Identification of biomarkers associated with specific sleep variables

Lead Investigator Robert Alexander, Takeda
Title of Proposed Research Identification of biomarkers associated with specific sleep variables
Vivli Data Request: 3287
Funding Source: None
Potential Conflicts of Interest: None
Summary of the Proposed Research
We want to explore whether a simple blood test can be used to predict objective measures of insomnia. Research subjects are frequently not accurate in self-reporting whether they have insomnia. As a result, studies that test novel hypnotics have to screen many subjects using expensive technology (polysomnography or PSG) in order identify appropriate study candidates with objective evidence of decreased sleep. PSG measures a subject’s brain waves, the oxygen level in the blood, heart rate and breathing, as well as eye and leg movements during the night. If blood-based biomarkers are associated with specific sleep variables, such as initial sleep onset or wake after sleep onset, they could be used in the screening process of clinical trials to more efficiently identify prospective study candidates and hasten the development of therapies for insomnia. Furthermore, understanding the relationships between blood biomarkers and sleep variables could shed light on biological mechanisms of insomnia.

Statistical Analysis Plan
Variable will be summarized by mean, median, min, max, and standard deviation. Univariate analysis of association between mean WASO and each blood sample cytokine will be performed as an initial look at the data and may be used for feature selection in machine learning models. A correlation heatmap of all blood sample predictors and all PSG outcomes will be produced. Machine learning methods such as random forests with cross validation will be used to assess ability to predict WASO and other PSG variables using blood samples and basic demographic variables. Model performance will be assessed using root mean squared error (RMSE).

Requested Studies:
A 28-Day, Polysomnographic and Subjective Assessment of Vestipitant (15mg/Day) for the Treatment of Primary Insomnia in Adult Outpatients
Sponsor: GlaxoSmithKline
Study ID: NCT00992160
Sponsor ID: 111364

Summary of Results
The protocol states, “Exploratory analyses may be additionally conducted by applying multi-analyte profiling techniques, that enables the simultaneous detection of hundreds of proteins with high sensitivity and accuracy to assess biomarkers (or clusters of biomarkers) that correlate with PSG or response outcome.” However, in the clinical data we received there is only data on four cytokines: IFN-Gamma, IL-10, IL-6, and TNF. In examining the relationship between polysomnography and these four cytokines, we did not find any significant associations and do not believe we have publishable results. We were disappointed in both the number of potential blood biomarkers in the dataset and in the results of our analysis; but we appreciate the work of Vivli and were glad to have the opportunity to explore the available data.