MICA: Development and Validation of a Transcriptomic-Based Model for Classifying and Predicting Treatment Response in Rheumatoid Arthritis (TRACT-RA)
Lead Research Organisation:
Queen Mary University of London
Department Name: William Harvey Research Institute
Abstract
Rheumatoid arthritis (RA) is the most common form of inflammatory arthritis affecting 1% of the population worldwide and around 500,000 people in the UK. It affects all ages and is characterised by inflammation of the joint lining (synovium) and destruction of cartilage and bone. In approximately 40% of patients current medications are ineffective leading to uncontrolled disease progression, loss of productivity and disability, causing patients suffering and costing ~£4.8billion/year in the UK.
RA is clinically a very diverse disease and at the time patients are diagnosed we can't predict what the likely outcome of the disease will be, or why in some patients treatments are effective whereas in others they are not. This lack of predictive markers leads to: delays in disease control; unnecessary exposure to potentially toxic drugs; large waste of valuable NHS/societal resources.
There are several ways of classifying patients using observable characteristics (known as the phenotype) and/or defining their detailed functional or molecular make-up (known as the endotype). We have a large, unique collection of samples (synovial tissue and blood) from over 750 patients at three distinctive stages of the disease and treated with different drugs: patients have early disease, patient with more advanced disease and patients with "resistant RA" in whom current medication are not effective.
The aim of this research is to develop a method to classify patients using a coding product of the DNA called RNA and this leads to the production of proteins the building blocks of our bodies. Looking at the RNA, known as the transcriptome, indicates which genes are active and in RA helps to identify those that cause inflammation and joint damage. We have already analysed (sequenced) the RNA in the disease tissue in our large collections and in the patients with early disease we have identified different patterns (referred to as signatures) associated with different sub-types of disease and how it progresses.
Our objectives are:
1) To find out whether the signatures we have found in early arthritis are also present at different disease stages i.e. in the established and the resistant RA patient groups, and whether these signatures are modified by treatment. Namely, whether they are conserved, so that we would only need to classify a patient once or whether they change over time and repeating testing is required.
2) To evaluate whether matching the way a medicine works to the signature identified in the patients tissue means that treatment is more effective than when they are not matched, so that we can work out which treatments work best for individual patients.
3) The sequencing of RNA to identify the signatures is complicated, expensive and time consuming so we will work with a company (NanoString) which has developed methods for doing similar tests in other diseases that can be used routinely in hospitals.
4) If we find important signatures we will look to see whether we can also identify these in blood samples which are easier to take than a biopsy.
5) We will apply mathematical modelling using the signature data and clinical information with the aim of developing a process to improve the care patients receive.
If we can identify patients who are more likely to develop aggressive forms of the disease this could be highlighted to their clinical team so that they are followed more closely and quickly offered new treatments to stop the disease deteriorating. Also, being able to predict the treatment likely to work best for a patient may save many years of 'trial and error' until the right treatment is found, this has the potential to greatly improve the quality of life of RA patients, prevent disability and save money to the NHS and society.
RA is clinically a very diverse disease and at the time patients are diagnosed we can't predict what the likely outcome of the disease will be, or why in some patients treatments are effective whereas in others they are not. This lack of predictive markers leads to: delays in disease control; unnecessary exposure to potentially toxic drugs; large waste of valuable NHS/societal resources.
There are several ways of classifying patients using observable characteristics (known as the phenotype) and/or defining their detailed functional or molecular make-up (known as the endotype). We have a large, unique collection of samples (synovial tissue and blood) from over 750 patients at three distinctive stages of the disease and treated with different drugs: patients have early disease, patient with more advanced disease and patients with "resistant RA" in whom current medication are not effective.
The aim of this research is to develop a method to classify patients using a coding product of the DNA called RNA and this leads to the production of proteins the building blocks of our bodies. Looking at the RNA, known as the transcriptome, indicates which genes are active and in RA helps to identify those that cause inflammation and joint damage. We have already analysed (sequenced) the RNA in the disease tissue in our large collections and in the patients with early disease we have identified different patterns (referred to as signatures) associated with different sub-types of disease and how it progresses.
Our objectives are:
1) To find out whether the signatures we have found in early arthritis are also present at different disease stages i.e. in the established and the resistant RA patient groups, and whether these signatures are modified by treatment. Namely, whether they are conserved, so that we would only need to classify a patient once or whether they change over time and repeating testing is required.
2) To evaluate whether matching the way a medicine works to the signature identified in the patients tissue means that treatment is more effective than when they are not matched, so that we can work out which treatments work best for individual patients.
3) The sequencing of RNA to identify the signatures is complicated, expensive and time consuming so we will work with a company (NanoString) which has developed methods for doing similar tests in other diseases that can be used routinely in hospitals.
4) If we find important signatures we will look to see whether we can also identify these in blood samples which are easier to take than a biopsy.
5) We will apply mathematical modelling using the signature data and clinical information with the aim of developing a process to improve the care patients receive.
If we can identify patients who are more likely to develop aggressive forms of the disease this could be highlighted to their clinical team so that they are followed more closely and quickly offered new treatments to stop the disease deteriorating. Also, being able to predict the treatment likely to work best for a patient may save many years of 'trial and error' until the right treatment is found, this has the potential to greatly improve the quality of life of RA patients, prevent disability and save money to the NHS and society.
Technical Summary
Rheumatoid Arthritis (RA) represents a huge public health problem. It is one of the most important chronic inflammatory disorders, leading to increased morbidity/mortality and a heavy health-economic burden. Although current medications have revolutionized the treatment of RA, they are not effective in ~40% of patients, the mechanisms of no-response remain unknown and there are no current predictors of treatment response. With MRC/Versus Arthritis support, by RNA-seq/NanoString analyses we have identified predictive signatures of prognosis and treatment response in the disease tissue (synovium) of early arthritis patients (PEAC) (Ref:6,7,8 in CfS) and delivered the first two biopsy-driven RCTs (STRAP and R4RA) worldwide.
The aim of the proposed project is to develop and validate a transcriptomic-based model building on the PEAC signatures to be evaluated in STRAP & R4RA.
The main hypothesis to be tested is that specific signatures in the synovium define disease and treatment-response subtypes (endotypes) that can predict prognosis and response to current targeted biologic therapies.
We will test this hypothesis by applying robust analytical pipelines to our unique RNA-seq datasets developed by our team (Ref:6,14 in CfS). Given that RNA-seq is not practical or economical for translation into clinical practice, we will evaluate the portability of synovial RNA-seq signatures into NanoString panels also at blood level in collaboration with two of their most senior scientists.
The development, validation and replication of our transcriptomic-based model using NanoString technologies, represent important steps toward future development of a clinical assay with potential major impact on patient care.
This may enable a new molecular definition of RA and the definition of drug target modules in the disease tissue as predictors of response, which may transform current management of RA from "trial and error" drug-prescribing to rational therapy allocation.
The aim of the proposed project is to develop and validate a transcriptomic-based model building on the PEAC signatures to be evaluated in STRAP & R4RA.
The main hypothesis to be tested is that specific signatures in the synovium define disease and treatment-response subtypes (endotypes) that can predict prognosis and response to current targeted biologic therapies.
We will test this hypothesis by applying robust analytical pipelines to our unique RNA-seq datasets developed by our team (Ref:6,14 in CfS). Given that RNA-seq is not practical or economical for translation into clinical practice, we will evaluate the portability of synovial RNA-seq signatures into NanoString panels also at blood level in collaboration with two of their most senior scientists.
The development, validation and replication of our transcriptomic-based model using NanoString technologies, represent important steps toward future development of a clinical assay with potential major impact on patient care.
This may enable a new molecular definition of RA and the definition of drug target modules in the disease tissue as predictors of response, which may transform current management of RA from "trial and error" drug-prescribing to rational therapy allocation.
Organisations
Publications
Boutet M
(2023)
Comparative analysis of late-stage rheumatoid arthritis and osteoarthritis reveals shared histopathological features
in Osteoarthritis and Cartilage
Sciacca E
(2023)
DEGGs: an R package with shiny app for the identification of differentially expressed gene-gene interactions in high-throughput sequencing data.
in Bioinformatics (Oxford, England)
Goldmann K
(2023)
Expression quantitative trait loci analysis in rheumatoid arthritis identifies tissue specific variants associated with severity and outcome
in Annals of the Rheumatic Diseases
Lewis MJ
(2023)
nestedcv: an R package for fast implementation of nested cross-validation with embedded feature selection designed for transcriptomics and high-dimensional data.
in Bioinformatics advances
Sciacca E
(2022)
Network analysis of synovial RNA sequencing identifies gene-gene interactions predictive of response in rheumatoid arthritis.
in Arthritis research & therapy
Lewis MJ
(2023)
Progress continues in prediction of the response to treatment of RA.
in Nature reviews. Rheumatology
Rivellese F
(2022)
Rituximab versus tocilizumab in rheumatoid arthritis: synovial biopsy-based biomarker analysis of the phase 4 R4RA randomized trial.
in Nature medicine
Iaquinta F
(2023)
Synovial biopsies for molecular definition of rheumatoid arthritis and treatment response phenotyping: where can we improve?
in Expert Review of Molecular Diagnostics
Schmidt M
(2022)
Transcriptomic and proteomic profiling of NaV1.8-expressing mouse nociceptors.
in Frontiers in molecular neuroscience
Description | Prediction Of Clinical Response To TNF Inhibitor Using Both Gene Expression And Epigenetic Profile |
Amount | £248,654 (GBP) |
Organisation | The Medical College of Saint Bartholomew's Hospital Trust |
Sector | Charity/Non Profit |
Country | United Kingdom |
Start | 04/2021 |
End | 07/2024 |
Description | TRACT-RA - NanoString |
Organisation | NanoString Technologies |
Country | United States |
Sector | Private |
PI Contribution | Providing samples and data from previous cohorts (PEAC/STRAP/R4RA) in the context of the TRACT-RA award |
Collaborator Contribution | This collaboration will investigate biomarkers of response/non-response, pathways analysis which may results in developing a diagnostic tool. |
Impact | Too early |
Start Year | 2021 |
Description | TRACT-RA - PrecisionLife |
Organisation | Precisionlife |
Country | United Kingdom |
Sector | Private |
PI Contribution | Providing data from previous cohorts (PEAC/STRAP/R4RA) in the context of the TRACT-RA award |
Collaborator Contribution | This collaboration will investigate biomarkers of response/non-response, pathways analysis which may results in developing a diagnostic tool. |
Impact | Too early |
Start Year | 2021 |
Description | TRACT-RA collaboration Universities |
Organisation | University of Edinburgh |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Provide data and samples from previous cohorts in the context of the TRACT-RA award |
Collaborator Contribution | Contribute to the analysis of the data in the context of the TRACT-RA award. UoM to perform serological and genotyping analysis. |
Impact | Too early |
Start Year | 2021 |
Description | TRACT-RA collaboration Universities |
Organisation | University of Manchester |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Provide data and samples from previous cohorts in the context of the TRACT-RA award |
Collaborator Contribution | Contribute to the analysis of the data in the context of the TRACT-RA award. UoM to perform serological and genotyping analysis. |
Impact | Too early |
Start Year | 2021 |
Title | METHOD FOR TREATING RHEUMATOID ARTHRITIS |
Description | A method for determining whether a Rheumatoid Arthritis (RA) patient is susceptible to treatment with a B cell targeted therapy, the method comprising the steps: (a) determining the level of one or more biomarker in one or more sample obtained from the patient, wherein the one or more biomarker is selected from Table 1; and (b) comparing the level of the one or more biomarker to one or more corresponding reference value; wherein the level of the one or more biomarker compared to the corresponding reference value is indicative of the susceptibility to treatment with a B cell targeted therapy. |
IP Reference | WO2022157506 |
Protection | Patent / Patent application |
Year Protection Granted | 2022 |
Licensed | Yes |
Impact | Patent has been licensed by Exagen Inc and is being used to develop a novel prognostic biomarker for stratifying optimal treatments in rheumatoid arthritis |
Title | R4RA RNA-Seq shiny web portal |
Description | Shiny web app for interrogating RNA-Sequencing data from synovial biopsies from the R4RA clinical trial. |
Type Of Technology | Webtool/Application |
Year Produced | 2021 |
Open Source License? | Yes |
Impact | This web app makes the RNA-Seq data from R4RA publicly available with an easy to use interface allowing researchers to query genes of interest and correlate gene expression in synovial biopsies against clinical parameters and clinical outcomes. |
URL | https://r4ra.hpc.qmul.ac.uk |
Title | STRAP shiny web app |
Description | Website using R/shiny interface to allow interactive analysis of RNA-Seq data from the STRAP trial and comparison with clinical metadata. |
Type Of Technology | Webtool/Application |
Year Produced | 2023 |
Open Source License? | Yes |
Impact | Allows public dissemination of data with simple and easy to use analysis of datasets. |
URL | https://strap.hpc.qmul.ac.uk/ |
Title | glmmSeq R package |
Description | Using mixed effects models to analyse longitudinal gene expression can highlight differences between sample groups over time. The most widely used differential gene expression tools are unable to fit linear mixed effect models, and are less optimal for analysing longitudinal data. This package provides negative binomial and Gaussian mixed effects models to fit gene expression and other biological data across repeated samples. This is particularly useful for investigating changes in RNA-Sequencing gene expression between groups of individuals over time, as described in: Rivellese, F., Surace, A. E., Goldmann, K., Sciacca, E., Cubuk, C., Giorli, G., ... Lewis, M. J., & Pitzalis, C. (2022) Nature medicine . |
Type Of Technology | Software |
Year Produced | 2022 |
Open Source License? | Yes |
Impact | Enabled analysis of longitudinal RNA-Seq data from synovial biopsies from the R4RA clinical trial. |
URL | https://cran.r-project.org/package=glmmSeq |
Title | nestedcv R package |
Description | Implements nested k*l-fold cross-validation for lasso and elastic-net regularised linear models via the 'glmnet' package and other machine learning models via the 'caret' package. Cross-validation of 'glmnet' alpha mixing parameter and embedded fast filter functions for feature selection are provided. |
Type Of Technology | Software |
Year Produced | 2022 |
Open Source License? | Yes |
Impact | Has been used to build predictive models from RNA-Seq data from the STRAP biopsy-driven RCT. |
URL | https://cran.r-project.org/package=nestedcv |