The purpose of this study was to assess the value of intravoxel incoherent motion and diffusion-weighted imaging for predicting the histologic grade of pancreatic neuroendocrine tumors (PNETs).
Materials and Methods
Forty patients with surgically diagnosed PNETs who underwent preoperative magnetic resonance imaging, including diffusion-weighted imaging with a series of 10 b values (0–1000 s/mm2), were included in this institutional review board–approved retrospective study. The apparent diffusion coefficient (ADCtotal), the intravoxel incoherent motion parameters (pure diffusion coefficient [D], pseudodiffusion coefficient [D*], and perfusion fraction [f]) were measured on the tumors. Histologic grading was performed on the basis of the World Health Organization 2010 classification system. Logistic regression analysis and receiver operating curve analysis were performed to identify the significant factors predicting the histologic grades.
Grades 2 and 3 tumors were significantly larger than grade 1 tumors (average 3.62 cm vs 2.17 cm in diameter; P = 0.001). Grades 2 and 3 tumors showed significantly lower D values than did grade 1 tumors (0.95 vs 1.21 × 10−3 mm2/s; P = 0.009), although the ADCtotal showed no significant difference. When any of the following 2 criteria was used, (a) tumor size smaller than 2.0 cm in diameter and (b) D value greater than 1.2 × 10−3 mm2/s, the sensitivity, specificity, and positive predictive value for diagnosing grade 1 PNETs were 76.92%, 100%, and 100%, respectively.
Pure diffusion coefficient (D) is possibly a better marker than ADCtotal is for differentiating grade 1 from grade 2 or 3 PNET and, combined with tumor size, can predict grade 1 PNET with a high specificity.