Figure 17.2 Two models of genetic influences contrasted by Mojtabai and Rieder (1998).
the overall structure of the family was poorly organized to cope with stress. Singer then attempted to predict the patient’s ratings from the ratings taken from the parent, and found that she could do so with a high degree of success. When given transcripts from the parents and the patients without knowing which transcript went with which, she was able to match them together, again with a high degree of accuracy.
Most subsequent studies have supported Singer and Wynne’s conclusion that communication deviance (to use their terminology) is found in the families of thought-disordered patients.19 For example, in two recent studies Nancy Docherty (whose work I discussed in some detail in Chapter 15) found that the speech of the parents of thought-disordered patients contains frequent referential failures.20 However, whereas the speech of patients becomes more incoherent as they become emotionally aroused, Docherty found their parents’ speech was not affected in this way.21
Communication deviance appears to be completely unrelated to expressed emotion, the type of critical or over-controlling behaviour sometimes observed in the families of psychotic patients, which I described in the last chapter.22 There is some evidence that it reflects, in part, genetic factors. For example, Orsola Gambini and colleagues at the University of Milan recently studied the speech of normal monozygotic and dizygotic twins. When they rated the speech for thought disorder, they found higher concordance rates in the MZ twins than in the DZ twins.23 In another recent study, Dennis Kinney and colleagues at Harvard University used an adoption study strategy to investigate the heritability of thought disorder, which they measured in Danish adoptees who had developed schizophrenia symptoms and adoptees not diagnosed as schizophrenic, before then tracing their biological parents through Denmark’s comprehensive adoption records. They found high levels of thought disorder in the parents of the schizophrenia patients, but not in the parents of the healthy adoptees.24
Of course, as we have already seen, strong genetic influences do not exclude the role of environmental factors. That such factors may play a decisive role in the case of thought disorder was revealed in another adoption study, this time carried out in Finland by a large team that included Lyman Wynne and Pekka Tienari, a respected psychiatric geneticist.25 The team used a refined version of Singer and Wynne’s methodology to assess disordered communication both in the adopted-away children of schizophrenia patients and in their adopting relatives. Adoptees who were regarded as being at high risk because their biological parents had been diagnosed as schizophrenic showed high levels of communication deviance only if their adopting parents showed evidence of communication deviance. Adoptees who were not at genetic risk of developing a psychotic illness did not develop thought disorder, even if their adopting parents showed evidence of communication deviance. These findings show that thought disorder results from an interaction between genes and the environment.
In contrast to these findings, genetic evidence on paranoid symptoms suggests that these are very little influenced by heredity. For example, studies of patients diagnosed as suffering from personality disorders have consistently yielded lower heritability estimates for the diagnosis of paranoid personality disorder than for any other DSM axis-2 diagnoses.26 Furthermore, it also appears that, even if genes make a minor contribution, they are unlikely to be the same genes as those involved in other complaints attributed to schizophrenia. In a series of studies carried out in the 1980s, Kenneth Kendler found that the relatives of patients given a diagnosis of delusional disorder rarely showed evidence of schizotypal traits,27 a finding that has been replicated many times in subsequent investigations.28
On this evidence at least, different psychotic complaints seem to be differentially influenced by genes. Of course, this still leaves us with Mojtabai and Rieder’s paradox that diagnoses seem to be more heritable than individual symptoms, to which I will return later.
What the new genetics tells us
In order to understand further the relationship between genes and symptoms it will be helpful to think about the mechanisms by which genes influence behaviour. This takes us into the complex territory of molecular genetics, where we will find evidence that will further discomfort neoKraepelinians. I will begin with a very brief primer covering some of the most important scientific concepts in this area.
Even when strong genetic effects can be demonstrated by family, twin or adoption studies, many intermediate steps must lie between genes and madness. Genes consist of sequences of chemical elements, known as nucleotides, joined together in long molecules of DNA, which in turn are contained in the chromosomes held in the nucleus of every cell. In humans there are twenty-three chromosomes, which together comprise the genetic material known as the human genome. However, most cells contain a double set, so that there are twenty-three pairs, with one member of each pair drawn from one parent and one from the other. The gametes – sperm and egg cells – are exceptions to this rule, as they contain only one version of each chromosome. During the initial formation of these cells, DNA may be swapped between corresponding sections of each pair of chromosomes, and one chromosome from each pair is then selected at random to enter each gamete. Of course, when sperm from the father and egg cells from the mother combine at fertilization the single sets of chromosomes in each gamete pair together, so that the resultant embryo contains twenty-three new chromosome pairs.
There are four types of nucleotides, labelled A (for adenine), G (guanine), C (cytosine) and T (thymine). The order in which these are arranged (sometimes described as the genetic code) determines the synthesis of proteins in each cell. Some sequences of nucleotides are involved directly in protein synthesis in co-operation with other biochemical systems; some sequences switch on or off the production of different proteins during different developmental periods and in different parts of the body. Genetic variation occurs because genes (sequences of nucleotides in particular regions – loci – on the chromosomes that are responsible for particular proteins) come in different varieties (differences in the sequence), known as alleles. Strictly speaking, therefore, rather than saying that someone has inherited the gene for a particular characteristic, we should say that they have inherited (one or two copies of) the relevant allele of that gene. For some characteristics, inheriting one crucial allele is sufficient, in which case the allele is said to be dominant. For others, the characteristic does not appear unless both of the corresponding alleles on a pair of chromosomes are of a particular type, in which case the allele is said to be recessive. The precise composition of alleles in the chromosomes is known as the individual’s genotype.
Ultimately, the synthesis of proteins controls the structure of the developing body. Thoughtful biologists have pointed out that unravelling this process of development is a task that will dwarf in complexity the much-vaunted efforts to map the human genome (that is, to determine the exact sequence of nucleotides on the DNA in the twenty-three chromosome pairs).29 Of course, it has long been recognized that single genes rarely exclusively determine characteristics in the adult person, either because interactions between different genes are involved, or because of environmental influences. When only a small proportion of individuals carrying crucial alleles develop a trait, the alleles are said to have low penetrance. In fact, the penetrance of an allele can be expressed as a simple percentage describing the proportion of carriers who are affected. Alleles also vary in their expressivity, which refers to the extent to which genetic traits, when present, vary in magnitude. The alleles responsible for some physical disorders, for example Marfan’s syndrome, which is characterized by various abnormalities in the skeleton and connective tissues, are 100 per cent penetrant (everyone carrying the crucial allele gets it to some degree) but variable in their expression (some people suffer from life-threatening defects in the wall of the aorta, whereas some people just look odd).30 Not surprisingly, given the evidence from twin and adoption studies reviewed ear
lier in this book, most psychiatric geneticists assume that alleles responsible for schizophrenia and bipolar disorder must be less than completely penetrant and variable in their expression. Geneticists sometimes use complex mathematical analyses of data from family, twin and adoption studies to derive estimates of the number of genes contributing to a particular disorder, whether the alleles are dominant or recessive, and the extent to which they are penetrant and variable in their expression. However, despite considerable efforts, no widely accepted model of the genetic contribution to psychosis has been generated this way.
In the case of the brain, some genes that cause undifferentiated cells in the embryo to become brain cells have been identified, together with genes that control the segmentation of the developing nervous system into different regions containing different kinds of neurones utilizing different chemical transmitter systems.31 However, although genes help to determine the broad structure of the nervous system, it is clear that they do not precisely determine the final ‘wiring diagram’ of the adult brain. It has been estimated that about 30,000 genes play a role in determining the structure of the human body, a number that is remarkably invariant across mammalian species that differ dramatically in the complexity of their nervous systems. By contrast, it has been estimated that the adult human brain contains about 100 trillion synapses, and that more than a quarter of a million of these are formed every second during the very early stages of development.32 Animals raised in impoverished environments compared to animals raised in enriched environments develop fewer synapses, and a lower density of blood capillaries (which supply neurones with nutrients),33 confirming that this process is influenced by experience. In a recent attempt to draw together evidence about this process, a group of distinguished researchers from Britain and the United States, led by Jeffrey Elman of the University of California at San Diego, concluded that brain development is determined by interactions with the environment at every stage, leading them to ponder whether the idea of ‘innate’ behaviour should be dispensed with altogether.34 Unfortunately, these advances in the understanding of the genetic control of neurodevelopment appear to have largely escaped the attention of psychiatric researchers.
Following the development of techniques for mapping DNA,35 psychiatric geneticists have hoped to identify genes, and eventually alleles of those genes, that determine the major psychiatric disorders. The most widely explored approach is known as linkage analysis, which capitalizes on existing knowledge about the location of certain genes (known as markers) on particular chromosomes. Linkage analysis is usually carried out on data collected from small numbers of families in which there are many affected cases, as this increases the probability that specific alleles are involved in the affected individuals’ psychiatric problems. The closer the physical location on a chromosome of a marker gene to a psychosis gene, the more likely the marker gene will be shuffled together with the psychosis gene during the interchange of DNA between chromosomes prior to conception. Therefore, if marker genes are found mostly in those family members who are affected by the illness, the gene responsible for the illness is likely to be located close to the marker gene on the relevant chromosome. Most linkage studies utilize a small number of markers located along a single chromosome. However, in some recent studies, markers spread evenly across all the chromosomes have been examined together, a technique known as genome scanning.36
There are several points to note about this approach. First, these techniques assume that researchers can accurately diagnose affected individuals. As most genetic researchers have accepted the validity of the neoKraepelinian system, it is not surprising that they have come unstuck on this very problem. Second, linkage analysis not only requires the use of complex biochemical techniques, but also sophisticated mathematical checks to ensure that any observed association between an illness and a marker is not the result of chance. Third, even if it is possible to locate psychosis genes in affected families, it is possible that the genes play no or little role in non-familial forms of psychosis (that is, in psychotic illnesses that appear sporadically, without a history of illness in the family). Finally, for reasons that should now be obvious, discovering the location of a psychosis gene is a long way from understanding how that gene plays some part in the genesis of psychotic behaviours and experiences. To make further progress, it will be necessary to understand how the implicated gene is involved in the synthesis of proteins, or how it influences the activity of other genes which do this. We would need to know the difference that the aberrant allele makes to that process, the role of the relevant proteins in brain development, the final consequences of any developmental abnormalities for the structure and function of different brain systems in adulthood, the cognitive processes that are supported by those brain systems and, last but not least, the role of those processes in patients’ complaints.
The earliest linkage studies of psychosis were hailed as major breakthroughs but soon turned out to be a source of embarrassment to biological psychiatrists. Janice Egeland and her colleagues in the United States chose to study a population with a restricted gene pool, the Old Order Amish living in southeastern Pennsylvania. The Amish (familiar to many people from their portrayal in Peter Weir’s film The Witness) are a religious sect that has shunned the trappings of modern life, has pursued a rural existence with the minimum use of mechanical aids, and whose members have rarely married outsiders. Psychiatric interviews were conducted and hospital records examined for a pedigree of 120 related individuals. Using the RDC, 19 received a diagnosis of major depression, 14 received a diagnosis of bipolar disorder, and 5 received a diagnosis of unipolar depression. In 1987, Egeland and her co-workers reported strong linkage between bipolar disorder and two markers on chromosome 11, indicating that agene for bipolar disorder lay somewhere on this chromosome.37 Only two years later, the investigators reported a second analysis of the pedigree, adding data from a further 37 members.38 These new data, on their own, had little impact on the findings. However, in the period between the two analyses, two individuals who had been considered unaffected in the original analysis developed bipolar symptoms. When these individuals were taken into account in their mathematical analysis, Egeland and her colleagues discovered that the association between bipolar disorder and their markers disappeared. The earlier finding had been spurious.
Reflecting on similar efforts to find genes for schizophrenia, American geneticist Cathy Barr described waking up one morning in 1988 to hear a radio news bulletin announcing that researchers had found linkage for the diagnosis. ‘The significance of such a finding caused me to bolt straight up from bed’, she recalled.39 She immediately reviewed her career plans and, soon afterwards, was carrying out her own research in psychiatric genetics. Most British researchers similarly learned of the first apparently successful linkage study of schizophrenia from a news bulletin. Returning home from work one night in 1988, I turned on my television set, to be greeted by a headline informing me that researchers had discovered the gene for schizophrenia on chromosome 5. The news bulletin announced that the results of a study conducted by investigators at the Middlesex Hospital would shortly appear in the prestigious journal Nature. When David Hill, a sceptical clinical psychologist, rang Nature on behalf of a British news magazine to inquire about the status of the report, he was told that it had not yet been considered by the journal.40 Presumably, the researchers’ enthusiasm to win the race to find a schizophrenia gene led them to assume that the journal would accept their report when it was eventually submitted for publication. In this respect, at least, they showed sound judgement, as their paper eventually appeared in Nature later in the year,41 alongside a report from a group of US and Swedish researchers claiming that a gene for schizophrenia did not exist on chromosome 5.42
The Middlesex group had obtained data from 104 people in five Icelandic and two British pedigrees whereas the US–Swedish team had obtained data from 81 people from a single Swedish pedigree but had used more powerful mapping techn
iques. An examination of the Middlesex report reveals causes for concern that should have been obvious to any competent reviewer (raising important questions about why Nature accepted the paper). For example, the researchers compared various different definitions of schizophrenia in order to maximize their linkage findings. Maximum linkage was achieved by including as ‘schizophrenic’ individuals who were diagnosed as suffering from depression, phobias, alcoholism and drug addiction. Although these individuals would not be considered remotely schizophrenic according to any definition of the disorder promulgated in the last 100 years, the researchers described them as ‘fringe phenotypes’ (that is, as individuals who were assumed to be genetically schizophrenic, but whose symptoms differed from the prototype because of presumed environmental influences). A subsequent study published by a group of researchers in Edinburgh used markers similar to those employed by the Middlesex group but obtained negative findings.43 After several further failures to replicate their findings, the Middlesex group reanalysed their data with an expanded sample and withdrew their original conclusions.44
Madness Explained Page 51