The relationship between CD88 levels and infection in patients. (A) Representative histograms showing fluorescence of healthy donor (green) or patient (red) neutrophils stained with anti-CD88:Alexa647, or isotype controls (blue). (B) Kaplan-Meier plot showing acquisition of infection as a function of time and patients alive while remaining within the ICU. Patients were censored for death or discharge without ICU-acquired infection, P = .01 by log-rank test. (B) Representative histograms showing fluorescence of healthy donor (green) or patient (red) neutrophils stained with anti-CD88:Alexa647, or isotype controls (blue). (C) The change in C3a concentrations over time, comparing the initial sample with the last sample before an end point (discharge, death, or infection) occurred. In the case of infection the “last sample” was censored for 2 days prior to infection. **P = .003 by Wilcoxon rank sum test for the difference between initial and end-point samples for patients with dysfunction (ie, low neutrophil CD88); n = 58 patients. (D) Neutrophil surface CD88 expression on the initial sample from patients admitted to ICU with and without sepsis. NS, P = .59 by Mann-Whitney U test. Data are shown as median and interquartile range (n = 60 patients, 23 with sepsis and 37 without). (E) Neutrophil surface CD88 expression before and after ICU-acquired infection. P = .97 by Kruskal-Wallis ANOVA (n = 22 patients).