Predicting dichotomised outcomes from high-dimensional data in biomedicine.

In many biomedical applications, we are more interested in the predicted probability that a numerical outcome is above a threshold than in the predicted value of the outcome. For example, it might be known that antibody levels above a certain threshold provide immunity against a disease, or a threshold for a disease severity score might reflect conversion from the presymptomatic to the symptomatic disease stage. Accordingly, biomedical researchers often convert numerical to binary outcomes (loss of information) to conduct logistic regression (probabilistic interpretation). We address this bad statistical practice by modelling the binary outcome with logistic regression, modelling the numerical outcome with linear regression, transforming the predicted values from linear regression to predicted probabilities, and combining the predicted probabilities from logistic and linear regression. Analysing high-dimensional simulated and experimental data, namely clinical data for predicting cognitive impairment, we obtain significantly improved predictions of dichotomised outcomes. Thus, the proposed approach effectively combines binary with numerical outcomes to improve binary classification in high-dimensional settings. An implementation is available in the R package cornet on GitHub (https://github.com/rauschenberger/cornet) and CRAN (https://CRAN.R-project.org/package=cornet).

Converging peripheral blood microRNA profiles in Parkinson's disease and progressive supranuclear palsy.

MicroRNAs act via targeted suppression of messenger RNA translation in the DNA-RNA-protein axis. The dysregulation of microRNA(s) reflects the epigenetic changes affecting the cellular processes in multiple disorders. To understand the complex effect of dysregulated microRNAs linked to neurodegeneration, we performed a cross-sectional microRNA expression analysis in idiopathic Parkinson's disease (n = 367), progressive supranuclear palsy (n = 35) and healthy controls (n = 416) from the Luxembourg Parkinson's Study, followed by prediction modelling, enriched pathway analysis and target simulation of dysregulated microRNAs using probabilistic Boolean modelling. Forty-six microRNAs were identified to be dysregulated in Parkinson's disease versus controls and 16 in progressive supranuclear palsy versus controls with 4 overlapping significantly dysregulated microRNAs between the comparisons. Predictive power of microRNA subsets (including up to 100 microRNAs) was modest for differentiating Parkinson's disease or progressive supranuclear palsy from controls (maximal cross-validated area under the receiver operating characteristic curve 0.76 and 0.86, respectively) and low for progressive supranuclear palsy versus Parkinson's disease (maximal cross-validated area under the receiver operating characteristic curve 0.63). The enriched pathway analysis revealed natural killer cell pathway to be dysregulated in both, Parkinson's disease and progressive supranuclear palsy versus controls, indicating that the immune system might play an important role in both diseases. Probabilistic Boolean modelling of pathway dynamics affected by dysregulated microRNAs in Parkinson's disease and progressive supranuclear palsy revealed partially overlapping dysregulation in activity of the transcription factor EB, endoplasmic reticulum stress signalling, calcium signalling pathway, dopaminergic transcription and peroxisome proliferator-activated receptor gamma coactivator-1α activity, though involving different mechanisms. These findings indicated a partially convergent (sub)cellular end-point dysfunction at multiple levels in Parkinson's disease and progressive supranuclear palsy, but with distinctive underlying molecular mechanisms.

Cognition and Other Non-Motor Symptoms in an At-Risk Cohort for Parkinson's Disease Defined by REM-Sleep Behavior Disorder and Hyposmia.

Background: REM-sleep behavior disorder (RBD) and other non-motor symptoms such as hyposmia were proposed by the Movement Disorder Society as research criteria for prodromal Parkinson's disease (P-PD). Global cognitive deficit was later added. Objective: To compare non-motor symptoms, focusing on cognition, between a P-PD group and a matched control group. Methods: In this cross-sectional, case-control study, in a first set of analyses, we performed extensive cognitive testing on people with (n = 76) and a control group without (n = 195) probable RBD and hyposmia. Furthermore, we assessed motor and non-motor symptoms related to Parkinson's Disease (PD). After propensity score matching, we compared 62 P-PD with 62 age- and sex-matched controls. In addition, we performed regression analyses on the total sample (n = 271). In a second set of analyses, we used, a.o., the CUPRO to evaluate retrograde procedural memory and visuo-constructive functions. Results: People with P-PD showed significantly poorer performances in global cognition, visuo-constructive and executive functions, mainly in mental flexibility (p < 0.001; p = 0.004; p = 0.003), despite similar educational levels (p = 0.415). We observed significantly more motor and non-motor symptoms (p < 0.001; p = 0.004), higher scores for depression (p = 0.004) and apathy (p < 0.001) as well as lower quality of life (p < 0.001) in P-PD. CONCLUSIONS: Our findings confirm that global cognitive, executive, and visuo-constructive deficits define the P-PD group. In addition, depression, apathy, and lower quality of life were more prevalent in P-PD. If replicated in other samples, executive and visuo-constructive deficits should be considered in non-motor P-PD. Determining specific patterns will support early recognition of PD, secondary prevention of complications and the development of neuroprotective treatments.

Comprehensive blood metabolomics profiling of Parkinson's disease reveals coordinated alterations in xanthine metabolism.

Parkinson's disease (PD) is a highly heterogeneous disorder influenced by several environmental and genetic factors. Effective disease-modifying therapies and robust early-stage biomarkers are still lacking, and an improved understanding of the molecular changes in PD could help to reveal new diagnostic markers and pharmaceutical targets. Here, we report results from a cohort-wide blood plasma metabolic profiling of PD patients and controls in the Luxembourg Parkinson's Study to detect disease-associated alterations at the level of systemic cellular process and network alterations. We identified statistically significant changes in both individual metabolite levels and global pathway activities in PD vs. controls and significant correlations with motor impairment scores. As a primary observation when investigating shared molecular sub-network alterations, we detect pronounced and coordinated increased metabolite abundances in xanthine metabolism in de novo patients, which are consistent with previous PD case/control transcriptomics data from an independent cohort in terms of known enzyme-metabolite network relationships. From the integrated metabolomics and transcriptomics network analysis, the enzyme hypoxanthine phosphoribosyltransferase 1 (HPRT1) is determined as a potential key regulator controlling the shared changes in xanthine metabolism and linking them to a mechanism that may contribute to pathological loss of cellular adenosine triphosphate (ATP) in PD. Overall, the investigations revealed significant PD-associated metabolome alterations, including pronounced changes in xanthine metabolism that are mechanistically congruent with alterations observed in independent transcriptomics data. The enzyme HPRT1 may merit further investigation as a main regulator of these network alterations and as a potential therapeutic target to address downstream molecular pathology in PD.

Determinants of patient-reported functional mobility in people with Parkinson's disease: A systematic review.

BACKGROUND: Information on determinants of patient-reported functional mobility is lacking but would inform the planning of healthcare, resources and strategies to promote functional mobility in people with Parkinson's disease (PD). RESEARCH QUESTION: To identify the determinants of patient-reported functional mobility of people with PD. METHODS: Eligible: Randomized Controlled Trials, cohort, case-control, or cross-sectional analyses in people PD without date or setting restrictions, published in English, German, or French. Excluded: instruments with under 50 % of items measuring mobility. On August 9th 2023 we last searched Medline, CINAHL and PsychInfo. We assessed risk of bias using the mixed-methods appraisal tool. Results were synthesized by tabulating the determinants by outcomes and study designs. RESULTS: Eleven studies published 2012-2023 were included (most in Swedish outpatient settings). Samples ranged from 9 to 255 participants. Follow-up varied from 1.5 to 36 months with attrition of 15-42 %. Heterogenic study designs complicated results synthesis. However, determinants related to environment seem to associate the strongest with patient-reported functional mobility, although determinants related to body structures and functions were most investigated. We identified disease duration, the ability to drive, caregiving, sex, age, cognitive impairment, postural instability and social participation as determinants of patient-reported functional mobility. DISCUSSION: Methodological quality of the studies was limited. No study reported an a priori power calculation. Three studies controlled for confounders. The included studies lack representativeness of the population of people living with PD. Standardized sets of outcomes could enable more systematic research synthesis. CONCLUSIONS: Future research should focus on activities, participation and environmental factors and improve methodological quality.

Penalized regression with multiple sources of prior effects.

MOTIVATION: In many high-dimensional prediction or classification tasks, complementary data on the features are available, e.g. prior biological knowledge on (epi)genetic markers. Here we consider tasks with numerical prior information that provide an insight into the importance (weight) and the direction (sign) of the feature effects, e.g. regression coefficients from previous studies. RESULTS: We propose an approach for integrating multiple sources of such prior information into penalized regression. If suitable co-data are available, this improves the predictive performance, as shown by simulation and application. AVAILABILITY AND IMPLEMENTATION: The proposed method is implemented in the R package transreg (https://github.com/lcsb-bds/transreg, https://cran.r-project.org/package=transreg).

Validation of a Parkinson's disease questionnaire-39-based functional mobility composite score (FMCS) in people with Parkinson's disease.

INTRODUCTION: Functional mobility is an important outcome for people with Parkinson's disease (PwP). Despite this, there is no established patient-reported outcome measure that serves as a gold standard for assessing patient-reported functional mobility in PwP. We aimed to validate the algorithm calculating the Parkinson's Disease Questionnaire-39 (PDQ-39) based Functional Mobility Composite Score (FMCS). METHODS: We designed a count-based algorithm to measure patient-reported functional mobility in PwP from items of the PDQ-39 subscales mobility and activities of daily living. Convergent validity of the algorithm calculating the PDQ-39-based FMCS was assessed using the objective Timed Up and Go (n = 253) and discriminative validity was assessed by comparing the FMCS with patient-reported (MDS-UPDRS II) and clinician-assessed (MDS-UPDRS III) motor symptoms as well as between disease stages (H&Y) and PIGD phenotypes (n = 736). Participants were between 22 and 92 years old, with a disease duration from 0 to 32 years and 64.9% in a H&Y 1-2 ranging from 1 to 5. RESULTS: Spearman correlation coefficients (rs) ranging from -0.45 to -0.77 (p < 0.001) indicated convergent validity. Hence, a t-test suggested sufficient ability of the FMCS to discriminate (p < 0.001) between patient-reported and clinician-assessed motor symptoms. More specifically, FMCS was more strongly associated with patient-reported MDS-UPDRS II (rs = -0.77) than clinician-reported MDS-UPDRS III (rs = -0.45) and can discriminate between disease stages as between PIGD phenotypes (p < 0.001). CONCLUSION: The FMCS is a valid composite score to assess functional mobility through patient reports in PwP for studying functional mobility in studies using the PDQ-39.

A powerful global test for spliceQTL effects.

Statistical methods to test for effects of single nucleotide polymorphisms (SNPs) on exon inclusion exist but often rely on testing of associations between multiple exon-SNP pairs, with sometimes subsequent summarization of results at the gene level. Such approaches require heavy multiple testing corrections and detect mostly events with large effect sizes. We propose here a test to find spliceQTL (splicing quantitative trait loci) effects that takes all exons and all SNPs into account simultaneously. For any chosen gene, this score-based test looks for an association between the set of exon expressions and the set of SNPs, via a random-effects model framework. It is efficient to compute and can be used if the number of SNPs is larger than the number of samples. In addition, the test is powerful in detecting effects that are relatively small for individual exon-SNP pairs but are observed for many pairs. Furthermore, test results are more often replicated across datasets than pairwise testing results. This makes our test more robust to exon-SNP pair-specific effects, which do not extend to multiple pairs within the same gene. We conclude that the test we propose here offers more power and better replicability in the search for spliceQTL effects.

Retrograde procedural memory is impaired in people with Parkinson's disease with freezing of gait.

Background: Freezing of gait (FOG), is associated with impairment of different cognitive functions. Previous studies hypothesized that FOG may be due to a loss of automaticity. Research question: To explore whether FOG is associated with impairment in cognitive functions, focusing on retrograde procedural memory, the memory responsible for the automatic, implicit stored procedures that have been acquired in earlier life stages. Methods: In this cross-sectional, case-control study, 288 people with typical Parkinson's disease (PD) from the Luxembourg Parkinson's Study were assigned to Freezers (FOG+) and non-Freezers (FOG-) based on the MDS-UPDRS 2.13 (self-reported FOG episodes) and 3.11 (FOG evaluated by clinicians during gait assessment). Both groups were matched on age, sex and disease duration. Global cognition (MoCA), retrograde procedural memory and visuo-constructive abilities (CUPRO), psychomotor speed and mental flexibility (TMT) were assessed. Furthermore, we repeated our analyses by additionally controlling for depression (BDI-I). Results: Besides lower global cognition (MoCA; p = 0.007) and mental flexibility (TMT-B and Delta-TMT; p < 0.001), FOG+ showed a lower performance in retrograde procedural memory (CUPRO-IS1; p < 0.001) compared to FOG-. After controlling additionally for depression, our main outcome variable CUPRO-IS1 remained significantly lower in FOG+ (p = 0.010). Conclusion: Our findings demonstrated that besides lower global cognition and mental flexibility scores, FOG+ showed lower performance in retrograde procedural memory compared to matched FOG-control patients, even when accounting for factors such as age, sex, disease duration or depression. Significance: In the context of limited treatment options, especially for non-invasive therapeutic approaches, these insights on procedural memory and FOG may lead to new hypotheses on FOG etiology and consequently the development of new treatment options.

Body-First Subtype of Parkinson's Disease with Probable REM-Sleep Behavior Disorder Is Associated with Non-Motor Dominant Phenotype.

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 (pRBD), sex, and the APOEɛ4 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 effect of pRBD on clinical outcomes adjusted for disease duration and age, (ii) the effect of sex on pRBD, and (iii) the association of APOEɛ4 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.

Retrograde Procedural Memory in Parkinson's Disease: A Cross-Sectional, Case-Control Study.

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 the CUPRO evaluation system, an extended evaluation system based on the Cube Copying Test, 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.

Biomarker discovery studies for patient stratification using machine learning analysis of omics data: a scoping review.

OBJECTIVE: To 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. DESIGN: Scoping review. METHODS: We 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. RESULTS: Overall, 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. CONCLUSIONS: While 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.

Predicting correlated outcomes from molecular data.

MOTIVATION: Multivariate (multi-target) regression has the potential to outperform univariate (single-target) regression at predicting correlated outcomes, which frequently occur in biomedical and clinical research. Here we implement multivariate lasso and ridge regression using stacked generalization. RESULTS: Our flexible approach leads to predictive and interpretable models in high-dimensional settings, with a single estimate for each input-output effect. In the simulation, we compare the predictive performance of several state-of-the-art methods for multivariate regression. In the application, we use clinical and genomic data to predict multiple motor and non-motor symptoms in Parkinson's disease patients. We conclude that stacked multivariate regression, with our adaptations, is a competitive method for predicting correlated outcomes. AVAILABILITY AND IMPLEMENTATION: The R package joinet is available on GitHub (https://github.com/rauschenberger/joinet) and cran (https://cran.r-project.org/package=joinet). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Predictive and interpretable models via the stacked elastic net.

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 regularization. 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) and CRAN (https://CRAN.R-project.org/package=starnet).

Testing for association between RNA-Seq and high-dimensional data.

BACKGROUND: Testing for association between RNA-Seq and other genomic data is challenging due to high variability of the former and high dimensionality of the latter. RESULTS: Using the negative binomial distribution and a random-effects model, we develop an omnibus test that overcomes both difficulties. It may be conceptualised as a test of overall significance in regression analysis, where the response variable is overdispersed and the number of explanatory variables exceeds the sample size. CONCLUSIONS: The proposed test can detect genetic and epigenetic alterations that affect gene expression. It can examine complex regulatory mechanisms of gene expression. The R package globalSeq is available from Bioconductor.