![[ Visit Client Website ]](images/banner.gif)
Studies such as TICORA, CAMERA and FinRACO suggest frequent quantitative monitoring of disease activity with resulting treatment changes improves patient outcomes. ACR and EULAR also recommend ongoing disease activity assessment. Current monitoring tools, while useful for longitudinal tracking, are suboptimal; laboratory tests such as ESR and CRP are non-specific and do not reflect the heterogeneous biology of RA, while symptom-based measures are subjective and have low reproducibility. We are developing a multi-protein biomarker index of RA disease activity using a rigorous, stepwise development program that comprehensively surveys the biological pathways underlying RA.
Method
Candidate serum protein biomarkers were selected from an extensive screen of literature, databases, and experimental data. Quantitative assays for 141 proteins were optimized for reproducibility, RF blocking, sensitivity, and dynamic range, resulting in 121 measurable proteins in RA patient serum.
Three serial studies were performed - only proteins with the strongest associations to disease activity continued to the next. The first study examined 121 proteins in 128 samples, the second 65 proteins in 320 samples, and the third 21 proteins in 255 samples. Associations between candidate biomarker levels and disease activity were assessed using univariate and multivariate statistics and a range of disease activity measures (including joint counts, DAS28ESR, DAS28CRP, CDAI and patient-reported outcomes). Major clinical covariates such as CCP status and treatment were taken into account by inclusion in statistical modeling and by analyzing patients by disease subtype.
Results
The 3 studies were progressively more enriched for proteins associated with disease activity. 8% of proteins had univariate correlations with DAS28 in the first study, 51% in the second and 60% in the third. The top 21 proteins, selected on the basis of univariate multivariate modeling, represent a diverse set of biological pathways implicated in RA pathogenesis. Statistical models with 4-11 protein biomarkers outperformed any individual biomarker at estimating disease activity. These models achieved average accuracy of 70% for assigning patients into low and high disease activity categories, and average correlations of 0.6 with DAS28 in 100 iterations of cross validation. Models developed in one cohort performed well in independent cohorts (correlation of 0.58-0.6 with DAS28).
Conclusion
Serum protein biomarkers contain biologically rich information reflecting RA disease activity. A robust, stepwise development path, using large cohorts from across the spectrum of care, increases likelihood of successfully validating a multi-marker assay. Developing a validated disease assessment biomarker in RA will enhance current measures and can provide value for patient care and outcomes.
Disclosure: Y. Shen, Crescendo Bioscience, Inc., 3 ; N. Knowlton, Crescendo Bioscience, Inc., 5 ; M. Turner, Crescendo Bioscience, Inc., 1, Crescendo Bioscience, Inc., 5 ; C. Sutton, Crescendo Bioscience, Inc., 3, Crescendo Bioscience, Inc., 1 ; D. Smith, Crescendo Bioscience, Inc., 3 ; D. Chernoff, Crescendo Bioscience, Inc., 3 ; L. Hesterberg, Crescendo Bioscience, Inc., 3 ; R. Roubenoff, Biogen Idec, 3 ; N. A. Shadick, Crescendo Bioscience, 2, Biogen Idec, 2 ; M. Weinblatt, crescendo, 2, Biogen Idec, 2, crescendo, 5, Biogen Idec, 5 ; G. Cavet, Crescendo Bioscience, Inc., 3, Crescendo Bioscience, 1 ; M. Centola, Crescendo Bioscience, 5 .