TL;DR
This paper introduces "SexBiasedDrug," a novel in silico model that predicts how drugs might affect men and women differently, focusing on cardiovascular health. By analyzing gene expression data and drug response signatures, the model successfully identified drugs with sex-specific effects, which were then validated in animal models and human clinical data, paving the way for more personalized medicine.
What Problem Does This Paper Address?
Despite significant physiological and pathophysiological differences between sexes, medical research and drug development have historically treated men and women as biologically similar. This has led to a lack of understanding regarding sex-specific drug responses, potentially resulting in suboptimal treatment efficacy and increased adverse effects. The paper aims to bridge this gap by developing a computational tool to predict these sex-specific drug effects, particularly in the context of cardiovascular diseases, which are a major cause of morbidity and mortality.
Methods
The core of the study is the development of the "SexBiasedDrug" computational pipeline. This pipeline operates in several stages:
- Sex-Biased Gene Identification: The model first analyzes gene expression profiles from healthy human cardiac tissues to identify genes that are expressed at significantly different levels between males and females. These are categorized as male-biased genes (MGs) and female-biased genes (FGs).
- Drug-Gene Association: It then leverages the Connectivity Map database, which contains gene expression signatures for thousands of drugs. For each drug, the model identifies genes that are significantly upregulated or downregulated upon treatment.
- Sex-Specific Effect Prediction: By comparing the sex-biased genes (MGs and FGs) with the drug-regulated genes, the pipeline uses Fisher's exact test to identify drugs that show a significant association with sex-biased gene expression patterns. This association indicates a potential for sex-specific drug effects.
- Validation: The predictions are then validated through experimental studies in animal models (rats and mice) and retrospective analysis of human clinical data.
Datasets
- GEO database (GSE57345-GPL11532): Used for obtaining normalized human cardiac gene expression profiles from healthy and failing hearts. Specifically, non-failing (normal control) heart samples were used to identify sex-biased genes.
- Connectivity Map database: This database, containing gene expression profiles for over 1000 drugs, was used to identify drug-induced gene expression signatures.
- Human Clinical Cohorts:
- A diabetes cohort from Luhe Hospital, Beijing (35 patients: 9 men, 26 women) treated with metoprolol.
- A health management cohort from Jidong community, Tangshan (153 patients: 86 men, 67 women) treated with metoprolol.
Key Insights and Ideas
- Sex Differences are Pervasive and Clinically Relevant: The study strongly emphasizes that biological sex is a critical factor influencing physiology, disease progression, and drug response, necessitating a move towards sex-specific medical approaches.
- In Silico Power for Drug Discovery: The paper demonstrates the significant potential of computational approaches, leveraging large-scale genomic datasets, to predict complex biological phenomena like sex-specific drug effects, which are difficult and costly to uncover through traditional experimental methods alone.
- Cardiovascular Focus is Just the Beginning: While the study focuses on cardiovascular effects due to their high impact, the developed "SexBiasedDrug" model is presented as a framework that can be extended to other therapeutic areas.
- Metoprolol's Sex-Specific Effect: The clinical validation with metoprolol provides compelling evidence that even drugs with similar chemical structures can exhibit differential efficacy between sexes, highlighting the importance of such predictions.
Key Results
- The "SexBiasedDrug" model identified sex-specific effects for several drugs.
- Acebutolol: In spontaneously hypertensive rats (SHRs), acebutolol significantly lowered heart rate, systolic blood pressure (SBP), and diastolic blood pressure (DBP) more effectively in females than in males.
- Tacrine: In mice, tacrine significantly decreased heart rate in males but not females. It also induced higher plasma cardiac troponin I (cTNI) levels and greater cardiomyocyte superoxide generation in females, indicating stronger cardiac toxicity in females.
- Metformin: As a negative control, metformin did not show significant sex-specific effects on heart rate or superoxide generation, validating the model's ability to distinguish between sex-specific and non-sex-specific drugs.
- Metoprolol (Clinical Validation): In two independent Chinese cohorts, patients receiving metoprolol showed a significantly greater decrease in diastolic blood pressure in females compared to males.
Limitations and Future Work
- Limited Sex-Specific Genomic Data: The authors acknowledge that the scarcity of sex-specific genomic and transcriptomic data, particularly for first-line drugs, limits the model's ability to predict responses for a wider range of medications.
- "Gene Signature" vs. Direct Targets: The model relies on gene expression signatures, which represent a "molecular phenotype" rather than directly identifying drug targets. This means it doesn't account for direct protein interactions or genetic variations in drug target DNA regions.
- Future Directions: The authors suggest that integrating GWAS (Genome-Wide Association Studies) data and eQTL (expression Quantitative Trait Loci) data could further improve the prediction accuracy of the model. They also propose that the model can be extended to predict sex-specific responses for drugs in other therapeutic areas.
Summary updated