Tu178 - The SPRED Study – Multiomic Prediction of Surgical Site Infections After Major Abdominal Surgery

Abstract Text:

Background: Postoperative complications, including surgical site infections (SSIs), occur in up to 44% of major surgeries, increasing mortality, hospital stays, and socioeconomic burden. Established clinical risk scores, such as the American College of Surgeons NSQIP (ACS) score, show limited predictive power. The SPRED study aimed to develop a predictive model for SSI after major abdominal surgery by integrating clinical and biological factors measured preoperatively [1].

Materials and

Methods: In a five-center prospective study of 164 adult patients undergoing elective major abdominal surgery in France, SSIs were assessed within 30 days post-surgery. Preoperative clinical data and a mass cytometry and plasma proteomic (Olink) approach quantified 880 immune cell and 726 proteomic features in blood samples. Sparse machine learning and data-driven feature selection [2] using the Stabl framework were applied to build a multivariable SSI prediction model. Model performance was evaluated using Monte Carlo cross-validation (5-folds, 10 repeats) and compared to the ACS risk score [3].

Results: The integrative model combining clinical and proteomic data outperformed the ACS model (AUROC = 0.79 [0.70–0.88] vs. 0.50 [0.38–0.63]). Stabl identified three proteins (CXCL14, NCAM1, FABP6) and two clinical factors (ASA score, surgical indication) as key predictors.

Conclusion: This integrative model provides superior accuracy in predicting SSIs compared to conventional factors, identifying biomarker candidates to personalize perioperative management.

[1] NCT05523713: Development and Validation of a Predictive Score for Surgical Site Infections (SPRED).
[2] Hédou J et al. Nat Biotechnol (2024). https://doi.org/10.1038/s41587-023-02033-x
[3] Bilimoria KY et al. J Am Coll Surg. 2013;217(5):833-42.e423. doi:10.1016/j.jamcollsurg.2013.07.385