**Background:** Pancreatic fistula (PF) remains the most common and serious complication after pancreaticoduodenectomy (PD), with reported incidences ranging from 5% to 30%. Pancreatic texture is a known key risk factor, but preoperative assessment of texture is subjective. CT value (Hounsfield units) reflects tissue density and may correlate with histological characteristics such as fibrosis, edema, and fatty infiltration. This study aimed to investigate the relationship between preoperative unenhanced pancreatic CT value and the incidence of postoperative PF.
**Methods:** The authors retrospectively reviewed 127 patients who underwent PD with end-to-side duct-to-mucosa pancreaticojejunostomy at Xinhua Hospital (Shanghai) between 2017 and 2021. The cohort included 68 males and 59 females, mean age 57 years (range 12–79). Indications included pancreatic head carcinoma (46.5%), benign pancreatic head tumor (6.3%), periampullary carcinoma (4.7%), duodenal carcinoma (14.2%), inferior common bile duct carcinoma (22.0%), and chronic pancreatitis (6.3%). Preoperative unenhanced CT scans (dual-source 64-row spiral CT, 6 mm slice thickness) were used to measure pancreatic CT values in the broadest layer of the pancreas, excluding tumor areas, splenic artery, and dilated pancreatic ducts. The CT value ratio of the pancreas to abdominal aorta (P_CT/A_CT) was calculated to calibrate across scans. PF was defined and graded per ISGPF criteria (amylase in drainage >3 times serum normal, volume >50 mL/d on postoperative day 3). Statistical analysis included independent t-tests, ANOVA, ROC curve analysis, chi-squared tests, and logistic regression.
**Key Results:** Postoperative PF occurred in 40 patients (31.5%): 11 grade A, 23 grade B, and 6 grade C. One patient died from intra-abdominal hemorrhage. The mean P_CT/A_CT was significantly lower in the PF group (0.886 ± 0.113) compared to the non-PF group (1.034 ± 0.138; t=5.89; P < 0.001). Variance analysis showed a significant rank correlation between P_CT/A_CT and PF severity (P = 0.008), with mean values of 0.95 ± 0.11 (grade A), 0.89 ± 0.09 (grade B), and 0.78 ± 0.13 (grade C). ROC analysis yielded an AUC of 0.837 (95% CI: 0.759–0.915; P < 0.0001) and an optimal cutoff of 0.99. On univariate analysis, BMI >25, undilated pancreatic duct, hypoproteinemia (Alb <30), and P_CT/A_CT <0.99 were associated with PF. Multivariate logistic regression identified P_CT/A_CT <0.99 (OR: 11.3; 95% CI: 4.0–31.6; P < 0.001) and undilated pancreatic duct (OR: 2.8; 95% CI: 1.1–6.8; P = 0.028) as independent preoperative predictors. Additionally, P_CT/A_CT was significantly lower in patients with soft pancreatic texture (P < 0.01), higher BMI (P < 0.05), and those who underwent preoperative ERCP (P < 0.05).
**Clinical Implications:** The preoperative P_CT/A_CT ratio is a simple, objective, and non-invasive tool that can indirectly reflect pancreatic histological characteristics—including fibrosis (higher CT value), edema, fatty infiltration, and acinar richness (lower CT value)—all of which influence PF risk. A cutoff of 0.99 provides good discriminatory power. This information can help surgeons identify high-risk patients before surgery and potentially tailor the pancreaticojejunostomy technique (e.g., choosing invaginated or binding anastomosis for low-CT-value/soft pancreases). The study is limited by its retrospective, single-center design and relatively small sample size. Prospective studies are needed to validate the cutoff and to develop a comprehensive preoperative CT-based risk evaluation system incorporating pancreatic duct diameter and enhancement patterns.