Madness Explained

by Richard P. Bental

  Linkage research has been pursued vigorously in the decade or so since these first reports. However, consistent findings have been elusive. In the case of bipolar disorder, efforts have focused not only on chromosome 11, but also on chromosomes 4, 5, 6, 9, 10, 13, 16, 18, 21, 22 and the X chromosome, but without consistent results.45 In the case of schizophrenia, numerous gene sites have been investigated, and special efforts have been made to study loci known to influence the structure and distribution of dopamine receptors in the brain.46 Sites on chromosomes 10, 13, 18 and 22, which have been tentatively linked to schizophrenia, appear to overlap those that have been tentatively linked to bipolar disorder,47 an observation that has special significance in the light of the persisting doubts about the distinction between the two diagnoses. Again, however, findings have been almost entirely inconsistent.48

  In a particularly challenging review of this kind of research, Tim Crow reflected on his own participation in a multinational collaborative project that aimed to replicate previously observed associations between schizophrenia and ten genetic markers.49 Although the main authors of the collaborative group’s research report stated that they had found evidence of linkage with genes on chromosomes 6 and 8,50 Crow disputed these findings on statistical grounds. In an accompanying table, he listed no fewer than twelve genetic loci that had been associated with schizophrenia by linkage analysis. At the time that he was writing in 1997, other researchers had failed to replicate these findings for eleven of the twelve gene sites. As I complete the final draft of this chapter in late 2002, it is still the case that replicable linkage results for major psychiatric disorders have remained elusive.51

  Frustrated by this ‘maddening hunt for madness genes’,52 many commentators have begun to reflect on the limitations of the molecular genetic approach.53 No doubt, the disappointing findings partly reflect the obstacles that must be overcome by investigators (the difficulty of diagnosing a homogeneous group of affected individuals; the difficulties involved in statistically evaluating the results). In the face of these obstacles, some researchers have even begun to search for genes that influence particular complaints. For example, a research group at the University of Milan reported a study of over 400 psychotic patients with a range of diagnoses, in which they attempted to determine which symptoms were associated with different alleles of a gene known to influence the structure of the dopamine D4 receptor in the brain. The gene is highly polymorphic (that is, there are many different alleles of it) and there is evidence that different variants produce different proteins in the receptor. It was found that patients with some of these alleles were much more likely to be deluded than patients with others.54 If we do not entirely reject the findings of the molecular geneticists, it is reasonable to reach three conclusions. First, the findings, such as they are, seem surprisingly consistent with critical reviews of the family, twin and adoption study data, which have argued that genetic investigator shave almost always overstated the evidence that psychosis is inherited.55 Second, it seems likely that psychotic disorders are heterogeneous at the molecular level (that is, genes that predispose some people to become psychotic may play no role in the psychotic experiences of other people).56 Third – and this is perhaps the most important conclusion – it seems likely that any genetic contribution to psychosis is caused by many genes of minor effect. Despite the limitations of existing methods, if just one or two genes played a major role, they would almost certainly have been identified by now. For this reason, it seems much more likely that many genes each make a small contribution towards vulnerability to symptoms.

  On the inheritance of symptoms: a genetic footnote

  Before leaving genetic studies of psychosis behind us, I would like to return briefly to the apparent paradox raised by Mojtabai and Rieder in their critique of the symptom-orientated approach. Recall their claim that diagnoses such as schizophrenia and manic depression are more heritable than individual symptoms. Although they argued that this observation is inconsistent with an analysis of madness that focuses on particular complaints, we can now see why this is not necessarily the case.

  Let us take the molecular geneticists’ observations seriously and suppose that there are no major genes for psychosis. Instead, let us suppose that there are many minor genes, and that each gene influences a particular complaint by means of a complex causal chain. As the heritability of complaints, in general, appears to be quite small, we can also assume that many other (perhaps environmental) factors play a role in determining whether a complaint is experienced by a person who inherits the relevant gene. (In other words, we can assume that the relevant alleles have low expressivity and/or low penetrance.)

  Mojtabai and Rieder’s paradox is resolved when we note that psychiatric studies mainly focus on people with many complaints. This is because the rules embodied in the DSM and similar systems require several complaints to be present before a diagnosis can be made, and also because people with many complaints are especially likely to experience distress and seek psychiatric treatment. The more complaints people have, the more psychosis genes they are likely to have inherited, and the more likely it will be that some of those genes will be shared by close relatives. Because of the role that other factors play in each complaint, an affected person’s close relatives may not experience the same complaint. However, if they have a sufficient number of psychosis genes, enough may be expressed as complaints to enable them to meet the criteria for a global diagnosis. As a consequence, the diagnosis will appear to be more heritable than each of the complaints.

  I offer this argument merely to establish that Mojtabai and Rieder’s genetic objection to the approach I am advocating is not as compelling as it at first appears. Although admittedly highly speculative, it probably understates the weakness of Mojtabai and Rieder’s position. As we have seen, genes do not directly cause particular kinds of experiences and behaviours but instead influence the development of the brain, which in turn affects the cognitive resources available to the individual. Furthermore, as I described in the last chapter, complaints may be functionally interconnected to each other in complex ways. When these added levels of complexity are taken into account, the Kraepelinian paradigm seems even less buttressed by the genetic evidence.

  Is There a Biological Time-Bomb?

  Earlier in this chapter, I introduced the biological time-bomb hypothesis, the idea that some kind of neurodevelopmental defect is responsible for psychotic breakdowns that occur in late adolescence. Theories of this kind have been stimulated by the recent discovery that the normal brain undergoes major structural changes throughout the second decade of life. This process was first detected in electro-encephalographic (brain-wave) studies, which revealed that the kind of electrical activity generated by the brain changes during this period.57 Anatomical studies subsequently revealed that these changes coincide with a decrease in the density of synapses (connections between neurones) in the human cortex,58 a process now known as ‘neural pruning’. Whereas early brain development is accompanied by a rapid increase in the brain’s connectivity, it now appears that late development is accompanied by some kind of ‘weeding out’ of connections that serve no purpose. (Not surprisingly, some brain scientists have drawn analogies between this process and the process of natural selection.)59

  Although there are several different versions of the biological time-bomb hypothesis, it is usual to distinguish between ‘early’ and ‘late’ theories. According to the former, damage to the brain or abnormal development very early in life (perhaps before birth) only leads to symptoms when the brain reaches a particular developmental stage; on this view, the primary cause of the psychotic patient’s neurodevelopmental deficits is the early insult experienced by the brain.60 According to late versions of the hypothesis, excessive neural pruning in the second decade of life leads directly to symptoms.61

  Arduous research

  The best way to test developmental theories of psychosis is to study children who later become psychotic. O
f course, the practical difficulties inherent in this kind of research are formidable.

  It is possible to collect some developmental data retrospectively, for example by asking parents to recall milestones or by examining school reports or other kinds of archives. A recent study of this kind examined psychological tests routinely administered to all Israeli adolescents in preparation for their induction into the military, and matched them against a national register of psychiatric patients. It was found that adolescents who were diagnosed as schizophrenic between four and ten years after their military assessment functioned more poorly socially, and scored considerably lower on intellectual tests, than peers who remained well.62

  American psychologist Elaine Walker at Emory University has reported a particularly interesting series of archive investigations, in which she analysed home movies taken by the parents of children who developed schizophrenia symptoms in later life. In one study, evidence of abnormal or clumsy movements was observed in seven future patients but not in their brothers or sisters who later remained well.63 A similar analysis of film clips of a larger group of future patients and their siblings revealed emotional problems in the future patients.64 In film clips taken between birth and adolescence, there was more evidence of negative emotion and, in the case of girls, less evidence of positive emotion in the future patients in comparison with their brothers and sisters.

  Other retrospective investigations have used educational records to show that children who later develop negative symptoms function poorly at school,65 and child-guidance records to show that unhappy adolescents who later develop positive symptoms are abnormally suspicious of other people.66 Although these findings suggest that social, cognitive and emotional problems may precede the onset of psychosis by decades, retrospectively collected information is rarely entirely reliable, limiting the conclusions that can be drawn from this kind of research.

  Prospective investigations are of greater scientific value but are inevitably more difficult. Because they require children to be followed up over many years, a single investigation may take up a researcher’s entire career. Studies of this kind require considerable forethought about the kinds of measures to be employed, as decades of effort can be wasted if the wrong kind of information is collected during the early stages of the project.

  The most common type of longitudinal investigation is the high-risk study, in which children who are believed to have a high risk of developing a future psychiatric disorder (usually because one or both of their parents have been diagnosed as suffering from a mental illness) are examined at regular intervals, together with comparison children who are believed to be at low risk. These studies aim to discover whether or not the high-risk children are especially likely to become psychotic in adulthood, and whether any indications of future illness can be detected in early life. One of the earliest studies of this kind was initiated by psychologist Sarnoff Mednick and psychiatrist Fini Schulsinger in Copenhagen in 1962 and has involved 207 children of schizophrenic mothers and 104 children of normal parents.67 The New York High-Risk Project, which commenced in 1971 under the direction of geneticist L. Erlenmeyer-Kimling and psychologist Barbara Cornblatt, has followed up two separate cohorts, each including some children whose parents had been diagnosed as schizophrenic, some whose parents suffered from affective disorders, and some whose parents were psychiatrically normal.68

  Although these studies might appear relatively simple in conception, interpretation of the findings has not always been straightforward, as important factors have often been confounded. For example, many of the children of schizophrenia patients recruited to the Copenhagen study had been given up for adoption or raised in institutions. It was therefore decided to compare them with children who had similar histories. It later transpired that the biological parents of many of the children in the control group were recidivist criminals. (In an effort to overcome these kinds of difficulties, a study was set up in Israel in which some high-risk and control children were recruited from kibbutzim.69 In the kibbutz system of communal living, which is unique to that country, children are raised collectively, live in dormitories, and spend comparatively little time in the exclusive presence of their parents.)

  Because of these kinds of confounds, researchers have usually employed complex statistical techniques (especially a group of procedures known collectively as path analysis) to try to tease out the way in which different possible predictor variables interact to influence the future development of psychotic complaints.70 However, these efforts have sometimes been undermined by the unexpectedly low rates of psychiatric disorder observed in many of the high-risk children as they have matured. For example, in the Copenhagen study, only 13 of the high-risk group had met the ICD-8 criteria for schizophrenia by the age of 25 years, although many more were said to show evidence of neurosis, personality disorder or ‘borderline states’.71 In the New York study, only 36 out of 208 individuals had shown evidence of any psychiatric disturbance by the time they reached the age of 20 years, most of these having come from the high-risk group. Of these, only 13 appeared to be psychotic, and they were drawn in equal proportions from the disturbed children in all three groups.72 Clearer evidence of the emergence of psychotic symptoms was reported when the children had reached 25 years of age, by which time 13 per cent of the children of parents diagnosed as suffering from schizophrenia had themselves met the criteria for the disorder.73

  In total, more than twenty high-risk studies designed to detect the developmental precursors of psychosis have been carried out, mostly focusing on schizophrenia. (Oddly, there have been no high-risk studies that have specifically focused on the children of bipolar parents, even though, as we saw in Chapter 4, family and twin studies have generally yielded stronger genetic effects for bipolar symptoms.) The results of these studies are not entirely consistent, but the broad picture that emerges has been drawn together by American psychologist Joan Asarnow74 and more recently by British psychiatrists Chris Hollis and Eric Taylor.75 Some studies have detected evidence of motor abnormalities in the first few years of life. During middle childhood, many future schizophrenia patients show evidence of poor social adjustment, and poor performance on attentional and other cognitive tests, and these differences from children who do not later become psychotic become more evident during adolescence. Overall, the balance of this evidence points to the role of early brain abnormalities in children who will later become psychotic, although a contribution from later maturational processes such as neural pruning cannot be ruled out.

  More detailed analyses of the data from the individual studies point to complex interactions between biological and environmental factors and suggest that different pathways may lead to different complaints. For example, the Copenhagen researchers identified those in their sample who had predominately positive or negative symp-toms.76 In the positive group there was strong evidence of inadequate parenting (six out of eight had been separated from their mothers for more than one and a half years during early childhood, one had been separated from her mother for more than five years, and one had been physically abused by her father). Similar evidence of inadequate parenting was absent in the negative symptom group, many of whom had suffered complicated births and showed enlarged cerebral ventricles on CT scans. In the New York study, poor social adjustment in adulthood (suspicious solitude, social insecurity or a lack of empathy) was associated with physical anhedonia and attentional impairments in childhood.77

  Almost certainly, the children recruited to high-risk studies are an unrepresentative sample of pre-psychotic children. After all, by far the majority of people who receive a diagnosis of schizophrenia or bipolar disorder do not have psychotic parents. This potential source of bias is avoided in a second type of longitudinal approach, in which individuals from the general population are followed up from birth. These kinds of investigations, known as cohort studies, are usually set up with broad aims in mind, and so measures of specific interest to psychiatrists and clinical psych
ologists are often not included at the outset. Two such studies carried out in the UK are the National Survey of Health and Development (NSHD) and the National Child Development Study (NCDS). The NSHD was originally a survey of all children born in England, Scotland and Wales between 3 and 9 March 1946, and was carried out to assess the need for maternity services in the planned National Health Service. Multiple and illegitimate births were excluded, and 13,687 births were studied in total. Regular contact has been maintained with a subset of 5362 individuals, who receive a birthday card from the project every year. As time has passed, various assessments have been made and additional data have been collected from educational and health services. The NCDS is a similar study, set up in 1958, designed to monitor the physical, social and educational development of children through to adulthood.

  British psychiatrists Peter Jones, Robin Murray and their colleagues have exploited these studies by identifying participants who had been admitted to psychiatric hospitals.78 Diagnoses were obtained by scrutinizing hospital case notes. From the data collected at an NSHD assessment at 2 years of age, it was found that the future psychotic patients were later to walk than their peers by an average of 1.2 months. Evidence of higher levels of clumsiness at 7 years in the pre-psychotic children was observed in the NCDS. In both studies, the future patients had, on average, slightly lower scores on IQ or similar tests when assessed between the ages of 7 and 16 years. However, perhaps the most interesting findings concerned speech. In the NCDS, qualitative reports of speech difficulties by mothers and teachers at 7 and 11 years discriminated between pre-psychotic and normal children, and a similar association between future psychosis and non-specific speech difficulties assessed by school doctors emerged from the NSHD. It would be interesting to know whether speech problems early in life are specifically associated with later thought disorder, but the NSHD and NCDS studies do not provide sufficient clinical information to answer this question.