Distinct genetic liability profiles define clinically relevant patient strata across common diseases
Lucia Trastulla, Georgii Dolgalev, Sylvain Moser, Laura T. Jiménez-Barrón, Till F. M. Andlauer, Moritz von Scheidt, Douglas M. Ruderfer, Stephan Ripke, Andrew McQuillin, Eli A. Stahl, Enrico Domenici, Rolf Adolfsson, Ingrid Agartz, Esben Agerbo, Margot Albus, Madeline Alexander, Farooq Amin, Silviu‐Alin Bacanu, Martin Begemann, Richard A. Belliveau, Judit Bene, Sarah E. Bergen, Elizabeth Bevilacqua, Tim B. Bigdeli, Donald W. Black, Douglas Blackwood, Anders D. Børglum, Elvira Bramon, Richard Bruggeman, Nancy G. Buccola, Randy L. Buckner, Brendan Bulik‐Sullivan, Joseph D. Buxbaum, William Byerley, Wiepke Cahn, Guiqing Cai, Dominique Campion, Rita M. Cantor, Vaughan J. Carr, Noa Carrera, Stanley V. Catts, Kimberley D. Chambert, Raymond C. K. Chan, Eric Chen, Ronald Y.l. Chen, Wei Cheng, Eric F.c. Cheung, Siow Ann Chong, Sven Cichon, C. Robert Cloninger, David Cohen, Nadine Cohen, David Collier, Paul Cormican, Nicholas Craddock, James J. Crowley, Mark J. Daly, Ariel Darvasi, Michael H. Davidson, Kenneth L. Davis, Franziska Degenhardt, Jurgen Del‐Favero, Ditte Demontis, Dimitris Dikeos, Timothy G. Dinan, Srdjan Djurovic, Gary Donohoe, Elodie Drapeau, Jubao Duan, Frank Dudbridge, Hannelore Ehrenreich, Peter Eichhammer, Johan G. Eriksson, Valentina Escott‐Price, Tōnu Esko, Laurent Essioux, Kai-How Farh, Martilias S. Farrell, Josef Frank, Lude Franke, Robert Freedman, Nelson B. Freimer, Joseph I. Friedman, Menachem Fromer, Pablo V. Gejman, Giulio Genovese, Lyudmila Georgieva, Ina Giegling, Michael Gill, Paola Giusti‐Rodríguez, Stephanie Godard, Jacqueline I. Goldstein, Srihari Gopal, Jacob Gratten, Hugh Gurling, Lieuwe de Haan, Christian Hammer, Marian L. Hamshere, Thomas Folkmann HansenNature Communications2024 · DOI
Stratified medicine holds great promise to tailor treatment to the needs of individual patients. While genetics holds great potential to aid patient stratification, it remains a major challenge to operationalize complex genetic risk factor profiles to deconstruct clinical heterogeneity. Contemporary approaches to this problem rely on polygenic risk scores (PRS), which provide only limited clinical utility and lack a clear biological foundation. To overcome these limitations, we develop the CASTom-iGEx approach to stratify individuals based on the aggregated impact of their genetic risk factor profiles on tissue specific gene expression levels. The paradigmatic application of this approach to coronary artery disease or schizophrenia patient cohorts identified diverse strata or biotypes. These biotypes are characterized by distinct endophenotype profiles as well as clinical parameters and are fundamentally distinct from PRS based groupings. In stark contrast to the latter, the CASTom-iGEx strategy discovers biologically meaningful and clinically actionable patient subgroups, where complex genetic liabilities are not randomly distributed across individuals but rather converge onto distinct disease relevant biological processes. These results support the notion of different patient biotypes characterized by partially distinct pathomechanisms. Thus, the universally applicable approach presented here has the potential to constitute an important component of future personalized medicine paradigms.
Introduction: The Challenge of Clinical Heterogeneity
Common diseases such as coronary artery disease (CAD) and schizophrenia (SCZ) are highly complex, arising from the cumulative effect of thousands of genetic variants. Currently, the Polygenic Risk Score (PRS) is the standard metric for quantifying this risk. However, PRS acts as a "black box," providing a single risk estimate without revealing the underlying biological drivers. Two patients with identical high PRS values may suffer from the same disease due to entirely different pathomechanisms—one driven by metabolic dysfunction and another by chronic inflammation. This clinical heterogeneity limits the ability of clinicians to tailor treatments effectively.
The CASTom-iGEx Framework
To address these limitations, researchers developed CASTom-iGEx (Cluster Analysis of Stratified Transcriptome-imputed Gene Expression). This framework moves beyond simple risk summation to model how genetic variations impact tissue-specific gene expression.
The workflow utilizes PriLer (Prior Learned elastic-net regression), a novel method that predicts gene expression by prioritizing genetic variants located in functional regions like cell-type-specific chromatin. By imputing expression across 34 tissues, the framework generates "T-scores" that quantify how an individual's predicted expression deviates from a healthy baseline. Finally, patients are grouped into "biotypes" using the Leiden clustering algorithm, with genes weighted by their known disease associations.
Discovery of Distinct Biological Biotypes
Applying CASTom-iGEx to large-scale datasets, including the UK Biobank and the CARDIoGRAM consortium, revealed that genetic liabilities are not randomly distributed. Instead, they converge onto specific biological processes within distinct patient subgroups.
UMAP visualization of patient biotypes identified by CASTom-iGEx, showing five distinct clusters based on genetic impact on tissue-specific gene expression.
In CAD, five distinct biotypes were identified. For instance, Group 2 was characterized by severe clinical outcomes and biological deficits in Golgi vesicle biogenesis and endocytosis. In contrast, Group 5 represented a "healthier" stratum with lower overall genetic liability. Similarly, in SCZ, four biotypes emerged, with Group 1 linked to cognitive impairment and Group 2 showing a high predisposition for metabolic syndrome.
Independence from Polygenic Risk Scores
A critical finding of the study is that these biological biotypes are fundamentally distinct from PRS-based groupings. Traditional PRS quartiles, which rank individuals from low to high risk, failed to capture the qualitative differences between biotypes.
Heatmap demonstrating the independence of CASTom-iGEx biotypes from traditional PRS quartiles, highlighting that PRS cannot resolve underlying biological pathomechanisms.
As shown in the analysis, individuals with high PRS are distributed across all biological biotypes. This indicates that PRS is "blind" to the specific pathomechanisms driving a patient's disease. CASTom-iGEx, however, explained 2 to 5 times more variance in clinical markers like LDL-cholesterol than PRS, demonstrating its superior ability to resolve clinically relevant strata.
Conclusion and Future Directions
The CASTom-iGEx approach provides a biologically grounded method for patient stratification, offering a path toward truly personalized medicine. By identifying biotypes with unique biological drivers, clinicians can move toward more targeted interventions.
Future work must address the current ancestry bias in genomic models, as performance currently drops in non-European populations. Additionally, integrating environmental factors and real-time clinical data into this genetic framework will be essential for a comprehensive multi-modal approach to precision healthcare.
Summary updated
Abstract
Stratified medicine holds great promise to tailor treatment to the needs of individual patients. While genetics holds great potential to aid patient stratification, it remains a major challenge to operationalize complex genetic risk factor profiles to deconstruct clinical heterogeneity. Contemporary approaches to this problem rely on polygenic risk scores (PRS), which provide only limited clinical utility and lack a clear biological foundation. To overcome these limitations, we develop the CASTom-iGEx approach to stratify individuals based on the aggregated impact of their genetic risk factor profiles on tissue specific gene expression levels. The paradigmatic application of this approach to coronary artery disease or schizophrenia patient cohorts identified diverse strata or biotypes. These biotypes are characterized by distinct endophenotype profiles as well as clinical parameters and are fundamentally distinct from PRS based groupings. In stark contrast to the latter, the CASTom-iGEx strategy discovers biologically meaningful and clinically actionable patient subgroups, where complex genetic liabilities are not randomly distributed across individuals but rather converge onto distinct disease relevant biological processes. These results support the notion of different patient biotypes characterized by partially distinct pathomechanisms. Thus, the universally applicable approach presented here has the potential to constitute an important component of future personalized medicine paradigms.
BibTeX
@inproceedings{trastulla2024distinct,
author = {Lucia Trastulla and Georgii Dolgalev and Sylvain Moser and Laura T. Jiménez-Barrón and Till F. M. Andlauer and Moritz von Scheidt and Douglas M. Ruderfer and Stephan Ripke and Andrew McQuillin and Eli A. Stahl and Enrico Domenici and Rolf Adolfsson and Ingrid Agartz and Esben Agerbo and Margot Albus and Madeline Alexander and Farooq Amin and Silviu‐Alin Bacanu and Martin Begemann and Richard A. Belliveau and Judit Bene and Sarah E. Bergen and Elizabeth Bevilacqua and Tim B. Bigdeli and Donald W. Black and Douglas Blackwood and Anders D. Børglum and Elvira Bramon and Richard Bruggeman and Nancy G. Buccola and Randy L. Buckner and Brendan Bulik‐Sullivan and Joseph D. Buxbaum and William Byerley and Wiepke Cahn and Guiqing Cai and Dominique Campion and Rita M. Cantor and Vaughan J. Carr and Noa Carrera and Stanley V. Catts and Kimberley D. Chambert and Raymond C. K. Chan and Eric Chen and Ronald Y.L. Chen and Wei Cheng and Eric F.C. Cheung and Siow Ann Chong and Sven Cichon and C. Robert Cloninger and David Cohen and Nadine Cohen and David Collier and Paul Cormican and Nicholas Craddock and James J. Crowley and Mark J. Daly and Ariel Darvasi and Michael H. Davidson and Kenneth L. Davis and Franziska Degenhardt and Jurgen Del‐Favero and Ditte Demontis and Dimitris Dikeos and Timothy G. Dinan and Srdjan Djurovic and Gary Donohoe and Elodie Drapeau and Jubao Duan and Frank Dudbridge and Hannelore Ehrenreich and Peter Eichhammer and Johan G. Eriksson and Valentina Escott‐Price and Tōnu Esko and Laurent Essioux and Kai-How Farh and Martilias S. Farrell and Josef Frank and Lude Franke and Robert Freedman and Nelson B. Freimer and Joseph I. Friedman and Menachem Fromer and Pablo V. Gejman and Giulio Genovese and Lyudmila Georgieva and Ina Giegling and Michael Gill and Paola Giusti‐Rodríguez and Stephanie Godard and Jacqueline I. Goldstein and Srihari Gopal and Jacob Gratten and Hugh Gurling and Lieuwe de Haan and Christian Hammer and Marian L. Hamshere and Thomas Folkmann Hansen and Thomas Folkmann Hansen},
title = {{Distinct genetic liability profiles define clinically relevant patient strata across common diseases}},
booktitle = {Nature Communications},
year = {2024},
doi = {10.1038/s41467-024-49338-2},
}