Tumoral and nontumoral pancreas: correlation between quantitative dynamic contrast-enhanced MR imaging and histopathologic parameters.

Purpose: To prospectively determine whether dynamic contrast material-enhanced (DCE) magnetic resonance (MR) quantitative parameters correlate with fibrosis and microvascular density (MVD) in malignant and benign solid pancreatic focal lesions and nontumoral pancreatic tissue.

Materials And Methods: The institutional review board approved the study; written informed consent was obtained. DCE MR was performed in 28 patients with surgically resectable focal pancreatic lesions. DCE MR quantitative parameters derived from one-compartment (OC) (transfer rate constant [K(trans)] and distribution fraction [ƒ]) and two-compartment (TC) (K(trans), tissue volume fraction occupied by extravascular extracellular space [v(i)], and tissue volume fraction occupied by vascular space [v(p)]) pharmacokinetic models were correlated with fibrosis content and MVD counts in focal lesions and nontumoral tissue (Spearman correlation coefficient [SCC]). Pharmacokinetic parameters were compared (Mann-Whitney test) between tumoral and nontumoral tissue. Diagnostic performance of DCE MR fibrosis detection was assessed (receiver operator characteristic curve analysis).

Results: K(trans) OC and K(trans) TC were significantly lower in primary malignant tumors compared with benign lesions (P = .023) and nontumoral pancreatic tissue downstream (P < .001) and upstream (P = .006); ƒ and v(i) were significantly higher in primary malignant tumors compared with nontumoral pancreatic tissue downstream (P = .012 and .018, respectively). Fibrosis was correlated negatively with K(trans) OC (SCC, -0.600) and K(trans) TC (SCC, -0.564) and positively with ƒ (SCC, 0.514) and v(i) (SCC, 0.464), with P < .001 (all comparisons). MVD was positively correlated with ƒ (SCC, 0.355; P = .019) and v(i) (SCC, 0.297; P = .038) but not with K(trans) OC (SCC, -0.140; P = .33) and K(trans) TC (SCC, -0.194; P = .181). Sensitivity and specificity for fibrosis detection were 65% (24 of 37) and 83% (10 of 12) for K(trans) OC (cutoff value, 0.35 min(-1)) and 76% (28 of 37) and 83% (10 of 12) for K(trans) TC (cutoff value, 0.29 min(-1)), respectively.

Conclusion: Quantitative DCE MR parameters, derived from pharmacokinetic models in malignant and benign pancreatic solid lesions and nontumoral pancreatic tissue, were significantly correlated with fibrosis and MVD.

Supplemental Material: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.11103515/-/DC1.