Association of VHL Genotype With Pancreatic Neuroendocrine Tumor Phenotype in Patients With von Hippel–Lindau Disease

Abstract

This study evaluates the natural history of von Hippel–Lindau disease–associated pancreatic lesions to determine the factors associated with pancreatic neuroendocrine tumor phenotype and prognosis.

Von Hippel–Lindau (VHL) disease is a familial cancer syndrome caused by a germline mutation in the VHL tumor suppressor gene (OMIM 193300). The prevalence of pancreatic neuroendocrine tumors (PNETs) in VHL disease ranges from 9% to 17% and is characterized by a better prognosis compared with sporadic PNETs. Several risk factors for PNET metastasis in VHL disease have been reported but have not been validated. We conducted a prospective study (NCT00062166) to evaluate the natural history of VHL disease–associated pancreatic lesions to determine what factors were associated with PNET phenotype and prognosis.

Methods

Patients were enrolled from January 1, 2010, to January 1, 2017, based on the following eligibility criteria: age of 12 years or older, a diagnosis of VHL disease based on germline VHL mutation or clinical criteria, and a pancreatic manifestation of VHL disease. Pancreatic protocol (2-mm section) abdominal computed tomography and pelvis computed tomography with intravenous contrast material were performed annually or every 2 years in patients with solid or cystic pancreatic lesions, respectively. Each imaging study was assessed by at least 2 independent reviewers. The criteria for surgical resection of pancreatic solid lesions were reported previously. Disease-free interval and progression-free interval were defined as the time from inclusion without PNET and with no tumor growth (>5 mm and >20%), respectively.

Patients diagnosed with VHL disease underwent genetic testing for mutations in the 3 coding exons and exon-intron boundaries of the VHL gene and for partial or complete deletion or duplication of the VHL gene in Clinical Laboratory Improvement Amendments–approved laboratories. This study was approved by the institutional review board of the National Cancer Institute, and written informed consent was obtained from all patients enrolled in the study. All data were deidentified.

Results

The current analysis included 229 patients (mean [SD] age, 49.6 [12.7] years, 110 [48.0%] male and 119 [52.0%] female); 54 patients had only cystic lesions, and 175 had solid pancreatic lesions consistent with PNETs (median of 2 PNETs per patient; range, 1-10; total, 489). Patients with VHL mutations in exon 3 vs exons 1 or 2 had a higher rate of PNETs on univariate (odds ratio, 3.0; 95% CI, 1.4-6.5; P = .006) and multivariable analyses (odds ratio, 2.8; 95% CI, 1.3-6.1; P = .01).

Among patients with PNETs (median follow-up, 53.0 months; range, 12-84 months), 29 (16.6%) required surgical intervention. VHL gene sequencing was performed in 156 patients, 7 (4.5%) of whom had metastatic disease during follow-up. Patients with missense vs other VHL mutation types had a larger PNET diameter (mean [SD] diameter, 1.6 [1.1] vs 1.4 [1.0] cm; P = .05, Mann-Whitney test), but this finding did not reach statistical significance (Table).

Table. Baseline Characteristics According to Mutation Type Among Patients With Solid Pancreatic Lesions^a.

Characteristic	Missense Mutations (n = 76)	Other Mutation Typesb (n = 80)	P Value
Age at enrollment, mean (SD), y	44.7 (12.8)	43.0 (13.5)	.40
Female	39 (51.3)	46 (57.5)	.40
Follow-up duration, mean (SD), mo	50.9 (21.3)	51.7 (19.8)	.80
VHL manifestations
Pheochromocytoma	28 (36.8)	5 (6.2)	<.001
Retinal or CNS hemangioblastomas	59 (77.6)	72 (90.0)	.04
Any renal involvement	58 (76.3)	70 (87.5)	.07
2010 WHO tumor gradec			.30
G1	11 (55.0)	3 (33.3)
G2	9 (45.0)	6 (66.7)
Lesion characteristics
No. of lesions per patient, mean (SD)	2.9 (2.0)	2.6 (1.9)	.30
Initial largest lesion diameter, mean (SD), cm	1.6 (1.1)	1.4 (1.0)	.05 (MW)
Lesion locations
Uncinate	40 (52.6)	40 (50.0)	.70
Head	55 (72.4)	47 (58.8)	.07
Body or neck	16 (21.1)	15 (18.8)	.70
Tail	22 (28.9)	26 (32.5)	.60
Disease-free interval, mean (SD), mo	7.4 (21.2)	16.8 (31.6)	.03
Progression-free interval, mean (SD), mo	45.4 (23.6)	46.2 (27.7)	.90

Abbreviations: CNS, central nervous system; MW, Mann-Whitney test; VHL, von Hippel–Lindau; WHO, World Health Organization.

^aData are presented as number (percentage) of patients unless otherwise indicated. Solid pancreatic lesions were demonstrated by computed tomography, enhancing on early arterial phase, typical of pancreatic neuroendocrine tumors.

^bAmong patients with other mutation types, 51 patients had full or partial deletion, 13 had nonsense mutation, 2 had gene rearrangement, and 14 had frameshift mutation.

^cPercentage among patients with available pathologic specimens.

Patients with a greatest tumor diameter less than 1.2 cm had a 100% negative predictive value for developing metastasis and requiring a surgical intervention during follow-up, whereas patients with a tumor diameter greater than 3.0 cm had a high risk of metastatic disease on univariate (hazard ratio [HR], 8.6; 95% CI, 2.1-34.9; P = .003) and multivariable analyses (HR, 8.6; 95% CI, 1.7-43.2; P = .009).

Among patients with a PNET diameter of 1.2 cm or greater and 3 cm or less, only those with a VHL missense mutation developed metastatic disease during follow-up (5 [12.5%] vs 0 patients with missense vs other type of VHL gene mutation; log-rank test, P = .04). Patients with a VHL missense mutation had a higher rate of requiring a surgical intervention compared with other mutation types(16 [40.0%] with missense vs 5 [16.1%] with other VHL gene mutation type; log-rank test, P = .04), and patients with a VHL gene mutation located in exon 3 vs exon 1 or 2 had a higher rate of requiring a surgical intervention during follow-up (13 [44.8%] vs 8 patients [23.5%], P = .02). Patients with a missense VHL mutation or any mutation type in exon 3 had a higher rate of surgical intervention compared with other VHL genotypes on univariate (HR, 9.2; 95% CI, 1.2-68.7; P = .03) and multivariable analyses (HR, 8.8; 95% CI, 1.2-66.3; P = .04).

Discussion

Our study findings support an evidence-based algorithm for risk stratification and surveillance of patients with VHL-associated pancreatic lesions (Figure). Such an approach would provide precision medicine for patients with VHL disease–associated pancreatic lesions based on VHL genotype and tumor size. These results should be validated in future studies because of the rate of metastatic disease observed in our cohort.

Figure. Suggested Algorithm for Risk Stratification of Patients With von Hippel–Lindau Disease With Pancreatic Solid Lesions on Anatomical Imaging.

Anatomical imaging was pancreas protocol computed tomography with an early arterial phase. Of 9 patients with metastatic disease, 3 had a largest lesion diameter of greater than 3 cm, 6 had smaller lesion diameters; of these 6 patients, 5 had a missense VHL gene mutation and 1 did not have a determined genotype. PNET indicates pancreatic neuroendocrine tumor.