Bacterial extracellular vesicle as a predictive biomarker for postoperative delirium status after spinal surgery: a prospective cohort study

Abstract

Background: Prognostic factors significantly associated with postoperative delirium (POD) have been reported discordantly, possibly due to heterogeneous cohorts. Here, bacteria extracellular vesicles (BEVs) were introduced to predict the POD status of a unique patient cohort.

Methods: One hundred twenty-eight patients who underwent spinal surgery participated in this prospective cohort study. Significant preoperative factors (i.e., baseline characteristics, and sequences of 16s rRNA genes from bloods and stools) between patients with and without delirium were subjected to random forest classifiers for prediction model, and potential metabolites that regulate the POD were inferred in silico .

Results: No significant differences were found between patients with and without delirium in terms of demographics, anthropometrics, intervention history or preoperative cognitive function scores, except for circulating BEVs; delirium group had less diverse BEVs dominated with EVs from Gammaproteobacteria, whereas more diverse BEVs enriched with EVs from Bacilli and Alphaproteobacteria were significantly associated with non-delirium. Compared to that with baseline characteristics or gut microbiome, prediction model using random forest classifier with the significant BEVs yielded the lowest error rate of 21.59%, and was validated with an independent data set, resulting in 80% accuracy. Moreover, EVs from Moraxellaceae and Acinetobacter showed the highest probabilities of prediction of the POD despite their low relative abundance, indicating the most significant prognostic markers for the POD. As the inference of a potential metabolites that regulate the POD, succinate and enterobacterial common antigens delivered from BEV cargo were expected to participate in pathogenic events, whereas S-methyl-5'-thioadenosine, 2-oxoglutarate, pyruvate, acetate and butyrate may play a neuroprotective role in the POD.

Conclusions: The profile of circulating preoperative BEVs is the key prognostic factor for distinguishing POD in elderly surgical patients with controlled baseline conditions. Metabolites of defensive and offensive mechanisms inferred from BEVs will be essential for developing next-generation POD prevention strategies.