Genetic and Epigenetic Alterations in the von Hippel-Lindau Gene: the Influence on Renal Cancer Prognosis

Renal cancer is estimated to account for 2.3% of all cancer-related deaths in Europe in 2006; this corresponds with ∼26,900 patients (1). The most common subtype of renal cancer is the clear-cell type which accounts for >75% of all cases (2). An early event during the evolution of clear-cell renal cell cancer (ccRCC) is loss of function of the von Hippel-Lindau (VHL) gene, a tumor suppressor gene located on 3p25, that is involved in cell cycle regulation, regulation of hypoxia-inducible genes, and fibronectin assembly in the extracellular matrix (3–6). Biallelic VHL inactivation is common in sporadic ccRCC (7, 8). It is estimated that ∼50% to 70% of clear-cell renal tumors harbor a mutation in VHL (3, 6). Mutations usually lead to a shortened, inactive protein. In another 10% to 20% of the clear-cell tumors, the promoter region of the VHL gene is hypermethylated (4, 5, 8, 9). In addition, loss of heterozygosity of the VHL gene has been detected in >90% of ccRCCs (10, 11). Reintroduction of the wild-type VHL protein in VHL-deficient cells can suppress tumor growth (12, 13). This finding suggests that the VHL gene has a critical role in tumorigenesis and is a possible target for treatment.

Renal cancer prognosis is estimated using several clinical variables such as tumor-node-metastasis (TNM) stage, tumor grade, presence of necrosis, and/or vascular invasion in the primary tumor (14). Lately, attention has been drawn to molecular markers in an attempt to add knowledge on molecular tumor features to the accuracy of ccRCC prognosis estimation (14). However, molecular markers have not yet gained much use in the management of patients with renal cancer (15). Few studies have assessed the influence of VHL alterations on renal cancer prognosis even though this information could be of critical importance in the identification of patient subgroups with a poor prognosis. Poorer overall and progression-free survival has been reported in patients with somatic loss of function VHL mutations in two studies (7, 15). One study assessed the influence of VHL through either somatic mutations or hypermethylation, reporting a possible trend for increased microvascular invasion based on mutation location, however, in a small study population (2). One other study analyzed the combined influence of VHL mutations and methylation and reported a better prognosis with VHL alterations (16). In our population, we previously reported an association, although not statistically significant, between mutations in VHL and tumor size (17).

False-positive results are common in research on molecular markers, often due to small population sizes or small effect sizes (18–20), and initial promising results can often not be replicated in later studies (20, 21). There is a clear need for more and larger studies on the influence of VHL alterations on renal cancer prognosis to determine whether this molecular marker is useful as an independent prognostic factor in renal cancer (2). The ability to identify patient subgroups with a poor prognosis based on molecular characteristics of the tumor would enable clinicians to anticipate this fact, and perhaps change treatment strategies, or develop new treatments targeting specific pathways (2). In the present study, we assessed the influence of VHL alterations through somatic mutations or hypermethylation in a large population-based case group.

From 1986 to 1997, 355 incident kidney cancer cases (ICD-O-3:C64.9) were identified within the Netherlands Cohort Study on diet and cancer. This cohort study started in 1986 with 120,852 healthy men and women. Incident cancer cases were identified by computerized linkage with the Netherlands Cancer registry and PALGA, a nationwide network and registry of histopathology and cytopathology (22). From the 355 kidney cancer cases, urothelial cell carcinomas were excluded and only histologically confirmed renal cell cancers were included (ICD-O: M8010-8119, 8140-8570), leaving 337 cases.

Tissue samples. Tumor material and healthy tissue samples from patients with kidney cancer were collected at the time of diagnosis after approval by the Ethical Review Board of Maastricht University, the Netherlands Cancer registry, and PALGA. Tissue collection has been described in detail elsewhere (17). For 273 of the 337 eligible cases, a PALGA record with information on the location of paraffin-embedded tissue samples was available. We were able to collect 251 tissue blocks from 51 pathology laboratories throughout the Netherlands.

Clinical characteristics. All HE-stained slides were revised by an experienced genitourinary pathologist resulting in a consistent classification of histology according to the WHO classification of tumors from 2002 (23). Nuclear grading was done by the same pathologist according to Fuhrman et al. (24). 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. Stage was defined by the Cancer Registry based on the fourth edition of the International Union Against Cancer-TNM (25), and was checked and completed in case of missing information using the pathology reports.

Material from 16 cases was discarded after revision because the material was not suitable for DNA analyses, leaving 235 cases with available DNA material. No statistically significant differences were found for baseline characteristics such as age, sex, and cancer stage between these cases and the 102 cases for which a tumor block was unavailable. Five cases were excluded from this study as they were diagnosed with renal cancer during obduction. As a result, DNA from 230 cases was available for further analysis. Because VHL aberrations are common in clear-cell renal carcinomas but rare in other histologic subtypes (17), we restricted our analyses to patients with clear-cell renal cancer. This resulted in 185 cases that were included for the present study.

Follow-up. Information on sex was retrieved from the Netherlands Cohort Study database, information on tumor localization and incidence date was retrieved from the Netherlands Cancer registry. Tumor size was retrieved using the pathology reports. Information on vital status until December 2005 was obtained from the Central Bureau of Genealogy and the municipal population registries, causes of death were retrieved through linkage with the Central Bureau for Statistics. The vital status and causes of death for all cases could be obtained.

Analyses were done on the 185 cases with clear-cell renal cancer. DNA isolation and mutation analyses have been described previously (17). Briefly, paraffin was removed with xylene and DNA was extracted by salt precipitation. The entire gene was amplified using six primer stets. Samples were first subjected to PCR single-strand conformational polymorphism analysis, which was followed by direct sequencing in case of aberrant or equivocal results. Mutations were identified by visual inspection of sequences provided by the ABI basecaller. Mutations were considered to be a loss of function mutation if the mutation resulted in an inactive protein.

DNA methylation in the CpG islands of the VHL gene promoter was determined by chemical modification of genomic DNA with sodium bisulfite and subsequent methylation-specific PCR as described in detail elsewhere (26). In brief, 500 ng of DNA was denatured by sodium hydroxide and modified by sodium bisulfite. DNA samples were then purified using Wizard DNA purification resin (Promega), again treated with sodium hydroxide, precipitated with ethanol and resuspended in water.

To facilitate methylation-specific PCR analysis on DNA retrieved from formalin-fixed, paraffin-embedded tissue, DNA was first amplified with flanking PCR primers that amplify bisulfite-modified DNA but do not preferentially amplify methylated or unmethylated DNA. The resulting fragment was used as a template for the methylation-specific PCR reaction (26–28).

All PCRs were done with controls for unmethylated alleles (DNA from normal lymphocytes), methylated alleles (normal lymphocyte DNA treated in vitro with SssI methyltransferase; New England Biolabs), and a control without DNA. Ten microliters of each methylation-specific PCR reaction was directly loaded onto nondenaturing 6% polyacrylamide gels, stained with ethidium bromide and visualized under UV illumination. Methylation-specific PCR analyses were successfully done for 147 out of 185 patients (79.5%).

VHL alteration was defined as a loss of function mutation (i.e., a mutation that causes a change in the protein versus silent or no mutations) or as a methylated VHL gene. Cause-specific survival was defined as the time from cancer diagnosis until renal cancer–related death or until the end of follow-up. Differences in baseline characteristics between the cases with and without VHL alterations were evaluated with Student's t tests and χ² tests. Kaplan-Meier curves and Wilcoxon tests were used to estimate the overall influence of VHL loss of function mutations, VHL hypermethylation and VHL alterations through mutations or methylation on cause-specific survival. Hazard ratios (HR) and corresponding 95% confidence intervals (CI) were assessed by use of Cox proportional hazard models. HRs were calculated for the association between VHL inactivation and survival for the total group of patients. In addition, analyses stratified for sex and cancer stage were done. Factors were considered possible confounders if they were known prognostic factors for renal cancer and influenced the crude HR. Possible confounders that were included in the model were sex, age at diagnosis, cancer stage, tumor size, and nuclear grading. The proportional hazard assumption was tested using the Schoenfeld residuals and the log(-log) hazards plots. All analyses were done by use of the statistical package STATA 9.0.

Table 1 presents baseline characteristics for the 185 cases that were included in the analyses and stratified for VHL alterations. Overall, the larger part of the population were male patients (59.5%) without a family history of renal cancer (98.9%). During follow-up, 52.4% (97 patients) died of a renal cancer–related cause. The median survival in the total population was 6 years. After 5 years, 105 (56.8%) cases were still alive, 59 (31.9%) were alive after 10 years, and 15 (8.1%) were still alive after 15 years.

In the total population, 52.4% (97 of 185) of the patients had a mutation in VHL that was considered a loss of function mutation, whereas 10.9% (16 of 147) of the patients showed hypermethylation in the VHL gene. Baseline characteristics did not differ between patients with and without VHL alterations. The frequency of loss of function mutations and methylation, separately or combined, did not differ between different cancer stages or between tumors with different sizes (data not shown).

Figure 1A shows the Kaplan-Meier curve for a loss of function mutation in VHL and cause-specific survival. No statistically significant difference was observed in survival between patients with and without a loss of function mutation in VHL (Wilcoxon, P = 0.392). The overall survival curve for VHL methylation is presented in Fig. 1B. Survival was not different for patients with or without methylation of the VHL gene (Wilcoxon, P = 0.439). The influence of total VHL alterations through mutations or methylation on cause-specific survival is presented in Fig. 1C. No difference was observed in cause-specific survival between patients with or without VHL alterations (Wilcoxon, P = 0.461).

Results from the univariate Cox proportional hazard analyses are presented in Table 2. VHL loss of function mutations and VHL methylation do not statistically significantly influence survival (mutation HR, 1.07; 95% CI, 0.72-1.61; methylation HR, 1.28; 95% CI, 0.64-2.59; and total VHL alterations HR, 1.12; 95% CI, 0.74-1.68). Results from the multivariate analyses are also presented in Table 2. Neither loss of function mutations nor methylation in the VHL gene had a statistically significant influence on cause-specific survival (VHL mutation HR, 1.02; 95% CI, 0.66-1.58; VHL methylation HR, 1.04; 95% CI, 0.48-2.22; total VHL alterations HR, 1.08; 95% CI, 0.69-1.68).

Clinicopathologic variables did have a significant influence on clear-cell renal cancer survival. Renal cancer mortality risk was increased with a 1-year increase in age at diagnosis (HR, 1.06; 95% CI, 1.01-1.12, P = 0.017) in the multivariate analyses (Table 2). Mortality risk was also increased for patients with a higher cancer stage (stage III: HR, 2.16; 95% CI, 1.25-3.74; stage IV: HR, 5.51; 95% CI, 2.96-10.27) and with a higher nuclear grade (grade 3: HR, 2.53; 95% CI, 1.28-4.98; grade 4: HR, 3.97; 95% CI, 1.93-8.19; Table 2) as compared with patients with the lowest cancer stage and a low nuclear grade. Sex and tumor size, although not statistically significant, also seemed to influence cause-specific survival (Table 2).

Analyses stratified for sex showed a small difference between men and women for ccRCC survival with or without loss of function mutations in VHL (men HR, 1.02; 95% CI, 0.55-1.88; women HR, 1.11; 95% CI, 0.54-2.29), with or without VHL methylation (men HR, 0.65; 95% CI, 0.19-2.23; women HR, 1.89; 95% CI, 0.64-5.61) and with or without VHL gene alterations (men HR, 1.04; 95% CI, 0.55-1.96; women HR, 1.18; 95% CI, 0.58-2.40). Results were, however, not statistically significant, and the interaction between sex and alterations in the VHL gene also did not reach statistical significance. Additional analyses stratified for cancer stage did not alter our conclusions. VHL alterations did not influence survival in cancer stages I, II, or III (stage I and II HR, 0.98; 95% CI, 0.45-2.16, P = 0.965; and stage III HR, 0.93; 95% CI, 0.42-2.04, P = 0.856). For stage IV patients, the hazard ratio for mortality was increased for patients with VHL alterations, but this was not statistically significant (HR, 1.64; 95% CI, 0.59-4.55, P = 0.340).

The proportional hazard assumption was tested using the Schoenfeld residuals. This statistical test revealed no violation of the proportional hazard assumption (P = 0.318). Inspection of the log(-log) hazard plots revealed that the distance between the independent curves was very small, and thus, the curves cross at several points. Plots show that the curves are slightly different early after diagnosis, but converge after a longer follow-up period. To inspect the influence of a shorter follow-up period, we calculated hazard ratios for the risk of dying within 5 years after diagnosis. Hazard ratios for this analysis showed a slightly increased risk of dying of renal cancer within 5 years after diagnosis, but this was not statistically significant (1.30; 95% CI, 0.76-2.24 for VHL mutations; 1.05; 95% CI, 0.45-2.44 for VHL methylation; and 1.30; 95% CI, 0.75-2.25 for VHL alterations).

In the current study, we tested the hypothesis that alterations in VHL were associated with a poorer cause-specific survival in ccRCC. We examined the influence of loss of function mutations and methylation, separately and combined, in the VHL gene on cause-specific survival in 185 ccRCC patients from a large population-based study (Netherlands Cohort Study) with a long follow-up. We observed no association between VHL alterations (i.e., loss of function mutations or gene methylation) and cause-specific survival in ccRCC, neither for mutations and methylation separately nor for those combined. Other factors that have previously been proven to be prognostic factors, such as TNM stage, Fuhrman grading, and tumor size (29–32) were also observed to be relevant for prognosis in our population. Our data suggest that VHL status does not provide any useful prognostic information for patients with ccRCC. This conclusion is in accordance with one of the previous studies on VHL alterations and survival (2).

Few studies investigated the clinical relevance of VHL gene alterations, although this information could be important in terms of prognosis. Moreover, previous studies on the influence of VHL alterations on survival yielded contradictory results. Two studies reported a poorer overall survival in patients with somatic loss of function VHL mutations (7, 15). Two studies assessed the combined influence of VHL mutations and methylation, one study reported a better prognosis with VHL alterations (16). However, this could not be replicated in another recent study (2).

In our population, we observed a 5-year survival rate of 56.8%, which is in accordance with survival rates observed previously in the Netherlands (33).⁶

Inactivation of the VHL gene is considered as an early event in the tumorigenesis of ccRCC (4). Previous studies report that ∼50% to 70% of clear-cell renal tumors harbor a mutation in VHL whereas another 10% to 20% seem to be hypermethylated (4, 5, 8). In our population, loss of function mutations in VHL were found in 52.4%, whereas in 10.9% of the ccRCC patients, we found aberrant methylation of VHL. Overall, 57.3% of the patients either had a functional mutation in VHL or aberrant methylation of VHL. Our findings are in accordance with the ranges mentioned in the literature.

A previous study observed an increased frequency of VHL mutations and methylation in higher cancer stages (11). However, we could not confirm this finding, as we observed no statistically significant differences in tumor stage or grade between cases with and without VHL alterations. A recent study could not replicate this finding either (2). In addition, we did not find a statistically significant difference between VHL alterations in tumors with different sizes. This observation supports the hypothesis that VHL alterations occur early in tumorigenesis as these alterations are found in tumors of all sizes. Previously, a better survival was seen in Japanese patients with stage I, II, and III cancer with VHL mutation or methylation (16), and particularly in patients with a higher grade and/or a higher stage. In our population, we did not observe any differences in cause-specific survival and VHL alterations.

For VHL mutational analyses, we included functional mutations only, i.e., mutations that cause disruption or a block of the protein. Previously, all mutations in this population-based case group were described (17). For the current study, we included functional mutations only because the relevance of silent mutations is unclear. Silent mutations, however, were not common in our population (8%). VHL methylation status could be determined for 79.5% of the cases. For the remaining cases, methylation status could not be determined due to lack of sufficient DNA.

Renal cancer cases from this study originated from the Netherlands Cohort Study on diet and cancer, a study that started in 1986 with >120,000 healthy subjects. Selection bias is unlikely due to this approach because we only included incident cancer cases from the total cohort. Tumor tissue was collected after surgical removal of the tumor. Thus, all patients in our population underwent surgical treatment. It is known that not all patients undergo surgery, a small number of patients, mostly those with stage IV cancer, do not receive surgical treatment. These patients were not included in our study because tumor tissue was not available for these patients.

Cancer stage in our population was based on the TNM classification. This classification has changed definitions over the last years. The patients in this study were classified according to the 1987 version of the TNM classification (25). Although we realize that the TNM classification used is an older classification, it was not possible to use the most recent version of the TNM (34) because of changes in the N-classification. We did include information on tumor size, which is included in the current T-classification, in our statistical analysis.

Even though our study is among the largest assessing the influence of VHL alteration on clear-cell RCC prognosis, the study population was relatively small. However, using an α of 5%, a power of 80%, a median survival of 6 years, and a follow-up until 2005, we had enough power to show a HR of 0.63 or 1.66 for dying because of renal cell cancer for cases with VHL alterations as compared with cases without VHL alterations (35).

Despite numerous studies on the relevance of genetic markers in oncology in general, the number of clinically useful markers is limited (21). Research on molecular markers is susceptible to publication bias and false-positive results, due to small population sizes and small effect sizes (18–20). Often, an initial study shows promising results that cannot be replicated in later studies (20, 21), as also seems to be true for the VHL gene in renal cancer prognosis. To our knowledge, this is one of the first studies to report the influence of both mutations and methylation in the VHL gene on the survival of patients with clear-cell renal cancer. Previous studies on this yielded contradictory results (2, 16). We did not observe an association between loss of function mutations and methylation in VHL on cause-specific survival in clear-cell renal cancer. Given these results, it seems that the inclusion of VHL status as a molecular variable for the assessment of prognosis or treatment of individual patients with ccRCC is of no value.

We are indebted to the participants of this study and further wish to thank the cancer registries (IKA, IKL, IKMN, IKN, IKO, IKR, IKST, IKW, IKZ, and VIKC), the Netherlands nationwide registry of pathology (PALGA) and the pathology laboratories for providing the tissue samples (for a complete list see ref. 15). We also thank Dr. E. Dorant, C.A. de Brouwer, Prof. Dr. A. Geurts van Kessel, and Prof. Dr. D.J. Ruiter for their preparatory work for this study; K. van Houwelingen and H. Gorissen for the laboratory analysis, Dr. A. Kester for statistical advice; S. van de Crommert, H. Brants, J. Nelissen, C. de Zwart, W. van Dijk, M. Jansen, and A. Pisters for assistance; and H. van Montfort, L. van den Bosch, and J. Berben for programming assistance.