Prevalence of von Hippel-Lindau gene mutations in sporadic renal cell carcinoma: results from the Netherlands cohort study

Background Biallelic von Hippel-Lindau (VHL) gene defects, a rate-limiting event in the carcinogenesis, occur in approximately 75% of sporadic clear-cell Renal Cell Carcinoma (RCC). We studied the VHL mutation status in a large population-based case group. Methods Cases were identified within the Netherlands cohort study on diet and cancer, which includes 120,852 men and women. After 11.3 years of follow-up, 337 incident cases with histologically confirmed epithelial cancers were identified. DNA was isolated from paraffin material collected from 51 pathology laboratories and revised by one pathologist, leaving material from 235 cases. VHL mutational status was assessed by SSCP followed by direct sequencing, after testing SSCP as a screening tool in a subsample. Results The number of mutations was significantly higher for clear-cell RCC compared to other histological types. We observed 131 mutations in 114 out of 187 patients (61%) with clear-cell RCC. The majority of mutations were truncating mutations (47%). The mean tumor size was 72.7 mm for mutated tumors compared to 65.3 mm for wildtype tumors (p = 0.06). No statistically significant differences were observed for nuclear grade, TNM distribution or stage. In other histological types, we observed 8 mutations in 7 out of 48 patients (15%), 1 mutation in 1 of 6 oncocytoma, 3 mutations in 2 of 7 chromophobe RCC, 2 mutations in 2 of 30 papillary RCC, no mutations in 1 collecting duct carcinoma and 2 mutations in 2 of 4 unclassified RCC. Conclusion VHL mutations were detected in 61% of sporadic clear-cell RCC. VHL mutated and wildtype clear-cell RCC did not differ with respect to most parameters.


Background
Historically, the classification of Renal Cell Cancer (RCC) was based on morphological features. The majority of RCC are of the clear cell type (~80%); other subtypes are papillary RCC (10%), chromophobe RRC (5%), collecting-duct carcinoma (1%) and unclassified RCC (3-5%). Based on the work of numerous investigators, it became evident that RCC could be divided into genetically distinct classes: this resulted in the so-called Heidelberg classification, which partly overlaps the former pathological classification of RCC based on morphological criteria. The most prominent common genetic aberration for clear-cell (conventional) RCC is loss of 3p. Characteristic for papillary RCC is trisomy of chromosomes 3q, 7,8,12,16,17,20, and loss of the Y chromosome, and chromophobe RCC is characterized by a combination of loss of heterozygosity at chromosomes 1,2,6,10,13,17, and 21 [1].
Von Hippel-Lindau disease (VHL) is a rare inherited disorder associated with, amongst others, an enhanced risk for clear-cell RCC [2]. The VHL gene responsible for this syndrome was identified through linkage analyses and molecular cloning and is located on chromosome 3p25. After its identification it became evident that the VHL gene is also involved in the development of sporadic clear-cell RCC. Together with loss of the homologous chromosome 3p allele (3p LOH), VHL mutations are ratelimiting events in the carcinogenesis of clear-cell RCC [3,4]. Mutations have been observed in the entire gene and usually lead to a truncated inactive protein [5]. The VHL gene is considered a tumor suppressor gene, involved in cell cycle regulation, regulation of hypoxia inducible genes and proper fibronectin assembly in extracellular matrix [6,7]. It was estimated that approximately 75% of all sporadic clear-cell RCC harbor biallelic VHL defects [8]. In approximately 19% of sporadic clear-cell RCC, methylation of the VHL gene promoter appeared to be involved [9]. In approximately 10%-20% of sporadic clear-cell RCC no alteration in the VHL alleles was detected, indicating that other genes are involved in clearcell RCC carcinogenesis, possibly affecting the same signaling pathway as VHL.
Several risk factors for developing RCC have been identified: tobacco smoking, obesity, drugs, such as phenacetin, hypertension and/or its medication, and occupational exposure to trichloroethylene, gasoline, petroleum products, asbestos, and iron processing fumes. The influence of dietary factors, such as vegetable, fruit vitamin C, carotenoid, meat and milk product consumption, is controversial [10]. Multiple and specific types of VHL mutations in RCC have been associated with exposure to the industrial solvent trichloroethylene [11,12]. Consumption of vegetables and citrus fruit decreased the frequency of VHL mutations among smokers and consumption of citrus fruit decreased VHL mutation frequency for all patients [13]. These findings and investigations in animals [14] suggest that mutational patterns in the VHL gene may serve as an etiological imprint to factors causing renal cancer. Thus, it may be possible to improve our etiological insight in particular risk factors when a more specific endpoint than "RCC" can be defined, e.g. based on histology and mutational status of a gene involved in tumor carcinogenesis.
We decided to determine the mutational status of the VHL gene of RCC cases identified within a population-based cohort of 120,852 men and women aged 55-69 which was recruited in the Netherlands to study associations between dietary habits, lifestyle and cancer occurrence. To validate whether SSCP could serve as a prescreening method, SSCP and direct sequencing was evaluated in a subset of 20 patients. In this article we report on histopathological and clinical parameters and the type and number of mutations in VHL observed in a large population-based sample of cases.

Study population
The RCC cases selected for this study were incident cases identified in the population of the Netherlands Cohort Study (NLCS) on diet and cancer. The NLCS has been described in detail elsewhere [15]. Briefly, this prospective study was initiated in 1986 and included 58,279 men and 62,573 women, 55 to 69 years of age, who completed a self-administered questionnaire on diet, family history of cancer, and other risk factors for cancer at baseline. The entire cohort is being monitored for cancer occurrence by annual record linkage to the Netherlands Cancer Registry and to PALGA, a nationwide network and registry of histo-and cytopathology [16]. Information on sex and family history of RCC (at baseline) was retrieved from the NLCS database, while information on age at diagnosis and tumor localization was retrieved from the Netherlands Cancer Registry. Tumor stage was classified according to the 1987 revision of the UICC-TNM classification [17], using information from the cancer registry and the pathology report. Tumor size was retrieved from the pathology report.

Tissue sample collection
Paraffin material was collected after approval by the Medical Ethical Committees of Maastricht University, PALGA and the Netherlands cancer registry. From 1986 to 1997 (11.3 years follow up) 355 kidney cancer cases (ICD-O-3: C64.9) were identified. Urothelial cell carcinomas were excluded and only histologically confirmed epithelial cancers were included (ICD-O: M8010-8119, 8140-8570), leaving 337 cases. The PALGA database was also used to identify the location of tumor tissue storage in the Dutch pathology laboratories. For 273 cases, a PALGA record with information on the location of paraffin material was available at the start of the collection of paraffin blocks. The tissue samples were distributed over 51 pathology laboratories throughout the Netherlands.
For 251/273 cases (92%) paraffin blocks were collected. Failure to retrieve material was the result of the refusal of the pathology laboratory to cooperate (3 laboratories with material for 10 cases), the unavailability of suitable material (i.e. only material from a biopsy, cytology or a metastasis was present) (8 cases), not being able to locate the paraffin block at the laboratory (3 cases), and for 1 case the reason was not recorded. Paraffin blocks of parallel normal tissue were collected if available. Collected archival tissue sample blocks were registered and coded using a unique identification number.

Revision
Genomic DNA was extracted from five 20-µm slices from each tumor and normal specimen. A flanking section was haematoxylin and eosin (HE) stained for histological purposes, e.g. grading and estimation of percentage of tumor cells. One experienced pathologist (CAHK) reviewed all HE stained slides. The RCC were classified according to the World Health Organization (WHO) classification of Tumours of 2002 [18]. Nuclear grading was performed according to Fuhrman [19]. Grading was based on the most atypical focus in the paraffin block used for DNA extraction, with a dimension of at least one high power field.
Material of 16 cases was discarded after revision. The collected material was unsuitable for analysis, because it con-cerned a biopsy (N = 2) or a metastasis (N = 2) or no tumor tissue was present (N = 4) or material contained less than 10% malignant cells (N = 7) (tumor samples had to contain at least 10% malignant cells to decrease the possibility of ignoring mutations). Material from one case was reclassified as urothelial cell carcinoma and subsequently excluded. As a result, tumor DNA from 235 cases was available for further analysis.

DNA isolation
Genomic DNA was prepared as follows: paraffin was removed with xylene and genomic DNA was extracted by salt-precipitation. Briefly, 450 µl of cell lysis solution (10 mM Tris/HCl (pH 7.4), 400 mM NaCl, 2 mM EDTA), 25 µl of 10% SDS and 50 µl of proteinase K solution (20 mg/ ml) were added to the tissue samples and incubated overnight at 55°C. Proteins were precipitated using 175 µl of saturated NaCl, followed by centrifugation (2 minutes, 13.200 rpm). DNA was precipitated by the addition of 0.6 volumes of iso-propanol, dissolved in TE (pH 7.4) and stored at -20°C. The DNA concentration and purity was measured at 260 and 280 nm.
Three µl of each sample was loaded on a 0.5x MDE gel, containing 0.6x TBE either with or without 10% (v/v) glycerol. Electrophoresis was performed at room temperature at 5W for MDE gels without glycerol, and at 7W for MDE gels with 10% (v/v) glycerol, using 0.6x TBE as electrophoresis buffer. After 20 hours of electrophoresis, the gels were transferred to Whatmann 3 MM paper (FIRMA) and dried on a gel dryer. The separated fragments were visualized by Hyperfilm-MP (FIRMA) exposure.
For sequence analysis, the PCR products were purified using the Wizard™ PCR preps purification system (Promega Corp.). Sequencing was performed at the local central sequence facility using BigDye Terminator and the ABI basecaller (Applied Biosystems). Mutations were identified by visual inspection of sequences provided by the ABI basecaller, and called when unequivocally present on the sense and/or antisense strand, and when the area under the curve of peaks changed by more than 5%, compared to the area under the curve of the normal signal.

Feasibility and comparability pilot
Twenty samples were analyzed by PCR-SSCP and direct sequencing to investigate the suitability of PCR-SSCP as a screening tool preceding direct sequencing. After evaluation, we decided to use PCR-SSCP as a screening instrument before direct sequencing. All remaining samples were subjected to PCR-SSCP, which was only followed by direct sequencing in case of an aberrant band pattern on the PCR-SSCP. Samples were also sequenced when equivocal PCR-SSCP results were obtained.

Statistical analyses
The overall frequency of VHL mutations as well as the type of mutation and affected exon and codon were computed for all 235 cases.

PCR-SSCP as a screening tool preceding direct sequencing
To evaluate whether PCR-SSCP could be used to distinguish between wildtype and mutated VHL, we determined the VHL status of 20 cases by SSCP and direct sequencing ( Table 2). SSCP and direct sequencing were mostly in agreement. In all cases where disparate results were obtained, this concerned aberrant signals in PCR-SSCP, followed by a negative result by direct sequencing ( Table  2). Most importantly, we never found a mutation in the sequence analysis after a negative PCR-SSCP (0% falsenegatives) ( Table 2). However, we are aware that the sampling size is limited, and it is known that SSCP as a screening tool is neither 100% sensitive nor specific. Therefore, we estimated the chance of a positive result on direct sequencing after a negative result on the SSCP (i.e., the chance that a mutation will be missed by using SSCP as a screening instrument). The point estimate for the percentage of false-negatives based on SSCP is 0%, with an upper confidence limit between 3.7% and 16.9%, based either on the individual analyses (N = 97), or on the number of cases (N = 20) (see Additional file 1: Calculation of the estimated upper 95% confidence limit). Different types of mutations were identified by SSCP followed by direct sequencing in this pilot. VHL gene mutations were mostly observed in clear-cell RCC, but also in unclassified RCC (2/4); papillary RCC (2/30); chromophobe RCC (3 mutations in 2/7 patients) and oncocytoma (1/6) (see Additional file 2: Description of observed mutations (N = 139), histological parameters and personal characteristics for 121 cases). The percentage of patients with a mutation was significantly higher for clear-cell tumors compared to tumors of other histological types (χ 2 : 32.9; p-value<0.001). In the remainder of this paper, we will only consider clear-cell RCC, leaving 204 outcomes (131 mutations (in 114 patients) and 73 wildtype) for 187 patients with 188 primary tumors. Nine outcomes were mutations in intronic sequences, possibly affecting splicing. The majority of observed mutations in coding regions lead to a truncated VHL product (insertion/deletion mutations leading to a frameshift and nonsense mutations, 62/122) or to a deleted/inserted or altered amino acid, 47/122). The percentage of point mutations was much higher in exon 1 (50.7%) than in exon 2 and 3 (29.6 and 23.1, respectively). Figure 1 shows the type of mutation plotted against the codon number. We observed 62 truncating mutations, which consisted of 4 nonsense mutations and 58 deletions or insertions leading to a shift of the reading frame ( Figure 1A), 15 insertions or deletions that did not affect the reading frame ( Figure 1B), 32 missense mutations ( Figure 1C) and 13 silent mutations ( Figure 1D).  Table 3. We did not observe a difference in nuclear grade, T, N, M, or stage between mutated and wildtype tumors.

Discussion
The von Hippel-Lindau (VHL) gene is a tumor suppressor gene predisposing to both sporadic clear-cell (conventional) RCC and von Hippel-Lindau disease. We analyzed 235 cases of primary sporadic RCC, of which 187 cases presented with the histological subtype clear-cell RCC, identified within the Netherlands cohort study on diet and cancer. VHL mutations were assessed by PCR-SSCP, followed by direct sequencing when aberrant signals were detected in SSCP. Screening by SSCP was considered appropriate, since we never detected a mutation by direct sequencing after a negative result on the SSCP in a pilot study. Furthermore, we estimated the upper 95% confidence limit for cases missed at 3.7% or 16.9%, based on the individual analyses (N = 97) or on the number of cases (N = 20). Therefore, mutated cases may have been overlooked, which placed them in the wildtype group. This would have resulted in a reduction of contrast between wildtype and mutated groups.
We observed 131 mutations in 114 out of 187 patients (61%) in patients with clear-cell tumors. Estimates reported in the literature [2,[20][21][22][23][24][25][26][27][28][29][30] range from 36% [25] to 54% [27], and therefore our estimate appears high. However, our estimate is close to the estimate of 75% as discussed in the review by Cohen [8]. This difference in the observed proportion of mutations in the VHL gene may be attributed to the use of different techniques, e.g., SSCP vs. direct sequencing, to contamination with normal Type of mutation plotted against the codon number  tissue components, or to the type of material analyzed (fresh or archival paraffin material). In our study, we used SSCP followed by direct sequencing, as most investigations did [2,[20][21][22][23][24][25][26][27][28][29]. To avoid ignoring VHL mutations through contamination with normal tissue, samples needed to contain at least 10% tumor cells. The vast majority of samples (180 out of 190) contained at least 50% viable tumor tissue. The sample with the lowest percentage of viable tumor tissue, in which we observed a mutation, contained 30% of viable tumor tissue. For 2 samples the percentage of viable tumor tissue was lower (10% and 20%), thus we cannot formally exclude the possibility that we overlooked an existing mutation. We identified cases in a population-based study after 11.3 years of follow-up; the time span and the number of patients and hospitals involved excluded the possibility to collect fresh material. Three other studies on sporadic RCC used paraffin embedded material and the percentages of mutations detected were 46% [29], 50% [22] and 54% [27].
The occurrence of various types of mutations is comparable to other studies of this size [21,24,26,27], but some differences are also apparent. The presence of silent mutations is not commonly reported; only Ma et al. [27] reported 13% of silent mutations, similar to 11% observed by us. It is possible that others do not report silent mutations because the investigators do not wish to designate these as true mutations since their relevance is unclear. The percentage of frameshift mutations was approximately 50% in most studies [21,26], comparable to the 48% observed in the current study. The percentage of in-frame deletions/insertions was higher in our study  [17] † Based on the review by one experienced pathologist (CAHK) ‡ According to Fuhrman [19] §1-2 vs. 3-4 ¶1-2 vs. 0 ** Excludes "X (unknown) " category (12%) as compared to the percentage in other studies (<5%) [21,26]. Compared to the distribution of mutation data present in the Universal VHL Mutation database with 747 mutations [31], the relative amount of point mutations is much higher in the database (64%), compared to the percentage of point mutations we observed (44%), as well as the frequency of nonsense mutations (11% vs. 3%). However, many of the mutations in the Mutation database concern germline mutations identified in VHLfamilies. It has been suggested that the mutational spectrum between germline and sporadic VHL mutations may differ [25], which could explain the disparity. We were not able to select records for sporadic tumors in the database, but Gallou et al. [26] described mutations for 145 sporadic cases from this database. Compared to our results, the percentage of nonsense mutations was higher (8% vs. 3%) and the percentage of in-frame deletions or insertions was lower (4% vs. 12%) [26].
Furthermore, we searched the Universal VHL Mutation database [31] for the mutations observed in the cases with a positive family history of RCC. The T>A (silent) mutation at codon 139 (sample id 1003) was not reported before, the deletion of 1 nucleotide at codon 146 (sample id 306) was recorded in the database twice, while the second 1 nucleotide deletion at codon 167 (sample id 1600) had not been reported before.
For 10 RCC cases, we observed different VHL genotypes in different tumor samples obtained from one tumor. Samples were re-analyzed to exclude technical errors. Again, different VHL genotypes, identical to the earlier genotypes, were observed, showing that tumor heterogeneity appears to play a role. This was an unexpected finding in view of the general acceptance that a mutation in the VHL gene is an early event in clear-cell RCC carcinogenesis, because one would then expect monogenetic VHL expression. Nevertheless, it is possible that genetic instability leads to additional mutational events, eventually leading to different VHL genotypes in clear-cell RCC. This was underscored by 13 cases showing 2 mutations within the same paraffin block and other studies have also reported multiple mutations per primary tumor [21,27,29].
The distribution of histological subtypes was comparable to the distribution described before [ [32], p2692]. Surprisingly, we also found VHL mutations in some of the non-clear-cell RCC samples that were collected. These results are in contrast with the current opinion that VHL mutations are exclusively restricted to clear-cell RCC. VHL mutations, however, have also been reported in chromophobe RCC, although these were clustered in the 5'UTR/ promoter region [22], and Brauch et al. [12] described a renal oncocytoma with a VHL mutation. To our knowledge, others have not yet described VHL mutations in pap-illary RCC. It is conceivable that genetic instability following tumor-initiating events leads to VHL mutations acquired at a later stage of tumor development. As expected, the relative occurrence of VHL mutations in the non-clear-cell RCC renal tumors was much lower than in clear-cell RCC, emphasizing the crucial role of VHL mutations in clear-cell RCC carcinogenesis.
The high percentage of mutations observed, the observation of mutational heterogeneity (different genotypes in 1 or different samples from the same tumor), and the observation of mutations in other histological types, raises the question whether paraffin material is suitable for analysis. It has been shown, however, that VHL gene mutations can be detected in archival paraffin material [33]. Identical mutations were observed in exon 2 in 3 cases for which archival paraffin material and tumor derived cell lines were assessed by SSCP and sequencing [33]. On the other hand, it has also been reported that artifacts may occur when archival paraffin material is used [34]. Williams et al. reported up to 1 mutation artifact per 500 bases recorded. The chance of such artificial mutations in formalin-fixed material was inversely correlated with the number of cells used in the PCR: the fewer the cells, the more artifacts. No artifacts were observed when 300 cells were used and only 1 artifact was observed in 150 cells (0.03%). The highest frequency of artifacts was observed when using 10 or 20 cells, and equaled 0.2 % [34]. The number of cells in our analyses, however, was large (>500 cells), since 5 20-µm slices of paraffin embedded material were used and biopsies were excluded.
The mean tumor size was larger for mutated tumors than for wildtype tumors, however not statistically significant. Two other studies did not show a difference in tumor size [24,28]. VHL gene mutations leading to reduced or inactive VHL protein (pVHL) could theoretically lead to a larger tumor size. In the absence of pVHL, hypoxia inducible factor becomes stabilized and upregulates a myriad of hypoxia-inducible genes, resulting in hypervascular tumors [35]. This should be a growth advantage. Our results and results from others [24,28] showed no association with nuclear grade (according to Fuhrman) [19]. It was difficult to compare TNM and stage, since different classifications were used in different studies and the TNM classification has changed definitions in recent years. We reviewed all cases and classified them according to the 1987 version of the TNM classification [17]. We did not observe a difference in T, N, M, or stage between mutated and wildtype tumors, as was confirmed by 2 other studies [24,28]. One study, however, reported a higher percentage of mutations in pT3 tumors with a VHL mutation or hypermethylation, compared to tumors without VHL alteration (64% vs. 40%) [21].
Phenotype should also be considered. Mutations may impair pVHL function. However, in order for pVHL to lose its function, biallelic VHL defects have to occur. We did not measure LOH, since it is well documented that LOH commonly occurs in RCC cases, with estimates ranging from 74% [30] to 93% [21]. Kondo et al. showed 97% LOH in mutated or hypermethylated samples compared to 81% LOH in wildtype samples [24]. Furthermore, silent mutations do not lead to an altered protein, but these types of mutations were not common (13/131) in our study. Also, pVHL formation may be inhibited by epigenetic silencing, which has been reported to occur in 19% of clear-cell RCC [9]. This could have resulted in misclassification in comparing pathological parameters, because samples with epigenetic silencing only were classified as wildtype tumors, as we were not able to investigate epigenetic silencing in the current study. Finally, cells in cell culture have been shown to produce two types of proteins: pVHL30 and pVHL19 (a results of internal translation from the second methionine within the VHL open reading frame (Met 54) [36][37][38]. The pVHL19 product has been shown to inhibit the production of hypoxia-inducible genes, but does not bind to fibronectin [36,37]. Reintroduction of pVHL19 has been shown to inhibit tumor formation in nude mice [37]. We detected mutations 5' of codon 54 in 11 cases; these are unusual and would be predicted to affect pVHL30, but not pVHL19 translation products. However in 5 of these 11 cases, a second mutation was observed 3'of codon 54. So, in 5 of these cases the other mutation may be relevant. For the other 6 cases, these mutations may not lead to a functionally inactive protein, and thus we may have misclassified these. However, as remarked before, the pVHL19 variant does not bind to fibronectin and thus may not contribute to the maintenance of the extracellular matrix.

Conclusion
VHL mutations were observed in 61% of sporadic clearcell RCC. No differences were observed in nuclear grade, TNM distribution or stage between VHL mutated and wildtype clear-cell RCC. However, the tumor size was larger for VHL mutated clear-cell RCC. This study forms a good basis to study possible associations between potential risk factors and VHL mutations in clear-cell RCC. Ultimately, this should lead to greater insight in the etiology of clear-cell RCC.