Abstract
Systemic lupus erythematosus (SLE) is a heterogeneous disease with unpredictable patterns of activity. Patients with similar activity levels may have different prognosis and molecular abnormalities. In this study, we aimed to measure the main differences in drug-induced gene expression signatures across SLE patients and to evaluate the potential for clinical data to build a machine learning classifier able to predict the SLE subset for individual patients. SLE transcriptomic data from two cohorts were compared with drug-induced gene signatures from the CLUE database to compute a connectivity score that reflects the capability of a drug to revert the patient signatures. Patient stratification based on drug connectivity scores revealed robust clusters of SLE patients identical to the clusters previously obtained through longitudinal gene expression data, implying that differential treatment depends on the cluster to which patients belongs. The best drug candidates found, mTOR inhibitors or those reducing oxidative stress, showed stronger cluster specificity. We report that drug patterns for reverting disease gene expression follow the cell-specificity of the disease clusters. We used 2 cohorts to train and test a logistic regression model that we employed to classify patients from 3 independent cohorts into the SLE subsets and provide a clinically useful model to predict subset assignment and drug efficacy.
Highlights
The heterogeneity of SLE is reflected in the variability in drug responsiveness between different patients, which results in a large proportion of patients showing no or partial response to therapies and where this trial-and-error approach clearly requires improvement[5]
We reported that previously established SLE subgroups[4], notably, the neutrophil and lymphocyte-driven subgroups, showed differences in drug-induced gene expression connectivity scores
Based on the heterogeneity of the SLE directed by neutrophils or lymphocytes, we should ideally consider two concepts when choosing a certain drug for a patient
Summary
The heterogeneity of SLE is reflected in the variability in drug responsiveness between different patients, which results in a large proportion of patients showing no or partial response to therapies and where this trial-and-error approach clearly requires improvement[5]. If clustering signifies a difference that may guide therapy selection, a classification model would be required for the physician to be able to assign a patient to the appropriate cluster or group of patients. Most importantly, such classifiers should be easy to use with routine laboratory and clinical information. The principal aim of this work was to investigate if the subtypes of SLE previously described showed differences in their association to drug-induced gene expression signatures analyzing the similarity scores among drug and subtype-based signatures. The model provides a data-driven approach that migh help clinicians for subgroup SLE patients
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
