User login
Machine-based learning of genetic and epigenetic characteristics of patients with rheumatoid arthritis could help to predict who is likely to benefit from the biologic drugs adalimumab and etanercept, according to results from a longitudinal, observational cohort study.
In the study, machine learning models created by researchers from Utrecht University in the Netherlands using different parameters predicted true-positive rates for response to adalimumab ranging from 76% to 90% and true-negative rates ranging from 70% to 89%, while for etanercept true-positive rates ranged from about 60% to 80% and true-negative rates ranged from about 82% to 98%.
“These results suggest that we can accurately predict the clinical response before adalimumab and etanercept treatment using molecular signatures-based machine learning models, although the prediction accuracy of these molecular signatures differs between cell types and treatments, underlining the need to study more than one drug, cell type, or epigenetic layers,” first author Weiyang Tao and colleagues wrote in Arthritis & Rheumatology. The ability to predict which tumor necrosis factor inhibitor (TNFi) is the first choice for treatment would be highly beneficial in reducing the time to effective treatment, which has been extensively proven to be a paramount factor for achieving long-sustained disease remission, they noted.
The researchers analyzed gene expression and epigenetic signatures in 80 patients with rheumatoid arthritis prior to treatment with adalimumab or etanercept and then examined patients’ response to treatment at 6 months. They then used that information to build a machine learning model to try to predict treatment response.
Overall, 47.5% of patients were treated with adalimumab, and 52.5% were treated with etanercept. Among the adalimumab group, 53% had a good or moderate response to treatment at 6 months, and among those treated with etanercept, 45% had a good or moderate response.
While there were no differences in baseline clinical parameters between responders and nonresponders, the study found significant genetic and epigenetic differences between patients.
They identified 549 genes that showed significantly different levels of expression between responders and nonresponders treated with adalimumab – in particular, genes involved in DNA and nucleotide binding – and 460 genes that were differentially expressed between etanercept responders and nonresponders, including genes involved in TNF-receptor signaling. However, only 2% of these differentially expressed genes were common in both the adalimumab and etanercept groups, suggesting treatment responses for these two medications have distinct gene signatures.
Looking at DNA methylation, researchers found 16,141 CpG positions – sites of DNA methylation – that were differentially methylated between adalimumab responders and nonresponders, 46% of which were hypermethylated among responders but not nonresponders. In the etanercept group, there were 17,026 differentially methylated sites in responders and nonresponders, 76.3% of which were hypermethylated among responders.
The researchers also noted that among the adalimumab responders, the hypermethylated sites were more likely to be found in the upstream and promoter regions of genes, and on CpG islands.
“Thus, on epigenetic level, we observed a distinct hypermethylation pattern between adalimumab and etanercept responders, suggesting the role of epigenetics in defining response towards adalimumab and to etanercept in PBMCs [peripheral blood mononuclear cells],” the authors wrote.
Given the differences in gene signatures seen in the adalimumab responders and etanercept responders, researchers speculated that different cell types might be involved in the responses to these two treatments. They undertook RNA sequencing on the variety of immune cell types known to be involved in rheumatoid arthritis, which revealed gene-expression differences between adalimumab responders and nonresponders in their CD4+ T cells but not in monocytes. However, the gene-expression differences between etanercept responders and nonresponders were seen in both CD4+ T cells and monocytes.
The study was supported by AbbVie, which manufactures adalimumab, and two authors were supported by the China Scholarship Council and the Netherlands Organization for Scientific Research. No conflicts of interest were declared.
SOURCE: Tao W et al. Arthritis Rheumatol. 2020 Sep 10. doi: 10.1002/art.41516.
Machine-based learning of genetic and epigenetic characteristics of patients with rheumatoid arthritis could help to predict who is likely to benefit from the biologic drugs adalimumab and etanercept, according to results from a longitudinal, observational cohort study.
In the study, machine learning models created by researchers from Utrecht University in the Netherlands using different parameters predicted true-positive rates for response to adalimumab ranging from 76% to 90% and true-negative rates ranging from 70% to 89%, while for etanercept true-positive rates ranged from about 60% to 80% and true-negative rates ranged from about 82% to 98%.
“These results suggest that we can accurately predict the clinical response before adalimumab and etanercept treatment using molecular signatures-based machine learning models, although the prediction accuracy of these molecular signatures differs between cell types and treatments, underlining the need to study more than one drug, cell type, or epigenetic layers,” first author Weiyang Tao and colleagues wrote in Arthritis & Rheumatology. The ability to predict which tumor necrosis factor inhibitor (TNFi) is the first choice for treatment would be highly beneficial in reducing the time to effective treatment, which has been extensively proven to be a paramount factor for achieving long-sustained disease remission, they noted.
The researchers analyzed gene expression and epigenetic signatures in 80 patients with rheumatoid arthritis prior to treatment with adalimumab or etanercept and then examined patients’ response to treatment at 6 months. They then used that information to build a machine learning model to try to predict treatment response.
Overall, 47.5% of patients were treated with adalimumab, and 52.5% were treated with etanercept. Among the adalimumab group, 53% had a good or moderate response to treatment at 6 months, and among those treated with etanercept, 45% had a good or moderate response.
While there were no differences in baseline clinical parameters between responders and nonresponders, the study found significant genetic and epigenetic differences between patients.
They identified 549 genes that showed significantly different levels of expression between responders and nonresponders treated with adalimumab – in particular, genes involved in DNA and nucleotide binding – and 460 genes that were differentially expressed between etanercept responders and nonresponders, including genes involved in TNF-receptor signaling. However, only 2% of these differentially expressed genes were common in both the adalimumab and etanercept groups, suggesting treatment responses for these two medications have distinct gene signatures.
Looking at DNA methylation, researchers found 16,141 CpG positions – sites of DNA methylation – that were differentially methylated between adalimumab responders and nonresponders, 46% of which were hypermethylated among responders but not nonresponders. In the etanercept group, there were 17,026 differentially methylated sites in responders and nonresponders, 76.3% of which were hypermethylated among responders.
The researchers also noted that among the adalimumab responders, the hypermethylated sites were more likely to be found in the upstream and promoter regions of genes, and on CpG islands.
“Thus, on epigenetic level, we observed a distinct hypermethylation pattern between adalimumab and etanercept responders, suggesting the role of epigenetics in defining response towards adalimumab and to etanercept in PBMCs [peripheral blood mononuclear cells],” the authors wrote.
Given the differences in gene signatures seen in the adalimumab responders and etanercept responders, researchers speculated that different cell types might be involved in the responses to these two treatments. They undertook RNA sequencing on the variety of immune cell types known to be involved in rheumatoid arthritis, which revealed gene-expression differences between adalimumab responders and nonresponders in their CD4+ T cells but not in monocytes. However, the gene-expression differences between etanercept responders and nonresponders were seen in both CD4+ T cells and monocytes.
The study was supported by AbbVie, which manufactures adalimumab, and two authors were supported by the China Scholarship Council and the Netherlands Organization for Scientific Research. No conflicts of interest were declared.
SOURCE: Tao W et al. Arthritis Rheumatol. 2020 Sep 10. doi: 10.1002/art.41516.
Machine-based learning of genetic and epigenetic characteristics of patients with rheumatoid arthritis could help to predict who is likely to benefit from the biologic drugs adalimumab and etanercept, according to results from a longitudinal, observational cohort study.
In the study, machine learning models created by researchers from Utrecht University in the Netherlands using different parameters predicted true-positive rates for response to adalimumab ranging from 76% to 90% and true-negative rates ranging from 70% to 89%, while for etanercept true-positive rates ranged from about 60% to 80% and true-negative rates ranged from about 82% to 98%.
“These results suggest that we can accurately predict the clinical response before adalimumab and etanercept treatment using molecular signatures-based machine learning models, although the prediction accuracy of these molecular signatures differs between cell types and treatments, underlining the need to study more than one drug, cell type, or epigenetic layers,” first author Weiyang Tao and colleagues wrote in Arthritis & Rheumatology. The ability to predict which tumor necrosis factor inhibitor (TNFi) is the first choice for treatment would be highly beneficial in reducing the time to effective treatment, which has been extensively proven to be a paramount factor for achieving long-sustained disease remission, they noted.
The researchers analyzed gene expression and epigenetic signatures in 80 patients with rheumatoid arthritis prior to treatment with adalimumab or etanercept and then examined patients’ response to treatment at 6 months. They then used that information to build a machine learning model to try to predict treatment response.
Overall, 47.5% of patients were treated with adalimumab, and 52.5% were treated with etanercept. Among the adalimumab group, 53% had a good or moderate response to treatment at 6 months, and among those treated with etanercept, 45% had a good or moderate response.
While there were no differences in baseline clinical parameters between responders and nonresponders, the study found significant genetic and epigenetic differences between patients.
They identified 549 genes that showed significantly different levels of expression between responders and nonresponders treated with adalimumab – in particular, genes involved in DNA and nucleotide binding – and 460 genes that were differentially expressed between etanercept responders and nonresponders, including genes involved in TNF-receptor signaling. However, only 2% of these differentially expressed genes were common in both the adalimumab and etanercept groups, suggesting treatment responses for these two medications have distinct gene signatures.
Looking at DNA methylation, researchers found 16,141 CpG positions – sites of DNA methylation – that were differentially methylated between adalimumab responders and nonresponders, 46% of which were hypermethylated among responders but not nonresponders. In the etanercept group, there were 17,026 differentially methylated sites in responders and nonresponders, 76.3% of which were hypermethylated among responders.
The researchers also noted that among the adalimumab responders, the hypermethylated sites were more likely to be found in the upstream and promoter regions of genes, and on CpG islands.
“Thus, on epigenetic level, we observed a distinct hypermethylation pattern between adalimumab and etanercept responders, suggesting the role of epigenetics in defining response towards adalimumab and to etanercept in PBMCs [peripheral blood mononuclear cells],” the authors wrote.
Given the differences in gene signatures seen in the adalimumab responders and etanercept responders, researchers speculated that different cell types might be involved in the responses to these two treatments. They undertook RNA sequencing on the variety of immune cell types known to be involved in rheumatoid arthritis, which revealed gene-expression differences between adalimumab responders and nonresponders in their CD4+ T cells but not in monocytes. However, the gene-expression differences between etanercept responders and nonresponders were seen in both CD4+ T cells and monocytes.
The study was supported by AbbVie, which manufactures adalimumab, and two authors were supported by the China Scholarship Council and the Netherlands Organization for Scientific Research. No conflicts of interest were declared.
SOURCE: Tao W et al. Arthritis Rheumatol. 2020 Sep 10. doi: 10.1002/art.41516.
FROM ARTHRITIS & RHEUMATOLOGY