Development and Validation of a Predictive Score for Surgical Site Infections (SPRED)

Development and Validation of a Predictive Score for Surgical Site Infections (SSI): a Prospective Preoperative Trial in Major Digestive Surgery

More than 8 millions surgical interventions are carried out each year in France. Postoperative complications, in particular infectious, can occur in 10 to 60% of cases and are the cause of postoperative revision in 30% of cases, an increase in mortality, length of stay, readmissions and lead to significant additional socio-economic costs. Currently, improvements in surgical practices have not reduced the incidence of surgical site complications. In this context, the development of predictive scores for the risk of post-operative complication becomes urgent in order to implement new interventions (pre-habilitation) or to modify surgical decisions (timing, approach) in order to reduce the risk of complications before surgery. Several recent studies highlights the importance of the immune response in postoperative prognosis. In particular, an imbalance between the adaptive and innate response involving MDSCs has been demonstrated in patients with postoperative complications.Thanks to new techniques for analyzing the immune system, in-depth analysis of the immune system before surgery is a very promising approach aimed at identifying predictive biomarkers of postoperative prognosis. Our team has developed and patented a multivariate model integrating mass cytometry data, proteomics and clinical data collected before surgery to accurately predict the occurrence of a surgical site complication (AUC = 0.94, p<10e-7) in a monocentric cohort of 43 patients to major abdominal surgery (Stanford University). The objective of the present study is to generalize and validate this preoperative predictive score of infectious complications of the surgical site in the 30 days following major digestive surgery on a larger workforce within a multicenter cohort and to validate this score at using a machine learning method.

Research hypothesis and expected impact: Postoperative complications are frequent and associated with excess mortality and increased costs for the health system. But, it is possible to avoid a significant number of these complications through prehabilitation programs, in particular to prepare patients at risk, and to reduce these postoperative events by 30%. However, it is currently not possible to predict, before surgery, which patients are at risk of developing a complication. Current predictive clinical scores such as the one developed by the American College of Surgeons are unsatisfactory (AUC = 68%). This study will be a reference study to define the groups of patients at risk of complications in order to develop, in a second step, personalized patient pathways in order to optimize their health before surgery and thus improve post-operative results.

The size of the cohort is 240 patients Population: Patients with major elective digestive surgery (eg, colon or colorectal resection, partial or total gastrectomy, pancreaticoduodenectomy, hepatectomy).

Primary Endpoint: Preoperative Prediction Score Performance of Surgical Site Infectious Complications Defined as Superficial, Deep and Organ Surgical Site Infection as per CDC 2021 definition within 30 days post-operatively. The performance of the score will be evaluate based on the evaluation of the F1 score criterion and the AUROC. F1: score ranges from 0 to 1, where 0 is the worst and 1 is the best possible score. AUROC: score ranges from 0.5 to 1 where 1 is the best score and 0.5 means the model is as good as random.

defined by prescribing antibiotics with one or more of the following: sputum sputumnew or changed, new or changed lung opacities on chest x-ray, fever > 38°C, leukocytes >12 × 109 /L within 30 postoperative days. The performance of the score will be evaluate based on the evaluation of the F1 score criterion and the AUROC. F1: score ranges from 0 to 1, where 0 is the worst and 1 is the best possible score. AUROC: score ranges from 0.5 to 1 where 1 is the best score and 0.5 means the model is as good as random.

Performance of the preoperative prediction score for urinary tract infections (according to the definition of the CDC - 2021) within 30 days post-operative. The performance of the score will be evaluate based on the evaluation of the F1 score criterion and the AUROC. F1: score ranges from 0 to 1, where 0 is the worst and 1 is the best possible score. AUROC: The score ranges from 0.5 to 1 where 1 is the best score and 0.5 means the model is as good as random.

Performance of the post-operative prediction score (D1) of infectious surgical site complications defined as superficial, deep and organ infection of the surgical site according to the CDC 2021 definition within within 30 days post-operative. The performance of the score will be evaluate based on the evaluation of the F1 score criterion and the AUROC. F1: score ranges from 0 to 1, where 0 is the worst and 1 is the best possible score. AUROC: Score ranges from 0.5 to 1 where 1 is the best score and 0.5 means the model is as good as random.

Data ( GHS, GHM, cost) will be collected upon the discharge of the patient from the hospital via the medical information departments (DIM) based on the PMSI of each establishment.

The score is calculated using a machine learning method integrating immune, plasma protein and clinical data. The aim is to validate and generalize the score result (AUC = 0,94, p<10e-7) of a multivariate model already developed in a monocentric cohort of 43 patients undergoing major abdominal surgery (Stanford University).

Inclusion Criteria: Patients will be included: Aged 18 and over Having undergone elective major digestive surgery: Major surgery defined according to the recent recommendations of the European Surgical Association - PMID: 32172309 by a rate of infectious or cognitive complications between 20 and 30% according to the ACS risk calculator Having expressed their non-opposition to participate in the study Being affiliated to a French health insurance Exclusion Criteria: Patients with the following criteria will not be included: Aged under 18 Having an ASA 4 or more, in palliative care Having an expected duration of hospitalization < 24 hours Not speaking French, illiterate patient Having expressed their opposition to participate in the study Current pregnancy or breastfeeding Absence of affiliation to social security plan Being deprived of liberty or under guardianship

Storesund A, Haugen AS, Hjortas M, Nortvedt MW, Flaatten H, Eide GE, Boermeester MA, Sevdalis N, Softeland E. Accuracy of surgical complication rate estimation using ICD-10 codes. Br J Surg. 2019 Feb;106(3):236-244. doi: 10.1002/bjs.10985. Epub 2018 Sep 18.

Hawn MT, Vick CC, Richman J, Holman W, Deierhoi RJ, Graham LA, Henderson WG, Itani KM. Surgical site infection prevention: time to move beyond the surgical care improvement program. Ann Surg. 2011 Sep;254(3):494-9; discussion 499-501. doi: 10.1097/SLA.0b013e31822c6929.

Gaudilliere B, Angst MS, Hotchkiss RS. Deep Immune Profiling in Trauma and Sepsis: Flow Is the Way to Go! Crit Care Med. 2017 Sep;45(9):1577-1578. doi: 10.1097/CCM.0000000000002594. No abstract available.

Zhu X, Herrera G, Ochoa JB. Immunosupression and infection after major surgery: a nutritional deficiency. Crit Care Clin. 2010 Jul;26(3):491-500, ix. doi: 10.1016/j.ccc.2010.04.004.

Gaudilliere B, Fragiadakis GK, Bruggner RV, Nicolau M, Finck R, Tingle M, Silva J, Ganio EA, Yeh CG, Maloney WJ, Huddleston JI, Goodman SB, Davis MM, Bendall SC, Fantl WJ, Angst MS, Nolan GP. Clinical recovery from surgery correlates with single-cell immune signatures. Sci Transl Med. 2014 Sep 24;6(255):255ra131. doi: 10.1126/scitranslmed.3009701.