A SIGNIFICANCE TEST FOR THE LASSO

Comparative analysis of histologically classified oligodendrogliomas reveals characteristic molecular differences between subgroups

Background

Molecular data of histologically classified oligodendrogliomas are available offering the possibility to stratify these human brain tumors into clinically relevant molecular subtypes.

Methods

Gene copy number, mutation, and expression data of 193 histologically classified oligodendrogliomas from The Cancer Genome Atlas (TCGA) were analyzed by well-established computational approaches (unsupervised clustering, statistical testing, network inference).

Results

We applied hierarchical clustering to tumor gene copy number profiles and revealed three molecular subgroups within histologically classified oligodendrogliomas. We further screened these subgroups for molecular glioma markers (1p/19q co-deletion, IDH mutation, gain of chromosome 7 and loss of chromosome 10) and found that our subgroups largely resemble known molecular glioma subtypes. We excluded glioblastoma-like tumors (7a10d subgroup) and derived a gene expression signature distinguishing histologically classified oligodendrogliomas with concurrent 1p/19q co-deletion and IDH mutation (1p/19q subgroup) from those with predominant IDH mutation alone (IDHme subgroup). Interestingly, many signature genes were part of signaling pathways involved in the regulation of cell proliferation, differentiation, migration, and cell-cell contacts. We further learned a gene regulatory network associated with the gene expression signature revealing novel putative major regulators with functions in cytoskeleton remodeling (e.g. APBB1IP, VAV1, ARPC1B), apoptosis (CCNL2, CREB3L1), and neural development (e.g. MYTIL, SCRT1, MEF2C) potentially contributing to the manifestation of differences between both subgroups. Moreover, we revealed characteristic expression differences of several HOX and SOX transcription factors suggesting the activity of different glioma stemness programs in both subgroups.

Conclusions

We show that gene copy number profiles alone are sufficient to derive molecular subgroups of histologically classified oligodendrogliomas that are well-embedded into general glioma classification schemes. Moreover, our revealed novel putative major regulators and characteristic stemness signatures indicate that different developmental programs might be active in these subgroups, providing a basis for future studies.

Electronic supplementary material

The online version of this article (10.1186/s12885-018-4251-7) contains supplementary material, which is available to authorized users.

Variable Selection for Confounding Adjustment in High-dimensional Covariate Spaces When Analyzing Healthcare Databases

BACKGROUND

Data-adaptive approaches to confounding adjustment may improve performance beyond expert knowledge when analyzing electronic healthcare databases and have additional practical advantages for analyzing multiple databases in rapid cycles. Improvements seemed possible if outcome predictors were reliably identified empirically and adjusted.

METHODS

In five cohort studies from diverse healthcare databases, we implemented a base-case high-dimensional propensity score algorithm with propensity score decile-adjusted outcome models to estimate treatment effects among prescription drug initiators. The original variable selection procedure based on the estimated bias of each variable using unadjusted associations between confounders and exposure (RRCE) and disease outcome (RRCD) was augmented by alternative strategies. These included using increasingly adjusted RRCD estimates, including models considering >1,500 variables jointly (Lasso, Bayesian logistic regression); using prediction statistics or likelihood-ratio statistics for covariate prioritization; directly estimating the propensity score with >1,500 variables (Lasso, Bayesian regression); or directly fitting an outcome model using all covariates jointly (Lasso, Ridge).

RESULTS

In five example studies, most tested augmentations of the base-case hdPS did not meaningfully change estimates in light of wide confidence intervals except for Bayesian regression and Lasso to estimate RRCD, which moved estimates minimally closer to the expectation in three of five examples. The direct outcome estimation with Lasso performed worst.

CONCLUSION

Overall, the basic heuristic of variable reduction in high-dimensional propensity score adjustment performed, as well as alternative approaches in diverse settings. Minor improvements in variable selection may be possible using Bayesian outcome regression to prioritize variables for propensity score estimation when outcomes are rare. See video abstract at, http://links.lww.com/EDE/B162.

EMD-regression for modelling multi-scale relationships, and application to weather-related cardiovascular mortality

In a number of environmental studies, relationships between nat4ural processes are often assessed through regression analyses, using time series data. Such data are often multi-scale and non-stationary, leading to a poor accuracy of the resulting regression models and therefore to results with moderate reliability. To deal with this issue, the present paper introduces the EMD-regression methodology consisting in applying the empirical mode decomposition (EMD) algorithm on data series and then using the resulting components in regression models. The proposed methodology presents a number of advantages. First, it accounts of the issues of non-stationarity associated to the data series. Second, this approach acts as a scan for the relationship between a response variable and the predictors at different time scales, providing new insights about this relationship. To illustrate the proposed methodology it is applied to study the relationship between weather and cardiovascular mortality in Montreal, Canada. The results shed new knowledge concerning the studied relationship. For instance, they show that the humidity can cause excess mortality at the monthly time scale, which is a scale not visible in classical models. A comparison is also conducted with state of the art methods which are the generalized additive models and distributed lag models, both widely used in weather-related health studies. The comparison shows that EMD-regression achieves better prediction performances and provides more details than classical models concerning the relationship.

Thermochemistry of gas-phase and surface species via LASSO-assisted subgraph selection

Graph theory-based regression techniques, such as group additivity, have widely been implemented for fast estimation of thermochemistry of large molecules.

Comparison of weighting approaches for genetic risk scores in gene-environment interaction studies

BACKGROUND

Weighted genetic risk scores (GRS), defined as weighted sums of risk alleles of single nucleotide polymorphisms (SNPs), are statistically powerful for detection gene-environment (GxE) interactions. To assign weights, the gold standard is to use external weights from an independent study. However, appropriate external weights are not always available. In such situations and in the presence of predominant marginal genetic effects, we have shown in a previous study that GRS with internal weights from marginal genetic effects ("GRS-marginal-internal") are a powerful and reliable alternative to single SNP approaches or the use of unweighted GRS. However, this approach might not be appropriate for detecting predominant interactions, i.e. interactions showing an effect stronger than the marginal genetic effect.

METHODS

In this paper, we present a weighting approach for such predominant interactions ("GRS-interaction-training") in which parts of the data are used to estimate the weights from the interaction terms and the remaining data are used to determine the GRS. We conducted a simulation study for the detection of GxE interactions in which we evaluated power, type I error and sign-misspecification. We compared this new weighting approach to the GRS-marginal-internal approach and to GRS with external weights.

RESULTS

Our simulation study showed that in the absence of external weights and with predominant interaction effects, the highest power was reached with the GRS-interaction-training approach. If marginal genetic effects were predominant, the GRS-marginal-internal approach was more appropriate. Furthermore, the power to detect interactions reached by the GRS-interaction-training approach was only slightly lower than the power achieved by GRS with external weights. The power of the GRS-interaction-training approach was confirmed in a real data application to the Traffic, Asthma and Genetics (TAG) Study (N = 4465 observations).

CONCLUSION

When appropriate external weights are unavailable, we recommend to use internal weights from the study population itself to construct weighted GRS for GxE interaction studies. If the SNPs were chosen because a strong marginal genetic effect was hypothesized, GRS-marginal-internal should be used. If the SNPs were chosen because of their collective impact on the biological mechanisms mediating the environmental effect (hypothesis of predominant interactions) GRS-interaction-training should be applied.

Tau and Aβ imaging, CSF measures, and cognition in Alzheimer's disease

Alzheimer's disease (AD) is characterized by two molecular pathologies: cerebral β-amyloidosis in the form of β-amyloid (Aβ) plaques and tauopathy in the form of neurofibrillary tangles, neuritic plaques, and neuropil threads. Until recently, only Aβ could be studied in humans using positron emission tomography (PET) imaging owing to a lack of tau PET imaging agents. Clinical pathological studies have linked tau pathology closely to the onset and progression of cognitive symptoms in patients with AD. We report PET imaging of tau and Aβ in a cohort of cognitively normal older adults and those with mild AD. Multivariate analyses identified unique disease-related stereotypical spatial patterns (topographies) for deposition of tau and Aβ. These PET imaging tau and Aβ topographies were spatially distinct but correlated with disease progression. Cerebrospinal fluid measures of tau, often used to stage preclinical AD, correlated with tau deposition in the temporal lobe. Tau deposition in the temporal lobe more closely tracked dementia status and was a better predictor of cognitive performance than Aβ deposition in any region of the brain. These data support models of AD where tau pathology closely tracks changes in brain function that are responsible for the onset of early symptoms in AD.

Model selection and prediction of outcomes in recent onset schizophrenia patients who undergo cognitive training

Predicting treatment outcomes in psychiatric populations remains a challenge, but is increasingly important in the pursuit of personalized medicine. Patients with schizophrenia have deficits in cognition, and targeted cognitive training (TCT) of auditory processing and working memory has been shown to improve some of these impairments; but little is known about the baseline patient characteristics predictive of cognitive improvement. Here we use a model selection and regression approach called least absolute shrinkage and selection operator (LASSO) to examine predictors of cognitive improvement in response to TCT for patients with recent onset schizophrenia. Forty-three individuals with recent onset schizophrenia randomized to undergo TCT were assessed at baseline on measures of cognition, symptoms, functioning, illness duration, and demographic variables. We carried out 10-fold cross-validation of LASSO for model selection and regression. We followed up on these results using linear models for statistical inference. No individual variable was found to correlate with improvement in global cognition using a Pearson correlation approach, and a linear model including all variables was also found not to be significant. However, the LASSO model identified baseline global cognition, education, and gender in a model predictive of improvement on global cognition following TCT. These findings offer guidelines for personalized approaches to cognitive training for patients with schizophrenia.

Testing and Confidence Intervals for High Dimensional Proportional Hazards Model

This paper proposes a decorrelation-based approach to test hypotheses and construct confidence intervals for the low dimensional component of high dimensional proportional hazards models. Motivated by the geometric projection principle, we propose new decorrelated score, Wald and partial likelihood ratio statistics. Without assuming model selection consistency, we prove the asymptotic normality of these test statistics, establish their semiparametric optimality. We also develop new procedures for constructing pointwise confidence intervals for the baseline hazard function and baseline survival function. Thorough numerical results are provided to back up our theory.

A Sequential Rejection Testing Method for High-Dimensional Regression with Correlated Variables

We propose a general, modular method for significance testing of groups (or clusters) of variables in a high-dimensional linear model. In presence of high correlations among the covariables, due to serious problems of identifiability, it is indispensable to focus on detecting groups of variables rather than singletons. We propose an inference method which allows to build in hierarchical structures. It relies on repeated sample splitting and sequential rejection, and we prove that it asymptotically controls the familywise error rate. It can be implemented on any collection of clusters and leads to improved power in comparison to more standard non-sequential rejection methods. We complement the theoretical analysis with empirical results for simulated and real data.

Characteristics of Persons Living With HIV Who Have Informal Caregivers in the cART Age of the Epidemic

People living with HIV (PLWH) are aging and many suffer with multimorbidities, making caregiving a relevant and important area of study. The purpose of our study was to understand the occurrence and role of informal caregivers in the current stage of the HIV epidemic. We conducted a Web-based survey with 1,373 PLWH to assess: how many had an informal, unpaid caregiver; the type of relationship with the informal caregiver; and the number of hours the caregiver provided support each day. Among respondents, 333 had an informal caregiver. Blacks, those with low income, individuals who ever had an AIDS diagnosis, those with basic cellphone service, and those living with other comorbid conditions were significantly more likely to have an informal caregiver. Given the demographic profile of those PLWH who were most likely to have caregivers, further study is needed to understand the needs of both caregivers and care recipients.

Iterative hard thresholding for model selection in genome-wide association studies

A genome-wide association study (GWAS) correlates marker and trait variation in a study sample. Each subject is genotyped at a multitude of SNPs (single nucleotide polymorphisms) spanning the genome. Here, we assume that subjects are randomly collected unrelateds and that trait values are normally distributed or can be transformed to normality. Over the past decade, geneticists have been remarkably successful in applying GWAS analysis to hundreds of traits. The massive amount of data produced in these studies present unique computational challenges. Penalized regression with the ℓ₁ penalty (LASSO) or minimax concave penalty (MCP) penalties is capable of selecting a handful of associated SNPs from millions of potential SNPs. Unfortunately, model selection can be corrupted by false positives and false negatives, obscuring the genetic underpinning of a trait. Here, we compare LASSO and MCP penalized regression to iterative hard thresholding (IHT). On GWAS regression data, IHT is better at model selection and comparable in speed to both methods of penalized regression. This conclusion holds for both simulated and real GWAS data. IHT fosters parallelization and scales well in problems with large numbers of causal markers. Our parallel implementation of IHT accommodates SNP genotype compression and exploits multiple CPU cores and graphics processing units (GPUs). This allows statistical geneticists to leverage commodity desktop computers in GWAS analysis and to avoid supercomputing.

AVAILABILITY

Source code is freely available at https://github.com/klkeys/IHT.jl.

regNet: an R package for network-based propagation of gene expression alterations

SUMMARY

Gene expression alterations and potentially underlying gene copy number mutations can be measured routinely in the wet lab, but it is still extremely challenging to quantify impacts of altered genes on clinically relevant characteristics to predict putative driver genes. We developed the R package regNet that utilizes gene expression and copy number data to learn regulatory networks for the quantification of potential impacts of individual gene expression alterations on user-defined target genes via network propagation. We demonstrate the value of regNet by identifying putative major regulators that distinguish pilocytic from diffuse astrocytomas and by predicting putative impacts of glioblastoma-specific gene copy number alterations on cell cycle pathway genes and patient survival.

AVAILABILITY AND IMPLEMENTATION

regNet is available for download at https://github.com/seifemi/regNet under GNU GPL-3.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Genome-wide identification of lineage and locus specific variation associated with pneumococcal carriage duration

Streptococcus pneumoniae is a leading cause of invasive disease in infants, especially in low-income settings. Asymptomatic carriage in the nasopharynx is a prerequisite for disease, but variability in its duration is currently only understood at the serotype level. Here we developed a model to calculate the duration of carriage episodes from longitudinal swab data, and combined these results with whole genome sequence data. We estimated that pneumococcal genomic variation accounted for 63% of the phenotype variation, whereas the host traits considered here (age and previous carriage) accounted for less than 5%. We further partitioned this heritability into both lineage and locus effects, and quantified the amount attributable to the largest sources of variation in carriage duration: serotype (17%), drug-resistance (9%) and other significant locus effects (7%). A pan-genome-wide association study identified prophage sequences as being associated with decreased carriage duration independent of serotype, potentially by disruption of the competence mechanism. These findings support theoretical models of pneumococcal competition and antibiotic resistance.

Topographies of Cortical and Subcortical Volume Loss in HIV and Aging in the cART Era

OBJECTIVES

Studies of HIV-associated brain atrophy often focus on a priori brain regions of interest, which can introduce bias. A data-driven, minimally biased approach was used to analyze changes in brain volumetrics associated with HIV and their relationship to aging, viral factors, combination antiretroviral therapy (cART), and gender, and smoking.

DESIGN

A cross-sectional study of 51 HIV-uninfected (HIV-) and 146 HIV-infected (HIV+) participants.

METHODS

Structural MRI of participants was analyzed using principal component analysis (PCA) to reduce dimensionality and determine topographies of volumetric changes. Neuropsychological (NP) assessment was examined using global and domain-specific scores. The effects of HIV disease factors (eg, viral load, CD4, etc.) on brain volumes and neuropsychological were investigated using penalized regression (LASSO).

RESULTS

Two components of interest were visualized using principal component analysis. An aging effect predominated for both components. The first component, a cortically weighted topography, accounted for a majority of variance across participants (43.5% of variance) and showed independent effects of HIV and smoking. A secondary, subcortically weighted topography (4.6%) showed HIV-status accentuated age-related volume loss. In HIV+ patients, the cortical topography correlated with global neuropsychological scores and nadir CD4, whereas subcortical volume loss was associated with recent viral load.

CONCLUSIONS

Cortical regions showed the most prominent volumetric changes because of aging and HIV. Within HIV+ participants, cortical volumes were associated with immune history, whereas subcortical changes correlated with current immune function. Cognitive function was primarily associated with cortical volume changes. Observed volumetric changes in chronic HIV+ patients may reflect both past infection history and current viral status.

Rat intersubjective decisions are encoded by frequency-specific oscillatory contexts

INTRODUCTION

It is unknown how the brain coordinates decisions to withstand personal costs in order to prevent other individuals' distress. Here we test whether local field potential (LFP) oscillations between brain regions create "neural contexts" that select specific brain functions and encode the outcomes of these types of intersubjective decisions.

METHODS

Rats participated in an "Intersubjective Avoidance Test" (IAT) that tested rats' willingness to enter an innately aversive chamber to prevent another rat from getting shocked. c-Fos immunoreactivity was used to screen for brain regions involved in IAT performance. Multi-site local field potential (LFP) recordings were collected simultaneously and bilaterally from five brain regions implicated in the c-Fos studies while rats made decisions in the IAT. Local field potential recordings were analyzed using an elastic net penalized regression framework.

RESULTS

Rats voluntarily entered an innately aversive chamber to prevent another rat from getting shocked, and c-Fos immunoreactivity in brain regions known to be involved in human empathy-including the anterior cingulate, insula, orbital frontal cortex, and amygdala-correlated with the magnitude of "intersubjective avoidance" each rat displayed. Local field potential recordings revealed that optimal accounts of rats' performance in the task require specific frequencies of LFP oscillations between brain regions in addition to specific frequencies of LFP oscillations within brain regions. Alpha and low gamma coherence between spatially distributed brain regions predicts more intersubjective avoidance, while theta and high gamma coherence between a separate subset of brain regions predicts less intersubjective avoidance. Phase relationship analyses indicated that choice-relevant coherence in the alpha range reflects information passed from the amygdala to cortical structures, while coherence in the theta range reflects information passed in the reverse direction.

CONCLUSION

These results indicate that the frequency-specific "neural context" surrounding brain regions involved in social cognition encodes outcomes of decisions that affect others, above and beyond signals from any set of brain regions in isolation.

Reward-related neural activity and structure predict future substance use in dysregulated youth

BACKGROUND

Identifying youth who may engage in future substance use could facilitate early identification of substance use disorder vulnerability. We aimed to identify biomarkers that predicted future substance use in psychiatrically un-well youth.

METHOD

LASSO regression for variable selection was used to predict substance use 24.3 months after neuroimaging assessment in 73 behaviorally and emotionally dysregulated youth aged 13.9 (s.d. = 2.0) years, 30 female, from three clinical sites in the Longitudinal Assessment of Manic Symptoms (LAMS) study. Predictor variables included neural activity during a reward task, cortical thickness, and clinical and demographic variables.

RESULTS

Future substance use was associated with higher left middle prefrontal cortex activity, lower left ventral anterior insula activity, thicker caudal anterior cingulate cortex, higher depression and lower mania scores, not using antipsychotic medication, more parental stress, older age. This combination of variables explained 60.4% of the variance in future substance use, and accurately classified 83.6%.

CONCLUSIONS

These variables explained a large proportion of the variance, were useful classifiers of future substance use, and showed the value of combining multiple domains to provide a comprehensive understanding of substance use development. This may be a step toward identifying neural measures that can identify future substance use disorder risk, and act as targets for therapeutic interventions.

Inflammatory markers in late pregnancy in association with postpartum depression-A nested case-control study

Recent studies indicate that the immune system adaptation during pregnancy could play a significant role in the pathophysiology of perinatal depression. The aim of this study was to investigate if inflammation markers in a late pregnancy plasma sample can predict the presence of depressive symptoms at eight weeks postpartum. Blood samples from 291 pregnant women (median and IQR for days to delivery, 13 and 7-23days respectively) comprising 63 individuals with postpartum depressive symptoms, as assessed by the Edinburgh postnatal depression scale (EPDS≥12) and/or the Mini International Neuropsychiatric Interview (M.I.N.I.) and 228 controls were analyzed with an inflammation protein panel using multiplex proximity extension assay technology, comprising of 92 inflammation-associated markers. A summary inflammation variable was also calculated. Logistic regression, LASSO and Elastic net analyses were implemented. Forty markers were lower in late pregnancy among women with depressive symptoms postpartum. The difference remained statistically significant for STAM-BP (or otherwise AMSH), AXIN-1, ADA, ST1A1 and IL-10, after Bonferroni correction. The summary inflammation variable was ranked as the second best variable, following personal history of depression, in predicting depressive symptoms postpartum. The protein-level findings for STAM-BP and ST1A1 were validated in relation to methylation status of loci in the respective genes in a different population, using openly available data. This explorative approach revealed differences in late pregnancy levels of inflammation markers between women presenting with depressive symptoms postpartum and controls, previously not described in the literature. Despite the fact that the results do not support the use of a single inflammation marker in late pregnancy for assessing risk of postpartum depression, the use of STAM-BP or the novel notion of a summary inflammation variable developed in this work might be used in combination with other biological markers in the future.

Longitudinal relationships among activity in attention redirection neural circuitry and symptom severity in youth

BACKGROUND

Changes in neural circuitry function may be associated with longitudinal changes in psychiatric symptom severity. Identification of these relationships may aid in elucidating the neural basis of psychiatric symptom evolution over time. We aimed to distinguish these relationships using data from the Longitudinal Assessment of Manic Symptoms (LAMS) cohort.

METHODS

Forty-one youth completed two study visits (mean=21.3 months). Elastic-net regression (Multiple response Gaussian family) identified emotional regulation neural circuitry that changed in association with changes in depression, mania, anxiety, affect lability, and positive mood and energy dysregulation, accounting for clinical and demographic variables.

RESULTS

Non-zero coefficients between change in the above symptom measures and change in activity over the inter-scan interval were identified in right amygdala and left ventrolateral prefrontal cortex. Differing patterns of neural activity change were associated with changes in each of the above symptoms over time. Specifically, from Scan1 to Scan2, worsening affective lability and depression severity were associated with increased right amygdala and left ventrolateral prefrontal cortical activity. Worsening anxiety and positive mood and energy dysregulation were associated with decreased right amygdala and increased left ventrolateral prefrontal cortical activity. Worsening mania was associated with increased right amygdala and decreased left ventrolateral prefrontal cortical activity. These changes in neural activity between scans accounted for 13.6% of the variance; that is 25% of the total explained variance (39.6%) in these measures.

CONCLUSIONS

Distinct neural mechanisms underlie changes in different mood and anxiety symptoms overtime.

Variable selection in discrete survival models including heterogeneity

Several variable selection procedures are available for continuous time-to-event data. However, if time is measured in a discrete way and therefore many ties occur models for continuous time are inadequate. We propose penalized likelihood methods that perform efficient variable selection in discrete survival modeling with explicit modeling of the heterogeneity in the population. The method is based on a combination of ridge and lasso type penalties that are tailored to the case of discrete survival. The performance is studied in simulation studies and an application to the birth of the first child.

A pathway linking reward circuitry, impulsive sensation-seeking and risky decision-making in young adults: identifying neural markers for new interventions

High trait impulsive sensation seeking (ISS) is common in 18-25-year olds, and is associated with risky decision-making and deleterious outcomes. We examined relationships among: activity in reward regions previously associated with ISS during an ISS-relevant context, uncertain reward expectancy (RE), using fMRI; ISS impulsivity and sensation-seeking subcomponents; and risky decision-making in 100, transdiagnostically recruited 18-25-year olds. ISS, anhedonia, anxiety, depression and mania were measured using self-report scales; clinician-administered scales also assessed the latter four. A post-scan risky decision-making task measured 'risky' (possible win/loss/mixed/neutral) fMRI-task versus 'sure thing' stimuli. 'Bias' reflected risky over safe choices. Uncertain RE-related activity in left ventrolateral prefrontal cortex and bilateral ventral striatum was positively associated with an ISS composite score, comprising impulsivity and sensation-seeking-fun-seeking subcomponents (ISSc; P⩽0.001). Bias positively associated with sensation seeking-experience seeking (ES; P=0.003). This relationship was moderated by ISSc (P=0.009): it was evident only in high ISSc individuals. Whole-brain analyses showed a positive relationship between: uncertain RE-related left ventrolateral prefrontal cortical activity and ISSc; uncertain RE-related visual attention and motor preparation neural network activity and ES; and uncertain RE-related dorsal anterior cingulate cortical activity and bias, specifically in high ISSc participants (all ps<0.05, peak-level, family-wise error corrected). We identify an indirect pathway linking greater levels of uncertain RE-related activity in reward, visual attention and motor networks with greater risky decision-making, via positive relationships with impulsivity, fun seeking and ES. These objective neural markers of high ISS can guide new treatment developments for young adults with high levels of this debilitating personality trait.

Application of penalized linear regression methods to the selection of environmental enteropathy biomarkers

BACKGROUND

Environmental Enteropathy (EE) is a subclinical condition caused by constant fecal-oral contamination and resulting in blunting of intestinal villi and intestinal inflammation. Of primary interest in the clinical research is to evaluate the association between non-invasive EE biomarkers and malnutrition in a cohort of Bangladeshi children. The challenges are that the number of biomarkers/covariates is relatively large, and some of them are highly correlated.

METHODS

Many variable selection methods are available in the literature, but which are most appropriate for EE biomarker selection remains unclear. In this study, different variable selection approaches were applied and the performance of these methods was assessed numerically through simulation studies, assuming the correlations among covariates were similar to those in the Bangladesh cohort. The suggested methods from simulations were applied to the Bangladesh cohort to select the most relevant biomarkers for the growth response, and bootstrapping methods were used to evaluate the consistency of selection results.

RESULTS

Through simulation studies, SCAD (Smoothly Clipped Absolute Deviation), Adaptive LASSO (Least Absolute Shrinkage and Selection Operator) and MCP (Minimax Concave Penalty) are the suggested variable selection methods, compared to traditional stepwise regression method. In the Bangladesh data, predictors such as mother weight, height-for-age z-score (HAZ) at week 18, and inflammation markers (Myeloperoxidase (MPO) at week 12 and soluable CD14 at week 18) are informative biomarkers associated with children's growth.

CONCLUSIONS

Penalized linear regression methods are plausible alternatives to traditional variable selection methods, and the suggested methods are applicable to other biomedical studies. The selected early-stage biomarkers offer a potential explanation for the burden of malnutrition problems in low-income countries, allow early identification of infants at risk, and suggest pathways for intervention.

TRIAL REGISTRATION

This study was retrospectively registered with ClinicalTrials.gov, number NCT01375647, on June 3, 2011.

Influence of maternal and socioeconomic factors on breast milk fatty acid composition in urban, low-income families

The lipid composition of breast milk may have a significant impact on early infant growth and cognitive development. Comprehensive breast milk data is lacking from low‐income populations in the Indian subcontinent impeding assessment of deficiencies and limiting development of maternal nutritional interventions. A single breast milk specimen was collected within 6 weeks postpartum from two low‐income maternal cohorts of exclusively breastfed infants, from Dhaka, Bangladesh (n = 683) and Kolkata, India (n = 372) and assayed for percentage composition of 26 fatty acids. Mature milk (>15 days) in Dhaka (n = 99) compared to Kolkata (n = 372) was higher in total saturated fatty acid (SFA; mean 48% vs. 44%) and disproportionately lower in ω3‐polyunsaturated fatty acid (PUFA), hence the ω6‐ and ω3‐PUFA ratio in Dhaka were almost double the value in Kolkata. In both sites, after adjusting for days of lactation, increased maternal education was associated with decreased SFA and PUFA, and increasing birth order or total pregnancies was associated with decreasing ω6‐PUFA or ω3‐PUFA by a factor of 0.95 for each birth and pregnancy. In Dhaka, household prosperity was associated with decreased SFA and PUFA and increased ω6‐ and ω3‐PUFA. Maternal height was associated with increased SFA and PUFA in Kolkata (1% increase per 1 cm), but body mass index showed no independent association with either ratio in either cohort. In summary, the socioeconomic factors of maternal education and household prosperity were associated with breast milk composition, although prosperity may only be important in higher cost of living communities. Associated maternal biological factors were height and infant birth order, but not adiposity. Further study is needed to elucidate the underlying mechanisms of these effects.

Weibull regression with Bayesian variable selection to identify prognostic tumour markers of breast cancer survival

As data-rich medical datasets are becoming routinely collected, there is a growing demand for regression methodology that facilitates variable selection over a large number of predictors. Bayesian variable selection algorithms offer an attractive solution, whereby a sparsity inducing prior allows inclusion of sets of predictors simultaneously, leading to adjusted effect estimates and inference of which covariates are most important. We present a new implementation of Bayesian variable selection, based on a Reversible Jump MCMC algorithm, for survival analysis under the Weibull regression model. A realistic simulation study is presented comparing against an alternative LASSO-based variable selection strategy in datasets of up to 20,000 covariates. Across half the scenarios, our new method achieved identical sensitivity and specificity to the LASSO strategy, and a marginal improvement otherwise. Runtimes were comparable for both approaches, taking approximately a day for 20,000 covariates. Subsequently, we present a real data application in which 119 protein-based markers are explored for association with breast cancer survival in a case cohort of 2287 patients with oestrogen receptor-positive disease. Evidence was found for three independent prognostic tumour markers of survival, one of which is novel. Our new approach demonstrated the best specificity.

Relationship between brain stem volume and aggression in children diagnosed with autism spectrum disorder

BACKGROUND

Aggressive behaviors are common in individuals diagnosed with autism spectrum disorder (ASD) and may be phenotypic indicators of different subtypes within ASD. In current research literature for non-ASD samples, aggression has been linked to several brain structures associated with emotion and behavioral control. However, few if any studies exist investigating brain volume differences in individuals with ASD who have comorbid aggression as indicated by standardized diagnostic and behavioral measures.

METHOD

We examined neuroimaging data from individuals rigorously diagnosed with ASD versus typically developing (TD) controls. We began with data from brain volume regions of interest (ROI) taken from previous literature on aggression including the brainstem, amygdala, orbitofrontal cortex, anterior cingulate cortex, and dorsolateral prefrontal cortex. We defined aggression status using the Irritability subscale of the Aberrant Behavior Checklist and used lasso logistic regression to select among these predictor variables. Brainstem volume was the only variable shown to be a predictor of aggression status.

RESULTS

We found that smaller brainstem volumes are associated with higher odds of being in the high aggression group.

CONCLUSIONS

Understanding brain differences in individuals with ASD who engage in aggressive behavior from those with ASD who do not can inform treatment approaches. Future research should investigate brainstem structure and function in ASD to identify possible mechanisms related to arousal and aggression.

Identification of biomarker-by-treatment interactions in randomized clinical trials with survival outcomes and high-dimensional spaces

Stratified medicine seeks to identify biomarkers or parsimonious gene signatures distinguishing patients that will benefit most from a targeted treatment. We evaluated 12 approaches in high-dimensional Cox models in randomized clinical trials: penalization of the biomarker main effects and biomarker-by-treatment interactions (full-lasso, three kinds of adaptive lasso, ridge+lasso and group-lasso); dimensionality reduction of the main effect matrix via linear combinations (PCA+lasso (where PCA is principal components analysis) or PLS+lasso (where PLS is partial least squares)); penalization of modified covariates or of the arm-specific biomarker effects (two-I model); gradient boosting; and univariate approach with control of multiple testing. We compared these methods via simulations, evaluating their selection abilities in null and alternative scenarios. We varied the number of biomarkers, of nonnull main effects and true biomarker-by-treatment interactions. We also proposed a novel measure evaluating the interaction strength of the developed gene signatures. In the null scenarios, the group-lasso, two-I model, and gradient boosting performed poorly in the presence of nonnull main effects, and performed well in alternative scenarios with also high interaction strength. The adaptive lasso with grouped weights was too conservative. The modified covariates, PCA+lasso, PLS+lasso, and ridge+lasso performed moderately. The full-lasso and adaptive lassos performed well, with the exception of the full-lasso in the presence of only nonnull main effects. The univariate approach performed poorly in alternative scenarios. We also illustrate the methods using gene expression data from 614 breast cancer patients treated with adjuvant chemotherapy.

On the impact of model selection on predictor identification and parameter inference

We assessed the ability of several penalized regression methods for linear and logistic models to identify outcome-associated predictors and the impact of predictor selection on parameter inference for practical sample sizes. We studied effect estimates obtained directly from penalized methods (Algorithm 1), or by refitting selected predictors with standard regression (Algorithm 2). For linear models, penalized linear regression, elastic net, smoothly clipped absolute deviation (SCAD), least angle regression and LASSO had a low false negative (FN) predictor selection rates but false positive (FP) rates above 20 % for all sample and effect sizes. Partial least squares regression had few FPs but many FNs. Only relaxo had low FP and FN rates. For logistic models, LASSO and penalized logistic regression had many FPs and few FNs for all sample and effect sizes. SCAD and adaptive logistic regression had low or moderate FP rates but many FNs. 95 % confidence interval coverage of predictors with null effects was approximately 100 % for Algorithm 1 for all methods, and 95 % for Algorithm 2 for large sample and effect sizes. Coverage was low only for penalized partial least squares (linear regression). For outcome-associated predictors, coverage was close to 95 % for Algorithm 2 for large sample and effect sizes for all methods except penalized partial least squares and penalized logistic regression. Coverage was sub-nominal for Algorithm 1. In conclusion, many methods performed comparably, and while Algorithm 2 is preferred to Algorithm 1 for estimation, it yields valid inference only for large effect and sample sizes.

Importance of rare gene copy number alterations for personalized tumor characterization and survival analysis

It has proven exceedingly difficult to ascertain rare copy number alterations (CNAs) that may have strong effects in individual tumors. We show that a regulatory network inferred from gene expression and gene copy number data of 768 human cancer cell lines can be used to quantify the impact of patient-specific CNAs on survival signature genes. A focused analysis of tumors from six tissues reveals that rare patient-specific gene CNAs often have stronger effects on signature genes than frequent gene CNAs. Further comparison to a related network-based approach shows that the integration of indirectly acting gene CNAs significantly improves the survival analysis.

A simulation based method for assessing the statistical significance of logistic regression models after common variable selection procedures

Classification models can demonstrate apparent prediction accuracy even when there is no underlying relationship between the predictors and the response. Variable selection procedures can lead to false positive variable selections and overestimation of true model performance. A simulation study was conducted using logistic regression with forward stepwise, best subsets, and LASSO variable selection methods with varying total sample sizes (20, 50, 100, 200) and numbers of random noise predictor variables (3, 5, 10, 15, 20, 50). Using our critical values can help reduce needless follow-up on variables having no true association with the outcome.