Rüdin’s 1916 Monograph: On the Inheritance and Primary Origin of Dementia Praecox

Abstract

In 1916, Ernst Rüdin published the first modern family study in the history of psychiatric genetics, the major goal of which was to test whether the pattern of risk in the siblings of dementia praecox (DP) probands followed Mendelian expectations. He utilized systematic ascertainment of probands and multisourced diagnostic assessments of probands and relatives, applying the narrow Kraepelinian concept of DP. In a novel step, he collaborated closely with a statistical geneticist—Wilhelm Weinberg—and applied his sibling, proband, and age correction methods. In his key sample—701 sibships when neither parent had DP—the morbid risk for DP in siblings was 4.48%, much lower than 25% expected for a recessive disorder. Risk for DP was increased by alcoholism or other mental disorders in parents. Other non-DP psychoses were common in both siblings and parents of DP probands. Rüdin discussed several alternative genetic models for DP including a 2-locus recessive, incomplete penetrance, and an oligogenic model. The high rates of other psychoses and psychopathic personalities in relatives might arise, he suggested, because these disorders shared genetic risks with DP. Rüdin established that DP, when carefully studied, ran in families, did not have a simple Mendelian genetic transmission pattern, and appeared likely to be genetically related to other non-DP psychotic disorders and perhaps some kinds of psychopathic personalities. This study, the most important in Rüdin’s career, should be viewed in the context of his later extensive support of and collaboration with Nazi eugenic policies.

Ernst Rüdin (1874–1952) was born in St. Gallen, Switzerland where his father was a textile merchant and his mother was a physician. He studied medicine at universities in several European countries, graduating from Zurich in 1898 after which his first job was as an assistant to Eugen Bleuler at the Burghölzli psychiatric hospital of the University of Zurich. In 1907, he went to work at the psychiatric clinic of the University of Munich in 1907 where he became an assistant to Emil Kraepelin, under whom he completed his doctoral thesis. When Kraepelin’s lifelong dream, the German Research Institute for Psychiatry (Deutsche Forschungsanstalt für Psychiatrie) was established in Munich in 1917, Kraepelin appointed Rüdin as head of its Department of Genealogical and Demographic Studies. Most of the leading next generation of psychiatric geneticists from across Europe came to train in this Department in the 1920s and early 1930s¹ which he directed up until the end of World War II.

This monograph,² universally recognized as an important milestone in the history of psychiatric genetics and the study of DP/schizophrenia,^3–5 began as follows:

The goal of the investigations which has made a modest start with this research, is the study of the structure of abnormal mental predispositions in the families of our mentally ill. Over time, by researching the families of our patients for the type and number of afflicted individuals and illness predispositions, a picture will be gained about their origin and inheritance. In the field of psychiatry, we are still a long way off from noteworthy or conclusive knowledge in these 2 mentioned directions. The difficulties of research in humans and especially in the field of the pathological predisposition of the nervous system are so great that we can only make progress incrementally.² P. III

This 172-page monograph, containing 61 tables and 56 figures, was divided into a foreword and 12 chapters. A complete English translation is in the appendix. A brief English review of this report has been published previously.⁶ Key quotes are put in the text, sometimes with added italics for emphasis. Other interesting, but less critical, quotes are placed in table 1.

Table 1.

Selected Quotes From the Rüdin’s Monograph

Quote #	Text
1	Proof in humans is extremely difficult and especially in the field of pathological characteristics requires specific conditions. It exists mainly in a subject-specific application of the statistical method on a material that has been collected according to specific principles. If one wants to specifically examine the rules of heredity of pathological characteristics in humans, one has a harder time than zoologists and botanists. Zoologists and botanists breed the material they require, in the quantities they require it. The medical practitioner has to make do with the material he obtains, which often enough is only made available due to a coincidence of the place, period, and other circumstances2 pp. 2–3.
2	Regarding the question of the lawfulness of the inheritance of pathological traits in humans, medical practitioners, and especially psychiatrists, have hitherto also created artificial difficulties which do not reside in the inadequacy of the material, but rather in the imperfect treatment of it, in preconceived notions about it. In the case of numerous colleagues, a fundamental mistake in the approach to questions of heredity was and still is the fact that their eyes seem to be hypnotically riveted to pedigrees with a particularly high burden2 p. 3.
3	To obtain Mendelian ratios from an inadequate study material, which is what is mostly available in the case of humans, Weinberg has devised the “sibling method” and the “proband method,” by means of which especially the lack of dementia praecox-free sibling series from which the psychiatrist suffers is corrected2 p. 12.
4	However, this is still lacking and I don’t dare to decide how many different, mental, psychopathic, or psychotic hereditary types occur in the sibships in addition to the dementia praecox siblings, let alone that I could allow myself to try to enumerate these different types according to their relative frequency and to force them into some or other Mendelian ratios2 p. 56.
5	We already know today that conditions occur in siblings of dementia praecox sufferers that are clearly different from the pronounced dementia praecox recognized as such by every psychiatrist, and also from complete mental health. One need only recall the “cranky psychopaths,” which are not so uncommon in such sibships, and which, as a result of some psychopathological points of contact with actual dementia praecox patients (alias schizophrenic psychotics) in the mechanism of their whole psychological and psychomotor behavior have been described as schizophrenic psychopaths. Nobody knows yet how they are to be assessed pathogenetically. That these are, as some maintain, only a lighter degree of dementia praecox is just as unproven as it would be to equate them completely with any other “psychopath.” What we do know is that this type of psychopath is found remarkably frequently in families in which pronounced dementia praecox is found2 p. 57.
6	The feature “dementia praecox–free,” by which the parents of the material used in Chapter 2 are characterized, of course, allows the presence of a number of positive features that have always been regarded by the psychiatrist as “burdening” factors2 p. 65.
7	In the overwhelming number of cases, only common (environmental) influences had an effect on those suffering from dementia praecox, and after a closer examination of the cases, one almost always gains the conviction that the onset of the disease is more deeply rooted in the personality and constitution itself which speaks for the fact that at least half of the sick are portrayed as always “different from their siblings and other people”2 p. 109.
8	In this context, one can understand the disposition to a disease as a dependency on one, possibly also several hereditary factors. The presence of such a factor either changes the constitution in such a way that … in the case of hereditary diseases, such a hereditary unit is the precondition (the conditional factor) for the hereditary factor to be active as a disease2 p. 110.
9	It has always been recognized that different-looking, ie, polymorphic psychoses can occur in one and the same family. Thus one spoke of polymorphic inheritance, since at the same time one was convinced that they arose on the basis of a common, general, and obviously hereditary predisposition… It is asserted that not a particular disposition to a particular form of mental disorder or a particular mental disorder is inherited, but that a general disposition is inherited, on the basis of which the most varied of endogenous and exogenous mental disorders can arise2 p. 139.
10	It would have to be done in such a way that, on the assumption that the individual psychoses and mental defects can substitute for each other in heredity, each mentally ill or defective person who is admitted to a given place at a given time, also would be examined with regard to the mental nature of his siblings and both of his parents. When processing the material, every psychotic who was in the institution at the relevant time would have to be counted as a proband, regardless of the type of psychosis, and every psychotic identified by research as a secondary case2 p. 141.
11	Where does, for example, psychopathy end and mental health begin? Or should psychopaths, hysterics, and alcoholics (and to what degree?) not be counted at all? And what about the exogenous psychoses, with the senile, organic, syphilitic psychoses, psychoses with brain tumors, with infectious illnesses etc? Are “convulsions” in childhood also the expression of a psychopathic or psychotic disposition (as is assumed in American studies [Rosanoff] and do they have to be included?2 p. 141.
12	… heredity research itself would obstruct an important source of instruction if, on the basis of partly outdated ideas about heredity, it decreed essential equivalence of illness processes [in close relatives] where, from a clinical and even etiological point of view, it is not the similarities but the differences that predominate2 p. 147.
13	We are therefore fairly well informed about the clinical forms of the current generation, but relatively poorly informed about well-characterized disease forms in the previous generation, so that it is particularly difficult to find a sufficiently large material of reliable dementia praecox diseases of the previous generation with the associated clinically obtained well-characterized offspring2 p. 156.
14	It seems, therefore, that the occurrence of other types of psychosis in the parents and siblings of those with dementia praecox is not an incidental side effect of the genesis of dementia praecox itself or its predisposition, but that it is in some way closely related to it2 p. 164.
15	Since the psychoses found in the parents and siblings are partly different, it seems to me first that the most probable assumption at the present stage of my investigations is that the mental disorder of the parent and dementia praecox, as well as the other psychoses of the associated children, are the products of complicated segregating hereditary processes,2 p. 168.
16	From this, it can be concluded that the development of dementia praecox, as well as other psychoses in general, requires the interaction of mutually pathological hereditary predispositions2 p.164.
17	Although, as we have heard, the assumption of the existence of the simplest recessive inheritance (a Mendelian pair of traits) for dementia praecox is to be rejected according to the present investigations, its recessivity in any form is made highly probable by the arguments already mentioned insofar as it must be regarded as a product of pathological factors that obviously complement each other from the 2 families of origin, the paternal as well as the maternal2 p. 165.
18	The assertion of a polymorphic inheritance is currently neither refuted nor proven, but its existence is probably possible on the basis of Mendelian views. In view of the relatively high frequency of other types of psychotic states in dementia praecox families (especially among parents and siblings of dementia praecox patients), it is not exactly unlikely2 p. 167.

We first review the findings presented and then discuss their historical significance and comment on Rüdin’s extensive collaboration with the National Socialists from 1933 to 1945. A footnote to the forward states:

The manuscript of the present monograph was sent to the publishing house on May 14, 1914, and printing was almost complete at the beginning of the war. As a result of the conditions created by the outbreak of war, the publication was delayed in an unforeseeable way² p. IV.

Before proceeding with a formal review of this paper, we wish to note that for readers interested in delving deeper into Rüdin work, we strongly recommend reading this monograph in the context of his major “position” paper outlining his view of an ideal family research program for psychiatry published in 1911⁷ which we have also explored in detail and translated into English.⁸ We also suggest readers review the introduction to this series which provides further background to this report.

Chapter I–Methodology

Rüdin begins this chapter with a concise summary of his understanding of the role of familial factors in the etiology of Dementia Praecox (DP) circa 1914:

If one has the opportunity to follow a large series of cases of adolescent insanity (dementia praecox) and their relationship to the health conditions of their relatives, one will at first not be able to avoid the impression that dementia praecox notably often does not occur as the only mental disturbance in a family. Conspicuously often in close or distant relations, there are other cases of dementia praecox, often also other pathological mental conditions. Compelling external reasons for the familial occurrence are initially not discoverable… the claim that dementia praecox arises due to detrimental psychological influences on the mind at this stage is lacking in any evidence. Finally, all the other numerous moments which have an influence from the outside, and have at times been blamed, do not adequately explain the familial occurrence of the disturbance… Faced with the absence of a unified external etiology, the familial occurrence at first glance primarily causes one to think of the role of heredity² p. 1.

Rüdin argues that heredity likely plays an important role in DP because (1) DP runs in families, (2) but other disorders also aggregate in relatives of DP patients, and (3) no good environmental explanation for the familial aggregation of the disorder has emerged.

What might be the best approach to further investigate the hereditary nature of DP? Rüdin summarizes his preferred method:

… an examination about the heredity of mental illnesses only deserves to be called scientific if it is permeated by the mathematical aspirations which have already led to beautiful results in experimental biology of inheritance in plants and animals² p. 1.

Notice that in this passage, Rüdin is not referring to the many possible meanings that “heredity” had throughout the 19th century, which included aspects of Lamarckian transmission of acquired characteristics as well as degeneration theory. Rather, he is referring to the specific theories of inheritance as postulated by Mendel and confirmed in the “experimental biology of inheritance in plants and animals.” He continues:

Since we, however, do not have access to deliberate breeding, we only have recourse in humans for the verification of the rules of heredity to the descendant and pedigree tables. However, the members of a set of relatives are afflicted by numerous destinies, which even in a large family branch only exceptionally will allow Mendel’s laws to emerging in any form purely.² pp. 1–2.

He then outlines his methodologic approach to this problem:

It is in the nature of heredity processes that an absolute law that is expressed in particular ratios will only show itself in a large statistical mass. In fact, this will have to be the greater, the more complicated the inheritance, in other words, the more numerous the competing hereditary factors are² p. 2.

In an approach that had not been previously taken by any researcher, Rüdin explicitly wants to approach the problem of clarifying the heredity of DP in the way in which Mendel approached his study of hybridization in peas. As noted by Rüdin a few pages later (and explicitly evaluated empirically in our last paper by Schulz), this approach assumes that DP is an appropriate “unit character” for Mendelian analysis.

By the time Rüdin was working, a few human traits and diseases had been found that followed Mendel’s laws (he specially mentions brachydactyly). But he was under no illusions about the difficulties of his task compared to the plant or animal breeder (table 1, quote 1). By taking this approach, Rüdin is rejecting 2 extensive prior traditions in psychiatric genetics, long antedating the rediscovery of Mendel’s laws: the “interesting pedigree” approach (quote 2) and the study of “heredity burden” (roughly positive versus negative family history).

In the first of many references to the Jewish-German physician/statistician Wilhelm Weinberg,⁹ whose methodological developments play a key role in this work, Rüdin summarizes the problem.

The mistake of biased consideration of heavily burdened families, which to Weinberg’s merit he has emphatically pointed out, must therefore particularly be avoided in the future.²

The problem of how to ascertain and calculate the risk of DP in a representative sample of the siblings of DP probands will preoccupy Rüdin in this work and represents to the best of our knowledge, the full detailed application of these ascertainment corrections for the study of any human trait or disorder. Then, Rüdin sets out his hypothesis:

If we take dementia praecox or the predispositions to it … and we initially assume that it proceeds according to the laws of segregating heredity … the assumption that dementia praecox is a recessive Mendelian disease in the sense of Mendel, is the most probable. Let us for the time being accept this, and further assume that it is a simple characteristic, which only competes with the characteristic “non-dementia praecox afflicted”. Let us further assume that adolescent dementia is not inherited in a gender-linked manner, that although it occurs clinically in the same form, it is not inherited according to different rules in every family, that there is no “change of dominance” … that there are no new causes which have nothing to do with heredity, and that dementia praecox in the hereditary process is not due to other mental disorders, and … there is no inheritance of a generally uniform disposition, a polymorphic inheritance² p. 4.

Rüdin makes 7 assumptions. DP is (1) a clear binary trait or a “unit-character” in a Mendelian sense, (2) transmitted as a simple recessive, (3) lacking modifier genes, (4) with no sex-specific penetrance, (5) no phenocopies, (6) no genetic heterogeneity, and (7) no cases arising from a broad predisposition to psychopathologies such as the “neuropathic constitution” recently examined by Rosanoff and Orr.¹⁰ If these assumptions hold, then:

we would have to find the known simplest ratios realized in the large statistical mass, i.e. no afflicted children should be expected if at least one parent is not only outwardly healthy, but also germ-healthy, ¼ afflicted if both parents are externally healthy, but germ-afflicted, half afflicted, if one parent is afflicted, the other parents only germ afflicted, albeit outwardly healthy, and ultimately all children afflicted when both parents are afflicted² pp. 4-5.

By “germ-healthy” and “germ afflicted,” Rüdin means a normal (“non-DP”) allele homozygote and a heterozygote. He then outlines an ideal theoretical study:

If dementia praecox really is a disease that is inherited recessively according to Mendel and as a simple characteristic … then a large psychiatric empirical material of sibling series, which contain healthy and afflicted in all ratios, especially sibling series in which there was never a disease of dementia praecox, must be obtained² p. 10.

But given the rarity of DP, such a study of unselected sibships from a general population sample is not feasible.

At the moment … I will essentially have to make do with sibling series in which at least one dementia praecox [case] has become manifest, i.e. with material that is definitely not necessarily representative of Mendelian numerical ratios, and even in the case of any size, and avoiding all selection of particularly heavily burdened sibling series in the collection, can never easily be representative² p. 11.

So, he understands that by studying only sibships containing a DP proband, he is missing the at risk sibships that happened by chance to have no ill members. He then makes a critical point:

… Weinberg has worked out new methods which seem to me to be perfectly worthy of being applied liberally to our material. In numerous conversations with … Dr. Weinberg, to whom I am very grateful for the sacrifice of time he has made, I have recognized that his statistical methods … meet the main needs of the psychiatrist … I have come to the conviction that today they belong to the necessary tools of the psychiatrist who deals with questions of inheritance, and since I have heard a lot that suggests that specialist colleagues are still quite ignorant of the fact that the cooperation of knowledgeable statisticians is necessary if good results are to be obtained² p. 11–12.

Rüdin here outlines the closeness of his collaboration with Weinberg. They had “numerous” discussions of the methodological and statistical aspects of this study. In the decade before Rüdin wrote this monograph, Pearson,¹¹ Heron¹² and Goring¹³ applied the then advanced statistical method of the tetrachoric correlation to data on insanity.¹⁴ But the collaboration between Rüdin and Weinberg was the first time that a psychiatrist with training in genetics collaborated with a leading statistical geneticist in the application of Mendelian models. For further details of the importance of Weinberg’s methods to Rüdin, see quote 3. The key papers describing these methods were published by Weinberg in 1912^15–17 so Rüdin was working at the cutting edge of the then young discipline of human Mendelian statistical genetics.

Rüdin spends many pages and tables illustrating Weinberg’s 2 key methods. While a confusing set of names have been applied to these statistical approaches, simple definitions and descriptions will suffice for our purposes. In the sibling method for recessive diseases where sibships are detected through an affected proband (as Rüdin did), Weinberg notes that the correct segregation ratio will only arise if the affected sibling through whom the family was ascertained (the proband), is not counted. Weinberg’s proband method states that to obtain the correct risk figures, if more than one affected sibling per family is ascertained, the family needs to be counted multiple times, once for each proband. Rüdin’s methods are clear:

For a long time, I have been aware of the falsification of the hereditary results, which must result from a biased collected material, and since I have been collecting with more available resources, I have therefore always worked according to principles that are in harmony with those statistically required, justified and substantiated by Weinberg² p. 22.

In the middle of these technical algebraic problems, Rüdin provides this interesting observation which suggests his interests expand considerably beyond finding the correct Mendelian model:

There is no longer any serious doubt concerning heredity in humans, that it follows Mendel’s rules just as it does in the case of other living beings. Far more important for us is the question of the extent to which there are deviations from the classic figures, because the size of this deviation is able to provide an indication of how far external influences besides hereditary factors play a role in the occurrence of a phenomenon, and such a determination is again only possible if we strive to use a method that is as precise as possible and avoids any unjustified concession p. 20.

Rüdin is being rigorous not only to determine the proper segregation ratio to fit the Mendelian models to DP but also because deviations from those expected ratios will inform us critically about other etiological risk factors for DP. To make this more explicit, Rüdin is suggesting that we will learn something quite important if DP conforms to Mendelian expectations but also if it deviates from such expectations.

After this long methodological introduction, Rüdin turns to a description of the ascertainment of his sibling series:

A large material series, which can be regarded as a representative selection, was obtained in such a way that, statistically speaking, randomly, ie, purely based on the certainty of the diagnosis, quite apart from all questions of [genetic] burden, all cases of dementia praecox disease were collected, which were admitted to these institutions since the existence of the psychiatric department of the hospital left of the Isar (1898–1904) and the Munich clinic [beginning in] (autumn 1904). Additionally, those cases which were recorded there in the first years of the existence of the old district mental asylum in Munich were reported to me in a joint consultation by Professor Kraepelin ….Here, too, selections were made purely for reasons of reliable diagnosis, so no attention was paid to burden…² p. 23.

He then describes his diagnostic approach:

My apprehension of dementia praecox as a concept of illness and a complex of symptoms is defined the same as Kraepelin’s. The forms that he recently delimited as paraphrenia were not included in the source material…. The patients were kept track of catamnestically up to the most recent time and cases which had been found to be misdiagnosed or which could not be diagnosed as conclusively due to insufficient observation time, were eliminated from the material² p. 24.

Rüdin only studied probands he considered to have certain DP (even excluding cases of paraphrenia, roughly modern paranoid schizophrenia without substantial personality deterioration¹⁸) and, very importantly, confirmed these diagnoses by their subsequent course of illness.

He then describes his detailed approach to collecting diagnostic information on the siblings.

The technique of collecting the material was generally such that not only the patients themselves, if they could provide information, but also as many of their closest and distant relatives as possible, be it personally or through correspondence, were asked about health as well as also genealogical data. Further sources were used: the registry offices, parish registers, estate files, guardianship files, divorce files, our own and foreign case histories and personal files of the patients, pedigrees and chronicles, criminal files, police files, house files of the prisons, criminal records, military files, accident files, and so on. In this manner, biased reports about the family's state of health were avoided where possible² p. 26.

Each proband had a file for their data and that of their siblings. On the advice of Weinberg, Rüdin established diagnostic cards for each proband for use in subsequent follow-up investigations. His material was divided into 2 parts: Those whose parents were DP-free at the end of the investigation (treated in chapter 2) and the much smaller group with parents with DP (chapter 3).

Chapter 2–Risk in Siblings From DP–Free Parents

Rüdin studied 701 sibships with parents both unaffected with DP containing 4823 individuals including the probands. At the end of 1912, when the fieldwork ended, he had identified, in the siblings, 44 secondary cases of DP, and 79 cases of other psychoses. However, 1506 siblings had died before reaching the age of 17 years. Furthermore, 1717 of these siblings were either still alive or had died in between the ages of 17–40 years. Only 539 had survived past the age of 40 years and were thus likely out of the risk period for DP.

Rüdin will eventually take us through 3 steps to get the correct morbid risk (MR) of DP in these siblings. First, he focuses on the sibling method to examine only secondary cases (ie, cases that are identified through a primary proband). Next, using the proband method, he accounts for the 18 sibships detected through 2 affected probands and the one sibship detected through 3 probands. Third, he will age correct the sample noting that to evaluate a Mendelian ratio for a disorder with a variable age of onset in adulthood, prevalence is not sufficient. Instead, it is necessary to estimate the MR which equals the prevalence if all members of the sample have survived through their age at risk. He uses 2 methods: A lifetable approach and Weinberg’s abridged age correction method. The latter method calculates a “lifetimes at risk” equivalent for all the siblings using the following approach. Those who are under the age of 17 years count zero, those who are aged 17–40 years count ½, and those who are over 40 counts as 1.0. The MR then equals the number of affected individuals divided by the total lifetimes at risk in the sample.

Rüdin next goes through a curious “teaching” exercise. He examined his data the wrong way—purposely breaking the rules that Weinberg established. He counts the affected probands as secondary cases and then miscalculates the lifetimes at risk by only excluding those who died before the age of 17 years, counting all the remaining siblings as having a full lifetime of risk. He finds a resulting MR of 23%, close to the expected Mendelian ratio of 25%. His point is to demonstrate that using incorrect methods, it is not difficult to find simple Mendelian ratios.

Then Rüdin describes his age correction methods for this sibling group. The level of detail he provides is extraordinary, including 10 tables and considerable text. He explores the impact of corrections for the multiply ascertained sibships and the possible effect of increased mortality of those with DP risk.

The first MR he presents derives from a computationally intense life table estimate which equaled 4.48%. He checks that by using 2 more approximate lifetable approaches for which he calculates 4.72% and 4.83%, respectively. He notes that these figures assume no increased mortality of those predisposed to DP. He then applies the abridged Weinberg method with the specific goal of determining “whether a certain, abbreviated procedure, which intends to take account of the insufficiently long observation of a part of the individuals, is able to deliver at least approximately correct results.² P.51.” This method produces an MR of 5.35%. Rüdin also notes the MR for “other psychoses” in these siblings as equal to 4.12% by his “exact” lifetable method.

Rüdin then summarizes these results and explores their implications:

… dementia praecox occurs in only 4.48% of children from dementia praecox-free parents. So only between 1/32 (3.1%) and 1/16 (6.25%) of the children from parents who were free of dementia praecox suffered from dementia praecox, which already suggests that the affliction or predisposition to it is more likely inherited recessively than based on the dominant mode. The assumption is that the trait “dementia praecox” and the trait “dementia praecox–free” form a simple Mendelian pair of traits, where dementia praecox is recessive, dementia praecox–free is dominant and thus 25% = ¼ of children with dementia praecox–free parents should be dementia praecox–afflicted, is therefore invalid. This would have been the simplest. A number of psychiatrists in our first era of psychiatric applications of Mendelian ideas seem to have wanted to adopt this idea. However, it is refuted by the present result. Dementia praecox does not seem to be a monohybrid trait, ie, not a trait that is only caused by one gene or only by a simple heredity factor. Things are obviously more complicated² pp. 51–52.

We would consider this passage as the first of many empirically well-justified statements about our inability to explain the pattern of segregation of major psychiatric illnesses in families by simple Mendelian models. However, he does not let the matter rest there. Rather, he goes on to propose a 2-locus recessive model for DP:

We can thus also assume that in dementia praecox-free parents and who have dementia praecox children, 2 Mendelian trait pairs are active, that their external, actual, or apparent health (ie, their absence of dementia praecox) comes about because they have both the dominant factors, which are perhaps not yet recognizable to us by their external characteristics … but that they also carry within them the 2 recessive factors latently² p. 52.

The expected risk to siblings for such a model equals 6.25%, so rather close to Rüdin’s observations. Rüdin then tabulates the 16 possible parental genotypes for a 2-locus recessive model and notes that those unaffected with DP could be lumped into 4 groups on the basis of whether they have 0, 1, 2, or 3 risk alleles (those with 4 will have DP). These groups would

have the most varied gametic composition, that is, represent the most varied genotypes, even though they also all look the same outwardly with regard to the characteristic “dementia praecox–free”² p. 55.

He continues

The psychiatrist may, therefore, first of all, have the thought … that a three fold mental difference could or should be expected here. Certainly, the idea is obvious and can be fruitful if it is supported by careful, unhurried, and unprejudiced investigation and research² p. 56.

Rüdin is suggesting the possibility of multiple classes of genetic liability to DP based on the number of risk alleles in a 2-locus model. He suspects that psychiatric differences might be found between these risk groups that might reflect what we would now call schizophrenia spectrum phenotypes. He continues to speculate about this possibility, and the difficulties in studying it in quote 4. However, before ending the chapter, he reflects on the accumulating experience in family studies of increased rates of potential personality disorder spectrum cases, whom he terms “cranky psychopaths,” in the siblings of DP probands (quote 5).

Chapter 3–Risk in Siblings From Affected by Unaffected Matings

Rüdin begins this chapter by noting the relative rarity of affected parent-offspring pairs with DP which has, he states, been commented upon by prior investigators. He notes “From the point of view of Mendelian recessive inheritance, this is understandable, in fact, quite self-evident² p. 58.” His results are consistent with this, as he found only 34 DP probands who had a parent with clear DP. This section is brief. He presents one MR estimate for these siblings, using the abridged age correction method, which was 6.18%, moderately higher than was seen in the children of DP-free parents. He also notes that the MR for other psychoses, calculated in the same manner was 10.3%, more than double that seen in the siblings from DP-free parents.

Chapter 4–Risk of DP in Siblings as a Function of Other Parental Disorders in Parents

Rüdin starts this chapter by stating that his previous category of “DP–free” could include a range of conditions in parents that might impact on the risk of DP in offspring (quote 6). He then reviews the rates of DP in the siblings of 2 non-DP forms of psychopathology in their parents: “non-DP mental illness” and alcoholism. He provides a clear definition for the latter (“clear chronic abuse of alcohol with the known consequences, attested by various sources² p. 65”), but not the former. He presents, for a range of parental combinations of these disorders, MR calculations using the abridged age correction method, both for DP and other psychoses, in the siblings from DP-free parents. He presents a detailed table for each combination. We review 3 representative analyses, contrasting each result with those found in all of the siblings.

In 109 sibships (784 siblings) where one of the parents had alcoholism without other mental illness, there was a substantially elevated risk for DP (7.80%) and a more modest increased risk for other psychoses (5.20%) compared to that found in the entire sample: 4.48 and 4.12%, respectively. In the 133 sibships (881 siblings) where one or both parents had a non-DP mental illness without alcoholism, an even greater increased risk was seen for both DP (8.21%) and other psychoses (8.21%). To see what would be found with more concentrated parental psychopathology, Rüdin then examined the 36 sibships (244 siblings) where at least one parent had a non-DP mental illness and one parent had alcoholism. Rates of illness in those siblings were even more strongly elevated for both DP (14.8%) and other psychoses (8.21%). Finally, Rüdin explored risks in 62 sibships (421 siblings) where parents were DP-free but cases of DP were known in other relatives (eg, aunts, uncles, and cousins). They too had a clearly increased risk for DP (8.1%) and other psychoses (6.8%).

From these results, Rüdin concludes:

… the frequency with which dementia praecox occurs in a sibling population does not depend solely on the frequency with which this particular disease occurs in the parents of that sibling population, but as it appears to depend, also on the frequency with which mental illnesses other than dementia praecox, occur in the parents² p. 75.

This is an important observation that is outside the “remit” of a standard Mendelian analysis and points to the fact that risk for DP in offspring is not just a result of the presence or absence of DP in parents. In considering the impact of these parental disorders on risk in offspring, he raises the possibility of nonhereditary sources:

… it is nevertheless possible that in the large average material of the 701 families … other factors than mere hereditary influences are active in the production of dementia praecox. One can consider alcoholism in the parents, briefly of germ intoxication, also of immaturity or “old age” of the germs coming to copulation, of disorders in pregnancy, and finally of individual damage that the individuals have suffered during their extrauterine life² p. 76.

By “germ intoxication,” Rüdin is probably referring to intrauterine effects of alcohol exposure.

Chapter 5–DP in the Half-Siblings of DP Cases

Rüdin approaches the analysis of the half-siblings of his DP probands as potentially providing “another means of answering the important question of whether or not DP is a recessive Mendelian trait² p. 79.” He notes that if DP is a dominant trait, and the common parent carries the risk allele, half-siblings would have a substantially elevated rate of illness. However, if DP is recessive, the risk to half-siblings should be far lower. Rüdin ascertained 498 half-siblings of DP probands from 134 sibships, and so had some power to address this question. The MR for DP and other psychoses were quite low: 0.56 and 1.70, respectively. Rüdin realized that a rigorous test of his Mendelian hypothesis within half-siblings would have to occur when the parent with 2 spouses had DP. He had too few such cases to attempt that analysis. Nonetheless, the scanty available evidence, as with the pattern of results on the entire sample, does not favor dominant transmission for DP.

In chapter 6, Rüdin examined the impact of birth order which we will not here discuss.

Chapter 7–The Mode of Transmission of DP

The early parts of this chapter focus on a further evaluation of the recessive and dominant Mendelian models for DP. Rüdin summarizes his results, concluding that they provide “important reasons to believe that the predisposition to DP is a recessive anomaly² p. 103.” In ruling out a dominant model, he emphasizes the low risk for DP in offspring of DP parents. He then brings in new data, from a sample available to him in Munich—a particularly old group of DP patients who had 81 children largely past their age at risk. With a careful analysis, using an abridged age correction, he calculates the MR for DP in their offspring as equaling 4.76% and for other psychoses 9.52%. He concludes that “The proportions [of affected] are so small that the idea of a dominance of DP can also be dismissed out of hand² p. 104.” Hoffmann, mentored in his study by Rüdin’s, will write a detailed monograph on this question reviewed next in this special issue.¹⁹

He adds 2 further arguments against a dominant genetic model for DP. First,

the fact that continuous inheritance of the disease over 2 generations is extremely rare … speaks against the mechanism of inheritance according to the dominant mode² p. 106.

Second, in a classical dominant model, all affected individuals have affected parents. However, for DP “the most frequent case of DP is that DP-free parents have DP-sick children² p.106”.

In the remainder of this chapter, Rüdin considers other possible Mendelian genetic models for DP indicating that the modes of transmission he considered were not limited to the most standard recessive and dominant models. First, he reviews what he terms “imperfect dominance,” referring to corn genetic studies where the heterozygote appears intermediate in appearance to the 2 homozygote forms. He does not favor this hypothesis.

In the case of dementia praecox, … it happens quite frequently, and indeed in the majority of cases, that not a single descendant of a patient with dementia praecox has the characteristics of dementia praecox² p. 108.

He then considers environmental and other genetic risk factors which might modify the penetrance. He does not favor discrete environmental causes for DP (quote 7) but gives more consideration to modifying genetic factors, suggesting the possibility that “some other hereditary factor could have obscured the disposition (epistasis)² p. 106.” He describes the possibility of “conditional factors … [those] hereditary units that must be present so that one or more factors can exert a certain effect (namely the generation of DP)² p. 109.” He expands on this idea in quote 8.

Finally, he suggests something similar to what we would now call oligogenic inheritance, perhaps requiring risk genes from both mother and father. He considers a model “… which also permits, indeed postulates, more complicated additions than those that occur in cases where only a simple Mendelizing feature pair comes into question.² p. 110–111” and then suggests that his results could “… suggest that DP is a product of apparently complementary pathogenic factors from both the families of origin, the paternal as well as the maternal² …p. 111.”

Chapters 8 and 9–Gender Effects and Anteposition

Nothing of substantial interest from a genetics perspective is provided in these 2 chapters. Rüdin finds no evidence for a major sex effect in the transmission of DP in chapter 8. Chapter 9 deals largely with artifacts that can produce evidence for anteposition —that is younger ages of onset for a disorder in parental versus offspring generations.

Chapter 10–Polymorphic Inheritance

Rüdin begins this chapter as follows:

The present paper is based on the assumption that the question of ratios should initially only be decided by taking into account similar clinical symptoms, namely all those conditions which we include under dementia praecox. A completely different approach would consist in disregarding the clinical differences and … compile all psychoses in the parents and their children, however different they may be, as if they were also clinically similar² p. 139.

He is raising a question traceable to the very beginning of psychiatric genetics in the late 18th century: Is what runs in families of the mentally ill the risk for specific psychiatric syndromes or a broad diathesis to mental illness which can take many clinical forms²⁰? That this remained an active area of debate in Rüdin’s day is demonstrated by the study of Mendelian transmission of mental illness by Rosanoff and Orr published in 1911—which examined a broad neuropathic constitution as their major phenotype^10,21—and a major 1913 article by Jolly on this particular topic.^22–24

Rüdin recognizes this earlier tradition (quote 9) and agrees that “it is absolutely necessary to pursue it in an exact manner² p. 140,” but he cannot do that with his own material designed to look specifically at DP transmission within families. Rüdin suggests that

the question of the inheritance of a uniform disposition to any mental disorder, formulated in this way, not for each individual type of disorder separately, can only be answered through long, arduous research² p. 141.

He outlines a possible approach in quote 10. He comments that the problems for such a project would “lie less in the statistical field of processing [the results] than in that of clinical appraisal.² p. 141.” He reviews, in quote 11, some of the differential diagnoses issues likely to be confronted. This demonstrates yet again how tightly interlinked conceptual/statistical and diagnostic issues are in psychiatric genetics.

He then turns to the one diagnostic category for which he has collected illustrative information, his category of “other psychoses”:

The fact that other psychoses … also occur in the families alongside pronounced dementia praecox, is certain…. The question now arises: Is this confluence or purely coincidence? Or are these other mental disorders perhaps due to other hereditary tendencies present in one or more marrying branches and different from the dementia praecox hereditary tendency? Or are they mainly exogenous, such as cerebral syphilis, alcohol psychoses, and the like, without any noteworthy genetic makeup? Or is this occurrence of the other psychoses and defective states perhaps internally connected with the occurrence of dementia praecox in the manner of the same … segregating hereditary process or an otherwise common cause?² pp. 143–144.

Rüdin suggests 4 hypotheses for his findings with other psychoses: (1) chance, (2) a result of a genetic liability distinct from DP, (3) a set of largely exogenous psychoses, or (4) conditions closely genetically related to DP. Rüdin favors the latter interpretation but states that further research would be needed to address this hypothesis, including more careful phenotypic characterization of the relatives. He also raises the question of how many of these other psychoses might, with more information or longer follow-up periods, be considered cases of DP.

In addition to discussing the excess rate of other psychoses in the relatives of DP probands, he also raises the question of

… the psychopathic personalities, and although they have not yet been definitively counted, they are quite common in these families, who must also be viewed as strongly deviating from the norm regarding nervous qualities, without being identified as … just related to dementia praecox² p. 144.

The nature of the relationship of these personality disorders, which appear in excess in the close relatives of DP, to a genetic liability to DP will be further addressed by papers from Rüdin’s colleagues reviewed in this special issue.

Chapter 11–Clinical Relationships of Mental Disorders in Families

Continuing the theme of chapter 10, Rüdin here focuses on the kinds of psychiatric illness observed in parents of individuals with DP. Rüdin cautions against assuming that, because disorders occur in close relatives, they are inevitably the same underlying condition (quote 12). This particularly applies to parents of DP patients, where he suggests that the majority of psychotic conditions observed are “not DP, at least not one that is readily recognizable as such² p. 147.”

Rüdin presents a variety of enumerations of “psychoses” in the parents of his DP probands. We will present the clearest of these, but the numbers do not map exactly onto earlier findings, in part, because he provided updated diagnoses on some parents “subsequent to the original statistical count² p. 148.” So, he was able to identify 189 proband parents with broadly defined “psychosis” but for 15 of them, information was lacking for any more definitive diagnosis. Of the remaining 174, 59 (34%) meet definite, likely, or suspected DP. Sixty-five parents had non-DP psychosis, with was largely divided between those who are with manic-depressive or organic psychoses, with a smaller proportion of alcoholic psychoses.

Rüdin was sufficiently interested in examining the kinds of psychotic disorders in parents of DP patients that he reports on a second sample “made available by the directors of other institutions at my request² p. 153”. His clearest breakdown of psychosis in parents of DP pro-bands is presented for the sum of his systematic series and these new families. Of the 276 psychotic parents in this group, 31% had DP, 32% other functional psychoses (of which the most common was manic-depression, depression, and hysteria), 22% organic (largely senile, arteriosclerotic, and paretic), 5% alcoholic, and the remaining 10% unknown. Rüdin comments here about the methodological problems of doing detailed clinical studies of DP across generations where good clinical records can often be obtained in the offspring generation but are much harder to come by for the parental generation (quote 13).

He summarizes these findings:

… conclusions can thus only be drawn with the greatest caution from the type of mental disorder in the parents to the type of mental disorder in the children, purely from the genealogical fact of the psychotic illness of one parent. [bold in original]² p. 155.

Chapter 12–Summary

Most of this section repeats the conclusions drawn above. In a few places, Rüdin gives a broader overview. He expresses more clearly than in earlier parts that (1) the frequency of other psychotic disorders in the parents and siblings of his DP probands suggests some kind of genetic relationship with DP (quotes 14 and 15), and (2) the likely role of interacting genetic risk factors in the etiology of DP (quote 16) that take some kind of “recessive” form with risks from both maternal and paternal side (quote 17) and the likelihood of some form of polymorphic inheritance (quote 18).

Discussion

Of the many points of interest in this historically important study, 7 are worthy of emphasis. First, Rüdin’s study of the Mendelian segregation ratio for DP in siblings of DP probands is, to the best of our knowledge, the first study of Mendelian segregation patterns of a psychiatric disorder conducted in collaboration with a statistical geneticist, using, at the time of its composition, cutting edge statistical methods. This kind of collaboration would prove quite fertile over the subsequent history of psychiatric genetics. Rüdin powerfully demonstrated the importance of the correct methodology in this monograph by demonstrating how incorrect application of ascertainment and age correction methods could produce false positive evidence for a single-locus recessive model for DP. Given Rüdin’s later intimate involvement with and support of the Nazi regime, it is both ironic and tragic that his close collaborator, whom he cites frequently, Wilhelm Weinberg, was Jewish.

To put Rüdin’s effort in historical context, we can usefully compare Rüdin’s study to that of 3 major prior attempts. In 1907, Heron tested a recessive model for the transmission of “insanity” by parental mating type.¹² Finding the risk for affected offspring in 2 of the 3 types of families (“both sane,” “one insane,” and “both insane”) differed widely from Mendelian expectations, he declared “no argument in favor of Mendelism seems possible on these figures.¹² p. 17” In 1911, Rosanoff and Orr published a report examining 73 pedigrees ascertained through patients admitted to Kings Park State Hospital with a psychotic disorder, most commonly with manic-depressive insanity or DP.¹⁰ Using clinical assessment methods learned from Davenport and his group at Cold Spring Harbor,²⁵ they explored rates of disorder as a function of presumed parental mating types to determine if ratios were consistent with a single-locus recessive model. They claimed success. However, their phenotype of interest was the “neuropathic constitution,” and most affected individuals were described as nervous, eccentric, or alcoholic. No age or proband corrections were attempted. Finally, in 1913, Jolly published a report focusing largely on the homogeneous versus heterogeneous transmission of forms of insanity in pedigrees.²² But he also comments on his collection of 52 pedigrees with DP cases, 24 of whom contained 2 or more affected members. By visual inspection, with no discussion of ascertainment or age corrections, he concludes that the families are most consistent with “recessive heredity.”²⁴

Second, this report stands out from contemporary work by the level of diagnostic rigor employed. Rüdin’s assessment methods were careful and use multiple sources of information. Not only did he try to interview the individual relatives, but he attempted to contact multiple other relatives as informants, and, if possible, consulted a wide array of official documents including health, police, and military files. Probands were excluded if insufficient diagnostic information was available. He employed a narrow construct of DP as articulated by his mentor E. Kraepelin and utilized follow-up methods to confirm the diagnostic assessment, particularly valuable given the central role of poor outcomes in the Kraepelinian concept of DP.

Third, Rüdin had, for his time period, a relatively broad vision of possible genetic models. When it was clear that his results were inconsistent with a single autosomal fully penetrant locus, he discussed a wide array of other possibilities including a 2-locus recessive model, a model with reduced penetrance resulting either from genetic or environmental effects and an oligogenic model.

Fourth, even though he set his task as understanding the transmission of narrowly defined DP as a Mendelian unit character, he did not dogmatically accept the idea that DP always had to run true within families. Indeed, he explored the possible heterogeneity of genetic transmission from several perspectives. He examined the impact of non-DP diagnoses in parents on risk for DP in siblings, showing substantial empirical evidence that nonDP psychopathology in parents impacted on offspring risk. He also explored the kinds of non-DP pathology seen in parents, noting that a majority of the psychotic disorders observed in them were not DP. He made similar observations in the siblings, noting increased rates of a range of non-DP disorders as well as psychopathic personalities. He suggested that the most likely reasons for this aggregation were that these syndromes had some sort of genetic relationship with DP.

Fifth, he did not restrict attention solely to genetic transmission within families but considered a range of environmental mechanisms. Sixth, he did not dismiss the broad diagnostic models used by Heron and especially Rosanoff, instead discussing how specifically these could be evaluated empirically. Indeed, in his conclusions, he stated this his data could be seen as somewhat supportive of this position.

Seventh, we would suggest that Rüdin was establishing a new paradigm in psychiatric genetic family studies which differed dramatically from the dominant paradigm of the 19^th century based on the assessment of “hereditary burden” as informally recorded in patient charts. This approach had the following main characteristics: (1) it turned the attention away from the proband and a global measure of his or her hereditary risk to a careful study of the proband and individual relatives of the proband, (2) systematic sampling of probands and statistical control of such features as ascertainment and age correction, (3) the acceptance of a specific diagnostic system, here the one proposed by Kraepelin, (4) careful attention to sources of diagnostic information individually applied to the proband and each relative—here siblings—to obtain a high-quality final diagnosis, (5) flexible approach to genetic modeling, and (6) the capacity to explore multiple definitions of the kinds of disorders than run in families.

When viewed from a historical perspective, this first systematic family study of DP had 3 major findings all of which would be replicated multiple times in subsequent investigations: (1) DP ran in families (Rüdin makes little mention of this but it was then known that rates of DP in the population was far below that found in his siblings), (2) DP did not have a simple genetic transmission pattern as expected from Mendel’s laws, and (3) DP appeared likely to be genetically related to non-DP psychotic disorders and perhaps some kinds of psychopathic personalities. We will see how much the 4 subsequent papers from the Rüdin school that we review in this issue further explore these and related themes.

Rüdin and the National Socialists

We cannot conclude this review without commenting on Rüdin’s relationship with the German National Socialists, a topic dealt with extensively by other authors. As we discussed in detail previously,⁸ it is widely agreed that from 1933 to 1945, both within and outside Germany, Rüdin functioned as one of the major German geneticists who supported and gave scientific prestige and credibility to the racial beliefs and policies of the Nazi party. From 1933 onward, he was a central participant in the development and promulgation of the Nazi program of involuntary sterilization of those with “hereditary” conditions which included mental handicap, schizophrenia, manic-depressive illness, and severe alcoholism. Over 300 000 individuals were sterilized under this program.

Rüdin had no direct role in the mass killings of the Holocaust. However, an active historical debate continues about his level of involvement in the Nazi T4 program of the murder of the mentally ill and handicapped. We have read conflicting conclusions from the historical record and here rely on what we considered the balanced conclusions of the American historian Sheila Weiss.²⁶ She writes

… from extensive research on the subject we know that Rüdin does not belong to this category of “victimizers” [active participants in the T4 program], his professional contacts with many of them notwithstanding …²⁶ p. 175.

She also notes that a detailed review of Rudin’s writings prior to 1933 provides no evidence that he supported “the killing of the mentally handicapped,²⁶ p. 176” and we agree. But Weiss also documents that several close colleagues urged Rüdin to protest the killings (including one of our authors, Luxenburger) and he refused “... never taking a public stand against the formal “euthanasia action²⁶ p. 177” although apparently protesting in private. Weiss concludes that his clear knowledge of the T4 program and his inaction, given his prominent position, certainly “implicates him in the crimes associated with the T4 project²⁶ p. 175.” We refer the interested reader to a range of references we supply in the introduction that discusses, from multiple perspectives, the role of Rüdin in Nazi eugenic policies. As overviews, we might suggest starting with these 2 articles^27,28 and then reading Weiss’s book.²⁶

Funding Statement

This supplement is sponsored by Professor Kenneth S. Kendler MD.