Impact of vascular abnormality on contrast-enhanced CT and high C-reactive protein levels on postoperative pancreatic hemorrhage after pancreaticoduodenectomy: A multi-institutional, retrospective analysis of 590 consecutive cases

2020 
Abstract Background /Objectives: This study aimed to elucidate the efficacy of CT findings and perioperative characteristics to predict post-pancreatectomy hemorrhage (PPH): a critical complication after pancreaticoduodenectomy. Methods The records of 590 consecutive patients who underwent pancreaticoduodenectomy at three institutes between 2012 and 2018 were included. The presence of a vascular wall abnormality or ascites with high density (vascular abnormality) on postoperative day (POD) 5–10 contrast-enhanced CT (early CT), perioperative characteristics, and any PPH or pseudoaneurysm formation (PPH events) were analyzed through a multivariate analysis. Results PPH events occurred in 48 out of 590 patients (8%). The vascular abnormality on early CT and the C-reactive protein (CRP) value on POD 3 were independent risk factors for PPH events after POD5 (vascular abnormality: odds ratio 6.42, p = 0.001; CRP on POD 3: odds ratio 1.17, p = 0.016). The sensitivity of vascular abnormality for PPH events was 24% (7/29), and the positive predictive value was 30% (7/23). The combination of vascular abnormality and a high CRP value (≥15.5 mg/dL) on postoperative day 3 had a higher positive predictive value of 64% (7/11) than the vascular abnormality alone. None of the seven PPH events that occurred more than one month after surgery were foreseen via early CT. Conclusion The combination of vascular abnormality and high CRP value was associated with increasing risk of PPH events after pancreaticoduodenectomy, but the low sensitivity of early CT must be noted as an important shortcoming. The normal findings on early CT could not eliminate the risk of late PPH.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    19
    References
    1
    Citations
    NaN
    KQI
    []