Hypermethylation of the thrombospondin-1 gene is associated with poor prognosis in penile squamous cell carcinoma

Authors


David Guerrero, Biomedical Research Center, Navarra Health Service, Irunlarrea st. 3, 31008 Pamplona, Navarra, Spain.
e-mail: dguerres@cfnavarra.es

Abstract

OBJECTIVE

To evaluate the presence of human papillomavirus (HPV) infection, the methylation status in the promoter region of thrombospondin-1 (TSP-1), RAS association domain family 1A (RASSF1-A) and p16 genes, and the expression of TSP-1, CD31, p16 and p53 proteins in patients diagnosed with penile cancer, and the possible associations between these variables and clinical and pathological features.

PATIENTS AND METHODS

HPV types, gene promoter hypermethylation and protein expression were analysed by reverse line blot, methylation-specific polymerase chain reaction, and immunohistochemistry, respectively, in 24 penile squamous cell carcinomas.

RESULTS

HPV infection was detected in 11 of 24 cases (46%), and TSP-1, RASSF1-A and p16 genes were hypermethylated in 46%, 42% and 38% of the tumours, respectively. TSP-1 hypermethylation was associated with unfavourable histological grade (grade 3; P = 0.033), vascular invasion (P = 0.023), weak expression of TSP-1 protein (P = 0.041), and shorter overall survival (P = 0.04). TSP-1 expression was not associated with microvessel density. However, RASSF1-A hypermethylation was more frequent in T1 tumours (P = 0.01), and p16 hypermethylation was not associated with any of the tested variables except for absence of p16 expression (P = 0.022).

CONCLUSION

In summary, the epigenetic inactivation of TSP-1 and RASSF1-A genes is associated with pathological variables and seems to be of prognostic significance in penile cancer.

Abbreviations
HPV

human papillomavirus

SCC

squamous cell carcinoma

RLB

reverse line blotting

IHC

immunohistochemistry

MVD

microvessel density

TSP-1

thrombospondin-1

TSG

tumour-suppressor gene

RASSF1-A

RAS association domain family 1-A

MS

methylation-specific (PCR)

DFS

disease-free survival

OS

overall survival.

INTRODUCTION

Penile cancer is rare in developed countries and accounts for <1% of all male cancers [1]; it includes several histological types, with squamous cell carcinoma (SCC) being the most prevalent [2]. The mean age at presentation of penile cancer is 60 years and risk factors include phimosis, chronic inflammatory conditions, lichen sclerosus, smoking, ultraviolet irradiation, a history of warts or condylomas, and lack of circumcision during childhood [3]. Human papillomavirus (HPV) infection is causally involved in the progression of anogenital tract cancers, and is present in 21–63% of patients with penile cancer [1,4]. The dissemination pattern of penile carcinoma is primarily to the regional lymph nodes, and lymphatic spread affects >40% of patients [5]. Systemic, blood-borne metastasis is a late event and can affect the liver, heart, lung and bone. Some of the pathological factors related to prognosis are the site of the primary tumour, tumour size, tumour depth, lymphatic spread, vascular invasion, histological type, cell adhesion, and proliferation status [2,6].

One of the proteins involved in cell adhesion is thrombospondin-1 (TSP-1), a cell-adhesion glycoprotein secreted by several types of normal cells and by tumour cells [7]. It is normally present in quiescent vessels, and its level decreases in vascular formation. Consequently, it is considered a potent endogenous inhibitor of angiogenesis [8], and has been advocated in the treatment of solid tumours [9]. However, the TSP-1 protein promotes cell invasion and metastasis in breast cancer [10], suggesting a dual role of this protein in cancer.

Cell proliferation has a pivotal role in cancer, and there are many genes involved in its regulation, e.g. RAS association domain family 1-A (RASSF1-A) and p16 genes. RASSF1-A acts as a tumour-suppressor gene (TSG) encoding for a RAS effector with additional functions in microtubule stabilization and apoptosis [11]. p16 is a TSG that encodes an inhibitor of cyclin-dependent kinases 4 and 6 involved in the activation of the cell cycle [12].

The silencing of the expression of TSGs contributes to tumour progression [13] and is caused by alterations such as loss of heterozygosity and hypermethylation of gene promoters. TSP-1, RASSF1-A and p16 are examples of genes that are hypermethylated in several types of cancer, including breast, cervical and lung [14,15]. The increase of hypermethylation of RASSF1-A with progression has been described recently in uterine [16] and urothelial carcinoma [17]. The contribution of promoter hypermethylation of p16 to the disruption of the p16/cyclin D/RB pathway and cancer progression was found to be present in cervical cancer at an early stage [18–20], and in vulvar carcinoma [21].

The objectives of the present study were to detect HPV in patients with penile cancer, to determine the methylation status of the promoters of TSP-1, RASSF1-A and p16 genes, to analyse the expression of TSP-1, CD31, p16 and p53 proteins, and to correlate these variables with clinicopathological variables over a long follow-up.

PATIENTS AND METHODS

This retrospective study included a series of 24 patients with penile SCC examined between January 1988 and February 2007. This study was approved by the local ethics committee, and the clinical and pathological data were reviewed by two pathologists. The histological grading, the histological typing and the pathological staging including vascular invasion were determined following previous criteria [2,22,23]. Lymph node surgery was performed on the basis of stage in most cases [2]. The follow-up was generally by a physical examination, CT and chest radiography every 6 months during the first 2 years and annually thereafter.

For DNA extraction and β-globin amplification, four 5-µm thick tissue sections were deparaffinized in xylene and subjected to digestion with proteinase K (10 mg/mL) (Roche, Germany) in 50–400 µL buffer at 56 °C overnight. Proteinase K was inactivated at 96 °C for 10 min; 2 µL of the supernatant obtained at 13 400g for 15 min was used directly for PCR. To assess DNA quality, the β-globin gene was amplified by PCR using primers BGPCO3 and BGPCO5 (Sigma, St Louis, MO, USA) that generates a 209 bp product [24].

To detection and genotype HPV DNA, a fragment of 150 bp of viral DNA was amplified by PCR, using GP5+/GP6+ biotinylated primers. The reaction was performed in a 40-µL mixture containing 5 µL of PCR buffer, 1.875 mm MgCl2, 0.25 mm deoxynucleotides, 25 pmol of each primer, and 1 unit of AmpliTaq DNA Polymerase (Applied Biosystems, Foster City, CA, USA). The mixture was heated for 4 min at 94 °C, followed by 40 cycles (1 min at 94 °C, 2 min at 40 °C, and 1.5 min at 72 °C), and a final extension for 4 min at 72 °C.

Subsequently, HPV genotypes were determined using a non-radioactive reverse line blotting (RLB) procedure, as described elsewhere [25]. The method is based on hybridization of the GP5+/6+ PCR products of up to 43 samples, including positive (HPV-16 positive) and negative samples (HPV-negative and water) controls, with specific oligonucleotide probes containing a 5′-amino group for 36 different HPV subtypes, on a carboxyl-coated Nylon membrane (Biodyne C, Pall Corp., East Hills, NY, USA) in a miniblotter (MN45, Immunetics, Boston, MA, USA). In all, 37 probes were used for the identification of high-risk types (HPV-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, -59, -68, -73, -82/MM4, and -82/IS39), probable high-risk types (HPV 26, -53 and -66) and low-risk types (HPV 6, -11, -34, -40, -42, -43, -44, -54, -55, -57, -61, -70, -71/CP8061, -72, -81/CP83104, -83/MM7, -84/MM8 and -CP6108), according to previous data [26]. These probes were kindly provided by Dr X. Bosch (ICO, Spain). Hybridization was followed by incubating the membrane with streptavidin-POD-conjugate (Roche) and enhanced chemiluminiscence detection (GE Healthcare, USA).

The methylation status of TSP-1, RASSF1-A and p16 genes was determined by methylation-specific PCR (MS-PCR), as previously described [27]. This method included a treatment with sodium bisulphite, chemically modifying unmethylated, but not methylated, cytosines to uracil. Specific primers (Sigma) used to amplify regions of interest were: For the TSP-1 gene, unmethylated reaction: 5′TTGAGTTTGTGTGGT GTAAGAGTAT3′ (forward primer) and 5′CCCCACTACCTAACACACAACT3′ (reverse primer); methylated reaction: 5′GTTCGCGTGG CGTAAGAGTAC3′ (forward primer) and 5′CGCTACCTAACGCGCAACT3′ (reverse primer); for the RASSF1-A gene, unmethylated reaction: 5′TTTGGTTGGAGTGT GTTAATGTG3′ (forward primer) and 5′-CAAACCCCACAAACTAAAAACAA-3′ (reverse primer); methylated reaction: 5′-GTGTTAACGC GTTGCGTATC-3′ (forward primer) and 5′-AACCCCGCGAACTAAAAACGA-3′ (reverse primer); for the p16 gene, unmethylated reaction: 5′TTATTAGAGGGTGGGGTGGATTGT3′ (forward primer) and 5′CAACCCCAAACCCACA ACCATAA3′ (reverse primer); methylated reaction: 5′TTATTAGAGGGTGGGGCGGATCGC3′ (forward primer) and 5′GACCCCCGAACCGCG ACCCTAA3′ (reverse primer). The annealing temperature for the unmethylated and methylated reactions was 58 °C and 60 °C for TSP-1 analysis, respectively, and 62 °C and 66 °C for p16. The annealing temperature for both methylated and unmethylated reactions was 60 °C for RASSF1-A. DNA from lymphocytes treated in vitro with SssI methylase (New England Biolabs, Ipswich, MA, USA) was used as a positive control for methylated alleles of the three genes. DNA from normal lymphocytes was used as a negative control. The presence of PCR products was verified with 2% agarose gel electrophoresis, stained with ethidium bromide and examined under ultraviolet illumination.

For immunohistochemistry (IHC), sections of 4 µm from paraffin-embedded tissue were rehydrated and antigen retrieved by incubating in sodium citrate buffer, pH 6.0, at 95 °C for 30 min. After endogenous peroxidase blocking with 3% H2O2, slides were incubated with mouse antibodies against TSP-1 (clone A6.1, dilution 1:50; Neomarkers, Fremont, CA, USA), p16 (clone 16P07, pre-diluted, LabVision, Thermo Fisher Scientific, Fremont, CA, USA) and p53 (clone BP53-12-1, 1:100; Novocastra, Newcastle, UK) for 45 and 30 min, respectively, at room temperature. Sections were then incubated with biotinylated polyclonal IgG at a dilution of 1/200 for 30 min at room temperature. Avidin-biotin-horseradish complex and diaminobenzidine were used for colour development; slides were counterstained with Harris haematoxylin. Incubation in PBS constituted the negative control. Placental tissue, a positive glioma and a positive colon tumour were used as the positive control for TSP-1, p16 and p53 proteins, respectively.

IHC results were semiquantitatively scored according to following previous criteria [28,29]. The intensity of TSP-1 in the cytoplasm and the p16 staining in cytoplasm and nucleus of tumour cells were scored into four categories, i.e. 0, negative (no positive cells); 1, weakly positive (1–50% of positive cells); 2, intermediate (51–75%); 3, strong (76–100%). Nuclear expression of p53 in tumour cells was scored positive (+) if present in ≥20% of the tumour cells.

The angiogenesis was studied using a monoclonal mouse CD31 antibody (clone M0823, dilution 1:30; Dako, Germany) incubated for 45 min at room temperature. Microvessel density (MVD) was determined at ×200 as the mean counts of CD31-positive endothelial cell clusters present in the three areas of the specimens with the most vessels, in accordance with a standardized procedure [30]. Large vessels with muscular walls or areas of fibrosis, necrosis and inflammation were excluded from the analysis. To evaluate results of MVD, the median value of the counts was used as the limit between low and high MVD. This strategy was also used for the analysis of invasion depth. Results were scored independently by two pathologists with no previous information about the cases.

Comparisons in distribution of pathological and clinical categorical variables were analysed by the chi-square test or Fisher’s exact test. Kaplan-Meier survival curves were constructed and differences in disease-free survival (DFS) and overall survival (OS) between groups were evaluated using the univariate log-rank test. We used a Cox proportional hazard model to calculate hazard rate and CIs; the level of statistical significance was set at P < 0.05.

RESULTS

The clinicopathological data of the patients are summarized in Table 1; the mean (sd, range) age was 67.6 (12.3, 28–81) years. All the patients were treated by primary surgical resection; 19 by partial penectomy and five by total penectomy (Fig. 1a). In all cases the surgical margins were negative for invasive carcinoma. Lymph nodes were available for examination in 14 of 24 tumours (58%); seven (29%) had inguinal lymph node metastasis at the time of surgery. None of the patients had distant metastasis at the time of surgery. No therapy was considered before the cancer progression.

Table 1.  The clinicopathological characteristics of the patients
No/age, yearsTumour size, mmStage*HistologicalInvasion depth, mmVascular invasionLocal recurrenceDistant metastasisDFS, monthsOS, monthsFollow-up, months
TypeGradeLNI
  • *

    pT stage according to TNM 1997;

  • †b, basaloid, cs, conventional squamous; v. verrucous; w, warty.

  • ‡Lymph node involvement; ne, not explored.

1/6410IIcsIne10NoYesNo108228228
2/6430IbIne1NoNoNo204204204
3/5830IIwINo10NoYesYes120144144
4/6925IcsIIne5NoNoNo132132132
5/6340IIcsIIINo15YesNoNo120120120
6/8125IcsIIINo1NoYesNo24120120
7/2815IcsIne0NoNoNo108108108
8/6020IcsIYes12NoNoNo108108108
9/6922IvIne0NoNoNo108108108
10/5615IIwINo2NoYesNo15108108
11/7120IcsIYes1NoYesYes729696
12/7130IIcsIIne10NoYesYes727272
13/7520IIIwIINo10YesNoNo727272
14/7135IIbIIIYes35YesNoNo727272
15/8130IIcsIne20NoNoNo484848
16/8060IIIcsIIIYes20YesNoNo484848
17/7012IcsIne1NoNoNo363636
18/8020IIcsIINo10NoYesYes363636
19/7915IcsIIne2NoYesNo222424
20/7320IIcsIIIYes15YesYesYes181818
21/4545IIwIIYes9NoYesYes121818
22/7335IIIcsIIIne20YesYesYes121212
23/6550IVcsIIYes30YesYesYes177
24/7711IIcsIINo5NoNoNo555
Figure 1.

a, Macroscopic view of the usual type of squamous penile tumour with exophytic growth. b, Microscopic view of a well-differentiated usual type of penile tumour (grade I) (×100, haematoxylin and eosin). c, IHC expression of TSP-1 protein in the cytoplasm of the tumour cells of the infiltrating component, showing strong (c, ×400) and weak expression (d, ×600). The high density of neoplastic vessels in the tumour is highlighted by CD31 staining (e, ×200). f, p53 expression in the nuclei of the tumour cells (×600); and strong (g, ×200) or no expression (h, ×600) of p16 in the nuclei and cytoplasm of the tumour cells.

The mean (range) tumour size was 26.2 (10–60) mm and the mean (median, range) invasion depth 9.78 (10, 0–35) mm. The tumour stages were T1, T2, T3 and T4 in nine (38%), 11 (46%), three (13%) and one patient (4%), respectively, and lymph node involvement was found in six of the 14 who had lymphadenectomy. The histological grade was I, II and III in 10 (42%), eight (33%) and six patients (25%), respectively (Fig. 1b). There were HPV-associated cell changes in the tissue surrounding the tumour (koilocytotic atypia and dysplastic lesions) and lichen sclerosus in two (8%) and five (21%) patients, respectively. Squamous hyperplasia was associated with cancer in eight cases (33%).

The values for the median follow-up, DFS and OS of the patients were 72, 60 and 72 months, respectively. During the follow-up 12 patients (50%) had a local recurrence and eight of these (35%) had both local recurrence and distant metastasis. The patients with local recurrence were treated with radiotherapy, and those with both local recurrence and distant metastasis were treated with radiotherapy and chemotherapy (5-fluorouracil).

Five patients (21%) died from penile cancer; four of these deaths were associated with ‘usual-type’ SCC and the other was associated with a warty carcinoma. One patient (4%) died from causes other than cancer.

All tumours were positive for β-globin amplification, and were included in the study. HPV was present in 11 of 24 patients (46%; Table 1), and HPV-16 was the most prevalent subtype (Fig. 2). HPV was less frequent in usual-type SCC (P = 0.085) and more frequent in patients with warty SCC (P = 0.031).

Figure 2.


RLB of the 24 penile tumours; the types tested are listed in the Methods. Ten of 24 patients (42%) were positive for HPV-16 (only), and one (4%) showed a combined infection of HPV-16 and -39 (patient 18); H2O, negative control (control of contamination); Ctl, negative control (HPV-negative patient); Ctl+, positive control (HPV-16).

To understand more about the molecular alterations associated with penile cancer, we analysed the methylation state of CpG in the promoter region of TSP-1, RASSF1-A and p16 genes, that were methylated in 11 (46%), 10 (42%) and nine tumours (38%), respectively (Fig. 3); the results are shown in Table 2. TSP-1 expression was scored as strong, intermediate, weak and negative in five (21%), nine (38%), eight (33%) and two (8%) tumours, respectively (Fig. 1c,d). Normal squamous epithelium in the suprabasal area was also positive for TSP-1 expression in all the histological types; in addition, keratinocytes above the basal layer in verrucous carcinoma were stained, and this staining was not evaluated due to its similarity with that of normal epithelium.

Figure 3.


MS-PCR of TSP-1, RASSF1-A and p16 genes in penile cancer. Both the unmethylated (U) and methylated (M) PCR products are shown for each case. TSP-1, samples 1 and 2 are methylated and unmethylated, respectively (a). RASSF1-A, samples 1 and 2 show unmethylated promoter, in contrast to sample 3 with methylated promoter (b); p16; patients 1 and 3 are unmethylated, and sample 2 is methylated (c). DNA from normal lymphocytes (NL) and in vitro methylated DNA (IVD) were used as the positive PCR controls for the unmethylated and the methylated reaction, respectively; H2O, negative control (control of contamination).

Table 2.  Molecular and IHC results
No.HPV subtypeTSP-1 methylationTSP-1 IHCRASSF1-A methylationp16 methylationp16 IHCp53 IHCMVD
  1. −, no expression. The median value of the counts was used as the limit between low (1) and high (2) MVD.

1/No2+NoNo1
2/No2+YesNo1++1
316No3+NoYes1
4/No2+YesYes+2
516Yes1+YesNo3+1
616Yes2+NoNo3++2
716No2+YesNo3+2
8/No1+NoNo+2
9/No2+YesYes1
1016Yes1+NoNo+1
11/YesYesNo+1
1216No3+NoNo3+2
1316, 39Yes1+NoYes+2
1416Yes1+NoNo3+1
15/Yes1+YesYes1+2
16/No3+NoNo1+2
1716No2+YesYes1+1
18/No3+NoYes+2
19/No1+YesNo1
20/Yes2+NoYes+2
2116Yes2+NoNo3+1
22/Yes1+NoNo+2
23/YesYesYes2
2426No3+NoNo3++1

Microvessel staining detected by CD31 expression was present in all tumours, with a median (range) of 15.8 (7.0–43.3) microvessels/mm2. There were few vessels in half the tumours, the remaining half showing a high MVD (Fig. 1e).

Nuclear accumulation of p53 protein was detected in 11 patients (46%; Fig. 1f); there was strong simultaneous nuclear and cytoplasm p16 staining in seven tumours (29%), weak staining in four (17%) and no expression in 13 (54%; Fig. 1g,h)

The associations between the hypermethylation of genes and pathological data are shown in Table 3. Neither the HPV presence nor p53 expression was associated with any of the molecular variables. Nevertheless, the methylation of TSP-1 was associated with unfavourable histological grade (grade 3; P = 0.033), and was notably linked to vascular invasion (P = 0.023). Remarkably, it was also more frequent in the tumours with an invasion depth of >10 mm (P = 0.082).

Table 3.  Correlation between hypermethylation of TSP-1, RASSF1-A and p16 genes and clinicopathological variables
VariablesPatients evaluatedTSP-1 methylationRASSF1-A methylationp16 methylation
YesNoP *YesNoP *YesNoP *
  • *

    From chi-square or Fisher’s exact test; significant values in bold. –, not significant.

No. of patients241113 1014 915 
Histological grade (differentiated)
 Well-moderately18 6120.0339 9810
 Poorly 6 5 1 1 5 1 5 
pT classification
 T1 9 2 70.0727 20.013 6
 T2–T415 9 6 312 6 9 
Invasion depth, mm
 ≤10165110.0827 9511
 >10 86 2 3 5 4 4 
Vascular invasion
 Yes 76 10.0232 53 4
 No17512 8 9 611 
Recurrence during follow-up (local-distant)
 Yes127 53 90.0984 8
 No124 8 7 5 5 7 

Eight of 10 tumours with weak or negative TSP-1 expression had the promoter methylated, by contrast with any tumour with strong expression (P = 0.041). There was strong expression of TSP-1 preferentially in T2-T4 tumours (P = 0.071) and it was not correlated with MVD (P = 0.32). An invasion depth of >10 mm was clearly associated with vascular invasion (P = 0.001) and tumour size (P = 0.028).

RASSF1-A hypermethylation was not associated with any variables except for T classification (P = 0.01; Table 3). The hypermethylation of p16 was also correlated with negative and weak expression of the p16 protein (P = 0.022). All the HPV-negative cases had weak or no p16 expression (P = 0.001). Lymph node involvement was not associated with any of the tested variables.

The follow-up data were examined to ascertain the influence of the variables on the prognosis of the patients; the results are shown in Table 4. There was a higher proportion of survivors in the HPV-related penile cancer group (10 of 11) than in patients with HPV-unrelated cancer (eight of 13; P = 0.17). The latter group also had more recurrences. Kaplan-Meier survival analysis also showed that the 5-year DFS and OS were significantly shorter in patients with TSP-1 hypermethylation (P = 0.031 and 0.04, respectively; Fig. 4). Conversely, TSP-1 expression had no prognostic significance. There were no differences in survival with the remaining variables, except for RASSF1-A hypermethylation, that seemed to be associated with a more prolonged DFS (P = 0.024).

Table 4.  Univariate log-rank test of variables for 5-year DFS and OS
VariablesPatients evaluated5-year DFS, nHazard ratio (95% CI)P5-year OS, nHazard ratio (95% CI)P
  1. –, not significant.

HPV presence
 Yes1161.02 (0.31–3.37)104.37 (0.51–37.47)
 No136   8  
TSP-1 methylation
 Yes1140.28 (0.08–0.99)0.031 60.14 (0.017–0.25)0.04
 No138  12  
RASSF1-A methylation
 Yes1074.8 (1.03–23.2)0.024 81.63 (0.29–8.97)
 No145  10  
p16 methylation
 Yes 940.91 (0.29–2.91) 60.56 (0.11–2.83)
 No158  12  
TSP-1 expression
 Strong-intermediate1482.13 (0.51–8.96) 91.74 (0.29–8.92)
 Weak-negative103   9  
p53 expression
 Positive1150.89 (0.27–2.94) 70.42 (0.077–2.28)
 Negative137  11  
Figure 4.

Kaplan-Meier plots stratified for TSP-1 promoter hypermethylation, showing the curves for DFS (a) and OS (b). Patients with methylated promoter had a shorter DFS and OS than those with unmethylated promoter.

DISCUSSION

Penile cancer is a rare urological cancer with a low incidence in our community (three cases per million people per year). Previous studies reported the presence of HPV in a variable percentage of patients with penile cancer [31,32], and the incidence rate and the prevalence of the different subtypes depend largely on the geographical area [33,34]. In the present study, HPV DNA was found in 44% of cancers, with HPV-16 as the most prevalent subtype, and was present preferentially in basaloid and/or warty tumours [2]. HPV infection can also influence the prognosis of patients with gynaecological cancers, e.g. cervical [35] and vulvar [36]. In the present patients HPV was not of prognostic significance, although there were few patients. Nevertheless, there was a tendency to better survival for HPV-positive patients, as previously reported [37]. Hence, the only HPV-positive men who died had bad prognostic variables, e.g. high tumour grade, local and distant recurrences during the follow-up, with a short survival time after surgical resection (18 months). Despite tumour depth being linked to a poor prognosis [22,38], it had no prognostic value in the present patients. The many local recurrences were probably due to the presence of other factors for a poor prognosis, e.g. histological grade and vascular invasion [2].

The study of molecular alterations and their involvement in the progression of penile cancer could identify more factors associated with a worse prognosis of the patients. In the present study, we analysed genes that are involved in cell adhesion and the cell cycle, with an often altered expression in urological and gynaecological cancer [11,14,16]. We report for the first time that hypermethylation of TSP-1 (leading to loss of TSP-1 expression) was clearly linked to a poor prognosis, and hence constitutes an important marker to detect more aggressive penile tumours. This alteration is also related to unfavourable histological grade and vascular invasion, and is more frequent in tumours with deep invasion (>10 mm), suggesting a pivotal role of this gene in penile cancer.

The tumours with strong expression of this protein detected by IHC were unmethylated and stage T2–T4. This finding is similar to that described in the tumour component of advanced urothelial [28] and vulvar [39] tumours, but contrasts with the decreased mRNA levels of TSP-1 in advanced cervical tumours [40]. The role of TSP-1 expression in cancer progression is currently unclear, because it was suggested to participate simultaneously in the promotion of tumour invasion and in the inhibition of angiogenesis [41]. In the present study the expression of TSP-1 in tumour cells did not seem to correlate with MVD, or any of the other pathological variables. Also, TSP-1 expression was considered a good prognosis factor in breast [42] and cervical cancer [43], probably due to the anti-angiogenic role of this protein in these types of cancer. Nevertheless, the present survival analysis confirms that this factor is unrelated to DFS and OS, according to previous findings in urothelial cancer [28]. Other factors might be involved in this mechanism, and additional studies with more patients are necessary to assess this issue [10].

We also found that RASSF1-A and p16 genes are hypermethylated in penile cancer. RASSF1-A hypermethylation was described in cervical cancer [19], head and neck cancer [44] and urothelial carcinoma [45]. In the present study the frequency of this alteration was similar to that reported for urothelial carcinoma, and the proportion of HPV-positive patients with unmethylated RASSF1-A promoter was higher than that for HPV-negative patients. In a few cases HPV and methylated promoter of RASSF1-A coexist, and this is at variance with what occurs in cervical carcinoma, in which both are mutually exclusive [46]. According to the present data, RASSF1-A hypermethylation was present in all T stages, mainly in T1, and this could explain why this alteration was a good prognosis factor for DFS. In this respect, RASSF1-A hypermethylation could be considered an early event in penile cancer progression.

The proportion of p16 hypermethylation in the present tumours was higher and lower than those reported previously in penile [47] and vulvar cancer [21], respectively. The alteration of p16 expression by hypermethylation and the coexistence between p16 expression and presence of HPV in penile carcinogenesis are confirmed in the present study [47]. Nevertheless, the inverse association between the expression of p16 and p53 proteins described in vulvar cancer [48] was not found in the present patients with penile cancer.

In the present study HPV and p53 expression were not correlated, differing from what occurs in cervical cancer [49] and penile cancer [50]. In penile cancer, p53 nuclear accumulation was described to be associated with tumour stage, nodal metastasis, cancer progression, and shorter survival of the patients [29,50]. In the present study we did not find these associations, although there were fewer patients. Also, there was no association between the expression of this protein and the lack of TSP-1 expression, as described in colon cancer [51]. This lack of association was also reported previously in urothelial carcinoma [28].

To our knowledge, this is the first report of epigenetic changes in TSP-1 and RASSF1-A genes in penile cancer. Additional studies with more patients are crucial to test the value of these genes in the prognosis of patients diagnosed with this type of tumour. In summary, promoter hypermethylation of TSP-1, RASSF1-A and p16 genes are common events in penile cancer and these epigenetic modifications are likely to contribute to tumour progression.

ACKNOWLEDGEMENTS

We are grateful to Dr M. Esteller from Spanish National Cancer Centre (CNIO, Madrid, Spain) and Dr J. Klaustermeier and M. Olivera from Institut Català d’Oncologia (ICO, Barcelona, Spain) for their invaluable support. We wish to thank MC. Caballero, M. Murillo and L. Serrano from Biomedical Research Center (Pamplona, Spain) for their technical assistance. This work was supported by a grant from the Department of Health of Government of Navarra (Spain).

CONFLICT OF INTEREST

None declared.

Ancillary