ObjectiveTo review biomarker discovery studies using omics data for patient stratification which led to clinically validated FDA-cleared tests or laboratory developed tests, in order to identify common characteristics and derive recommendations for future biomarker projects.DesignScoping review.MethodsWe searched PubMed, EMBASE and Web of Science to obtain a comprehensive list of articles from the biomedical literature published between January 2000 and July 2021, describing clinically validated biomarker signatures for patient stratification, derived using statistical learning approaches. All documents were screened to retain only peer-reviewed research articles, review articles or opinion articles, covering supervised and unsupervised machine learning applications for omics-based patient stratification. Two reviewers independently confirmed the eligibility. Disagreements were solved by consensus. We focused the final analysis on omics-based biomarkers which achieved the highest level of validation, that is, clinical approval of the developed molecular signature as a laboratory developed test or FDA approved tests.ResultsOverall, 352 articles fulfilled the eligibility criteria. The analysis of validated biomarker signatures identified multiple common methodological and practical features that may explain the successful test development and guide future biomarker projects. These include study design choices to ensure sufficient statistical power for model building and external testing, suitable combinations of non-targeted and targeted measurement technologies, the integration of prior biological knowledge, strict filtering and inclusion/exclusion criteria, and the adequacy of statistical and machine learning methods for discovery and validation.ConclusionsWhile most clinically validated biomarker models derived from omics data have been developed for personalised oncology, first applications for non-cancer diseases show the potential of multivariate omics biomarker design for other complex disorders. Distinctive characteristics of prior success stories, such as early filtering and robust discovery approaches, continuous improvements in assay design and experimental measurement technology, and rigorous multicohort validation approaches, enable the derivation of specific recommendations for future studies.

References to equations in the main document are preceeded by 'MD Eq'. Computing times CV single penalty ridgeBelow we present computing times for plain CV, fast CV using SVD, and approximated leaveone-out CV (LOOCV) as discussed in (Meijer and Goeman, 2013). Table 1 presents results for LOOCV and 10-fold CV. Data sets used are: 1) the methylation data set dataFarkas as available in the R-package GRridge (Van de Wiel et al., 2016), which has dimensions n × p = 37 × 40, 000 and binary response; 2) the MESO miRNA data set presented in the main document, with dimensions 84 × 1, 398 and survival response; 3) as 2) but mRNA, with dimensions 84 × 19, 252; 4) the KIRC miRNA data set presented in the main document, with dimensions 506 × 1, 487 and survival response; 5) as 4) but mRNA, with dimensions 506 × 19, 431.

No abstract

This paper introduces the paired lasso: a generalisation of the lasso for paired covariate settings. Our aim is to predict a single response from two high-dimensional covariate sets. We assume a one-to-one correspondence between the covariate sets, with each covariate in one set forming a pair with a covariate in the other set. Paired covariates arise, for example, when two transformations of the same data are available. It is often unknown which of the two covariate sets leads to better predictions, or whether the two covariate sets complement each other. The paired lasso addresses this problem by weighting the covariates to improve the selection from the covariate sets and the covariate pairs. It thereby combines information from both covariate sets and accounts for the paired structure. We tested the paired lasso on more than 2000 classification problems with experimental genomics data, and found that for estimating sparse but predictive models, the paired lasso outperforms the standard and the adaptive lasso. The R package is available from cran.

Motivation Machine learning in the biomedical sciences should ideally provide predictive and interpretable models. When predicting outcomes from clinical or molecular features, applied researchers often want to know which features have effects, whether these effects are positive or negative, and how strong these effects are. Regression analysis includes this information in the coefficients but typically renders less predictive models than more advanced machine learning techniques. Results Here we propose an interpretable meta-learning approach for high-dimensional regression. The elastic net provides a compromise between estimating weak effects for many features and strong effects for some features. It has a mixing parameter to weight between ridge and lasso regularisation. Instead of selecting one weighting by tuning, we combine multiple weightings by stacking. We do this in a way that increases predictivity without sacrificing interpretability. Availability and Implementation The R package starnet is available on GitHub: https://github.com/rauschenberger/starnet. Supplementary information Supplementary data are available at Bioinformatics online.

Several phenotypic differences observed in Parkinson’s disease (PD) patients have been linked to age at onset (AAO). We endeavoured to find out whether these differences are due to the ageing process itself by using a combined dataset of idiopathic PD (n = 430) and healthy controls (HC; n = 556) excluding carriers of known PD-linked genetic mutations in both groups. We found several significant effects of AAO on motor and non-motor symptoms in PD, but when comparing the effects of age on these symptoms with HC (using age at assessment, AAA), only positive associations of AAA with burden of motor symptoms and cognitive impairment were significantly different between PD vs HC. Furthermore, we explored a potential effect of polygenic risk score (PRS) on clinical phenotype and identified a significant inverse correlation of AAO and PRS in PD. No significant association between PRS and severity of clinical symptoms was found. We conclude that the observed non-motor phenotypic differences in PD based on AAO are largely driven by the ageing process itself and not by a specific profile of neurodegeneration linked to AAO in the idiopathic PD patients.

Background: The hypothesis of body-first vs. brain-first subtype of PD has been proposed with REM-Sleep behavior disorder (RBD) defining the former. The body-first PD presumes an involvement of the brainstem in the pathogenic process with higher burden of autonomic dysfunction. Objective: To identify distinctive clinical subtypes of idiopathic Parkinson’s disease (iPD) in line with the formerly proposed concept of body-first vs. brain-first subtypes in PD, we analyzed the presence of probable RBD, sex, and the APOE ɛ4 carrier status as potential sub-group stratifiers. Methods: A total of 400 iPD patients were included in the cross-sectional analysis from the baseline dataset with a completed RBD Screening Questionnaire (RBDSQ) for classifying as pRBD by using the cut-off RBDSQ≥6. Multiple regression models were applied to explore (i) the effects of pRBD on clinical outcomes adjusted for disease duration and age, (ii) the effect of sex on pRBD, and (iii) the association of APOE ɛ4 and pRBD. Results: iPD-pRBD was significantly associated with autonomic dysfunction (SCOPA-AUT), level of depressive symptoms (BDI-I), MDS-UPDRS I, hallucinations, and constipation, whereas significantly negatively associated with quality of life (PDQ-39) and sleep (PDSS). No significant association between sex and pRBD or APOE ɛ4 and pRBD in iPD was found nor did we determine a significant effect of APOE ɛ4 on the PD phenotype. Conclusion: We identified an RBD-specific PD endophenotype, characterized by predominant autonomic dysfunction, hallucinations, and depression, corroborating the concept of a distinctive body-first subtype of PD. We did not observe a significant association between APOE ɛ4 and pRBD suggesting both factors having an independent effect on cognitive decline in iPD.

Background: The analysis of the procedural memory is particularly relevant in neurodegenerative disorders like Parkinson’s disease, due to the central role of the basal ganglia in procedural memory. It has been shown that anterograde procedural memory, the ability to learn a new skill, is impaired in Parkinson’s disease. However, retrograde procedural memory, the long-term retention and execution of skills learned in earlier life stages, has not yet been systematically investigated in Parkinson’s disease. Objective: This study aims to investigate retrograde procedural memory in people with Parkinson’s disease. We hypothesized that retrograde procedural memory is impaired in people with Parkinson’s disease compared to an age- and gender-matched control group. Methods: First, we developed an extended evaluation system based on the Cube Copying Test, the CUPRO evaluation system, to distinguish the cube copying procedure, representing functioning of retrograde procedural memory, and the final result, representing the visuo-constructive abilities. Development of the evaluation system included tests of discriminant validity. Results: Comparing people with typical Parkinson’s disease (n = 201) with age- and gender-matched control subjects (n = 201), we identified cube copying performance to be significantly impaired in people with Parkinson’s disease (p = 0.008). No significant correlation was observed between retrograde procedural memory and disease duration. Conclusion: We demonstrated lower cube copying performance in people with Parkinson’s disease compared to control subjects, which suggests an impaired functioning of retrograde procedural memory in Parkinson’s disease.