The ToxCast pipeline: updates to curve-fitting approaches and database structure

Introduction: The US Environmental Protection Agency Toxicity Forecaster (ToxCast) program makes in vitro medium- and high-throughput screening assay data publicly available for prioritization and hazard characterization of thousands of chemicals. The assays employ a variety of technologies to evaluate the effects of chemical exposure on diverse biological targets, from distinct proteins to more complex cellular processes like mitochondrial toxicity, nuclear receptor signaling, immune responses, and developmental toxicity. The ToxCast data pipeline (tcpl) is an open-source R package that stores, manages, curve-fits, and visualizes ToxCast data and populates the linked MySQL Database, invitrodb. Methods: Herein we describe major updates to tcpl and invitrodb to accommodate a new curve-fitting approach. The original tcpl curve-fitting models (constant, Hill, and gain-loss models) have been expanded to include Polynomial 1 (Linear), Polynomial 2 (Quadratic), Power, Exponential 2, Exponential 3, Exponential 4, and Exponential 5 based on BMDExpress and encoded by the R package dependency, tcplfit2. Inclusion of these models impacted invitrodb (beta version v4.0) and tcpl v3 in several ways: (1) long-format storage of generic modeling parameters to permit additional curve-fitting models; (2) updated logic for winning model selection; (3) continuous hit calling logic; and (4) removal of redundant endpoints as a result of bidirectional fitting. Results and discussion: Overall, the hit call and potency estimates were largely consistent between invitrodb v3.5 and 4.0. Tcpl and invitrodb provide a standard for consistent and reproducible curve-fitting and data management for diverse, targeted in vitro assay data with readily available documentation, thus enabling sharing and use of these data in myriad toxicology applications. The software and database updates described herein promote comparability across multiple tiers of data within the US Environmental Protection Agency CompTox Blueprint.

0348 Sleep EEG-Based Brain Age Index is Reduced Under Continuous Positive Airway Pressure Treatment

Introduction

Continuous positive airway pressure (CPAP) is a treatment for apnea. With long-term CPAP, changes in electroencephalogram (EEG) include increased delta power (1 - 4Hz) and sigma power (11 - 15Hz, spindle). However, the short-term EEG response to CPAP in a split-night study is less quantified. We recently developed a “brain age” model using sleep EEG features. The brain age index (BAI) is defined as the difference between chronological age and brain age (BA - CA). Here we first quantify how BAI changes during CPAP in the same patient, and then investigate how much brain age features during the diagnostic part can predict the reduction in apnea-hypopnea index (AHI) during CPAP.

Methods

The dataset consisted of 160 subjects. The average age was 59 years with 53% male, 24% female and 23% unknown. We extracted 480 features including band powers, and then computed the BAIs for both diagnostic and CPAP parts. To predict the reduction in AHI during CPAP, we fit a Bayesian regression model using the brain age features, demographics, and sleep parameters during the diagnostic part, and assessed the feature importance using dominance analysis.

Results

The BAI from the diagnostic part is significantly reduced compared to BAI during CPAP for the same subject (paired t-test, p < 0.01). The diagnostic part has an average BAI 2.24 years; and the CPAP part -4.75 years. The brain age features that are increased during CPAP include sigma powers in N2 and N3. The prediction of AHI reduction has Pearson’s correlation 0.85. The features predictive of reduced AHI are the diagnostic AHI (explained variance 69%), followed by high/low waveforms during N2 (e.g. K-complex, measured by kurtosis) (8.6%), delta power during REM (4.5%) and N1 (2%). The feature predictive of increased AHI is frontal alpha power during quiet awake (2.6%).

Conclusion

The average BAI is reduced during CPAP. BAI provides a novel view of the acute response to CPAP in sleep EEG. Future study with more CPAP failure patients has the potential of predicting CPAP failure.

Support

MBW is supported by Glenn Foundation for Medical Research. RJT is supported by Category I AASM Foundation.

0758 Quantification of Late REM Periods in Patients With Prolonged Sleep Duration

Introduction

Evaluation of hypersomnia includes polysomnography followed by mean sleep latency testing (MSLT). As consistent with guidelines as applied in most centers, the overnight portion of the study will be terminated to begin sleep latency testing. For patients with prolonged sleep duration, this interruption could result in REM sleep on nap testing that reflects continuation of their biological night, rather than abnormalities in REM sleep pressure/regulation.

Methods

We reviewed 42 consecutive extended (unrestricted) sleep studies for patients with a total sleep time greater than 600 minutes. For studies with sleep onset before midnight, we evaluated for REM period onset after 6AM, the number of REM periods after 6AM and 8AM, and the time of the final REM period onset.

Results

42 hypnograms were reviewed for patients undergoing evaluation of hypersomnia, median age 32 years (range 19-92) with a median total sleep time of 663 minutes (range 602-832), of these 28/42 (67%) had sleep onset before midnight (12 AM) and were included in the analysis. 27/28 (96%) of hypnograms reviewed had REM sleep after 6 AM, 24/28 (86%) had REM sleep after 8 AM, with the onset of the final REM period ranging from 4:46 AM-12:30 PM for patients with sleep onset time before midnight (12 AM).

Conclusion

These data suggest that termination of overnight polysomnography to complete mean sleep latency testing, as is standard in most sleep labs, may influence the presence of REM sleep on MSLT for patients with prolonged total sleep duration. These results may have implications for the interpretation of MSLT for patients with long sleep duration, and may explain why a given individual may test as type II narcolepsy or idiopathic hypersomnia unpredictably on repeat testing.

Support

Sleep Medicine Fellowship at BIDMC

A new method to analyze dGEMRIC measurements in femoroacetabular impingement: preliminary validation against arthroscopic findings

OBJECTIVE

To validate a new method to analyze delayed gadolinium-enhanced magnetic resonance imaging of cartilage (dGEMRIC) measurements in the hip for early assessment of cartilage defects in femoroacetabular impingement (FAI).

METHODS

We performed a retrospective review of 10 hips in 10 FAI patients, who underwent hip arthroscopy. T1-weighted images and dGEMRIC T(1) maps were acquired at 1.5 T on coronal planes, including the anterior-superior, superior, posterior-superior hip cartilage. For all slices, a region of interest (ROI) was defined over the central portion of the femoral cartilage, assumed to be healthy, and T1 values (x) were transformed to standard scores (z) using z = (x -μ)/σ, where μ and σ are the average and standard deviation of T1 in the femoral ROI. Diagnostic performance of the resulting standardized dGEMRIC maps was evaluated against intraoperative findings and compared with that of a previously proposed dGEMRIC analysis as well as morphologic assessment.

RESULTS

Assuming z = -2 or z = -3 as the threshold between normal and degenerated cartilage, sensitivity, specificity and accuracy were 88%, 51% and 62%, and 71%, 63% and 65%, respectively. By using T1 = 500 ms as single threshold for all dGEMRIC T1 maps, these values became 47%, 58% and 55%, whereas they were 47%, 79% and 70% for morphologic evaluation.

CONCLUSIONS

Standardized dGEMRIC can increase the sensitivity in detecting abnormal cartilage in FAI and has the potential to improve the clinical interpretation of dGEMRIC measurements in FAI, by removing the effect of inter- and intra-patient T1 variability.

Factor structure of the Grief Experience Questionnaire (GEQ)

The Grief Experience Questionnaire (GEQ; T.W. Barrett & T.B. Scott, 1989) is a self-report measure of grief responses, including some that have been associated with grief after a suicide (e.g., feelings of rejection, responsibility, shame, stigmatization, etc.). In this study, a sample of 350 university students who had experienced the death of a significant other completed the GEQ. A principal components analysis with varimax rotation yielded an 8-factor solution with satisfactory psychometric properties. Results clearly document that the GEQ is a multidimensional measure of grief phenomenology. It is concluded that although the GEQ has broad applicability, the scale as revised herein may have a special relevance to suicide bereavement, and may be of use in both research and clinically based applications.

Survivors of suicide do grieve differently: empirical support for a common sense proposition

Previous empirical investigations have produced mixed results on the question of whether mode of death differentially affects grief. To further investigate the influence of suicide on grief, 350 previously bereaved university students completed a questionnaire package consisting of several standardized measures. Participants were separated into four groups based on the mode of death experienced as either survivors of suicide (n = 34), accident (n = 57), unanticipated natural (n = 102), or anticipated natural (n = 157) deaths. Hierarchical multiple regression analyses indicated that suicide survivors, compared against the other groups, experienced more frequent feelings of rejection, responsibility, "unique" reactions, and more total grief reactions. Trends indicating increased levels of shame and perceived stigmatization were also evident. Aggregate factors of death "naturalness" and "expectedness" showed less influence than mode of death in influencing grief. Overall, results support previous clinical and research findings and intuitive logic in demonstrating that the grief experienced by suicide survivors includes elements that are less frequently seen in the case of nonsuicidal deaths.

Using functional health patterns to predict outcome with seniors

Two powerful forces in Canada's health care system today are the rapidly growing seniors population and the trend toward community care. Recently, five hospitals in southwestern Ontario initiated a program designed to avert the admission of seniors to hospital and provide them with supports at home to maintain their health. Simultaneous research was conducted to assess the impact and effectiveness of the program.