Distinguishing prognostic and predictive biomarkers: an information theoretic approach

Sechidis, Konstantinos; Papangelou, Konstantinos; Metcalfe, Paul; Svensson, David; Weatherall, James Owen; Brown, Gavin

doi:10.1093/bioinformatics/bty357

Cited by 64 publications

(63 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Biomarkers that can predict responses to treatment would be valuable in supporting the paradigm shift towards personalized medicine [24][25][26]. In this study, univariate analysis identified several biomarkers that individually predicted responses to sarilumab treatment.…”

Section: Discussionmentioning

confidence: 99%

Sarilumab and adalimumab differential effects on bone remodelling and cardiovascular risk biomarkers, and predictions of treatment outcomes

et al. 2020

View full text Add to dashboard Cite

Background: Interleukin-6 (IL-6) is a pleiotropic cytokine that plays a key role in the pathogenesis of rheumatoid arthritis. Sarilumab is a human monoclonal antibody that binds membrane-bound and soluble IL-6 receptor-α to inhibit IL-6 signalling. The aim of this study was to compare the effects of sarilumab and adalimumab (a tumour necrosis factor alpha inhibitor) monotherapy on levels of circulating biomarkers associated with the acute-phase response, bone remodelling, atherothrombosis, anaemia of chronic disease and markers purported to reflect synovial lymphoid and myeloid cell infiltrates, as well as the potential of these biomarkers to differentially predict clinical and patient-reported outcomes with sarilumab vs. adalimumab. Methods: In this post hoc analysis, serum samples were analysed at baseline and prespecified post-treatment timepoints up to week 24 in adults with moderate-to-severe active rheumatoid arthritis intolerant of or inadequate responders to methotrexate from the MONARCH trial (NCT02332590). Results: Greater reductions in C-reactive protein (CRP; − 94.0% vs.-24.0%), serum amyloid A (SAA; − 83.2% vs.-17.4%), total receptor activator of nuclear factor-κB ligand (RANKL; − 18.3% vs. 10.5%) and lipoprotein (a) (− 41.0% vs.-2.8%) were observed at week 24 with sarilumab vs. adalimumab, respectively (adjusted p < 0.0001). Greater increases in procollagen type 1 N-terminal propeptide (P1NP) were observed with sarilumab vs. adalimumab at week 24 (22.8% vs. 6.2%, p = 0.027). Patients with high baseline SAA, CRP and matrix metalloproteinase-3 (MMP-3) were more likely to achieve clinical efficacy, including American College of Rheumatology 20% improvement criteria and Disease Activity Score (28 joints)-CRP < 3.2, and report improvements in patient-reported outcomes, including Health Assessment Questionnaire-Disability Index and pain visual analogue scale, with sarilumab than adalimumab. Conclusion: Sarilumab was associated with greater positive effects on bone remodelling and decreases in biomarkers of the acute-phase response, synovial inflammation and cardiovascular risk vs. adalimumab. High baseline concentrations of SAA, CRP and MMP-3 are predictive of clinical and patient-reported outcome responses to sarilumab treatment and prospective validation is warranted to confirm these results.

show abstract

Section: Discussionmentioning

confidence: 99%

Sarilumab and adalimumab differential effects on bone remodelling and cardiovascular risk biomarkers, and predictions of treatment outcomes

et al. 2020

View full text Add to dashboard Cite

show abstract

“…(1) we get the ML estimatorÎ ML (X ; Y ). It is well known that this estimator is biased (Steuer et al 2002), while there are many works that tried to derive expressions for correcting this bias. As Brillinger (2004) mentioned, the bias correction expressions are messy, and for that reason non-parametric procedures, such as jackknife (JK) or bootstrap, are preferable for estimating information theoretic terms.…”

Section: Frequentist Estimatorsmentioning

confidence: 99%

Efficient feature selection using shrinkage estimators

et al. 2019

Self Cite

View full text Add to dashboard Cite

Information theoretic feature selection methods quantify the importance of each feature by estimating mutual information terms to capture: the relevancy, the redundancy and the complementarity. These terms are commonly estimated by maximum likelihood, while an under-explored area of research is how to use shrinkage methods instead. Our work suggests a novel shrinkage method for data-efficient estimation of information theoretic terms. The small sample behaviour makes it particularly suitable for estimation of discrete distributions with large number of categories (bins). Using our novel estimators we derive a framework for generating feature selection criteria that capture any high-order feature interaction for redundancy and complementarity. We perform a thorough empirical study across datasets from diverse sources and using various evaluation measures. Our first finding is that our shrinkage based methods achieve better results, while they keep the same computational cost as the simple maximum likelihood based methods. Furthermore, under our framework we derive efficient novel high-order criteria that outperform state-of-the-art methods in various tasks.

show abstract

“…This can be done by performing sample size determination for observing given MI quantities with a particular statistical power [ 30 ]. Finally, by connecting the problem of multi-target FS with the problem of biomarker discovery in clinical trials with multiple endpoints, we can potentially use Group-JMI-Rand for deriving prognostic and predictive biomarkers in multiple endpoint trials [ 31 ].…”

Section: Discussionmentioning

confidence: 99%

Information Theoretic Multi-Target Feature Selection via Output Space Quantization

Sechidis

Spyromitros-Xioufis

Vlahavas

2019

Entropy

Self Cite

View full text Add to dashboard Cite

A key challenge in information theoretic feature selection is to estimate mutual information expressions that capture three desirable terms-the relevancy of a feature with the output, the redundancy and the complementarity between groups of features. The challenge becomes more pronounced in multi-target problems, where the output space is multi-dimensional. Our work presents an algorithm that captures these three desirable terms and is suitable for the well-known multi-target prediction settings of multi-label/dimensional classification and multivariate regression. We achieve this by combining two ideas-deriving low-order information theoretic approximations for the input space and using quantization algorithms for deriving low-dimensional approximations of the output space. Under the above framework we derive a novel criterion, Group-JMI-Rand, which captures various high-order target interactions. In an extensive experimental study we showed that our suggested criterion achieves competing performance against various other information theoretic feature selection criteria suggested in the literature.

show abstract

Distinguishing prognostic and predictive biomarkers: an information theoretic approach

Cited by 64 publications

References 25 publications

Sarilumab and adalimumab differential effects on bone remodelling and cardiovascular risk biomarkers, and predictions of treatment outcomes

Sarilumab and adalimumab differential effects on bone remodelling and cardiovascular risk biomarkers, and predictions of treatment outcomes

Efficient feature selection using shrinkage estimators

Information Theoretic Multi-Target Feature Selection via Output Space Quantization

Contact Info

Product

Resources

About