D25-02 Serial Cytokines and Delirium Prevalence: An Exploratory Analysis of the Mends2 Trial

Abstract Rationale Delirium is prevalent in mechanically ventilated patients in the intensive care unit (ICU), yet understanding of its underlying pathophysiologic mechanisms remains incomplete. We conducted an exploratory analysis to investigate the relationship between peripheral cytokines and delirium in a large cohort of mechanically ventilated ICU patients. Methods This investigation included N = 422 patients enrolled in the MENDS2 prospective interventional trial Boncyk, NEJM 2021 comparing dexmedetomidine versus propofol sedation in mechanically ventilated patients. Samples were collected on days 1 (N = 402), 3 (N = 369), 5 (N = 311), and 7 (N = 272). Serum levels of fifteen cytokines (Figure 1A) were measured. Delirium assessments were performed twice daily with the Confusion Assessment Method for the ICU (CAM-ICU). Cytokines were standardized (z-score) and screened using univariate logistic regression (Figure 1A). Cytokines that passed screening (p 0.05) were included in a multivariable model, with independent associations with delirium identified via forward stepwise logistic regression using the Bayesian Information Criterion, for parsimony and to reduce overfitting. Model stability was verified by repeating stepwise selection on all cytokines without pre-screening. Analysis was conducted separately for days 1, 3, 5, and 7 to identify cytokines independently associated with delirium on each day. Results MCP-1 (2.12 1.34-3.40, p 0.001), IL-6 (1.51 1.15-2.10, p = 0.002), and GM-CSF (0.69 0.53-0.85, p 0.001) were identified as independently associated with delirium via both pre-screening and total cytokine inclusion methods (Figure 1B). On analyses by day (Figure 1C), MCP-1 (2.58 1.21-5.95, p = 0.009) and GM-CSF (0.70 0.44-0.96, p = 0.027) were associated with delirium on day 1. IL-6 (6.63 2.84-18.55, p 0.001) and GM-CSF (0.47 0.24-0.83, p = 0.005) were associated with delirium on day 3. MCP-1 (3.03 1.32-7.44, p = 0.008) and IL-12p40 (0.38 0.11-0.85, p = 0.009) were associated with delirium on day 5. Only MCP-1 (3.81 1.36-11.40, p = 0.011) was associated with delirium on day 7. Conclusions In this cohort of ICU patients with serial sampling, our exploratory analysis identified elevated MCP-1 and IL-6 as independently associated with delirium, while GM-CSF was inversely associated with delirium. Analysis by day also identified IL-12p40 as inversely associated with delirium on day 5. MCP-1 is of particular interest considering prior evidence that it can alter blood-brain barrier integrity, promote monocyte extravasation into the CNS, and activate resident glial cells, possibly providing a mechanistic link between systemic inflammation and acute neuroinflammatory cascades. Conversely, the inverse associations of GM-CSF and IL-12p40 observed here may reflect protective or compensatory immune responses. These data are hypothesis-generating for future work to help decipher cause and effect. This abstract is funded by: R21 AG080420