verticalSpacer

Institute for Response-Genetics (e.V.)

Prof. Dr. Hans H. Stassen, Chairman

(Formerly Associated Institute of the University of Zurich)

IFRG Logo
IFRG Emblem
Hspace
Trithemius

Towards a Molecular-Genetic Model of Vulnerability

Our studies with monozygotic twins concordant and discordant for schizophrenic or bipolar disorders have made it clear that the genotype is not a sufficient condition for the development of a psychiatric disorder. A single "biogenic deficit" or a combination of "biogenic imbalances" does not necessarily cause psychiatric disorders. Rather, empirical evidence suggests that in many cases an exogenous trigger ultimately initiates the onset of a schizophrenic, bipolar, or depressive disorder. Indeed, susceptible persons (e.g. former patients) can function perfectly well in daily life if they take the necessary precautions. Likewise, it can be assumed that there exists a non-genetic vulnerability as well, built up e.g. through lifestyle, diet, consumption behavior, and lack of exercise, which can lead to psychiatric disorders if stimulated by exogenous triggers. Consequently, the genotype does not constitute a necessary condition either.

Method of Approach

In a first step, we searched the genetic vectors derived from our sample of 1,698 subjects for genotype patterns (100 specifically selected genes, 592 SNPs) that exhibited unique characteristics in the target populations when compared to the healthy control population (n=267). Target populations were «Depressives» (n=596), «Schizophrenics» (n=363), «Bipolars» (n=134), «Schizoaffectives» (n=62), and «Alzheimer's» (n=75). Half of the healthy control subjects (52.8%) were first-degree relatives of the patients of the target populations, so that the contribution of biological ethnicity to discrimination could be estimated. With this knowledge, we were able to construct classifiers in the 2nd step by means of Neural Net models (NNs) that separated the target populations from the control population. A false-positive rate of "0" was part of the optimization criteria.

Unspecific Vulnerability across Clinical Diagnoses

For all target populations we found a rate of some 90% correctly classified patients along with a 10% subgroup labeled as "unknown". The only exception was the group of «Alzheimer's» patients where obviously one or more relevant genes were missing in our study's gene catalog. The classifiers were composed of 6-7 genes: 4 core genes that were common to all classifiers, plus 2-3 accessory genes that depended on the respective target population. The genes were not independent but correlated with each other, in the range between r=0.0333 (SLC6A1/GRIK3) and r=0.3409 (SLC6A1/STAT4). Our results suggest the presence of an unspecific vulnerability among psychiatric patients that is largely independent of clinical diagnosis, while a few specific characteristics enable discrimination between diagnostic groups. It is highly unlikely that the results can be explained entirely by differences in biological ethnicity, as half of the healthy control subjects were first-degree relatives of patients from the target populations. It is equally unlikely that the genotypes involved have a direct causal relationship to psychiatric disorders as genes code for proteins or RNA ("gene products") which may interact in a variety of ways and influence the phenotype only after a longer cascade of intermediate steps. Therefore, we are in all probability dealing with highly sensitive disease markers whose significance and potential application remain to be clarified. To this end, we have translated the complex classifiers into SAS 9.4 macros that can be used routinely.

We found significant differences in genetic diversity between target populations and healthy controls. Yet unexpectedly, the postulated reduction of genetic diversity among «Schizophrenics» (80 % of male patients with a diagnosis of schizophrenic disorders have no offspring) did not reach statistical significance and is the object of further investigations.

vSpacer
Molecular-Genetic Neural Net Analysis: Classification of Patients
Neural Net analyses yielded classifiers that allowed us to distinguish between the target populations and healthy controls at a rate of 90% correctly classified patients, along with a 10% subgroup labeled as "unknown". The only exception was the «Alzheimer's» group where obviously one or more relevant genes were missing in our study's gene catalog.
 
In all probability, the above results suggest highly sensitive disease markers whose significance and potential application remain to be clarified.
Hspacer
[ Mail to Webmaster ] k454910@ifrg.ch