Abstract
CONTEXT: Patients with von Hippel-Lindau (vHL) disease caused by a missense VHL mutation have a more severe phenotype compared with other VHL mutation types. OBJECTIVE: To define pancreatic neuroendocrine tumor (PNET) aggressiveness according to VHL genotype. DESIGN: A prospective natural history study. SETTING: The National Institutes of Health clinical center. PATIENTS: Patients with vHL disease, pancreatic manifestations, and germline missense VHL gene mutations. INTERVENTION: In-silico prediction of VHL mutation via five computational prediction models. Patients with >80% prediction for disease-causing mutations in all models [high predicted risk (HPR)] were compared with others [low predicted risk (LPR)]. MAIN OUTCOME MEASURE: Rates of metastases, surgical intervention, and disease progression. RESULTS: Sixty-nine patients were included: 2 developed metastases, 12 needed surgery, and 31 had disease progression during a median follow-up of 60 months (range 13 to 84 months). Thirteen patients were excluded for low prediction reliability. In the remaining 56 patients (45 with PNETs, 11 with pancreatic cysts), the HPR group (n = 13) had a higher rate of disease progression than the LPR group (n = 43) in multivariable analysis (hazard ratio 3.6; 95% confidence interval, 1.1 to 11.9; P = 0.037). The HPR group also had a higher risk of developing metastases (P = 0.015). Among patients with codon 167 hotspot mutations (n = 26), those in the HPR group had a higher risk for disease progression (P = 0.03) than other patients. CONCLUSIONS: Computational models for predicting the impact of missense VHL gene mutations may be used as a prognostic factor in patients with PNETs in the context of vHL disease.