NOTE: Extended to 7/31/2019 per NSSC information (Ed., 8/7/18)

NOTE: Extended to 7/31/2018 per K. Ohnesorge/JSC (Ed., 7/14/17)

NOTE: Element change to Human Factors & Behavioral Performance; previously Behavioral Health & Performance (Ed., 1/18/17)

NOTE: End date shows as 7/31/2017 per NSSC information (Ed., 6/2/15)

In Phase 1 of this project, we conducted a comprehensive and standardized analysis of individualized neurocognitive and psychological responses to sleep loss and circadian desynchrony from existing laboratory data in over 150 subjects and more than a thousand bed-days incorporating multiple measures of neurocognitive performance. These analyses have confirmed the best candidates for predicting neurocognitive impairment in simulated sleep and circadian challenges, common in space missions that are studied routinely in our laboratory.

In Phase 2 of this project, we have conducted a series of short (6-day), targeted laboratory studies to test prospectively the predictive power of selected biomarkers and to measure novel biomarkers absent from our original data sets in order to inform operational analog studies testing the predictive value of these biomarkers in operational conditions. Our innovative goal is to generate sets of biomarker panels from lipidomic and/or metabolomic biomarker profiles to identify and predict interindividual differences in performance under acute sleep deprivation and circadian misalignment.

In Phase 3 of this project, we aimed to conduct a pilot study testing the feasibility and utility of implementing the predictive biomarkers for neurocognitive impairment in Antarctica as a high fidelity space analog.

Phase 2. We completed data collection for the laboratory study outlined for Phase 2 of this project. We consented and screened a total of 60 potential candidates. Forty-five were excluded due to not meeting study inclusion/exclusion criteria and we have completed studies on 15 individuals. Based on our Phase 1 analysis, participants were screened for performance prior to admission and excluded if they had 2 or more lapses on at least 2 screening test sessions. Among the 15 participants who completed the study, analysis of their performance data indicates that 11 participants were resilient, 3 participants were intermediate, and 1 participant was vulnerable at 20 hours awake during the 50-hour constant routine.

Each participant was studied for 6 days in the laboratory. During days 1-3, participants completed baseline measures under stable well-rested conditions (8 h sleep per night) to provide individual control data for assessing the degree of impairment due to sleep loss or circadian desynchrony. On days 4-6, participants underwent 50 hours of acute sleep deprivation under constant routine conditions in order to assess inter-individual responses to acute sleep deprivation and being awake at an adverse circadian phase. Participants were allowed a final 10-hour recovery sleep before discharge. Throughout the 6-day study, we collected tests of performance and alertness every 1-2 hours while awake. Blood samples were taken at 60-minute intervals, and saliva was collected every hour during wake as backup for the plasma samples. Our initial analyses have focused on every 4th sample (every 4 hours) from Day 2 (to document the diurnal pattern of the biomarkers under well-rested conditions) through the end of the constant routine (CR). We completed lipidomics and metabolomics assays in the four-hourly samples in 12 participants (n=6F, 35.8 ± 8.1 yrs).

Biomarker candidate characterization for circadian phase.

Lipids. The number of unique lipid species that were identified in males and females were 335 and 351, respectively, which included 4 lipid classes. There was evidence of robust endogenous circadian regulation of the plasma lipidome with approximately 35% and 19% being rhythmic under ambulatory and constant routine conditions, respectively, in males and 24% and 27% in females. Triglycerides (TG) were the most abundant subclass of lipids that were rhythmic in both sexes under both behavioral states. Given that no lipids were circadian in all 12 participants, we lowered the threshold to include at least 50% of the population to being rhythmic for each lipid, and identified 29 unique lipids in the males and 76 in the females. Of these, only 19 unique lipids were circadian in both males and females. Taken together, these results suggest large inter-individual and sex-specific differences in the circadian regulation of the plasma lipidome. Identifying lipidomics biomarkers of circadian phase may require developing models that account for potential sex differences.

Metabolites. The number of plasma metabolites that were chromatographically identified in males and females were 117 and 124, respectively. There was evidence of robust endogenous circadian regulation of plasma metabolites in both males and females. Approximately 36% and 25% were rhythmic under ambulatory and constant routine conditions, respectively, in males and 33% and 23%, respectively in females. Of these, only 6 unique metabolites were circadian in both males and females. As for lipids, there are large inter-individual and sex-specific differences in the circadian regulation of plasma metabolites. The majority (~60%) of the metabolites remained unchanged between baseline and constant routine regardless of whether they were rhythmic or not.

Biomarker candidate characterization for time awake (sleepiness).

Lipids. Similar to circadian regulation, there was evidence of extensive inter-individual variability in linear changes. In the males, approximately 5% and 32% were linear under baseline and constant routine conditions, respectively, with 6% and 34% for the females. The highest percentage of linear lipid profiles were seen in the TG subclass in both males and females, approximately 44%. The Lysophospholipid (LPC) subclass had a higher percent change in females (~45%) compared to males (~17%), highlights potential sex differences in the lipids that change linearly with time awake. Of the lipid features that changed linearly, approximately 50% of them increased in the males during the 16-h wake episode under ambulatory conditions versus ~70% increased in females. Under CR conditions, ~60% increased in males and ~67% in females.

Metabolites. As in the lipids, the proportion of metabolites that were linear during baseline and CR conditions were similar for males (6% and 32%) and females (6% and 35%, respectively), although the individual candidates were not the same: Of the 80 metabolites that were linear between males and females during CR in at least 50% of the population, only 10 were shared by males and females. The direction of change was similar between males and females with ~70% decreasing during baseline and 55% decreasing during constant routine.

Predictive biomarker models.

Prediction of endogenous circadian phase and phase angle of entrainment. A preliminary predictive model was developed using the concentrations of lipid species that were (1) rhythmic under both ambulatory and CR conditions; (2) were rhythmic under CR in at least 2 participants; and (3) the phase of the lipid rhythm did not differ by more than 5 hours between individuals. The number of unique lipid species that met these criteria across all 12 participants was 52 and all were included in developing the predictive model. Partial Least Squares Regression (PLSR) yielded a predictive model that was able to use the lipid concentrations measured in a single blood sample collected ~4 h after waking on the day to predict endogenous circadian phase and phase angle of entrainment to be within, on average, ~20 and ~42 minutes of actual, respectively. While these analyses are preliminary, they are encouraging in that they demonstrate reasonable accuracy in predicting circadian phase, albeit the data are likely overfit. We are currently evaluating the model using additional time points, and fewer lipid features to further test the robustness of these predictions.

Prediction of neurobehavioral performance impairment due to sleep loss. The number of unique lipid species common across all 12 participants initially included in the preliminary predictive model for sleepiness was 289. Five of the 12 participants were identified as vulnerable to sleep loss at 40 hours awake on CR; the remaining participants were identified as resilient. All participants were high-level performers during ambulatory baseline, which was defined as exhibiting only 0 or 1 lapse in attention on the PVT at 8 hours awake following an 8-hour time-in-bed opportunity. Supervised learning methods yielded predictive models that were able to correctly discriminate between resilient and vulnerable individuals using just 1-2 lipids measured at a single time point. Specifically, both k-nearest neighbor and naïve Bayes classifiers trained on lipidomic data at 8 hours awake on CR were able to discriminate resilient from vulnerable individuals with accuracies ranging from 83%-100% when tested on data collected at 8 hours awake two days earlier under ambulatory baseline. We are currently developing additional predictive models at additional time points, including predictions for less extreme acute sleep loss (e.g., 24 hours). We are also developing similar models using metabolomics data.

Phase 3:

We conducted similar analyses to those that we conducted in the Phase 1 analysis to identify whether baseline performance alone could accurately discriminate response to sleep loss among individuals overwintering in Antarctica. Unlike the laboratory data in which participants were kept awake for up to 50 hours, however, few individuals overwintering in Antarctica in our previously collected dataset reported wakefulness beyond 18 hours, and therefore our analysis was limited to predicting response to sleep loss >16 hours awake only. As was observed in the Phase 1 analysis, although baseline performance could identify high-level from poor performers at baseline, we could not further discriminate the response to sleep loss phenotype among high-level performers from baseline performance alone; we found that we were able to discriminate the response phenotype at >16 hours of wakefulness from baseline performance with an overall accuracy of ~64%.

NOTE: Extended to 7/31/2019 per NSSC information (Ed., 8/7/18)

NOTE: Extended to 7/31/2018 per K. Ohnesorge/JSC (Ed., 7/14/17)

NOTE: Element change to Human Factors & Behavioral Performance; previously Behavioral Health & Performance (Ed., 1/18/17)

NOTE: End date shows as 7/31/2017 per NSSC information (Ed., 6/2/15)

In this project, we will first identify candidate biomarkers in a retrospective analysis from neurocognitive and psychological measures collected over the past 17 years in our laboratory analog. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new space analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.

We have completed Phase 1 of the project, namely the development of a preliminary candidate biomarker panel that can predict an individual’s response to sleep loss phenotype (i.e., resilient, intermediate, or vulnerable, as defined by performance at 20 hours of wakefulness) from neurobehavioral performance collected in the individual at baseline after a habitual (8 hour) night of sleep 2 days earlier. We have compiled and analyzed retrospective performance and alertness data from over 160 subjects (over 1400 days of intensive physiological monitoring) who were studied under identical laboratory conditions. The first goal of our Phase 1 analysis was to develop a metric to define response to sleep loss phenotype within an individual. Prior definitions of response to sleep loss have been based almost exclusively on performance on the psychomotor vigilance task (PVT). Thus, we developed other metrics incorporating data from several neurobehavioral tests, including the Karolinska Sleepiness Scale (KSS), non-numeric bipolar Visual Analog Scales (VAS), auditory PVT (aPVT), 4-minute addition calculation test (ADD), 1.5-minute digit symbol substitution test (DSST), and 10-minute auditory and visual versions of the PVT, to determine whether data from other neurobehavioral performance tasks provide additional information about response types.

We explored four metrics for defining the response to sleep loss phenotype in these individuals: a metric based on visual PVT lapses only (metric 1); a metric derived from principal component analysis of 74 variables from all 6 neurobehavioral tests (metric 2); a metric derived from principal component analysis of 8 representative variables from all 6 neurobehavioral tests (metric 3); and a metric derived from principal component analysis of 12 representative variables from all neurobehavioral tests excluding the two versions of the PVT (metric 4).

Comparison of the metrics showed that similar information is obtained from the three metrics that include data from the PVT (metrics 1-3) compared with data from other neurobehavioral tasks (metric 4). Therefore, we determined that metric 1, using lapses on the visual PVT, was sufficient for defining response to sleep loss phenotype. The final definitions of each phenotype are as follows: resilient (fewer than 6 lapses), intermediate (between 6 and 30 lapses), and vulnerable (more than 30 lapses). This metric can discriminate ~50% of our subject population (studied under inpatient laboratory conditions) as non-resilient to sleep loss (i.e., intermediate or vulnerable response to sleep loss) at ~20 h awake from one neurobehavioral performance assessment during rested conditions (i.e., at ~8 hours of wake after a habitual 8-hour night of sleep) with ~100% accuracy. Our findings suggest that these individuals who are poor performers during rested conditions exhibit non-sleepiness-related performance impairment that is exacerbated by subsequent sleep loss. Among the remaining ~50% of our subject population, who are high-level performers during rested conditions, the distribution of response to sleep loss phenotype is expected to be approximately equal across resilient, intermediate, and vulnerable phenotypes. A model developed to discriminate response to sleep loss phenotype at 20 h awake in these high-level performers from a single neurobehavioral performance assessment 2 days earlier during rested conditions has an overall accuracy of ~64%, suggesting that neurobehavioral performance alone is insufficient to discriminate response to sleep loss phenotype in these individuals.

Our metric that discriminates high-level from poor performers during rested conditions has been implemented as a screening tool to select participants for our 6-day Phase 2 study; only high-level performers have been included as participants in our study. These high-level performers who also met all our other inclusion/exclusion criteria have been enrolled in our 6-day laboratory-based study for Phase 2 of this project. In this laboratory study, we have collected multiple biomarker samples during baseline and sleep deprivation to determine whether there are one or more “omics”-based measures that provide a greater sensitivity and specificity to identify individuals who are resilient versus non-resilient to sleepiness-related performance impairment. We started recruiting for this study in January 2016, and have consented and screened 60 potential candidates in total. Forty-five were excluded due to not meeting study inclusion/exclusion criteria, including 8 participants who were identified as poor performers from the 35 participants who completed our initial screening test battery. We have completed studies on 15 individuals. Enrollment in this study is now complete.

We have conducted analyses of lipidomic and metabolomic biomarker profiles in a subset of participants who have completed the study. Specifically, lipidomic analysis of blood samples collected every 4 hours during both baseline (days 2-3) and constant routine (days 4-5) has been completed in 14 subjects (6 females). Lipids were extracted from human plasma using dichloromethane:methanol. The lipid extracts were then analyzed by reversed-phase liquid chromatography-mass spectrometry (LC-MS). Full-scan MS analysis was done on a stand-alone orbitrap mass spectrometer (Exactive, Thermo Scientific) in both the positive and negative ionization modes. Chromatographic alignment and framing to select features was performed using SIEVE 2.0. Features that represented lipids were then identified using an in-house lipid database. Metabolomic analysis has also been conducted on these blood samples using a CoulArray platform, and we are currently in the process of identifying metabolite peaks for further analysis.

Initial analysis of lipidomics data has identified 340 unique lipids, ~42% of which exhibit a rhythm during rested conditions and ~28% of which exhibit a rhythm during constant routine when subjects were sleep deprived. In the coming year, we will continue to analyze these lipidomic and metabolomic markers at both the group and individual level to determine a candidate biomarker panel that identifies response to sleep loss phenotype.

In preparation for Phase 3, we have compiled the data from over 50 expeditioners who were studied during the Antarctic winter. These subjects were instructed to complete the ANAM-ICE performance battery approximately once per month several times per day for 1-2 days. Our analysis was conducted primarily on 28 subjects from this cohort who had completed a test battery both at baseline (defined as a test session completed during the first month of the overwintering period after a night of at least 7 hours of sleep between the first 4 to 10 hours awake) and after at least 16 hours of wakefulness during a subsequent month of testing. As in Phase 1, we have started to test different metrics from the ANAM-ICE performance battery to define “resilient,” “intermediate,” or “vulnerable” response phenotypes after more than 16 hours of wakefulness. One metric that we have developed is based on performance on the simple reaction time test of the ANAM-ICE battery. We first normalized each subject’s response based on mean and standard deviation of normative data from each subject’s age group [Vincent et al., 2012]. Then we computed the 25th and 75th percentiles of these normalized values across all subjects across all test sessions. Scores in the top 25th percentile were classified as resilient, scores in the bottom 25th percentile were classified as vulnerable, and scores between the 25th and 75th percentile were classified as intermediate. We then compared the response at baseline to responses from test sessions at >16 hours of wakefulness. We found that we were able to discriminate the response phenotype at >16 hours of wakefulness from baseline performance with an overall accuracy of ~65%. We are continuing to develop this model by incorporating information from other neurobehavioral tests at baseline to improve its overall accuracy. We plan to prospectively apply this model to baseline data collected from expeditioners who will be overwintering in Antarctica.

Reference:

Vincent, A.S., et al., Automated Neuropsychological Assessment Metrics (v4) Traumatic Brain Injury Battery: military normative data. Military Medicine, 2012. 177(3): p. 256-69.

NOTE: Extended to 7/31/2018 per K. Ohnesorge/JSC (Ed., 7/14/17)

NOTE: Element change to Human Factors & Behavioral Performance; previously Behavioral Health & Performance (Ed., 1/18/17)

NOTE: End date shows as 7/31/2017 per NSSC information (Ed., 6/2/15)

In this project, we will first identify candidate biomarkers in a retrospective analysis from neurocognitive and psychological measures collected in more than 1,500 subjects that have been studied over the past 17 years in our laboratory analog and from existing extensive longitudinal data in over 50 individuals overwintering in Antarctica, which is a high fidelity space analog. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new space analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.

In the first four years of this grant, we have completed the development of a preliminary candidate biomarker panel that can predict an individual’s response to sleep loss phenotype (i.e., resilient, intermediate, or vulnerable, as defined by performance at 20 hours of wakefulness) from neurobehavioral performance collected in the individual at baseline after a habitual (8 hour) night of sleep 2 days earlier. From our analysis, we have identified a metric that can discriminate ~50% of our subject population (studied under inpatient laboratory conditions) as non-resilient to sleep loss (i.e., intermediate or vulnerable response to sleep loss) at ~20 h awake from one neurobehavioral performance assessment during rested conditions (i.e., at ~8 hours of wake after a habitual 8-hour night of sleep) with ~100% accuracy. Our findings suggest that these individuals who are poor performers during rested conditions exhibit non-sleepiness-related performance impairment that is exacerbated by subsequent sleep loss. Among the remaining ~50% of our subject population, who are high-level performers during rested conditions, the distribution of response to sleep loss phenotype is expected to be approximately equal across resilient, intermediate, and vulnerable phenotypes. A model developed to discriminate response to sleep loss phenotype at 20 h awake in these high-level performers from a single neurobehavioral performance assessment 2 days earlier during rested conditions has an overall accuracy of ~64%, suggesting that neurobehavioral performance alone is insufficient to discriminate response to sleep loss phenotype in these individuals.

Our metric that discriminates high-level from poor performers during rested conditions has been implemented as a screening tool to select participants for our 6-day Phase 2 study; only high-level performers have been included as participants in our study. These high-level performers who also met all our other inclusion/exclusion criteria have been enrolled in our 6-day laboratory-based study for Phase 2 of this project. In this laboratory study, we are collecting multiple biomarker samples during baseline and sleep deprivation to determine whether there are one or more “omics”-based measures that provide a greater sensitivity and specificity to identify individuals who are resilient versus non-resilient to sleepiness-related performance impairment. We started recruiting for this study in January 2016, and have consented and screened 60 potential candidates in total. Forty-five were excluded due to not meeting study inclusion/exclusion criteria, including 8 participants who were identified as poor performers from the 35 participants who completed our initial screening test battery. To date, we have completed studies on 15 individuals. Enrollment in this study is now complete. Initial analysis of lipidomics data has identified 340 unique lipids, ~42% of which exhibit a rhythm during rested conditions and ~28% of which exhibit a rhythm during constant routine when subjects were sleep deprived. We expect to complete lipidomic and metabolomic analysis on all subjects in the next year of the project in support of Phase 2 of this study.

As part of our Phase 1 analysis, we also have conducted an analysis of retrospective neurobehavioral performance data in expeditioners overwintering in Antarctica. We have developed a definition of response to sleep loss phenotype in this population based on simple reaction time performance at >16 hours of wakefulness, and developed a model to predict this phenotype from baseline neurobehavioral performance data collected 1-6 months earlier. Our current model predicts response to sleep loss with an overall accuracy of 65%. We are continuing to develop this model by incorporating information from other neurobehavioral tests at baseline to improve its overall accuracy. We plan to prospectively apply this model to baseline data collected from expeditioners who will be overwintering in Antarctica.

NOTE: Element change to Human Factors & Behavioral Performance; previously Behavioral Health & Performance (Ed., 1/18/17)

NOTE: End date shows as 7/31/2017 per NSSC information (Ed., 6/2/15)

In this project, we will first identify candidate biomarkers in a retrospective analysis from neurocognitive and psychological measures collected in more than 1,500 subjects that have been studied over the past 17 years in our laboratory analog and from existing extensive longitudinal data in over 50 individuals overwintering in Antarctica, which is a high fidelity space analog. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new space analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.

In the first three years of this grant, we have completed the development of a preliminary candidate biomarker panel that can predict an individual’s response to sleep loss phenotype (i.e., resilient, intermediate, or vulnerable, as defined by performance at 20 hours of wakefulness) from neurobehavioral performance collected in the individual at baseline after a habitual (8 hour) night of sleep 2 days earlier. From our analysis, we have identified a metric that can discriminate ~50% of our subject population (studied under inpatient laboratory conditions) as non-resilient to sleep loss (i.e., intermediate or vulnerable response to sleep loss) at ~20 h awake from one neurobehavioral performance assessment during rested conditions (i.e., at ~8 hours of wake after a habitual 8-hour night of sleep) with ~100% accuracy. Our findings suggest that these individuals who are poor performers during rested conditions exhibit non-sleepiness-related performance impairment that is exacerbated by subsequent sleep loss. Among the remaining ~50% of our subject population, who are high-level performers during rested conditions, the distribution of response to sleep loss phenotype is expected to be approximately equal across resilient, intermediate, and vulnerable phenotypes. A model developed to discriminate response to sleep loss phenotype at 20 h awake in these high-level performers from a single neurobehavioral performance assessment 2 days earlier during rested conditions has an overall accuracy of ~64%, suggesting that neurobehavioral performance alone is insufficient to discriminate response to sleep loss phenotype in these individuals.

Our metric that discriminates high-level from poor performers during rested conditions has been implemented as a screening tool to select participants for our 6-day Phase 2 study; only high-level performers have been included as participants in our study. These high-level performers who also met all our other inclusion/exclusion criteria have been enrolled in our 6-day laboratory-based study for Phase 2 of this project. In this laboratory study, we are collecting multiple biomarker samples during baseline and sleep deprivation to determine whether there are one or more “omics”-based measures that provide a greater sensitivity and specificity to identify individuals who are resilient versus non-resilient to sleepiness-related performance impairment. We started recruiting for this study in January 2016, and to date we have consented and screened 58 potential candidates. Forty-five were excluded due to not meeting study inclusion/exclusion criteria, including 6 participants who were identified as poor performers from the 35 participants who completed our initial screening test battery. To date, we have completed studies on 13 individuals. Initial analysis of lipidomics data has identified 264 unique lipids, 26% of which exhibit a rhythm during rested conditions and 44% of which exhibit a rhythm during constant routine when subjects were sleep deprived. We expect to study an additional 3 participants by August 2017 and complete lipidomic and metabolomic analysis on all subjects in the next year of the project in support of Phase 2 of this study.

As part of our Phase 1 analysis, we also have conducted an analysis of retrospective neurobehavioral performance data in expeditioners overwintering in Antarctica. We have developed a definition of response to sleep loss phenotype in this population based on simple reaction time performance at >16 hours of wakefulness, and developed a model to predict this phenotype from baseline neurobehavioral performance data collected 1-6 months earlier. Our current model predicts response to sleep loss with an overall accuracy of 65%. We are continuing to develop this model by incorporating information from other neurobehavioral tests at baseline to improve its overall accuracy. We plan to prospectively apply this model to baseline data collected from expeditioners who will be overwintering in Antarctica.

NOTE: End date shows as 7/31/2017 per NSSC information (Ed., 6/2/15)

In this project, we will first identify candidate biomarkers in a retrospective analysis from neurocognitive and psychological measures collected in more than 1,500 subjects that have been studied over the past 17 years in our laboratory analog and from existing extensive longitudinal data in over 50 individuals overwintering in Antarctica, which is a high fidelity space analog. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new space analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.

In the first two years of this grant, we have completed the development of a preliminary candidate biomarker panel that can predict an individual’s response to sleep loss phenotype (i.e., resilient, intermediate, or vulnerable, as defined by performance at 20 hours of wakefulness) from neurobehavioral performance collected in the individual at baseline after a habitual (8 hour) night of sleep 2 days earlier. From our analysis, we have identified a metric that can discriminate ~50% of our subject population (studied under inpatient laboratory conditions) as non-resilient to sleep loss (i.e., intermediate or vulnerable response to sleep loss) at ~20 h awake from one test session at baseline (i.e., at ~8 hours of wake after a habitual 8-hour night of sleep) with ~100% accuracy. Our findings suggest that these individuals who are poor performers at baseline exhibit non-sleepiness-related performance impairment that is exacerbated by sleep loss. Among the remaining ~50% of our subject population, who are high-level performers at baseline, the distribution of response to sleep loss phenotype is expected to be approximately equal across resilient, intermediate, and vulnerable phenotypes. A model developed to discriminate response to sleep loss phenotype at 20 h awake in these high-level performers from baseline data 2 days earlier has an overall accuracy of ~64%, suggesting that neurobehavioral performance alone is insufficient to discriminate response to sleep loss phenotype in these individuals.

Our preliminary candidate biomarker panel that discriminates high-level from poor performers at baseline has been implemented as a screening tool to select subjects for our 6-day Phase 2 study; only high-level performers will be included as subjects in our study. These high-level performing subjects who pass our initial screening test battery and also meet all our other inclusion/exclusion criteria will be enrolled in our 6-day laboratory-based study in Phase 2 of this project. In this laboratory study, we will collect multiple biomarker samples during baseline and sleep deprivation to determine whether there are one or more “omics”-based measures that provide a greater sensitivity and specificity to identify individuals who are resilient and vulnerable to sleepiness-related performance impairment. We started recruiting for this study in January 2016, and to date we have consented and screened 38 potential candidates. Twenty-four were excluded due to not meeting study inclusion/exclusion criteria and we have completed studies on 6 individuals. We expect to complete studies on an additional 8 individuals by September 2016. We are continuing our recruiting and screening efforts, and will study up to 20 subjects in Phase 2.

As part of our Phase 1 analysis, we also have conducted an analysis of retrospective neurobehavioral performance data in expeditioners overwintering in Antarctica. We have developed a definition of response to sleep loss phenotype in this population based on simple reaction time performance at >16 hours of wakefulness, and developed a model to predict this phenotype from baseline neurobehavioral performance data collected 1-6 months earlier. Our current model predicts response to sleep loss with an overall accuracy of 65%. We are continuing to develop this model by incorporating information from other neurobehavioral tests at baseline to improve its overall accuracy. We plan to prospectively apply this model to baseline data collected from expeditioners who will be overwintering in Antarctica.

In this project, we will first identify candidate biomarkers in a retrospective analysis from neurocognitive and psychological measures collected in more than 1,500 subjects that have been studied over the past 17 years in our laboratory analog and from existing extensive longitudinal data in over 50 individuals overwintering in Antarctica, which is a high fidelity space analog. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new space analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.

In the first year of this grant, we have made significant progress in the identification of potential biomarkers from objective and subjective measures of cognitive performance and alertness. To date, we have compiled and analyzed retrospective performance and alertness data from over 160 subjects who were studied under identical laboratory conditions. From these data, we have developed a preliminary candidate biomarker panel that can predict whether an individual will be resilient or vulnerable to sleepiness-related performance impairment from information collected in the individual at baseline after a habitual (8 hour) night of sleep. This preliminary candidate biomarker panel will be used as a screening tool to select subjects for our 6-day Phase 2 study.

The development of this preliminary candidate biomarker panel involved three goals. The first goal was to develop a definition of “good” vs. “poor” performance and alertness after a habitual night of sleep and less than 16 hours of wakefulness to identify those individuals that may have performance and alertness impairments unrelated to sleep loss. The second goal was to develop a definition of resilient, intermediate, and vulnerable performance impairment related to sleep loss at more than 16 hours of wakefulness. The third goal was to determine a priori which of the individuals who appear to be good performers at baseline (i.e., not susceptible to non-sleep-related performance impairment at less than 16 hours of wakefulness) were susceptible to sleep-related performance impairment at more than 16 hours of wakefulness. Our analysis revealed that among those individuals who perform well under baseline conditions, a subset remain “resilient” after more than 16 hours awake and the remaining subjects exhibit varying degrees of sleepiness-related performance impairment, some of which meet our definition for “vulnerable.”

The model that we have developed has been validated to predict the most “resilient” and “vulnerable” individuals with a specificity of >85%. Potential subjects that are predicted as “resilient” or “vulnerable” during our screening assessments and also meet all our other inclusion/exclusion criteria will be enrolled in our 6-day laboratory-based study in Phase 2 of this project. In this laboratory study, we will collect multiple biomarker samples during baseline and sleep deprivation to determine whether there are one or more “omics”-based measures that provide a greater sensitivity and specificity to identify individuals who are resilient and vulnerable to sleepiness-related performance impairment. We are completing the development of our screening tool and expect to start recruiting for this study by June 2015, with our first subject completing all screening procedures and entering the laboratory-based portion of the study in July 2015. We will study up to 20 subjects in Phase 2.

We are also conducting the same Phase 1 analyses on our retrospective data in expeditioners overwintering in Antarctica. We will develop a similar model to predict the degree of resiliency or vulnerability in sleepiness-related performance impairment among this population of individuals, and apply this model to baseline data collected from expeditioners who will be overwintering in Antarctica from March to October of 2016. We expect to deploy this model and start baseline data collection in these individuals as early as September 2015.

To date, there are no biomarkers that are approved to determine individual astronauts’ vulnerabilities to sleep loss and circadian rhythm disruption, although there are many candidates to test. In a recent NASA Evidence Review addressing HRP Sleep Gap4, we identified and evaluated approximately 80 potential biomarkers of response to sleep loss or circadian desynchrony, ranging from simple subjective assessments of alertness through to genetic polymorphisms. While there are few data specifically characterizing individual responses to sleep loss and circadian desynchrony, fortunately, many of these biomarkers have been studied by our research group in detailed laboratory conditions of thousands of subjects over tens of thousands of bed-days in federally-funded studies, but have never been examined with respect to individual differences. Furthermore, in a previous NASA-supported study (ROSES-2008), we studied 51 Antarctic expeditioners over 6 months during the Antarctic winter (March to September) at Australian Antarctic stations Casey, Davis, and Mawson (54% of the overwintering population). A large number of subjective and objective measures of stress, affect, safety, team interactions, and well-being were recorded, in addition to subjective and objective measures of individual behavioral health outcomes including sleep, circadian rhythms, urinary cortisol, and melatonin levels, a range of neurocognitive performance outcomes and alertness. The large dataset is available to test the predictive value of a range of behavioral, performance, sleep, and circadian biomarkers on neurocognitive impairment and going forward, will allow our research team, in collaboration with HRP, to select only those measures that yield significant information for future study, and will allow capacity to add new measures not yet assessed under these operational conditions.

Based on this review, we have devised a protocol to assess the sensitivity and specificity of a core set of these biomarkers to predict neurocognitive and psychological responses to the sleep deprivation and circadian misalignment inherent in ISS operations. First, we will identify candidate biomarkers from our laboratory analog, where individuals are housed in individual, semi-isolated, window-less suites, and from existing extensive longitudinal data in over 50 Antarctic expeditioners. These, and previously untested biomarkers, will then be validated in new limited laboratory analog studies to inform a new analog study that will test the feasibility and utility of a set of core biomarkers to predict and assess variation in neurocognitive and psychological function associated with living in Antarctica during a 12-month mission. These innovative studies have a high potential to identify reliable biomarkers that are suitable for operational use.