of European ancestry.

Gene and environment together

Environmental factors often interact with genotype. Folic acid supplementation before and early in pregnancy reduces the risk of neural tube defects (spina bifida, anencephaly) by roughly half or more; the general recommendation is 0.4 mg daily, with a higher dose (commonly 4 mg) for women who have had an affected pregnancy. Smoking multiplies the lung disease risk of people with alpha 1 antitrypsin deficiency. Phenylketonuria itself, a single gene disorder, shows that even a strongly genetic condition can be completely prevented by changing the environment (the diet).

Section 3Genetic testing: diagnostic, carrier, prenatal and newborn screening

Screening versus diagnostic testing

Two words organize this section. A screening test is offered to a whole population or a large group with no particular reason for concern, to find the few people at increased risk; it is designed to be safe, cheap and sensitive, and a positive result must be confirmed. A diagnostic test is done in a person already suspected of having a condition and aims to give a definitive answer. Within genetic testing, the main purposes are:

Laboratory methods

The method must match the type of variant suspected. A karyotype (G banded chromosomes) detects aneuploidy and large rearrangements of about 5 to 10 megabases or more, including balanced translocations. FISH uses fluorescent probes to detect specific deletions or count chromosomes rapidly in interphase cells. A chromosomal microarray detects copy number gains and losses down to tens of kilobases and is the first line test for unexplained developmental delay or multiple anomalies, but it cannot detect balanced rearrangements. Targeted variant testing looks for one known familial variant. Sequencing ranges from a single gene to a multigene panel, the exome (about 1 to 2 percent of the genome that codes for protein) or the whole genome. Repeat expansions require special methods (PCR sizing or Southern blot), and methylation tests detect imprinting disorders.

Sequencing a genome finds millions of variants, so interpretation is the hard part. Variants are classified on a five tier scale: pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign and benign, using evidence such as population frequency, predicted effect on the protein, segregation in families, functional studies and whether the variant arose de novo. A VUS should not be used to make clinical decisions, and many are reclassified, usually as benign, as evidence accumulates.

Measuring test performance

Every test is judged by four numbers. Sensitivity is the proportion of affected people who test positive (true positives divided by all affected). Specificity is the proportion of unaffected people who test negative. These are properties of the test. The positive predictive value (PPV) is the proportion of positive results that are truly affected, and the negative predictive value (NPV) is the proportion of negative results that are truly unaffected. Predictive values depend heavily on how common the condition is in the people tested, and this is where intuition most often fails.

Prenatal screening and diagnosis

The risk of a pregnancy with a chromosome aneuploidy rises with maternal age because of nondisjunction in aging oocytes; the risk of trisomy 21 at term is about 1 in 1,500 at age 20, 1 in 350 at age 35 and 1 in 100 at age 40. Today screening is offered to all pregnant people regardless of age.

• Cell free DNA (cfDNA) screening, often called noninvasive prenatal testing, sequences DNA fragments in maternal plasma, of which roughly 10 percent come from the placenta. Overrepresentation of chromosome 21 fragments signals trisomy 21. It can be done from about 10 weeks and detects about 99 percent of trisomy 21. Because the DNA comes from the placenta, confined placental mosaicism and maternal conditions can cause false results, and a low fetal fraction (as in maternal obesity) can make the test fail.
• First trimester combined screening (11 to 14 weeks) combines maternal age, ultrasound measurement of the nuchal translucency and two serum markers (PAPP-A and free beta hCG), detecting about 85 percent of trisomy 21.
• Second trimester serum screening (the quad screen, about 15 to 22 weeks) measures AFP, hCG, unconjugated estriol and inhibin A. In trisomy 21, AFP and estriol are low and hCG and inhibin A are high. An elevated AFP alone suggests an open neural tube defect.
• Chorionic villus sampling (CVS, about 10 to 13 weeks) samples placental tissue and gives an earlier result. Amniocentesis (from about 15 weeks) samples fetal cells from amniotic fluid. Both are diagnostic, allow karyotype, microarray or targeted testing, and carry a small procedure related miscarriage risk, now estimated well below 1 percent.

Preimplantation genetic testing (PGT) tests one or a few cells from embryos created by in vitro fertilization, so that only unaffected embryos are transferred. It is used by couples who know they are at risk for a single gene disorder (PGT-M) or carry a structural rearrangement (PGT-SR), and it avoids decisions about an established pregnancy, though it requires IVF.

Carrier screening

Carrier screening identifies couples at risk before or early in pregnancy. Historically it was targeted by ancestry: Tay-Sachs screening in Ashkenazi Jewish communities, which reduced the incidence of the disease by more than 90 percent, beta thalassemia screening in Mediterranean populations and sickle cell screening in people of African ancestry. Ancestry is an imperfect guide in mixed populations, so many programs now offer pan-ethnic or expanded carrier screening covering dozens to hundreds of recessive conditions for everyone. Spinal muscular atrophy (deletion of SMN1, carrier frequency about 1 in 40 to 1 in 60) shows a pitfall: some carriers have two copies of SMN1 on one chromosome and none on the other, so a copy number test reads them as normal ("silent carriers").

Newborn screening

Newborn screening is the largest genetic testing program in the world. A few drops of blood from a heel stick, usually collected 24 to 48 hours after birth, are dried on filter paper and tested for treatable disorders that cause irreversible harm if not caught early. Tandem mass spectrometry detects dozens of amino acid, organic acid and fatty acid oxidation disorders from a single spot. Programs also include hearing screening and pulse oximetry for critical congenital heart disease.

Section 4Genetic counseling and risk calculation

What genetic counseling is

Genetic counseling is a communication process that helps people understand and adapt to the medical, psychological and family implications of genetic contributions to disease. It integrates three things: interpreting family and medical histories to assess risk; educating about inheritance, testing, management, prevention and resources; and counseling to promote informed choices and adaptation. A typical session includes taking a three generation pedigree, confirming diagnoses from records, explaining the condition and its inheritance, calculating risks, discussing testing options with their benefits and limits, obtaining informed consent and, after testing, explaining results and arranging follow-up.

Traditionally, counseling has been nondirective: the counselor provides balanced information and supports the client's decision, especially about reproduction, rather than recommending a choice. In practice, when a result points to a clearly effective intervention (such as risk reducing surgery for a BRCA1 carrier or colonoscopy in Lynch syndrome), counselors give clear medical recommendations while respecting the person's autonomy. Counseling also attends to emotions such as guilt in a parent who "passed on" a variant, anxiety about uncertain results and the burden of telling relatives.

Mendelian risks and conditional probability

Many counseling risks come straight from Mendelian rules, but the most common error is to forget information that is already known. The classic example is the healthy sibling of a person with an autosomal recessive disease. Both parents are obligate carriers, so at conception the sibling had a 1/4 chance of being affected, 1/2 of being a carrier and 1/4 of being a noncarrier. But the sibling is known to be unaffected, so the affected outcome is eliminated, and the carrier risk is (1/2)/(1/2 + 1/4) = 2/3, not 1/2.

Bayesian risk calculation

When there is additional information, such as unaffected children or a negative test, counselors use Bayes' theorem to combine it with the starting risk. The method is a short table with four rows:

1. Prior probability: the risk based on the pedigree alone, for each hypothesis (for example, carrier and noncarrier).
2. Conditional probability: the chance of the observed information (such as three unaffected sons) under each hypothesis.
3. Joint probability: prior multiplied by conditional for each hypothesis.
4. Posterior probability: each joint probability divided by the sum of all joint probabilities.

The same method handles negative carrier tests and late onset disease. For Ben in the cystic fibrosis problem, the prior is 1/25 carrier and 24/25 noncarrier; the conditional probability of a negative result is 0.10 if he is a carrier (the test misses 10 percent) and 1 if he is not. Joint: 0.004 and 0.96; posterior carrier risk = 0.004/0.964 = 0.0041, about 1/241. For Huntington disease, suppose the child of an affected parent is unaffected at an age by which half of carriers show symptoms. Prior: 1/2 and 1/2. Conditional probability of being unaffected: 1/2 if a carrier, 1 if not. Joint: 1/4 and 1/2. Posterior risk = (1/4)/(3/4) = 1/3.

Consanguinity and empirical risks

The coefficient of relationship is the proportion of genes two relatives share by descent: 1/2 for first degree relatives (parents, siblings, children), 1/4 for second degree and 1/8 for first cousins. The coefficient of inbreeding (F) of a child is the probability that both alleles at a locus are identical by descent; for the child of first cousins, F = 1/16. The background risk of a serious birth defect or genetic disorder in any pregnancy is about 3 to 4 percent; for first cousin parents it is increased by about 2 to 3 percentage points, mostly from autosomal recessive disease.

For multifactorial conditions there is no Mendelian formula, so counselors use empirical risks, observed recurrence figures from family studies. A useful rule (Edwards' approximation) estimates the risk to first degree relatives as roughly the square root of the population incidence: for a population incidence of 1/1,000, the first degree recurrence risk is about √0.001 = 0.032, or about 3 percent.

Section 5Ethical, legal and social issues

Principles that guide genetic medicine

Genetic information is unusual medical information. It is predictive (it can forecast disease decades ahead), familial (a result about one person is also information about parents, siblings and children), permanent (it does not change) and increasingly identifying (a genome is a unique fingerprint). These features raise ethical, legal and social issues, often abbreviated ELSI, that go beyond those of most medical tests. Four standard principles of bioethics frame the discussion.

Informed consent and the right not to know

Before a genetic test, a person should understand its purpose, what it can and cannot detect, the possible results (positive, negative, uncertain, unexpected), the implications for relatives, the possibility of findings unrelated to the original question, how samples and data will be stored and used, and the protections against discrimination. Because there is no cure for some conditions, the right not to know is also respected. Predictive testing for Huntington disease follows a careful protocol, with counseling before and after testing and time to reflect, because a positive result foretells an untreatable disease, and some at risk people choose never to be tested.

Testing children

Children cannot give consent themselves, so the guiding question is whether testing during childhood brings a medical benefit to the child. Testing is appropriate when a condition begins in childhood or when childhood interventions help: RET testing in families with multiple endocrine neoplasia type 2 (preventive thyroid removal in early childhood) or APC testing in familial adenomatous polyposis (colonoscopy beginning around age 10 to 12). Predictive testing for adult onset conditions without childhood interventions, such as Huntington disease or BRCA1 variants, is generally deferred until the child can decide as an adult, preserving their future autonomy. Carrier testing of children is likewise usually deferred.

Confidentiality and the family

Because relatives share variants, a result can be lifesaving for them. Clinicians encourage patients to share results, often with a family letter, and offer cascade testing to relatives. Rarely, a patient refuses to tell relatives at high risk of a serious, preventable condition. Most professional guidance holds that confidentiality may be breached only in exceptional circumstances, when serious harm is likely, preventable and the relatives are identifiable, and only after attempts to persuade the patient have failed.

Genetic discrimination and privacy

A major fear is that insurers or employers will use genetic information against people. In the United States, the Genetic Information Nondiscrimination Act (GINA) prohibits health insurers from using genetic information to set eligibility or premiums and prohibits employers from using it in hiring, firing or promotion. Its limits are frequently tested: GINA does not cover life insurance, disability insurance or long term care insurance, does not apply to employers with fewer than 15 employees, and does not protect people once they have a manifest disease (other laws, such as health care law that bars preexisting condition exclusions, help there). Other countries have different laws or voluntary agreements with insurers.

Privacy is harder to guarantee than it once seemed. Genomes cannot be truly anonymized: researchers have shown that names can sometimes be inferred by matching Y chromosome markers to surnames in genealogy databases. Direct to consumer (DTC) tests sell ancestry and some health information without a clinician, and consumer databases have been used in forensic genetic genealogy, identifying crime suspects through distant relatives who uploaded their own DNA. DTC health results can also mislead: a test that checks a few common BRCA variants misses most pathogenic variants, so a "negative" result is not reassuring, and raw data interpretations can have high false positive rates.

Secondary findings

Exome and genome sequencing can reveal variants unrelated to the reason for testing. Professional guidance recommends that laboratories offer to report pathogenic variants in a defined list of medically actionable genes (for example BRCA1, BRCA2, the Lynch syndrome genes, LDLR, and genes for inherited arrhythmias and cardiomyopathies), because early knowledge allows prevention. About 2 to 3 percent of people sequenced carry such a finding. Patients may opt out of receiving them, consistent with the right not to know.

Equity, history and the future

Genetics has a troubled history. The eugenics movement of the early twentieth century used crude ideas of heredity to justify forced sterilization laws and immigration restrictions, and these ideas fed into atrocities. Modern genetics emphasizes voluntary, informed individual choice, but critics note that prenatal screening raises questions about how society values people with disabilities. Equity remains a present concern: most GWAS participants have been of European ancestry, so variant databases and polygenic scores perform worse for others, and people of African ancestry more often receive uncertain results.

Gene therapy now treats several genetic diseases. Somatic therapies, which alter only the patient's cells, include viral vector gene replacement for spinal muscular atrophy and inherited retinal dystrophy, and ex vivo CRISPR editing of blood stem cells that reactivates fetal hemoglobin in sickle cell disease and beta thalassemia. Germline editing, which would change embryos and all their descendants, is a different matter. A widely condemned experiment in which embryos were edited and brought to birth led to near universal calls that germline editing not be used for reproduction while safety (off target edits, mosaicism) and ethical questions remain unresolved, especially since PGT already lets most at risk couples avoid passing on a variant.

Words worth knowing