# Genetic predisposition to unwanted side effects under antidepressants and antipsychotics: a molecular-genetic study of 902 patients over 6 weeks

## TL;DR

This study explored how genetic factors influence side effects in patients taking multiple antidepressants or antipsychotics, finding that genetics play a more complex role than previously thought.

## Contribution

The study introduces a novel approach using multidimensional gene vectors and AI to analyze genetic diversity in relation to side effect clusters under polypharmacy.

## Key findings

- 61.9–68.1% of patients reported moderate to severe side effects, with 52.1% experiencing significant weight gain.
- Standard AI and neural networks failed to explain a clinically relevant proportion of phenotypic variance in side effects.
- Genetic factors appear to contribute to side effect variability in complex, previously unrecognized ways.

## Abstract

This project aimed at (1) detailing the complex side effect patterns of 902 inpatients treated for major depression or schizophrenia under polypharmacy regimens; (2) developing a quantitative side effect model that accounts for the various facets of clinically observable adverse events; and (3) analyzing irregularities in genetic diversity through multidimensional “gene vectors” in order to reveal possible interrelations with side effect clusters. The patients’ acute medication, their time course of recovery, and their side effects were assessed by up to 8 repeated measurements. The genotyping included 100 candidate genes with genotypic patterns computed from 549 Single Nucleotide Polymorphisms (SNPs). Between 61.9% and 68.1% of study patients reported moderate to severe side effects, while response rates were with 29.5–35.7% quite modest. Half of the patients (52.1%) experienced weight gains of ≥ 2 kg. On the phenotype level, up to 30% of the observed variance could be “explained” by regression models with the dominating factor “number of concurrent drugs”. On the genotype level, we relied on standard Artificial Intelligence (AI) procedures along with multilayer Neural Nets (NNs) to search for combinations of multidimensional genotypic patterns that were characteristic of patients with severe side effects, while being rare (< 10%) among patients without side effects. These analyses failed to explain a clinically relevant proportion of the observed phenotypic variance. The 14 cytochromes analyzed were found to play no more than a minor role. While type and severity of side effects under polypharmacy were primarily determined by the overall medication “load”, the actually observed side effect patterns varied considerably between patients receiving the same medication “load”, thus stressing the role of genetics. Our results suggested that the role of genetics in the development of severe side effects under polypharmacy is by far more complex than previously assumed, related to a completely different set of genes, or that there exists genotypic heterogeneity such that multiple pathways on the genotype level lead to the same clinical picture on the phenotype level.

The online version contains supplementary material available at 10.1007/s00406-025-02062-4.