Despite the lack of guidance available for practitioners, extensive polypharmacy has become the primary method of treating patients with severe and chronic mood, anxiety, psychotic or behavioral disorders. This ground-breaking new book provides an overview of psychopharmacology knowledge and decision-making strategies, integrating findings from evidence-based trials with real-world clinical presentations. It adopts the approach and mind-set of a clinical investigator and reveals how prescribers can practice 'bespoke psychopharmacology', tailoring care to the individualized needs of patients.
Pharmacogenetics: When Relevant, When Not
Your genetics is not your destiny.
The field of combinatorial pharmacogenomics is fast-evolving. One of the main controversies facing practitioners is understanding how much evidence-based findings from the literature are or are not yet ready for prime-time implementation in routine patient care – or, how much the deliverables remain in the gestational stage, and still belong more at the bench than the bedside. In fact, even experts can find it hard to agree about the distinct role of pharmacogenetics in routine clinical practice. Our main objectives in this chapter are twofold: (1) to help readers acquire a greater sense of literacy for interpreting pharmacogenetics data for themselves, and (2) to make prescribers think more about the neurobiology behind the decisions they make. In the case of pharmacogenetics, that means understanding how to interpret a pharmacogenetics test report, how to separate efficacy and tolerability as distinct drug effects, how to use pharmacogenetics to affirm hypotheses about possible reasons for poor response, and how to integrate pharmacogenetic data alongside other previously discussed moderators and mediators of treatment outcome. We will highlight points where even we, the authors, may differ in our points of view, and the reasons behind divergent perspectives, to help the reader form his or her own conclusions about the evidence base and utility for “bedside” pharmacogenetic testing.
▢ Define pharmacogenetics, pharmacogenomics, allele, single nucleotide polymorphism (SNP), familiality versus heritability, and complex versus Mendelian traits
▢ Understand the difference between safety and efficacy pharmacogenetics, the distinction between pharmacokinetic and pharmacodynamic gene variants, and the relevance of poor metabolizer, extensive metabolizer, and ultra-rapid metabolizer genotypes and phenotypes
▢ Know how to interpret the findings of a pharmacogenetics test report and the clinical significance of identified polymorphic variants
▢ Understand the strengths and limitations of currently available technology for utilizing pharmacogenetics to inform prescribing decisions in clinical practice
In Chapter 5, we discussed the growing concept of precision medicine and its forerunner term, “personalized medicine,” as the initiative to craft individually tailored treatments. For psychiatry, precision medicine represents the goal of utilizing a given patient’s unique clinical and biological profile in order to broker the best fit with a particular drug regimen. That theme is woven throughout this book, as the means by which clinicians must interpret large-scale clinical trials and decide whether and how their findings apply to an individual case. In the minds of many psychopharmacologists, pharmacogenetics and pharmacogenomics represent a key component, if not the key component, of that endeavor, based on assumptions that everyone’s unique genetic architecture must figure critically in how they will respond to a drug – and that without such information, efforts toward devising an appropriate pharmacotherapy regimen are merely trial and error. This chapter will explore the basis for these propositions and critically examine the evidence for if, when, and how pharmacogenetic testing may or may not be useful to personalized psychopharmacology.
Pharmacogenetics refers to how the variation in a single gene influences response to one drug. Pharmacogenomics more broadly refers to how variation in many genes within the entire genome affects drug response.
A first consideration involves the fundamental question of whether, or how much, drug response is under genetic influence, and if so, how significant its contribution is relative to the myriad other factors that influence pharmacotherapy outcomes. Many clinicians often presume that the effect a particular drug has in a given patient’s first-degree relative will, more likely than not, be observed in the patient now before us. Is this an evidence-based concept? The literature answers affirmatively only to a very limited degree. We must first discriminate between familiality and heritability.
There is debate and uncertainty about just how much family history informs drug response. There is little formal study of drug response within families. Consequently, clinicians vary in how strongly they perceive drug response to be a true phenotype (meaning that it can run in families, and is strongly under genetic influence).
Whether or not drug response “runs in families” is quite understudied. As noted in Chapter 5, lithium responsivity in bipolar disorder has shown an approximate two-thirds concordance between probands and first-degree relatives, with similar concordance rates in MDD with at least some SSRIs. Beyond these limited data, little to no evidence is available that informs whether or not a family member’s response to a psychotropic drug is familial.
Familiality refers to whether an observed phenomenon tends to run in families. Examples include speaking the same language, eating turkey instead of lasagna on Thanksgiving, and entering the same profession as one’s parents.
The probability that susceptibility to a particular psychiatric disorder (or, a proneness to a particular drug response) runs in families is sometimes confused with the separate issue of how much a particular trait results from the passage of genetic information across generations. Heritability refers only to how much variability in a trait (say, a phenomenon like mood or psychosis) within a population is caused solely by variations in genes among members of the population. The term tells us nothing about the likelihood that mood or psychosis per se is inherited; only that “high heritability” of mood or psychosis means that its variation from one person to another is heavily influenced by genetic rather than environmental factors. Heritability estimates (h2) from twin/family studies are high for many psychiatric conditions.
Heritability refers to how much the variability of a trait within a population comes from genetic as opposed to environmental effects. Examples include eye color, hammertoes, dimples, and freckles.
Note that textbooks and primary literature sources will vary in their reported heritability rates across psychiatric disorders – many sources often cite autism as “the most heritable psychiatric disorder” (e.g., Sandin et al., 2014), although absolutes are difficult if not impossible to know with certainty. For exemplary purposes, approximate heritability (h2) estimates, alongside relative risks (RRs) for disease when present in a first-degree relative, are summarized in Box 8.1.
|Autism spectrum disorders||0.80–0.90|
|Major depressive disorder||0.37|
a As reported by Sullivan et al., 2012
The genetic components of most if not all forms of psychopathology represent complex traits, meaning the coalescence of many genes that each exert small influences on observable phenomena. Unlike Mendelian genetics, where a single gene can express its dominance in the form of a complete syndrome (e.g., Duchenne’s muscular dystrophy, or cystic fibrosis, or Tay Sachs disease), the manifestations of many genes of small effect may contribute to the expression of dimensional phenotypes (such as impulsivity, or attentional processing, or autonomic hyperarousal, or sleep disruptions, or other dimensional characteristics, as described in Chapter 2). In the “cleanest” case, an effective medication is a comprehensive antisyndromal therapy (remember from Chapter 2 the differential diagnosis and treatment of increased abdominal girth); but more often than not, incomplete responses to a particular drug may reflect disparate components of a complex psychiatric syndrome, potentially targeting some but not necessarily all of its elements (such as lithium for impulsivity and suicidal behavior, psychostimulants for slowed attentional processing or low motivation, serotonergic antidepressants for anxiety and depressed mood). Assuming all these elements can fit neatly into a single rubric, then assuming that a single medication will remedy the whole ensemble of features, and then further assuming that the drug’s effects are governed in large part by genetic influences, makes for a lot of assuming.
“Complex traits” refer to heritable phenomena that are governed not by principles of Mendelian genetics but, rather, by multiple genes that are each thought to exert small effects on an overall phenotype. Most if not all heritable psychiatric phenomena involve complex traits.
Because psychiatric disorders involve multiple genes having small effects, variants of just one gene are unlikely to exert a large influence over a complex phenotype. Combinatorial pharmacogenetics involves examining a collection of several or more genes, but possibly dozens or even hundreds (from the roughly 20 000–25 000 human genes) may be necessary to understand meaningful effects.
Genetic factors generally are thought to confer diatheses or susceptibilities to psychiatric conditions, rather than to directly cause a condition itself. (Were that not the case, concordance rates among monozygotic twins would be 100%, which is far from the reality of psychiatric genetics.) Analogous to oncogenes and proto-oncogenes, psychiatric genetic predispositions constitute the first “hit” or susceptibility factor, ultimately then activated and expressed by one or more subsequent environmental interactions (second or third or more “hits”) that might include exposures such as trauma or abuse, substance misuse, medical illnesses (e.g., strokes, cancer), or poor capacity to manage high-stress life events, among other occurrences.
Before embarking further, let us refresh our knowledge of some fundamental genetics terminology relevant to our further discussion of pharmacogenetics (Box 8.2).
Allele: A variant or alternative form of a gene located at a specific position (locus) on a chromosome.
Association: Genetic association means that the frequency of a particular genotype (or SNP) is seen more often than would be expected by chance in connection with a particular trait of interest (such as drug response).
Genotype: The combination of alleles for a particular gene or locus.
Haplotype: A set of SNPs that are inherited together (for example, the major histocompatibility complex (MHC)).
Homozygote: A gene that has two identical alleles.
Heterozygote: A genotype having two different alleles.
Hardy–Weinberg equilibrium (HWE): The principle that allele or genotype frequencies within a population remain constant across generations in the absence of disturbing factors (e.g., mutations that introduce new alleles into the population). In candidate gene association studies, it is a quality control measure to show that cases and controls within a population are in HWE. Otherwise, if allele or genotype frequencies deviate significantly from HWE, it means that undetected factors in a population (such as racial or ethnic disparities – called “population structure” or “stratification”) are likely present that could account for observed genetic differences. Without demonstrated HWE, reported associations between SNPs and traits could be spurious and invalid.
Linkage disequilibrium (LD): When two separate genes are in LD, it means that alleles at the different loci are associated nonrandomly (i.e., more often than if they were unlinked); the genotype at one locus is linked with (not independent of) the genotype at a second locus.
Locus: A fixed position of a gene on a chromosome.
SNP: A single nucleotide polymorphism is a variation in a single base pair within a DNA sequence. Substituting one nucleotide for another (say, T(hymidine) for G(uanine)) results in the transcription of a different amino acid in the gene product encoded by a particular gene. If there is a known physiological consequence for such a variation, the SNP is described as being functional.
Variable number of tandem repeat (VNTR): VNTRs are adjacent patterns of nucleotide sequences that repeat within a DNA sequence.
Genetic association studies of complex traits typically focus on one of two approaches:
(a) Candidate gene studies: if one suspects that a particular trait might be associated with a specific enzyme, protein, receptor, or other gene product, that association can be examined or tested when the gene coding for the protein or enzyme of interest has a known, functionally important allelic variant (polymorphism), known as a single nucleotide polymorphism (SNP). For example, the enzyme catechol-O-methyltransferase (COMT, which degrades dopamine to its metabolite 3-methoxytyramine), is encoded by a gene (the COMT gene) for which there is a known SNP involving the substitution of a valine for a methionine molecule at position 158 – the so-called Val158Met polymorphism. When the Val amino acid occurs at this position, the resulting COMT product breaks down dopamine up to four times faster than if the Met variant occurs at this location. Consequently, a Met/Met homozygote for the Val158Met COMT SNP may have increased synaptic availability of dopamine in his or her prefrontal cortex, which in turn may lead to enhanced executive functioning.
SNPs, in general, are often identified by an accession or “rs” number, which denotes their “reference SNP cluster identification.” Results from candidate gene studies involving individual SNPs are sometimes hard to replicate because allele frequencies can vary across racial or other ancestral subgroups within a population – a phenomenon called population stratification, as noted in Box 8.2. Another substantial limitation of many candidate gene studies is the need for large enough sample sizes to provide sufficient statistical power to detect what are often small effects.
REMINDER: Val/Val means overactive.
A heritable trait that is not outwardly visible is called a “hidden” or endophenotype. Examples include verbal fluency and working memory, proneness toward impulsive aggression, and social intuition (sometimes called “theory of mind”).
To learn about the functional importance of any gene, if you know its rs number you can look it up easily at www.snpedia.com.
(b) Genome-wide association studies (or GWAS): here, rather than studying the possible association between a given trait (such as disease susceptibility, or drug response) and SNPs of an individual candidate gene of interest, one studies SNPs across the entire genome (typically examining over one million SNPs) comparing a group of cases (i.e., the phenotype of interest, such as a clinical diagnosis, or a particular drug response or adverse effect of a drug) with controls (where the phenotype of interest is absent) on allelic frequencies of all SNPs between cases and controls. Odds ratios (ORs) with associated p-values are reported, SNP by SNP, to show whether an allele of interest is over-represented in cases versus controls. However, because so many statistical tests get performed, the α level for significance must be set extremely high – typically at least 5 × 10−8. That means having a p-value of at least 0.00000005. That means many thousands of subjects in a GWAS to generate enough statistical power to detect a significant association. Consequently, when a signal is detected in a GWAS, the finding often goes unreplicated because the study may be underpowered. Statistical underpowering, along with sample heterogeneity (e.g., population stratification), pose major limitations for declaring existing pharmacogenetic technology as ready for prime-time everyday clinical practice.
It takes an enormous amount of statistical power (sample size) to detect multiple small effects.
As complex traits, most if not all psychiatric conditions represent an interplay of biological, psychological, developmental, and environmental factors. Similar multideterminate factors are thought to contribute to psychotropic drug response, with genetics being only one of many such contributors. To the surprise and dismay of many clinicians and patients, the present state of pharmacogenetic testing does not provide robust information that tells “what drugs work and what drugs don’t” for a given individual. What then is the practical relevance of pharmacogenetics, if not a crystal ball for predicting global treatment outcome?
In the universe of pharmacotherapy outcomes, true phenotypes (that is, phenomena which, by definition, are the product of genetic influences) are traditionally divided into drug effects related to metabolism and adverse effects (usually involving genetic variants of pharmacokinetic enzymes, and termed “safety pharmacogenetics”) and drug effects related to intended therapeutic effects (focusing on genetic variants of pharmacodynamic genes, and termed “efficacy pharmacogenetics”). We will consider each of these as separate domains.