The study of human inheritance often tends to focus on relatively simple traits. But as we explored in Chapter 2 and elsewhere, there is a broad avenue of molecular and developmental events that connect the DNA with a phenotype. When the work of Gregor Mendel was “rediscovered” in 1900, various researchers attempted to repeat and confirm his observations using a variety of plants and animals, including humans. Some studies supported the Mendelian genetic models. But many other cases were unsuccessful either because the organisms they chose had unusual genetic characteristics, like honeybee drones that are haploid, or because the traits they looked at did not have the simple phenotypic basis that we see in those studied by Mendel. Some people attribute these latter studies to bad luck in choosing an experimental system. But in fact they were setting the stage to explore a parallel, and very important, dimension of genetic complexity.
This is illustrated historically by the work of people like Francis Galton, a cousin to Charles Darwin, and Karl Pearson, one of the founders of modern statistics. A key trait they studied, human height (Figure 10-1), is now known to be influenced both by a number of genes segregating independently and by many environmental variables. The genotypes were not simple, so the application of Mendelian rules could not be detected. Galton’s initial work actually predates the rediscovery of Mendel’s papers by several decades, and indeed it was hotly debated whether quantitative traits have the same underlying genetic basis as seen in simpler Mendelian traits like flower color and seed shape. But far from being a dead end, work by pioneers like these led to the establishment of an important field of genetics, biometrical or quantitative genetics, based on the statistical analysis of genetic relationships.
Figure 10-1. Human height was one of the earliest quantitative characters studied in detail. (a) Height distribution (inches) for 175 students in 1914 attending the Connecticut Agricultural College. (b) Graphical presentation of these student heights showing their close fit to a normal distribution. (a: Reprinted with permission from Albert and Blakeslee: Corn and Man. Journal of Heredity. 1914;5:51. Oxford University Press. b: Reprinted with permission from Brooker RJ: Genetics: Analysis & Principles, 3rd ed. New York: McGrawHill, 2008.)
Significant advances over the last couple of decades have seen the emergence of quantitative genetics as a field with important biomedical perspectives. It is leading to a deeper understanding of the way traits are influenced by quantitative trait loci (QTLs). This appreciation of the multilayered genetic underpinnings of normal phenotypic diversity and a large array of significant medical conditions reflects the continuing maturity of genetics as an explanatory science. In this chapter, we will discuss the basis of phenotype assessments and some of the ways this knowledge can be applied to benefit patients.
Part 1: Background and Systems Integration
Quantitative versus Qualitative Traits
Most of the phenotypes we have discussed to this point are relatively simple, in that they can be traced to mutations in one or a couple of genes and they generally follow predictable rules of transmission and expression. On the other hand, many important traits are influenced not by one gene, but by a battery of genes whose effects are over-shadowed by unknown environmental factors. In fact, if we analyze almost any trait in detail, we soon realize that all genes work within a complex network of biochemical interactions and environmental conditions. One only has to think back to phenomena like variable expressivity to see examples. But the fundamental inheritance mechanisms are the same.
How can one make genetic predictions about traits that are so complicated? The key, as usual, is to acknowledge that genetic effects on human development are complex. But there are underlying patterns and relationships that allow us to work effectively with traits that might initially appear almost intractable.
The genetic boundary between “qualitative” and “quantitative” is blurred. But we can establish a context by defining a qualitative trait as one that can be categorized by a “quality” into two or more defined phenotypic groupings, like pigmented skin and albino. A quantitative trait is one that must be described by some quantity along a spectrum of expressions, like the range of skin pigmentation from dark to light.
This distinction is not simply descriptive. It distinguishes two completely different ways in which phenotypes must be analyzed. A qualitative trait, for example, distributes individuals into defined groups. When traits like this are analyzed statistically, one would present them in clear phenotypic categories (e.g., green versus yellow seeds) and use statistics like the chi square test to evaluate a fit between the observed proportions and a theoretical expectation. For a quantitative trait, on the other hand, the phenotypes are defined by their placement on some scale (e.g., height or degree of pigmentation). In this case, a comparison among groups is dependent on quantitative measurements like phenotypic averages and range of expression. The appropriate statistical tests are quite different. Groups are compared by t-tests, analyses of variance, or other approaches.
In the following sections we will discuss some of these statistical tests. But our goal will be larger. We will explore the ways one can think about quantitative traits and their genotype × environment interactions. The objective is not just to document them mathematically, but instead to clarify the uncertainties behind complex information that patients and family members want to understand.
Quantitative Description of Phenotypic Distributions
One can visualize the array of phenotypes of a typical quantitative trait as a normal distribution. Such a distribution can be described using standard statistics. Even exceptions like threshold characters (Figure 10-2), in which a certain level of gene product must be reached before the trait is expressed phenotypically, can be interpreted in terms of an underlying distribution of developmental capacities.
Figure 10-2. A normal distribution of biochemical product of gene activity can underlay a trait that is only expressed when a certain threshold amount is reached. Such traits may appear to be discrete, even though the underlying genetic mechanism is quantitative.
The mean (
) is the sum of all measurements (ΣX) divided by the number of individuals in the sample (N). But by itself, the mean overlooks an important element needed for prediction. It overlooks the variation seen among the individuals averaged in the sample. Variance (V) is a measure of this dispersal of the trait. More specifically, it is the average squared deviation of each data point from the mean.
The value N – 1 is called the degrees of freedom and is used to assess the strength of conclusion to be drawn in many statistical tests. One can think about this measure by a simple thought experiment. Let’s imagine gathering a group of data, such as child birth weights. The first child gives one measurement of weight, and so does the second child. So the average weight is calculated by adding the two weights and dividing by 2. But when assessing variance in weight, the first baby gives no indication about how much variability there is in such a measurement. The difference between baby 1 and baby 2, however, gives the first estimate of variation, the third baby gives a second estimate, and so forth. Thus, there are always one fewer estimates of trait variation than there are of trait mean.
The square root of the variance is called the standard deviation, symbolized s.d., s, or δ. Similarly, the variance can be symbolized as Vx (i.e., the variance of x) or as s2 or δ2, since it is the square of the standard deviation. Generally, s and s2 are used when referring to a measured characteristic of a specific sample, and δ and δ2 are used when referring to the underlying theoretical population of data. You should always expect the mean and standard deviation (or variance) to be presented together in order to have a complete description of a quantitative trait’s phenotypic expression.
One of the especially useful applications of the standard deviation is often called the “empirical rule.” This rule is that the standard deviation will subdivide a normal distribution into predictable subgroups of the population (Figure 10-3), no matter what trait is being considered. Specifically, one standard deviation above and below the mean of a normal distribution will contain approximately 68% of the data points, while two standard deviations will contain about 95% and three standard deviations will contain 99.7%. For an example of its application, consider performance on a standard IQ test, on which the average score is 100 and the standard deviation is 15 IQ points. A typical population would be expected to have 68% of its members with an IQ score of 85 to 115 points. Half that number, 34%, would be expected to have an IQ between 100 (the mean) and 115. For more accurate evaluation of distribution differences, data are typically converted into units of standard deviation and compared to a table of standard normal deviations.
Figure 10-3. This rule is that the standard deviation will subdivide a normal distribution into predictable subgroups of the population. (Reprinted with permission from Brooker RJ: Genetics: Analysis & Principles, 3rd ed. New York: McGraw-Hill, 2008.)
Polygenic Inheritance and Heritability
As the name suggests, “polygenic” refers to “many genes.” The genetic basis of polygenic traits is now often found in the literature as quantitative trait loci or QTLs. Analytical techniques are making it possible to identify some of the loci that contribute most strongly to certain quantitative traits. But it is often enough just to measure the proportion of phenotypic variation that can be traced to segregating genetic differences among individuals, versus how much is due to environmental factors. It is the genetic component that is most influential in determining the trait’s expression in the next generation.
Heritability is an important measure, but it is often misunderstood. One should not make the mistake of equating high heritability with high importance. It is a measure of the effect of segregating genetic variation, and critically important traits like those affecting survival or reproductive success (i.e., so-called “fitness traits”) seldom tolerate high levels of variety. Directional selection to maximize an individual’s survival and reproductive success has selected against less beneficial genetic variation that might have been present in the gene pool. For example, the heritability of IQ test scores in some populations is about 0.60, which means that about 60% of the phenotypic difference among individuals can be explained by genetic allele difference. On the other hand, the heritability for an important fitness trait like egg production in chickens or Drosophila is estimated to be only about 0.21 and 0.18, respectively. Thus, heritability simply measures the proportion of phenotypic variation that can be explained by genetic differences among individuals, it is not a statement about the biological importance of the characteristic.
Heritability is symbolized h2, not because there is such a thing as its square root (there is no “h”), but because it is the ratio of two variances (s2). For this part of the discussion, however, we will follow the style that uses V to represent variance. Specifically, heritability is the ratio of the proportion of all phenotypic variance (VP) that can be explained by genetic segregation (VG), i.e., to genetic differences among members of the sample. VP in turn has various components, the two most important being genetic segregation and environmental effects (VE) (Figure 10-4). Thus, for practical purposes,
Figure 10-4. An example (using weight as a sample trait) of the effect of the number of alleles and environmental variation on quantitative genetic distributions. (Reprinted with permission from Brooker RJ: Genetics: Analysis & Principles, 3rd ed. New York: McGraw-Hill, 2008.)
There is one other aspect of heritability that is critically important to remember. Heritability is a measure of the proportion of phenotypic variation that can be accounted for by genetic differences among members of the sample. But we know that gene pools vary from one time to another and from one population to another. For that reason, heritability is valid only in making predictions about genetic makeup or responses to selection for the population in which it is measured. It cannot validly be generalized from one population to another. Ignoring this simple population genetic fact led to many inaccurate and destructive statements about the genetic differences among races in traits like IQ, since comparisons often generalized heritabilities in unfounded ways.
Genotype × Environment Interactions
Environmental differences can be as important as genetic differences in phenotypic expression. Indeed, the same genotype can be expressed in quite different ways as a relevant environmental factor changes. This type of relationship is called a genotype × environment (G × E) interaction.
A classic example of a G × E interaction is the temperature-sensitive activity of melanin production in Siamese kittens. The enzyme is catalytically more active in cooler conditions, so more pigment is synthesized in the tips of the ears and the paws. Temperature, nutritional components, light wavelength and intensity, and even internal conditions like age and sex can influence the expression of a given genotype.
Genetic Heterogeneity
We pointed out earlier how a single genotype can result in different phenotypes because of environmental influences. The opposite can also be true. The same phenotype can be traceable to genetically different backgrounds. Genetic heterogeneity refers to the situation in which a condition or disease could be traced to different underlying genotypes. Different genes or different alleles of a single gene might be involved in each case analyzed. Thus, tools like microarrays or DNA sequencing that uncover the individual makeup of each case are important aides to accurate diagnosis. Given the developmental complexity of even the simplest anatomical structure, cases of genetic heterogeneity should become even more commonly recognized.
One example of genetic heterogeneity is found in the condition retinitis pigmentosa (RP). This is a collection of literally dozens of genes or gene regions at which mutations affect photoreceptor structure or longevity in the eye. This is locus heterogeneity. Some mutations are, not surprisingly, more common than others. About 10% of the RP cases, for example, can be traced to mutations in the rhodopsin gene on chromosome 3.
A related situation is found in allelic heterogeneity. In these situations, different alleles have distinctly different phenotypic consequences. In fact, in some instances the conditions were thought to be linked but genetically unrelated until sequencing and other molecular techniques uncovered their relationship. There are a growing number of examples of this phenomenon. One is associated with different mutations in the gene coding for β-globin. One mutation causes sickle cell disease while others result in various forms of β-thalassemia.
Concordance in Twins
Twins provide a special opportunity to explore the repeatability of genetic expression. Monozygotic (MZ) or identical twins share the same genotype and largely the same developmental environment. Dizygotic (DZ) or fraternal twins are probably as similar as are monozygotic twins in their shared developmental environment, but they differ genetically as much as any other typical sibling pair. By measuring the degree of phenotypic agreement, concordance, between monozygotic versus dizygotic twins, it is, therefore, possible to estimate the genetic influence on a trait. Some examples are shown in Table 10-1.
Table 10-1. Trait Concordances in Monozygotic (MZ) and Dizygotic (DZ) Twins
It is important to note that although monozygotic twins start off as genetically identical, things can change quickly. Once the twinning event has occurred and the two zygotes resume/continue their development, divergences can occur. Postconception mutations can and often do occur. Since these are largely random events, what occurs in one MZ twin is not what will happen in the other. Thus, even though they started off as genetically identical, there are likely to be multiple genetic differences in the twins by delivery. As was discussed in Chapter 7 (Mutations) many of these genetic differences may affect no observable phenotypic change—and thus the twins still appear “identical.” However, on occasion one twin may acquire a mutation that does result in a phenotypic change that is present in one MZ twin but not the other (Figure 10-5). The twins are thus said to be discordant for that trait.
Figure 10-5. (a) Monozygotic twins discordant for Beckwith Wiedemann syndrome (macrosomia, macroglossia, omphalocoele). Note the difference in the phenotype of the twins. (b, c, d) Twin A showing features of Beckwith-Wiedeman syndrome–-(b) Preuricular ear tags, (c) Macroglossia, and (d) Ompahlocoele (repaired).
A caveat in twin studies is that it is sometimes difficult to distinguish monozygotic from dizygotic pairs. Tests like DNA fingerprinting may be needed to establish the degree of genetic similarity. In addition, monozygotic twins share the same genotype, but they also share many potentially important environmental conditions like nutritional level and cases of infectious disease. Comparing identical twins raised together with the rare examples of identical twins raised apart might provide a partial answer. But when adopted independently, identical twins are often raised in families with similar environments, such as city living versus rural, education level of the adoptive parents, or number of other children in the household. Twin data are, therefore, informative but not without limitation.
It is also useful to keep in mind that one can use other examples of shared genotype to quantify the genetic influence on a trait, or its recurrence risk. Recurrence risk is related to the degree of shared genotype expected among relatives. For a condition traceable to a single gene, the recurrence risk for siblings is 0.50 and it decreases by half for each step of separation in the pedigree (e.g., aunt – nephew = 0.25; first cousins = 0.125).
Part 2: Medical Genetics
Thus far we have discussed cytogenetic abnormalities (Chapter 5) and single gene disorders (Chapter 6) as major categories of etiologies of human genetic disorders. These inheritance modes would best be characterized as uni-factorial. That is, a single key genetic change is responsible for the majority of the phenotype. Historically, these conditions were the first described and are among the best understood genetic conditions. However, these conditions are also relatively rare when all human medical disorders are considered. In general, the more common diseases tend to have more complex modes of inheritance. Still, they will have significant genetic contributions to explain their occurrence (Table 10-2). In the first part of this chapter the concepts of gene-environment interactions were explained. Here we will discuss the applications of these concepts in clinical practice.
Table 10-2. Common Medical Disorders With a Proven Genetic Basis
Multifactorial Inheritance
“Multifactorial” as the name implies means “many factors.” This simply means that both genetic and environmental factors have significant contributions to the phenotype. It is important to note that in a literal sense, all medical conditions are “multifactorial.” It is hard to imagine any condition that does not have some degree of genetic basis to it and some degree of environmental modification of the phenotype as well (i.e., the Mendelian and chromosomal disorders as discussed can/do have environmental influences that can modify the phenotype somewhat). By convention the simplest model or mechanism is designated for a given condition to give the most accurate predictions of outcomes (recurrence risk/familial pattern/transmission mode). Thus, Mendelian disorders and chromosomal disorders are not conventionally classified as having “multifactorial inheritance.”
So, then, what constitutes multifactorial inheritance? The major features include:
1. Genetic variability exists yet no uni-factorial mechanism can be identified to explain the transmission pattern.
2. Family studies indicate an increased risk for relatives to be affected.
3. Environmental factors may exert a significant influence on the phenotype.
Conditions that exhibit multifactorial inheritance usually involve complicated pathophysiologic or morphogenetic processes. This means that the search for an etiology becomes significantly more complicated. The search is not for “the gene” responsible for the condition. Rather, the pattern that best describes what is observed may involve multiple different genes interacting with more than one environmental factor. As described in the first part of this chapter, the expression of multifactorial traits involves a biologic threshold. The concept is that every individual has a set of liabilities toward a given condition. These liabilities are both the genetic and environmental factors. Each individual has their own unique combination of protective or predisposing genes and favorable or unfavorable environmental influences (Figure 10-6). For any given person, an increasing number of liabilities push that person toward the biologic threshold. When the cumulative contributions of all genetic and environmental liabilities exceed a certain threshold, the capacity of the organism to buffer against the liabilities is overcome, and the trait is observed (Figure 10-2).
Figure 10-6. Graphic representation of the concept of cumulative liabilities in multifactorial inheritance. As the number of adverse genetic and environmental factors accumulate, the chances of disease expression likewise increases.
As multifactorial traits are observed in families, several general characteristics are observed in their transmission. The basic principles of inheritance exhibited by multifactorial traits include:
1.The condition does not segregate through the family in a recognizable Mendelian (single gene) or other uni-factorial manner.
2. The recurrence risk of the condition is increased in relatives as compared to its occurrence in the general population.
3. There is a nonlinear decrease in the recurrence risks with an increasing distance of relationship. In general, recurrence risks are higher in first or second degree relatives. Once the distance of relationship becomes greater than third degree, the risk has fallen back to the general population baseline risk.
4. There is an increased risk with an increasing number of affected individuals.
5. Within the spectrum of variably expressed conditions, there is an increased risk seen with an increased severity of the disorder seen in the affected individuals.
6. Many multifactorial traits show a sex bias (i.e., they occur more often in one sex than the other) (Table 10-3). Interestingly, an increased recurrence risk is noted if the affected person(s) is of the less commonly affected sex.
Table 10-3. Multifactorial Conditions Reported With a Significant Sex Bias
For most multifactorial traits we do not know enough about the genetic and environmental factors involved to be of practical use in the clinical setting. For most, then, etiologic-specific counseling is not possible, and this is often confusing and frustrating to patients. However, there is still helpful information that can be provided. If a multifactorial condition is common enough to gather reasonable population data, empiric recurrence risk data may be available and provided to the family. Empiric risk counseling is the application of observational population data when little is known about underlying factors. The recurrence risk for a particular family is based upon what has been observed in other, similar families. It involves identifying recurrences in defined sub-populations that are condition and situation specific.
Neural tube defects (NTDs), for example, are congenital malformations of the embryonic neural tube. Failure of the neural tube to close properly will lead to anomalies of the brain and/or spinal cord and the surrounding bony structures (Figure 10-7). There is a range of expression depending upon where/how much of the neural tube fails to close. They are relatively common birth defects, occurring in about 1 to 2 per 1000 births. Neural tube defects exhibit multifactorial inheritance. Because they occur commonly enough, empiric risk data have been obtainable (Table 10-4). Such data are applicable in the clinic setting. Such a scenario might happen something like this: a young couple is seen in which their child was born with a neural tube defect. The pregnancy was uncomplicated. A careful review of the family history reveals no other known family member with a neural tube defect as far back as anyone can remember. Examination of the child does not demonstrate anything else wrong with the child other than the NTD. The couple now wants to know “what are their chances of having another child with an NTD?” In referring to the data in Table 10-4, it can be determined that the answer to their question is 5%. Although this may seem quite simple, there is still a lot of useful information embedded within. As per the discussion above, several deductions based on the multifactorial model can be made. First, the 5% risk clearly does not suggest a Mendelian (single gene) disorder. Second, this risk can be compared to the baseline (population) risk. That is, the recurrence risk is “only” 5% (and for the family that means a 95% chance that a subsequent child will not have an NTD). Still, a 5% recurrence risk as compared to a general population incidence of 1/1000 represents a 50-fold increase.
Figure 10-7. Lower back of an infant with an inferior neural tube defect (i.e., spina bifida/meningomyelocoele).
Table 10-4. Empiric Recurrence Risks for Neural Defects (NTD)
Depending on the amount of data available, even more information can be discerned from empiric data. Cleft lip and cleft palate are among the most common structural congenital anomalies seen in humans (Figure 10-8). Orofacial cleft occurs in 1 to 2 per 1000 live births. About half of the patients born with a cleft have cleft in conjunction with other structural anomalies (i.e., “complex” clefts). The other patients just have a cleft; that is isolated or nonsyndromic clefts. The occurrence of isolated clefts can be explained by a multifactorial model as described earlier. Although intense research continues around defining the etiology of clefts, much is still not known. In general genetic testing for nonsyndromic clefts is not readily available. As such, when a patient with a non-syndromic cleft is seen, genetic counseling is provided using empiric data. Because clefting is a relatively common condition, and because multiple large population surveys have been published, extensive empiric data exist (Table 10-5). The data in this table are regularly used when counseling with families of children with clefts. The data in this table highlight the major features of multifactorial inheritance (relationship to baseline incidence, relationship to affected individual, number of affected persons, severity of expression, and sex bias).
Figure 10-8 Patients exhibiting the spectrum of cleft lip and palate. (a) Bilateral cleft lip and palate. (b) Unilateral cleft lip with cleft palate. (c) Unilateral cleft lip with alveolar notching. (d) Cleft palate. (e) Submucous (occult) cleft palate.
Table 10-5. Recurrence Risk Data for Cleft Lip With/Without Cleft Palate
A Better Understanding of Multifactorial Inheritance
Currently, for NTDs and many other conditions, this type of information is still the only clinically available recurrence risk information that can be given to families. In fact, this is what is routinely used in clinics on a regular basis. While it is helpful to have such information available to share with families, it is not ideal. It should be noted that by the very nature of the information, it is a population average. In reality, the 5% recurrence risk quoted for the hypothetical couple above is actually not their actual recurrence risk but rather an average risk for a group of couples with similar circumstances. The individual risk for a given couple could be quite low (as it often is) or sometimes could be significantly increased. However, in the absence of etiologic specific information it cannot be determined which is true for a specific case.
Clearly what is needed is a better understanding of the underlying genetic and environmental factors that contribute to the expression of any given multifactorial trait. The more that is discovered about the factors that are involved, the better (more specific and more accurate) are the predictions that can be made. As mentioned previously most multifactorial conditions involve complex, interacting physiologic processes. Typically each of these processes will have multiple components to them—each separately genetically regulated. Figure 10-9 shows how this might look. Thus, by earlier ways of thinking, the question might have been asked: “what gene causes condition X?” In hindsight, the reason for this was simply oversimplified view of multifactorial inheritance that was prevailing at the time. A quick look at the figure shows that the more accurate question would be: “which of the several possible genes is at work in a particular family with condition X?”
Figure 10-9. Diagram of multiple physiologic processes each with multiple contributing genes for a hypothetical disorder (condition X) that exhibits multifactorial inheritance.
A good example of this can be seen in how our understanding of diabetes and the genetics underlying it have changed over time. Diabetes mellitus (DM) is a metabolic disorder characterized by carbohydrate intolerance (high serum blood glucose levels). There are several types of diabetes classified by the purported physiologic mechanism. The most common type is type II occurring in almost 10% of all adults in the United States. The primary pathogenic change in type II DM is insulin resistance. The inheritance pattern of type II DM is best described as multifactorial. In the 1980s, ambitious researchers set out to find “the gene” that causes type II DM. Despite valiant efforts, little progress was made. Early studies identified candidate genes based upon the understanding of the condition’s physiology. It thus was logical to suspect that mutations in genes like the insulin receptor or the insulin gene itself might be responsible for type II DM. As these possibilities were explored, it became quickly evident that what might be a very logical assumption was in fact wrong. In fact, most of the candidate genes that researchers have deduced have proven to not be significant (Table 10-6).
Table 10-6. Candidate Genes for Type II Diabetes Mellitus
Over the past 30 years, much headway has been made in understanding the etiology of type II DM. Using powerful genome scanning techniques, many major genetic factors that predispose to type II DM have been identified (Table 10-6). It is fascinating to look at this list and realize that discovering these linkages would probably never have happened without whole genome analysis. Logic simply would not have led researchers to the answers. Once these factors are identified, they can then be linked to the primary physiologic processes involved. For our example of type II DM, the major physiologic processes involved appear to be resistance to:
1. insulin stimulation (not mediated by the insulin receptor),
2. beta cell constitution and mass, and
3. beta cell fatigue.
The beta cells are the cells in the pancreas that make and secrete insulin. Of course these genetic predispositions must interact with environmental modifiers. Over the decades, the environmental risk factors for diabetes have not changed. Obesity, decreased physical activity, and age are clear modifiers of the genetic background. The relationship of all of these factors to the overall occurrence of type II DM can be visualized as in Figures 10-10 and 10-11. The final step then is to apply this knowledge in the clinical realm. Genetic testing would need to be performed to identify which gene(s) were contributory for a given factor. Recurrence risk counseling could then be given as etiologic specific information. Therapies would be designed to address the specific pathogenic mechanism that was disrupted. For diabetes and many of the other so-called common disorders, that end point is within sight.
Figure 10-10. Influences on the occurrence of type II diabetes mellitus. Three major physiologic processes interface with environmental factors.
Figure 10-11. Diagram of multiple physiologic processes each with multiple contributing genes for type II diabetes mellitus as an illustrative multifactorial disorder.
Polygenic (Oligogenic) Inheritance
Polygenic inheritance, as the name implies, means “many genes.” But environmental influences are also important in determining the final phenotype. Although many geneticists use the terms ‘multifactorial’ and ‘polygenic’ interchangeably, here we want to focus on the genetic component. The concept of polygenic inheritance is that there are multiple genes that individually contribute to the phenotype in a cumulative manner. The additive effect of the overall genetic contribution then determines a relative “size” or degree of expression. Traditionally, polygenic inheritance has been used to describe the expression of quantitative traits (e.g., height, weight, head circumference, blood pressure, and so forth). Clinical observations of polygenic traits usually identify mathematical relationships of the traits within families. Height, for example, is one of the best understood polygenic traits. Simple observations of people quickly identify the heritable nature of height. It is intuitive that taller people tend to have taller children (Figure 10-12). In the event of one tall parent and one short parent, the children usually end up somewhere in between.
Figure 10-12. (a) Two adult men, fathers of one daughter each. (b) It shouldn’t be hard to guess which daughter goes with which father.
Multiple large auxologic studies scanning many decades have identified several important features of the inheritance of height. First, there is a clear sexual dimorphism. Simply, males tend to be taller than females. Within families with the same parents, the male children as a rule are taller than the female children. Second, the primary determinant of a child’s height is the heights of the parents. The heights of other, more distant, relatives have little influence or predictability on the height of a given child. Clinically this relationship can be represented by the formulas below:
Albeit simple, these two formulas provide an accurate estimate of final adult height of a child given the parental heights. The expected height then is simply an average of the parent’s height with an adjustment for the sex of the child. The first question that usually follows is: “so then how come all male siblings of the same parents aren’t the same height?” The answer is simply that this number is a calculated mean height of the children. The rest of the equation is a standard deviation of 5 cm around this mean.
Let’s apply this then. Assume a mother that is 5′4″ tall (162.5 cm) and a father that is 6′1″ tall (185.4 cm). Using the formula mentioned earlier, the target height (expected mean) of their male children would be 180.5 cm. If 1 SD is 5 cm, then 95% (mean + 2 SDs) of their male children would be predicted to be between 170.5 cm and 190.5 cm. This information has great clinical utility. When children are seen for a short stature evaluation, there are two major and complementary questions that need to be answered:
1. What is the target height of the child (i.e., how tall do we think they should be when growth is complete), and
2. What is the predicted height (i.e., how tall do we think they will be when growth is complete).
Comparison of these two answers allow the determination of which patient has “normal” versus pathologic short stature.
Similar relationships also exist for other quantitative traits in people. Intelligence as estimated by IQ testing is another such example. The relationship of a child’s IQ is also close to an average of the parents’ IQs. Notably, however, the sexual dimorphism observed for height does not apply to IQ (i.e., females have predicted IQs that are the same as their male full siblings).
Early thinking on polygenic inheritance envisioned numerous genes (maybe in the hundreds) each with a small additive effect to the phenotype. The final expression then would be the cumulative “score” of all of the pluses and minuses toward the phenotype. That is, a person with 80 positive height genes and 20 negative height genes would be expected to be above of above average height. Current evidence suggests a slightly different situation. For most polygenic conditions, there are actually a small number (maybe 3-5) major gene influences that account for the majority of the phenotype. Just a few genes segregating independently can generate a statistically normal distribution of the trait (Figure 10-13). The remainder of the phenotype (and a smoother distribution) can be accounted for by environmental buffering and the relatively smaller contributions of any number of other “minor” modifying genetic changes.
Figure 10-13. Phenotypic distributions produced in a representative quantitative trait in which only two segregating genes account for 90% of the genetic effect. (Reprinted with permission from Thoday JM and Thompson JN: The number of segregating genes implied by continuous variation. Genetica. 1976;46:335-344.)
Genetic Susceptibility to Environmental Factors
A particularly important aspect of gene-environment interactions is the concept of genetic susceptibility. Genetic susceptibility refers to genetic changes that an individual possesses that alter his or her response to specific environmental exposures. Understanding how a person’s genome influences their response to the environment has tremendous implications for targeted treatments and, even more importantly, prevention. Table 10-7 lists just a few examples of known genetic susceptibilities. In reviewing this list, we hope you are impressed by how significant these susceptibilities can be.
Table 10-7. Examples of Reported Genes in Which Mutations Produce an Increased Susceptibility to Environmental Factors
Let’s use the first one on the list, factor V Leiden, to highlight this concept. A single nucleotide polymorphism in the factor V clotting gene called the Leiden mutation occurs in about 7% of the general population. This particular mutation makes factor V more resistant to degradation—so it hangs around longer than usual. The consequence of this is an increased tendency to form pathologic blood clots. Persons who are heterozygous for the Leiden mutation have significantly increased risks for spontaneous thrombi, abnormal clotting in association with precipitating events such as surgery or trauma and even with oral contraceptive use. In fact, some estimates suggest that this mutation potentially accounts for 40% of all pathologic thrombi. Since the Leiden mutation occurs in 7% of the general population, it is easy to extrapolate that about 0.1% of the population will be homozygous for the mutation. Homozygotes for the Leiden mutation have an order of magnitude greater increased risk of thrombi. Consider, then, what the risk might be in a middle-aged woman, who is homozygous for the factor V Leiden mutation and who also smokes and then starts oral contraceptives! The hope is that this type of knowledge can be used in preventative strategies utilizing genetic screening to identify genetic susceptibilities, identify and avoid high risk exposures, and prevent morbidity and mortality.
Part 3: Clinical Correlation
In Chapter 3 teratogens were briefly introduced. Here is a good place to look at these in a little more detail. Teratogens are environmental exposures that a woman may encounter during pregnancy and that can adversely affect fetal development. In the not-so-distant past, the prevailing understanding was that the womb was an almost impervious barrier that protected the developing baby safely inside the mother. Beginning in the 1960s, the science of teratology emerged and quickly expanded. It is now understood that the fetus is not protected from many environmental exposures that the mother may encounter. Table 10-8 lists just a few of the more important human teratogens. As a group, teratogens are extremely important as they represent one cause of congenital anomalies (birth defects) that are completely preventable.
Table 10-8. Major Human Teratogens
Exposure to alcohol (ethanol) may easily be the most common teratogenic exposure in our society. It is amazing that fetal alcohol syndrome (FAS) was not medically defined until the mid-1970s. Since that time, a tremendous body of literature has emerged on the effects of fetal exposure to alcohol in the womb. As it turns out, there is a wide range of outcomes in fetal exposures to alcohol. At the severe end of the spectrum is FAS. The features of FAS include characteristic facial changes (most notably smooth philtrum, thin upper lip, and short palpebral fissures), microcephaly, decreased linear growth, and several types of structural anomalies (Figure 3-31). Children with FAS also have cognitive and behavioral problems. However, not all children exposed to significant amounts of alcohol in the womb will turn out to have FAS. In fact, what is observed in children with significant in utero alcohol exposures is that about one third will have FAS, one third will have neurodevelopmental and behavioral problems that can be attributed to the exposure, and one third will have no apparent effects of the exposure. Taken as a whole, these can be referred to as the spectrum of alcohol-related birth defects.
Salient to the theme of this chapter (gene-environment interactions), research suggests that the explanation for the wide range of outcomes after in utero alcohol exposure is indeed genetic susceptibility. Differences in both the fetal and maternal genome seem to affect the final outcome. Some of these genetic differences have been shown to be genetic changes that influence the metabolism and elimination of alcohol. For now, it is simply not predictable as to which fetus will have which outcome. By far, the most important point to take from this is: there currently is no known safe amount of alcohol exposure for a given fetus. It is sobering to know that there are still medical practitioners who will “prescribe” alcohol for pregnant women to “calm their nerves” or who tell women it is permissible to drink after the first trimester of pregnancy has passed. It is critical for all health care professionals to understand the appropriate stance on this issue. The only recommendation that should be made is the complete avoidance of alcohol throughout pregnancy for all women. It is also important to stress that most pregnancies in the United States are not identified until 6 to 8 weeks of gestation. If steps are going to be taken to avoid such exposures, education and public health measures are needed to reach all women of child-bearing age. The importance of this cannot be overstated. It has been estimated that one-third of all neurodevelopmental and neurobehavioral disabilities could be eliminated by simply avoiding fetal exposures to alcohol!
Although much remains to be learned about the actual genetic changes that alter susceptibility of the fetus to alcohol, there are other examples for which more detailed information exists. For example, fetal hydantoin syndrome (FHS) is another teratogenic syndrome analogous to FAS. FHS is seen in children exposed to the anticonvulsant medication hydantoin or its derivatives during gestation. Children with FHS have characteristic craniofacial features, growth disturbance, neurodevelopmental delays, limb anomalies, nail hypoplasia, and hirsutism (Figure 10-14). Similar to what has been observed in FAS, not every child exposed to hydantoin derivatives during gestation will have FHS; in fact only about one-third will. In trying to sort out the question: “why do only some of the children with this exposure have FHS?” researchers focused on the metabolism of the drug (Figure 10-15). Ultimately, the genetic susceptibility of FHS was shown to be due to mutations in the gene for the enzyme epoxide hydrolase (the second step in the drug’s metabolism). With almost 100% predictability, it can be determined which children will have FHS when exposed to hydantoins in the womb—those with defects in epoxide hydrolase. Thus for FHS, the susceptibility to the teratogenicity of the drug can be demonstrated to be an autosomal recessive condition that operates at the fetal level. This exciting discovery was reported in 1990 and was one of the first examples where the molecular basis of the susceptibility to an environmental exposure was proven.
Figure 10-14. Infant with fetal hydantoin syndrome. (Reprinted with permission from Buehler BA, Bick D, Delimont D: Prenatal prediction of risk of the fetal hydantoin syndrome. N Engl J Med. 1993 Nov 25;329(22):1660-1661.)
Figure 10-15. First two metabolic steps in the degradation of phenytoin. (© Dilantin).
Board-Format Practice Questions
1. Suppose that a condition, blueism, occurs commonly in a population of individuals. After careful genetic analysis of hundreds of people in the population, you discover three different genes that are located on three different chromosomes (blueism 1, blueism 2, and blueism 3). A mutation in any one of these genes can cause this condition. From this information you can conclude that:
A. blueism exhibits multifactorial inheritance.
B. blueism exhibits polygenic inheritance.
C. there are marked environmental influences on blueism.
D. there is no environmental influence on blueism.
E. blueism exhibits genetic heterogeneity.
2. Which of the following is true about conditions that show a multifactorial inheritance pattern?
A. The recurrence risk is lower if more than one family member is affected.
B. If the expression of the disease in the proband is more severe, the recurrence risk is lower.
C. The recurrence risk is higher if the proband is of the less commonly affected sex.
D. The recurrence risk for the disease is quite high even in remotely related relatives.
E. Environmental influences are not important.
3. Assume that nose size is inherited in a polygenic manner. Based on this you would predict that:
A. at least 50 different genes contribute to the size of the nose.
B. a person with a big nose who mates with a person with a small nose would most likely have children with normal sized noses.
C. there would be a high threshold effect for persons with few liabilities for a large nose.
D. a survey of the population would likely show a bimodal curve with most persons having either a large or a small nose.
E. women would tend to have smaller noses than men.
4. Pyloric stenosis (PS) is a condition that is associated with hypertrophy (enlargement) of the muscle of the pyloris (outlet of the stomach). Infants with PS usually have severe vomiting beginning around 2 to 6 months of life. PS is inherited in a multifactorial manner. It is more common in boys. A couple’s first born child (a male) was born with PS. It began when he was 2 months old. They come to you with questions about recurrence risk. Correct information to give them would include:
A. their recurrence risk would be lower if the first (affected) child had been a female.
B. if the PS had been less severe (onset at 6 months), the recurrence risk would be higher.
C. if there are any other affected relatives, the risk would be lower.
D. the recurrence risk is about 20% to 25%.
E. the inheritance of pyloric stenosis could also be called monogenic.
5. In regards to in utero alcohol exposure:
A. it is an uncommon occurrence.
B. it has no real clinical importance.
C. all children exposed to alcohol in the womb will have FAS.
D. the best medical recommendation is the complete avoidance of alcohol for all stages of pregnancy.
E. there are no genetic effects on the impact of such exposures.