Characterizing indeterminate liver lesions in patients with localized pancreatic cancer at the time of diagnosis.

BACKGROUND

In patients with newly diagnosed pancreatic cancer, the classification of indeterminate liver lesions is an unanswered clinical dilemma as misclassification of these lesions can impact the assignment of clinical stage and subsequent treatment planning. Our objective was to design a standardized classification system to more accurately define the risk of malignancy in indeterminate liver lesions.

METHODS

In this retrospective study, patients with localized, non-metastatic pancreatic cancer were identified and pre-treatment computed tomography (CT) scans were evaluated for the presence or absence of liver lesions. Liver lesions were defined as definitely benign (1) or indeterminate (2). Indeterminate lesions were further sub-classified as either indeterminate probably benign (2B) or indeterminate possibly malignant (2M). The index liver lesion was evaluated on follow-up imaging for stability or unequivocal disease progression.

RESULTS

From 2008 to 2015, 304 patients with localized, non-metastatic pancreatic cancer were identified and 125 (41%) patients had liver lesions. Of the 125 patients, the liver lesions in 35 (28%) were classified as definitely benign and in 90 (72%) patients they were classified as indeterminate. The 90 patients with indeterminate lesions included 80 (89%) classified as indeterminate probably benign (2B) and 10 (11%) classified as indeterminate possibly malignant (2M). After a median follow-up of 56 weeks, no patient with a definitely benign lesion had metastatic disease progression of the index lesion. Of the 90 patients with indeterminate liver lesions, the index lesion progressed to unequivocal liver metastasis in 8 (9%) patients; 5 (6%) of the 80 lesions classified as indeterminate probably benign (2B), and 3 (30%) of the ten lesions classified as indeterminate possibly malignant (2M). The sensitivity of the classification system was 38% and the specificity was 91%. The positive predictive value was 30% and the negative predictive value was 94%.

CONCLUSIONS

A significant proportion of patients with localized pancreatic cancer will have liver lesions identified at the time of diagnosis and most of these lesions will have indeterminate characteristics. A classification system which further stratifies indeterminate liver lesions by malignant potential can assist clinicians in determining optimal treatment plan and is associated with a high negative predictive value.