Patients with suspected CSF infection
During a 1-year period, 30 patients required EVD because of intraventricular hemorrhage with a variety of causes: subarachnoid hemorrhages by aneurysm rupture (n = 20), parenchymal hematoma (n = 6), operated tumors (n = 3), and severe brain trauma (n = 1). Patients' mean age was 46 (range, 44-52) years, and they presented with a mean SAPSII of 42 (range, 34-49) at admission. Four patients died during the study of causes related to the severity of their intracranial vascular pathology. During the same period of time, we also tested CSF samples obtained by lumbar puncture from patients with suspected infection a mean of 8 (range, 7-12) days after trauma or surgery (n = 2 subarachnoid hemorrhages by aneurysm rupture, n = 1 parenchyma hematoma, n = 5 operated tumors, and n = 6 severe brain trauma).
Global ROS production in CSF from patients with suspected infection
According to the protocol, CSF was sampled by EVD in response to fever (> 38°C). Fifteen patients met the criteria for noninfected CSF, seven for infected CSF, and eight for negative CSF. The microorganisms observed in the seven infected CSF cases were Pseudomonas aeruginosa, Staphylococcus aureus, Staphylococcus epidermidis, Enterococcus faecalis, Acinetobacter baumanii, and Corynebacterium. Table summarizes the patients' clinical and biological characteristics. Ventricular drainage continued for a longer duration in patients with infected CSF. According to the definition, cell count in CSF was higher in "infected" and "noninfected" groups than in the "negative" group; the highest count was recorded in the infected group. No difference in percentage of PMNs was reported between groups.
Clinical and biological variables, separated by CSF inflammation and infection status
At the basal state, ROS production was significantly higher in the infected group than in the noninfected and negative groups (Figure ). After stimulation with PMA (Figure ), higher ROS production was observed in CSF from the infected group compared with the negative and noninfected groups. Basal ROS production did not differ between the noninfected group and the negative group; however, it was higher in the noninfected group than in the negative group after PMA stimulation. When ROS production values were normalized to PMN number (per cell; Figure ), we observed the same trends, but with smaller differences. To validate the sensibility of using ROS to diagnose infection, we reported the follow-up in three patients after treatment with antibiotics and demonstrated a rapid decrease of ROS production (Figure ).
Quantity of ROS produced by cells in 250 μl of cerebrospinal fluid (CSF) under basal conditions (A) or after stimulation by phorbol 12-myristate 13-acetate (PMA) (B).
Quantity of ROS normalized per cell under basal conditions (A), and after PMA stimulation (B). Samples were grouped according to infection status: negative CSF (white, n = 8), noninfected CSF (gray, n = 15), and infected CSF (black, n = 7).
Repetitive measurement of ROS in 250 μl of cerebrospinal fluid on several days: follow-up in three patients and impact of antibiotic treatment. The arrow represents the start of antibiotic treatment (day 0).
In CSF obtained by lumbar puncture, the leukocyte count was higher in infected samples (2590/mm3, IQR 1295-2925) than in noninfected (1200/mm3, IQR 943-4828; P = 0.0233) and in negative samples (20/mm3, IQR 13-35; P = 0.0339). Noninfected CSF had a lower leukocyte count than negative samples (P = 0.0167). ROS measured in CSF differed significantly among the three groups: negative CSF (75.5 × 103, IQR 49.0 × 103-203.4 × 103); noninfected CSF (5,760 × 103, IQR 3,545 × 103-11,582.5 × 103); and infected CSF (52,500 × 103, IQR 23,700 × 103-84,050 × 103; P = 0.0073, Kruskal-Wallis test). A similar pattern was observed when the leukocyte counts were normalized to PMN number and after PMA stimulation (data not shown).
PMN phenotype in noninfected and infected CSF during EVD
Eleven patients with EVD were sampled for CSF PMN characterization. Flow cytometric analysis clearly revealed two distinct populations of PMNs with respect to CD16 expression: CD16high
cells (Figure ). Repeated analysis of CSF over time demonstrated that the proportion of CD16high
cells decreased in CSF between day 5 and day 15 (Figure ). CSF PMNs also showed increased annexin V expression at the cell surface, especially in the CD16low
population (Figure ), suggesting that they became apoptotic [11
CD16 expression on polymorphonuclear neutrophils in cerebrospinal fluid over time in the same patient. Two cell populations are distinguished by differing CD16 expression and granulosity (SSC): CD16high and CD16low.
Representative annexin V staining in the CD16high and CD16low polymorphonuclear neutrophil (PMN) populations. CD16low PMN exhibited the highest annexin V level.
CD11b and CD62L are adhesion molecules, and their level of expression correlates with activation status. CD11b expression was similar in PMNs from CSF (33,900 AB/C, IQR 26,200-39,100) and from blood in healthy volunteers (47,900 AB/C, IQR 17,474-77,242); the percentages of CD62L-positive PMNs were similar in CSF (32% (6-41)) and healthy blood (12% (7-22)). PMNs in CSF did not present enhanced markers of activation.
One patient presented with an infection on day 10. The infected CSF was characterized by an increased number of CD16high
cells, high CD11b expression (119,100 AB/C), and a low percentage of CD62L-expressing PMNs (4%). The heightened expression of CD16 and CD11b was probably related to the recruitment of new leukocytes. The presentation of PMNs with high CD11b levels and low CD62L in cases with infection was in agreement with the classical activated phenotype described in the literature [12
CSF analysis by luminometry under different experimental conditions
Different experimental procedures were conducted to study the impact of various experimental conditions on the results. We performed analyses at different temperatures and with different time delays before measurements. CSF samples collected during patients' follow-up were analyzed following HBSS lavage plus centrifugation, and the results were compared with those obtained without centrifugation. ROS production measured by luminometry was always higher after washing and centrifugation than without (basal AUC RLU: 16,101 (IQR 9,835-35,358) versus 76,336 (IQR 41,222-277,007), P < 0.0001; PMA-stimulated AUC RLU: 315,592 (IQR 90,668-1,504,078) versus 683,190 (IQR 169,492-3,676,343), P = 0.0002). Neither the duration of delay before sample treatment (2 versus 4 h) nor the storage temperature (4°C versus room temperature) affected ROS production under basal or PMA-stimulated conditions, as measured by luminometry.