Lasso regularization for left-censored Gaussian outcome and high-dimensional predictors

Background

Biological assays for the quantification of markers may suffer from a lack of sensitivity and thus from an analytical detection limit. This is the case of human immunodeficiency virus (HIV) viral load. Below this threshold the exact value is unknown and values are consequently left-censored. Statistical methods have been proposed to deal with left-censoring but few are adapted in the context of high-dimensional data.

Methods

We propose to reverse the Buckley-James least squares algorithm to handle left-censored data enhanced with a Lasso regularization to accommodate high-dimensional predictors. We present a Lasso-regularized Buckley-James least squares method with both non-parametric imputation using Kaplan-Meier and parametric imputation based on the Gaussian distribution, which is typically assumed for HIV viral load data after logarithmic transformation. Cross-validation for parameter-tuning is based on an appropriate loss function that takes into account the different contributions of censored and uncensored observations. We specify how these techniques can be easily implemented using available R packages. The Lasso-regularized Buckley-James least square method was compared to simple imputation strategies to predict the response to antiretroviral therapy measured by HIV viral load according to the HIV genotypic mutations. We used a dataset composed of several clinical trials and cohorts from the Forum for Collaborative HIV Research (HIV Med. 2008;7:27-40). The proposed methods were also assessed on simulated data mimicking the observed data.

Results

Approaches accounting for left-censoring outperformed simple imputation methods in a high-dimensional setting. The Gaussian Buckley-James method with cross-validation based on the appropriate loss function showed the lowest prediction error on simulated data and, using real data, the most valid results according to the current literature on HIV mutations.

Conclusions

The proposed approach deals with high-dimensional predictors and left-censored outcomes and has shown its interest for predicting HIV viral load according to HIV mutations.

     

Cerebral-cortical networking and activation increase as a function of cognitive-motor task difficulty

Excessive increases in task difficulty typically result in marked attenuation of cognitive-motor performance. The psychomotor efficiency hypothesis suggests that poor performance is mediated by non-essential neural activity and cerebral cortical networking (inefficient cortical dynamics). This phenomenon may underlie the inverse relationship between excessive task difficulty and performance. However, investigation of the psychomotor efficiency hypothesis as it relates to task difficulty has not been conducted. The present study used electroencephalography (EEG) to examine cerebral cortical dynamics while participants were challenged with both Easy and Hard conditions during a cognitive-motor task (Tetris^®). In accord with the psychomotor efficiency hypothesis, it was predicted that with increases in task difficulty, participants would demonstrate greater ‘neural effort,’ as indexed by EEG spectral power and cortical networking (i.e., EEG coherence) between the premotor (motor planning) region and sensory, executive, and motor regions. Increases in neural activation and cortical networking were observed during the Hard condition relative to the Easy condition, thus supporting the psychomotor efficiency hypothesis. To further determine the unique contributions of cognitive versus sensory-motor demands, a control experiment was conducted in which cognitive demand was increased while sensory-motor demand was held constant. This experiment revealed that regionally specific neural activation was influenced by changes in cognitive demand, whereas cortical networking to the motor planning region was sensitive only to changes in sensory-motor demand. Crucially, the present study is the first, to our knowledge, to characterize the separate impact of cognitive versus sensory-motor demands on cerebral cortical dynamics. The findings further inform the dynamics of the cortical processes that underlie the quality of cognitive-motor performance particularly with regard to task difficulty. A broader understanding of the brain and muscle interactions during varying levels of challenge may inform the design of effective training protocols aimed at optimizing cognitive-motor performance.     

A Multimedia Mobile Phone–Based Youth Smoking Cessation Intervention: Findings From Content Development and Piloting Studies

Background

While most young people who smoke want to quit, few access cessation support services. Mobile phone–based cessation programs are ideal for young people: mobile phones are the most common means of peer communication, and messages can be delivered in an anonymous manner, anywhere, anytime. Following the success of our text messaging smoking cessation program, we developed an innovative multimedia mobile phone smoking cessation intervention.

Objective

The aim of the study was to develop and pilot test a youth-oriented multimedia smoking cessation intervention delivered solely by mobile phone.

Methods

Development included creating content and building the technology platform. Content development was overseen by an expert group who advised on youth development principles, observational learning (from social cognitive theory), effective smoking cessation interventions, and social marketing. Young people participated in three content development phases (consultation via focus groups and an online survey, content pre-testing, and selection of role models). Video and text messages were then developed, incorporating the findings from this research. Information technology systems were established to support the delivery of the multimedia messages by mobile phone. A pilot study using an abbreviated 4-week program of video and text content tested the reliability of the systems and the acceptability of the intervention.

Results

Approximately 180 young people participated in the consultation phase. There was a high priority placed on music for relaxation (75%) and an interest in interacting with others in the program (40% would read messages, 36% would read a blog). Findings from the pre-testing phase (n = 41) included the importance of selecting “real” and “honest” role models with believable stories, and an interest in animations (37%). Of the 15 participants who took part in the pilot study, 13 (87%) were available for follow-up interviews at 4 weeks: 12 participants liked the program or liked it most of the time and found the role model to be believable; 7 liked the role model video messages (5 were unsure); 8 used the extra assistance for cravings; and 9 were happy with two messages per day. Nine participants (60%) stopped smoking during the program. Some technical challenges were encountered during the pilot study.

Conclusions

A multimedia mobile phone smoking cessation program is technically feasible, and the content developed is appropriate for this medium and is acceptable to our target population. These results have informed the design of a 6-month intervention currently being evaluated for its effectiveness in increasing smoking cessation rates in young people.     

Rapid screening and confirmatory methods for biochemical diagnosis of human prion disease

Transmissible spongiform encephalopathies (TSEs) are characterised by accumulation of an abnormal isoform of prion protein (PrP^sc), mainly in the brain but also in various peripheral tissues. Home-made assays consisting of non-standardised protocols are used currently for laboratory diagnosis of human TSE. The purpose of the present study was to test the ability of two commercial assays, TeSeE™ CJD ELISA and TeSeE™ Western blot, to detect PrPsc in cerebral and lymphoid tissues of TSE patients. Both tests detected a PrPsc-significant signal in the brains of 54 affected patients and not in 51 controls, yielding 100% specificity and 100% sensitivity. Furthermore, three post-mortem spleens and two pre-mortem tonsils from three patients with variant Creutzfeldt-Jakob disease (vCJD) were detected correctly. The expected PrPsc molecular patterns were found in TSE patient brain tissue and in the tonsils and spleens of the three vCJD patients. In conclusion, these rapid and robust in vitro tools were suitable for routine human TSE diagnosis and characterisation. CJD could also be diagnosed during the patient's lifetime by detection of PrPsc in the tonsil. A diagnostic strategy associating TeSeE™ CJD ELISA screening to biochemical confirmation by TeSeE™ Western blot is proposed.     

CMV plasma DNAemia and risk of cancer among HIV-infected patients: A case–control study nested in the ANRS CO3 Aquitaine Cohort,France, 2002–2007

BackgroundCMV reactivation, which enhances immune senescence, could be associated with a higher risk of cancer.ObjectivesWe compared the prevalence of positive CMV DNAemia in HIV-infected patients with and without cancer.Study designThis case–control study, nested in the ANRS-CO3 Aquitaine Cohort, included patients with a first diagnosis of cancer (2002–2007) as cases. Two controls were matched per case.Cancer risk was estimated using conditional logistic regression models, an Odds Ratio (OR) of 2 could be detected with 80% power. The variables considered were: ≥1 positive CMV DNAemia, CD4+ and CD8+ counts, HIV plasma load. Plasma CMV DNA was retrospectively quantified within the 3-year period preceding the endpoint.ResultsThe 143 cases (93 non-AIDS-related and 50 AIDS-related cancers) and 284 controls had a median age of 47 years (IQR: 41–56). At the time of diagnosis or censorship, for cases and controls, median values were respectively, for CD4+ count: 327 cells/mm³ (IQR: 164–514) and 416 (IQR: 275–582), and for HIV plasma load: 2.6 log₁₀ copies/mL (IQR: 1.7–4.7) and 1.7 log₁₀ copies/mL (IQR: 1.7–3.3). We performed 2056 CMV PCR; 14 cases (9.8% [95% CI: 4.9–14.7]) and 19 controls (6.7% [CI: 3.8–9.6]) presented ≥1 positive PCR. CMV DNAemia was not associated with the risk of cancer (unadjusted and adjusted p-values = 0.19 and 0.54, respectively). HIV load >500 copies/mL was independently associated with a higher risk of cancer (OR = 2.02; p = 0.002; 95% CI: 1.29–3.17).ConclusionThis large case–control study did not show any differential exposure to positive CMV plasma DNAemia between cancer cases and controls.