Introduction

Von Hippel-Lindau (VHL) disease (OMM #193300) is an autosomal dominant inheritable familial cancer syndrome. It predisposes to a variety of malignant and benign tumours such as cerebellar and spinal haemangioblastoma (HB), pheochromocytoma, renal clear cell carcinoma (RCC), pancreatic endocrine tumours, or visceral cysts.1 Clinical diagnosis of VHL in a patient without a familial history requires the occurrence of two tumours (two HBs or an HB and a visceral tumour), or a single VHL tumour in case of confirmed familial history of VHL.1 VHL is caused by a mutation, micro- or macro-deletion in the VHL tumour suppressor gene mapped to chromosome 3p25-26 (RefSeq NC_000003.11). In a patient with classical VHL disease, a germline abnormality of VHL can be identified in 892 to 95%1 of the cases if the molecular analysis comprises both nucleotide sequencing and PCR-based deletion search. Despite clinical evidence of the disease, this combination of molecular tools can fail to identify a germline mutation of VHL thus suggesting low levels of mosaic mutations. Mosaicism is the consequence of a somatic mutation occuring early in the embryo development that leads to co-existence of two cell populations in patient’s body. The prevalence of mosaicism in VHL, that is largely under-explored owing to the difficulty of its detection, is currently estimated to be around 5%.3, 4 However, the identification of mosaicism is important since it will influence both patient monitoring and genetic counseling in the relatives.

Here, we report a series of eight patients whose phenotype was suggestive of VHL and who were negative for a VHL abnormality both by PCR-Sanger sequencing and by quantitative multiplex PCR of short fluorescent fragments (QMPSF). Since next-generation sequencing (NGS) allows the detection of genetic variants in cell sub-populations,5, 6 we decided to evaluate its efficiency in the identification of mosaic VHL mutations in a hospital laboratory setting.

Materials and methods

NGS and data analysis

DNA was isolated from peripheral blood mononuclear cells using the EZ1 DNA Blood kit (Qiagen, Hilden, Germany). Library of amplicons was prepared according to the manufacturer’s instruction and sequenced with a ROCHE 454 GS Junior (Roche Diagnostics, Meylan, France), which uses the pyrosequencing technology.7, 8 Robotic protocols were developed on HAMILTON StarLet 8 platforms (Hamilton Robotic, Villebon sur Yvette, France) to perform multiplex amplification of the three exons of VHL (RefSeq NC_000003.11) using primers as follows: exon 1 Forward (F): 5′–CCCGGGTGGTCTGGATCG-3′ and Reverse (R): 5′-CTGGATGTGTCCTGCCTCAAG-3′ (amplicon size: 489 bp), exon 2F: 5′-GTGGCTCTTTAACAACCTTTGC-3′ and R: 5′-TGGAATAACGTGCCTGACATCA-3′ (amplicon size: 272 bp), exon 3F: 5′-CAGTAGTACAGGTAGTTGTTG-3′ and R: 5′-ATCAAAAGCTGAGATGAAACAG-3′ (amplicon size: 315 bp). Then, we performed ‘AmPure’ purification with magnetic beads, quantification with a spectro-fluorimeter SAFAS (SAFAS Monaco, Monaco), normalization of the amplicons and pooling to prepare the normalized library. During the multiplex amplification step, the different patients DNA was ‘tagged’ with specific Multiplex Identifiers (MID). A high-depth coverage (1000 × ) was defined to detect mutational events poorly represented into the normalized library generated from DNA samples. The sequencing reads were aligned against the human chromosome 3 reference sequence (NC_000003.11) using SeqNext (JSI medical Systems Gmbh, Kippenheim, Germany) and GenSearch (PhenoSystems, Wallonia, Belgium) Softwares. VHL mutations were defined using NCBI NM_000551.3 as a reference for VHL transcript.

Confirmation techniques for mosaicism detection

Pyrosequencing was carried out using a PyroMark MD according to the manufacturer’s protocol (Qiagen S.A.S., Courtaboeuf, France). The Primers were designed using the PSQ Assay Design v1.0 Software (Pyrosequencing AB, Uppsala, Sweden). Data obtained were analysed with the PyroMark MD v1.0 Software (Pyrosequencing AB).

SNPs genotyping was performed using ABI PRISM 3130 × 1 Genetic Analyzer with 50-cm capillary Array and the ABI PRISM SnaPshot Multiplex Kit according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA).

Results and discussion

To conduct our study, we built up a panel of eight patients whose clinical and molecular characteristics are detailed in Table 1. All patients harbour clinical or imaging features of VHL, but only two of them (P1 and P8) had a classical VHL disease. Three patients had a positive familial history but a familial VHL mutation was identified for only one (P8).

Table 1 Summary of clinical and molecular characteristics of the patients included in the panel

Two NGS runs were carried out on Roche 454 (Table 2). Patient P8 DNA was simultaneously analysed with his daughter’s DNA who had a heterozygous p.(R161*) VHL mutation. Results show that patient P1 who has developed three tumours (Figure 1a) presents a mosaic VHL mutation (c.500G>A, p.(R167Q)) with a frequency of the mutated allele of 5.7% thus explaining that it was undetectable by Sanger nucleotide sequencing (Figure 1b). To confirm these results, we developed alternative methods using specific pyrosequencing and SnaPshot protocols. Both methodologies confirmed that patient P1 is mosaic for the p.(R167Q) mutation (Figure 1c and d). No mosaic mutation was found in patients P2–P7. Finally, patient P8 who had two RCC was mosaic for the mutation p.(R161*) with a mutated allele frequency of 1.7% (Table 2). We did not develop alternative methods to confirm it for two reasons. First, the mutation was identical to those previously identified in his affected daughter. Second, the number of reads for each strand was above 350, that is the lower limit allowing the identification of 1% mosaic mutation with statistical confidence as recently shown by Izawa et al.6

Table 2 Summary of the analysis of NGS data by SeqNext software
Figure 1
figure 1

Clinical and molecular features of patient P1. (a) Abdominal CT scan of the P1 patient showing (1) a right adrenal pheochromocytoma (33 × 47 mm) and (2) a pancreatic neuro-endocrine tumour. (b) Sanger sequence analysis of VHL. A zoom is performed on VHL exon 3 around the codon 167 that is highlighted by a curly bracket. (c) Profile of pyrosequencing of VHL around codon 167 for a wild-type patient (1) and for the patient P1 with the c.500G>A VHL mosaic mutation (2). Arrow shows the abnormal nucleotide. VHL transcript reference number: NCBI NM_000551.3. (d) Profile of Snap shot analysis for a wild-type patient (1) and for the patient P1 with the c.500G>A VHL mosaic mutation (2). Arrow shows incorporation of the abnormal nucleotide, A in green (Forward sense) and T in red (Reverse sense).

Previous data showed that Roche 454 sequencing systems allow the detection of rare variants in blood extracted DNA samples, even for those below the Sanger sequencing limit of detection.9 Our data expand these results by demonstrating that in a hospital laboratory setting, the GS Junior with a high-depth coverage is a valuable alternative to conventional molecular tools for the detection of mosaic mutations in the VHL gene.3, 4, 10

By using these tools, Wu et al3 and Santarpia et al10 previously identified mosaic mutations in patients with a mild phenotype and a late onset of the disease (50 and 43 years old, respectively). Our results do not support these observations since patient P1 had a severe phenotype with the first tumour diagnosed at 16 years. Therefore, we suggest that the screening for mosaic VHL mutation should not be restricted to criteria of mild phenotype and late onset of the disease. Sgambati et al,4 evaluated the frequency of mosaicism by molecular screening of clinically asymptomatic parents of patients with a putative de novo VHL mutation. They found a 5% mosaicism rate (2 families on 42). On the basis of our preliminary data, it seems necessary to re-evaluate the true frequency of mosaic mutations in VHL by high sensitive NGS due to its clinical implications in the care of patients.

In conclusion, our study highlights the usefulness of clinical NGS as a diagnostic tool able to detect the presence of mosaic VHL mutation and it confirms the possibility to use pyrosequencing or SnaPshot analyses as confirmation techniques. This work also suggests that the research of mosaicism should not be restricted to patients with mild phenotype or late onset of the disease.