Prediction of clinically relevant postoperative pancreatic fistula after pancreatoduodenectomy based on multifrequency magnetic resonance elastography

Background

Clinically relevant postoperative pancreatic fistula (CR-POPF) is the major complication of pancreatoduodenectomy, and the pancreatic texture is one of the potential affecting factors. Multifrequency magnetic resonance elastography (MRE) is a novel technique for measuring tissue stiffness, but its value in predicting CR-POPF preoperatively has not been well documented.

Methods

A total of 70 patients who underwent multifrequency MRE before pancreatoduodenectomy between July 2021 and April 2024 were retrospectively recruited into the study. The parameters of MRE, shear wave speed (c) and phase angle (φ), and clinical data were collected. Logistic regression and the receiver operating characteristic curve analyses were used to assess the performance of multifrequency MRE in predicting CR-POPF.

Results

CR-POPF was developed in 14 of 70 patients (20%), all categorized as grade B. The CR-POPF group had significantly lower c (1.339 ± 0.210 m/s) and longer hospital stays (21 [IQR, 15.50−37.75] days) than the no-CR-POPF group. The MRE parameters, c and φ, were moderately correlated with pancreas stiffness (eta² for φ = 0.189 and eta² for c = 0.106). Dilated major pancreatic duct (MPD ≥ 3 mm) and higher c were independently associated with a lower risk of CR-POPF in univariant and multivariant analyses (odds ratio [OR] for c, 0.041 [95% CI, 0.002–0.879]; OR for dilated MPD, 0.129 [95% CI, 0.022–0.768]). The area under the curve (AUC) of the predictive model based on c and MPD diameter was 0.786, which was better than the fistula risk score (FRS) (AUC = 0.587) and alternative FRS (AUC = 0.556) in our center, with the DeLong test P = .028 and P = .002, respectively.

Conclusion

The MRE parameters were associated with pancreatic stiffness, and c was an independent predictor of CR-POPF after pancreatoduodenectomy.