My research centers on identifying genetic and genomic determinants that shape the molecular and clinical heterogeneity of complex inflammatory skin diseases. My lab has been focusing our studies on psoriasis, atopic dermatitis, and prurigo nodularis.
- Genetic architecture of complex diseases and their causal relationships. We applied statistical genetic techniques to identify the shared genetic signals between psoriasis vs type-2 diabetes (Patrick, et al. 2021. J Invest Derm); cardiovascular disorders (Patrick, et al. et al. 2022. Nat Comm); and multiple sclerosis (Patrick, et al. 2022. Annals of Neurology). Our Mendelian randomization analysis also highlighted the causal inference between these comorbidities.
- Precision medicine. My group applies machine-learning approaches to identify biomarkers for translational applications. We have developed a model utilizing genetic variants to provide risk assessment for psoriatic arthritis among psoriatic patients (Patrick, et al. 2018. Nat Comm), achieving over 90% precision among the top cases we predicted to have psoriatic arthritis (PsA). By using genomic data modeling, we also highlighted expression profiles in non-lesional skin of psoriasis can be used to successfully model the drug response of anti-TNF treatment, with AUROC >0.75 (Tsoi et al. 2021. J Allerg Clin Immunol). Our recent work conducted a large-scale retrospective pharmacogenetic study to reveal that eQTL of KLK7 is associated with drug response in psoriasis. Using a functional genomic approach, we further highlighted how KLK7 can modulate the TNF stimulation effect in keratinocytes (Zhang, et al. 2023. Brit J Derm).
- Transcriptomic analysis and pipeline to characterize transcript expression patterns in inflammatory skin tissue. We developed a computational pipeline to identify and characterize expressed lncRNAs in the psoriatic and normal skin biopsies. We identified unique tissue-specific expression and epigenetic patterns for the identified novel lncRNAs (Tsoi, et al. 2015. Genome Biol). By applying advanced machine-learning techniques to 834 RNA-seq samples from different in vivo and in vitro experiments, we predicted lncRNAs for specific cytokine pathways. Our approach has high accuracy (AUROC of 0.79); and by using a unique combination of statistical analyses, we revealed new insights into the impact of lncRNAs in cutaneous diseases (Patrick, et al. 2023. JCI Insight).
Please describe one or two of your most interesting projects.
* Causal Relationship and Shared Genetic Loci between Psoriasis and Type 2 Diabetes through Trans-Disease Meta-Analysis. J Invest Dermatol.141(6): 1493-1502, 06/2021.
This study investigates the genetic overlap and causal relationships between psoriasis and type 2 diabetes (T2D). Using trans-disease meta-analysis (TDMA), we identified shared genetic loci that suggest a potential common pathophysiological pathway between the two diseases. This pioneer work highlights a clear genetic link between T2D and psoriasis independent of confounding factors, such as BMI. Applying robust statistical techniques for TDMA to data from over one million individuals, we identified four shared genetic signals, two of which are new findings for each disease, and we then evaluated them in a hospital-based dataset (42,112 individuals). Our study confirmed the significance of these loci using five different TDMA techniques, in addition to rigorous selection criteria. These findings suggest potential roles of NFκB mediator TRAF6 as a hub protein connecting the 4 identified loci. The findings of this work can serve as a starting point for developing new, more effective medicine for T2D and psoriasis, as well as other immune-mediated diseases. Moreover, we applied multi-variable Mendelian randomization techniques to demonstrate psoriasis has a significant causal effect on T2D, independent of BMI and other confounders.
* Retrospective pharmacogenetic study of psoriasis highlights the role of KLK7 in tumour necrosis factor signalling. Br J Dermatol.190(1): 70-79, 12/2023.
Treatment responses for psoriasis can vary, and it is important to understand factors associated with this variation. Linking to our Michigan psoriasis genetic cohort, we have collected the drug responses of six different treatment options, enabling us to conduct retrospective genome-wide pharmacogenetic studies. We associated genetic markers with self-evaluated treatment responses from six different options using linear regression. We further utilized an integrative approach incorporating epigenomic, transcriptomic, and a longitudinal clinical cohort to provide biological implications for the top signals associated with treatment response. We identified two novel genetic markers, one associated with anti-TNF biologics, and the other associated with methotrexate. These two markers are also associated with cutaneous mRNA expression levels of two known psoriasis-related genes: KLK7 and CD200. We further confirmed that the expression level of KLK7 is increased in the psoriatic epidermis and responses to anti-TNF treatment. We also found that keratinocytes have decrease in pro-inflammatory responses to TNF, by inhibiting the expression of KLK7. These findings highlight the role of KLK7 in modulating TNF signaling in keratinocytes, offer valuable insights for personalizing treatment in psoriasis patients, and present potential targets for loci and genes for future psoriasis drug development.