Schizophrenia: Problems of Heterogeneity

in Clinical Genetics ed. A. Sorsby, pp. 345‑49 (London: Butterworth, 1953). Reprinted in Man, Mind, Heredity, cit. (with added title)


    Since Rüdin (1916) first showed that among the sibs of schizophrenics there was an excess of schizophrenic illness, the question of the mode of inheritance has been debated with a vigour and enthusiasm unmatched elsewhere in psychiatric genetics. Weinberg, the statistician who handled Rildin's data, proposed a complex theory involving more than one gene. A later expert, Luxenburger (1928, 1930, 1935b), who carried out the first systematic twin survey in psychiatry, on schizophrenic propositi, was inclined to support a hypothesis attributing the disorder to a recessive autosomal gene. Lenz (1937), whose interests embraced the whole range of human genetics, supported the notion of dominance. The latest in the field to take a definite standpoint is Kallmann (1938, 1946), and he bases his opinion on the widest experience of actual investigation. He is strongly in favour of the hypothesis of recessivity.


     Kallmann's contributions of a major kind are twofold. He made a very extensive family survey (1938) covering the sibs, half‑sibs, children and grandchildren of 1,087 schizophrenic propositi, the study differing in a significant particular from similar studies which had been made before. In earlier studies of the children of schizo­phrenics, families were almost automatically taken in which there were known to be one or more children. This involved a degree of biased selection of the parents, parents being automatically overrepresented whose illness came on late or was so mild that reproduction could still occur after the illness had begun. Kailmann started from a hospital population, without selection, and without knowledge whether there were children or not. His material contained a fairer selection of the graver forms of psychosis than had been obtained before. His second major contribution has been the collection and examination of a very large series of twins from the New York State mental hospitals. This series includes 953 schizophrenic propositi, as well as much smaller numbers of manic‑depressive, involutional, and senile psychoses. A detailed statistical study of the first 794 schizophrenic pairs appeared in 1946, when 2,741 sibs as well as parents, consorts, and other relatives had been investigated. The material was gathered by a systematic attempt to discover all the twins in a total mental hospital population of some 85,000 persons. From these studies the follow­ing percentage figures relating to the incidence of schizophrenia among various classes of relatives are of interest: (1) unrelated: general population, 0.85; stepsibs, 1.8; husbands and wives, 2.1; (2) relatives: first cousins, 2.6; nephews and nieces, 3.9; grandchildren, 4.3; half‑sibs, 7.0‑7.6; parents, 9.2‑10.3; full sibs, 11.5‑14.3; dizygotic co‑twins, 12.5; children, 16.4; children of two schizophrenic parents, 68.1; same‑sexed dizygotic twins, 17.6; monozygotic twins who have been separated for five years or more, 77.6; monozygotic twins who have not been so separated, 91.5. The increasing likelihood of schizophrenia with increasing nearness of blood rela­tionship to a schizophrenic is very striking.

    Kallmann considers that his findings and the findings that have been made by others support the hypothesis of single‑factor inheritance of a recessive type, but with a lower than 100 per cent rate of manifestation, that is, some homozygotes never develop schizophrenia but will nearly always be schizoid personalities. Mani­festation in the heterozygotic state is not entirely suppressed, but may also be shown in some schizoid traits of personality. Nonspecific genetical factors are postu­lated to account for variation in the clinical form of the disease and its degree of malignancy. Kailmann and his co‑workers have found that a robust bodily physique tends to mitigate the severity of the disease, or to protect against it, while loss of weight and an asthenic physique have the opposite tendency.

Some Difficulties

     The hypothesis of recessivity has been criticized by Koller (1939), and his argu­ment is an important one. Although in the detail in which he presented it, it is complex, it can be quite simply presented. If the schizophrenic is homozygous for gene s, he will only produce schizophrenic children by mating with a heterozygote, or with another individual homozygous like himself. Schizophrenics compose ap­proximately 1 per cent of the population, so that the frequency of s in the popula­tion is approximately 0.1, and the frequency of heterozygotes approximately 0.18, if one assumes that mating is at random. The frequency of persons homozygotic for s among the children of schizophrenics should be, accordingly, and again assuming random mating, 0.5 X 0.18 + 0.01 = 0.10. The frequency of homozygotes among the sibs of schizophrenics, which is not affected to the same degree by the gene frequency, may be taken as 0.25. If s is recessive the frequency of homozy­gosity for s, and therefore the frequency of schizophrenia, should be higher among the sibs of schizophrenics than among their children. Assortative mating would only equalize the two frequencies, or lead one to expect a higher frequency of schizo­phrenia in the children than in the sibs, if it was of a very extreme degree, if in fact a very large proportion of schizophrenics married others also homozygotic for s. Sev­eral studies of the consorts of schizophrenics and their families have been published and it is clear that in them the incidence of schizophrenia is only very slightly greater than in the general population. The incidence of schizophrenia in the con­sorts themselves has, for instance, been found to be 2 per cent. Assortative mating therefore can be neglected for this purpose.

    The actual observations, however, do not fit these expectations. The incidence of schizophrenia has been found by Kalimann to be higher in the children than in the sibs. This is therefore a strong indication of dominance. On the dominance hypothesis one would of course expect an equal incidence in children and sibs, but a difference of the extent which actually exists would not be a serious objection to the theory as it may result from a sampling error or be explicable in other ways.

    One of the findings which Kallmann has adduced as evidence in favour of reces­sivity is that approximately 5.7 per cent of schizophrenics have been found to be the children of consanguineous matings. The relation between gene frequency (p), the incidence of cousin‑marriage in the general population (a) and the incidence of cousin‑marriage between the parents of recessive homozygotes (c) is given by Dahl­berg's formula, which may be written p = a(1‑c)/16c. If for a we write 0.006, which is Bell's estimate of the frequency of cousin‑marriage in England and Wales (the incidence in the United States of America is unknown to the writer), and if we assume that about two‑thirds of Kallmann's consanguineous matings were first-­cousin marriages and give the value of 0.042 for c, the value of p proves to be 0.0 08554, and the value of p2 0.000073. However, the expectation of schizophrenia for a member of the general population, and the frequency of hypothetical homozy­gotes is about 0.009. The theory therefore leads us to the supposition that some 125 distinct autosomal recessive genes are involved.

    This number is not, perhaps, quite such an unlikely one as it would seem to be at a first glance. There might well be a large number of different genotypes in schizo­phrenia, just as there almost certainly are in the hodge‑podge of clinically inade­quately differentiated imbecilities and idiocies. There are, however, factual grounds for considering this consequence of the recessivity hypothesis improbable. A fair number of families have now been collected in which two schizophrenic parents produced children. The incidence of schizophrenia among these children has been estimated by Schulz (1940) as 38 per cent, or 45 per cent if doubtful cases were included. An earlier estimate made by Luxenburger was 66 per cent, and Kallmann's (1938) estimate is 68 per cent. Whatever the true figure, it is quite a high one: but if there are 125 different genes involved, it would be extremely unlikely for any two schizophrenic parents to be of the same genotype, and the incidence of schizo­phrenia in their children should be little higher than the incidence of schizophrenia in the children of one schizophrenic and one normal parent.

    The incidence of schizophrenia in the children of schizophrenics is rather more easy to reconcile with the dominance hypothesis. If dominant genes are involved, we must calculate their manifestation rates as approximating to 0.3. The children of two parents both of whom carry a dominant gene, but at different bei, should contain equal proportions of persons carrying one gene, carrying the other, carrying both, and carrying neither. The expectation of normality should therefore be (0.7 + 0.7 + 0.49 + 1)/4 = 0.72, and we should expect an incidence of schizo­phrenia of 0.28. If, however, the genes were at the same bei, the homozygous dominant would almost certainly be abnormal and, unless the double dose proved lethal, the incidence of schizophrenia would be 0.40.

The Possibility of Heterogeneity

     None the less, Kallmann's finding of a greatly increased rate of consanguineous marriage among the parents of schizophrenics would seem to be a direct proof of the operation of recessive genes. The only reasonable way in which to reconcile all the conflicting evidence is on the supposition of heterogeneity. In view of the fact that schizophrenia is a great deal more common than any single genetically determined disorder is otherwise known to be, heterogeneity is inherently probable. As far as the writer is aware, there is no evidence to the contrary which should be allowed great weight. The principal objection to the hypothesis of heterogeneity is that so far it has proved impossible to identify genetically distinct forms of the disorder. A brief glance at the work which has been done on this aspect may be allowed.

    If schizophrenia is genetically heterogeneous, we may expect to find some forms which are not genetically determined at all. There is some evidence that they exist. It is an accepted finding of clinical psychiatry, since the work of Kretschmer, that schizophreniform reactions may occur, that is, states of mental disturbance clinically resembling schizophrenia, but produced by some form of physical or mental stress, accessible to treatment by psychological means and by adjustment of the environ­ment, and likely to recover without, as in the great majority of cases of schizo­phrenia, leaving some sequelae behind. Schulz (1932) found in his investigation into the sibs of schizophrenics that, if he singled out a group of propositi whose illness came on after a head injury, the incidence of schizophrenia in their sibs was much lower than in the families of the remaining propositi. The work of Leonhard (1934) showed that in some rare families a clinically distinct form of psychosis, with many schizophrenic‑like features, may be identified with possession of a dominant gene. Leonhard, following Kleist (1921), believes that it is possible to distinguish between "typical" forms of schizophrenia, which take a progressive and malignant course, and "atypical" forms with a better prognosis. An experiment was staged in which the clinical differentiation was done by Leonhard, without knowledge of family relationships, and the familial investigation was done by Schulz without knowledge of the clinical findings in the propositi. One of the interesting findings that emerged was that the incidence of schizophrenia in the parents of the propositi was in the typical group 1 per cent, in the parents of the atypical group 6 per cent (Schulz and Leonhard, 1940). This does suggest a preponderance of recessive types in the former, and a proportion of disorders due to a dominant gene in the latter.

    On the other hand the attempt to identify the clinically recognized subforms of schizophrenia ‑ catatonia, hebephrenia, paranoid schizophrenia and schizophrenia simplex ‑ with distinct genotypes has failed. Some suggestive points have emerged, but do not take us far. Kailmann's figures, derived from his investigation of the children of schizophrenics, show, though he did not observe the point himself, a noteworthy correlation between the form taken by the illness in the propositus and in the relative. Furthermore, he found that the incidence of schizophrenia was more than twice as high in the children of catatonic and hebephrenic propositi as in the children of paranoid and simplex cases.

    The matter must be left here, with the important questions still all unanswered, although the available evidence has been far from exhaustively discussed.