Abstract
The molecular mechanisms underlying penile carcinoma are still poorly understood, and the detection of genetic markers would be of great benefit for these patients. In this study, we assessed the genomic profile aiming at identifying potential prognostic biomarkers in penile carcinoma. Globally, 46 penile carcinoma samples were considered to evaluate DNA copy-number alterations via array comparative genomic hybridization (aCGH) combined with human papillomavirus (HPV) genotyping. Specific genes were investigated by using qPCR, FISH, and RT-qPCR. Genomic alterations mapped at 3p and 8p were related to worse prognostic features, including advanced T and clinical stage, recurrence and death from the disease. Losses of 3p21.1–p14.3 and gains of 3q25.31–q29 were associated with reduced cancer-specific and disease-free survival. Genomic alterations detected for chromosome 3 (LAMP3, PPARG, TNFSF10 genes) and 8 (DLC1) were evaluated by qPCR. DLC1 and PPARG losses were associated with poor prognosis characteristics. Losses of DLC1 were an independent risk factor for recurrence on multivariate analysis. The gene-expression analysis showed downexpression of DLC1 and PPARG and overexpression of LAMP3 and TNFSF10 genes. Chromosome Y losses and MYC gene (8q24) gains were confirmed by FISH. HPV infection was detected in 34.8% of the samples, and 19 differential genomic regions were obtained related to viral status. At first time, we described recurrent copy-number alterations and its potential prognostic value in penile carcinomas. We also showed a specific genomic profile according to HPV infection, supporting the hypothesis that penile tumors present distinct etiologies according to virus status. Cancer Prev Res; 8(2); 149–56. ©2014 AACR.
Introduction
Penile carcinoma is an aggressive and mutilating disease, which presents a high incidence in developing countries. In Brazil, 2.9 to 6.8/100,000 inhabitants are affected, representing one of the highest incidences in the world (1). Several risk factors have been associated with this tumor etiology, including phimosis, poor hygiene, and human papillomavirus (HPV) infection (2).
The incidence of HPV in penile carcinoma varies according to histologic subtype, with an overall prevalence of 50% of tumors (3, 4), and the molecular mechanisms behind penile carcinogenesis are still largely unknown. It has been reported amplification and/or high expression of MYC levels in HPV-positive cases (5), high levels of ANXA1 protein in high-risk HPV cases (6), strong positive correlation between viral presence and p16INK4A expression, and negative correlation with RB1 protein expression (7), as well as high pEGFR expression in HPV negative samples and HER3 positivity in HPV-positive cases (8).
The diagnosis of penile carcinoma is frequently delayed. Consequently, regional lymph node metastasis and advanced tumors are often observed, resulting in total or partial amputation of the organ with a profound psychologic and social impact on the patient, as well as reduced survival (9). Therefore, the detection of reliable genetic markers would be of great benefit to the patient.
Genetic and epigenetic studies in penile carcinoma are extremely limited. Mutations for PIK3CA, HRAS, KRAS, TP53, and CDKN2A, as well as methylation of CpG islands in particular genes, including FHIT, RUNX3, CDKN2A, and THBS-1, have been reported (10, 11). Even fewer studies have described an association between genetic alterations and clinical findings. Allelic losses on chromosomes 4, 6, 9, 12, and 13 were associated with lower survival and T stage (12).
In this study, DNA copy-number alterations were evaluated by array comparative genomic hybridization (aCGH) combined with HPV infection status, aiming to identify potential molecular markers in penile carcinoma and evaluate the viral role in penile tumors biology.
Materials and Methods
Patients and tissue specimens
Forty-six fresh-frozen penile squamous cell carcinomas (SCC) were collected at A.C. Camargo Cancer Center and Barretos Cancer Hospital (São Paulo, Brazil) between 2000 and 2010. The patients were advised of the procedures and provided written informed consent. This study was approved by the Human Research Ethics Committees at AC Camargo Cancer Center (CEP 1,230/2009) and Barretos Cancer Hospital (363/2010). The cases were classified histopathologically, according to the recommendations of the World Health Organization (13) and the International Union Against Cancer (14). Clinical and pathologic data are summarized in Supplementary Table S1. The cases were evaluated by an expert pathologist and classified as usual (N = 41), papillary (N = 2), sarcomatoid (N = 1), warty (N = 1), and mixed usual/papillary (N = 1) SCC. None of the patients have received neoadjuvant radiotherapy or chemotherapy before the surgery. The follow-up time ranged from 0.1 to 67.4 months with a median of 9.7 months (Supplementary Table S2). Ten samples obtained from peripheral lymphocytes from healthy individuals were included as reference for qPCR. Fifteen normal glans obtained from necropsies were used as control samples for transcripts evaluation.
Nucleic acids extraction and HPV genotyping
Genomic DNA was isolated using a standard phenol-chloroform (Invitrogen) extraction and ethanol precipitation procedure. Total RNA was extracted from using TRizol reagent (Life Technologies). cDNA synthesis was performed in final volume of 20 μL containing 1 μg of RNA treated with Dnase I (Life Technologies); 200 U of SuperScript III reverse transcriptase (Life Technologies); 4 μL of SuperScript First-Strand Buffer 5X; 1 μL of dNTP 10 mmol/L each (Life Technologies); 1 μL of Oligo-(dT)18 (500 ng/μL; Life Technologies); 1 μL of random hexamers (100 ng/μL; Life Technologies); 1 μL of DTT 0.1 mol/L (Life Technologies), and 40 U RNAse Out (Life Technologies). Reverse transcription was carried out for 60 minutes at 50°C and subsequently inactivated for 15 minutes at 70°C. The cDNA was stored at −70°C. HPV status was assessed in all penile carcinoma (N = 46) by the Linear Array HPV Genotyping Kit (Roche Molecular Diagnostics) according to the manufacturer's recommendations.
Array-based comparative genomic hybridization
High-quality genomic DNA (500 ng) from study specimens (N = 38), and a reference sample (male genomic DNA; Promega), were hybridized on Agilent Human 4 × 44 K CGH Microarrays (Agilent Technologies), according to the manufacturer's instructions. Array images were acquired using a DNA microarray scanner with Surescan High-Resolution Technology (Agilent Technologies) and Scan Control (version 8.1, Agilent Technologies) software. The data were analyzed using Nexus Copy Number software (version 6.0, Biodiscovery Inc.). Copy-number alteration was defined as exceeding the significance threshold of 1 × 10−6 and containing at least three consecutive altered probes per segment. Thresholds were defined as the average log2 CGH fluorescence ratio for copy gains ≥0.3, high copy-number gains ≥0.6, losses ≤−0.3, and homozygous losses ≤−1.0. The Fast Adaptive States Segmentation Technique 2 (FASST2) algorithm and the Significance Testing for Aberrant Copy-number (STAC) statistical method were used to identify nonrandom genomic copy-number alterations (15). Alterations detected in at least 20% of the samples were evaluated in more details. Hierarchical clustering analysis was performed using the Complete Linkage Hierarchical algorithm. Genomic data were deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE50134 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE).
Quantitative real-time PCR
qPCR analysis was used to confirm genomic alterations detected by aCGH for DLC1 (8p22), PPARG (3p25), LAMP3 (3q26.3–q27), and TNFSF10 (3q26) genes in 41 penile carcinoma samples (35 aCGH-dependent and six microarray independent samples). The effect of copy-number alterations on gene transcripts was tested in 36 penile carcinoma samples (34 aCGH-dependent and two independent samples) for the four genes by RT-qPCR. Nine primer sets were designed to amplify the altered regions detected by aCGH for each gene. Eight primers flanking microarray altered probes were also constructed intending to determine the extension of the genomic alteration in a certain gene. The primers were designed using Primer-BLAST tool (http://www.ncbi.nlm.nih.gov/tools/primer-blast/; Supplementary Table S3). GAPDH was used as a reference gene for qPCR and GUSB and HMBS for RT-qPCR amplifications. PCR experiments were performed in an ABI Prism 7500 Sequence Detection System (Applied Biosystems) using SYBR Mix (Applied Biosystems). The reactions were carried out in duplicates in the automated PCR equipment Qiagility (Qiagen). RT-qPCR experiments followed the MIQE guidelines (16). The relative quantification was calculated using the model proposed by Pfaffl (17). The thresholds to classify a fragment as not altered (0.65–1.36), involved in losses (<0.65) and gains (>1.36) were established from the relative copy-number values obtained in control samples (blood samples from healthy individuals). Gains or losses presented in more than 20% of penile carcinoma samples evaluated by qPCR were considered as significant.
Fluorescence in situ hybridization
FISH was performed to evaluate MYC gene (8q24) gains and chromosome Y losses in seven penile carcinoma samples using even-archived formalin-fixed paraffin-embedded penile carcinoma tumor samples, sectioned to a thickness of 4 μm. The AneuVysion Multicolor DNA Probe (Vysis CEP 18/X/Y; Abbott Molecular) was used to evaluate Y chromosome aneusomy. Gains of the MYC gene were assessed using MYC/CEN-8 FISH Probe Mix (Dako). A fluorescence microscope (Olympus BX 61, Olympus Optical), equipped with a CCD camera (Photometrics CH 250) was used to analyze FISH results. At least 50 nonoverlapping tumor cell nuclei were evaluated. Absence of Y chromosome alteration was considered for 1:1:2 (Y:X:18) interphase nuclei labeling. An MYC/CEN-8 ratio ≥1.5 was considered for gains and ≥2.5 was positive for MYC amplification.
Statistical analysis
The Fisher exact test (P ≤ 0.05) was applied to compare genomic alterations with clinicopathologic features. Mann–Whitney (P ≤ 0.05) was used for transcriptional data analysis. Cancer-specific survival (CSS) and disease-free survival (DFS) were calculated using the Kaplan–Meier method and the log-rank test (P ≤ 0.05). The follow-up interval was calculated in months from the date of first medical assessment to the date of last follow-up or death by the disease/recurrence. Significant variables in univariate analysis were included in the multivariate analysis model (Cox regression). Statistical analysis was carried out using SPSS version 17.0 (SPSS) and GraphPad Prism 5 (GraphPad Software Inc.) softwares.
Results
Genome-wide profiling
Twenty-eight copy-number alterations were detected in more than 20% of the penile carcinoma cases analyzed. Recurrent copy-number gains were detected on 3q, 5p, 8q, 9p, 20p, and 21p and losses on 3p, 8p, 9p, 21p, and Y (Table 1). The most frequent alterations were found on chromosomes 3, 8, and Y (26% to 66%).
Frequent genomic alterations (>20% of cases) detected in 38 penile tumors evaluated by aCGH
Genomic region . | Start (bp) . | End (bp) . | Event . | Genes . | Frequency (%) . |
---|---|---|---|---|---|
3p26.3–p25.1 | 993,748 | 14,191,316 | Loss | 143 | 26 |
3p24.3–p22.2 | 15,550,927 | 37,778,461 | Loss | 109 | 42 |
3p22.2 | 38,062,745 | 38,411,982 | Loss | 9 | 26 |
3p22.2–p21.32 | 38,941,951 | 44,629,118 | Loss | 74 | 34 |
3p21.1–p14.3 | 54,190,324 | 57,080,457 | Loss | 15 | 26 |
3p14.1 | 66,457,075 | 70,311,198 | Loss | 23 | 32 |
3p13–p11.1 | 72,491,521 | 90,264,177 | Loss | 44 | 34 |
3q12.3–q13.13 | 104,132,986 | 110,515,632 | Gain | 34 | 29 |
3q13.13–q13.31 | 111,130,268 | 115,247,042 | Gain | 46 | 34 |
3q13.31–q25.31 | 116,240,263 | 157,523,060 | Gain | 428 | 42 |
3q25.31–q29 | 157,878,836 | 199,501,827 | Gain | 382 | 42 |
5p13.3 | 31,077,746 | 32,825,265 | Gain | 13 | 26 |
5p15.33 | 0 | 1,305,409 | Gain | 33 | 26 |
8p23.3–p11.21 | 0 | 43,175,310 | Loss | 444 | 50 |
8q11.1–q11.23 | 47,062,121 | 55,033,387 | Gain | 28 | 34 |
8q12.1 | 55,682,239 | 56,316,934 | Gain | 2 | 26 |
8q12.1–q21.13 | 56,967,220 | 82,092,323 | Gain | 172 | 34 |
8q21.2–q24.3 | 87,183,119 | 146,274,826 | Gain | 461 | 66 |
8q22.3 | 101,799,120 | 103,732,868 | Gain | 16 | 61 |
9p21.3 | 21,905,405 | 22,067,827 | Loss | 6 | 26 |
9p12–p11.2 | 42,014,069 | 42,702,421 | Gain | 9 | 26 |
9q33.3–q34.11 | 129,284,384 | 130,899,653 | Gain | 79 | 29 |
20p12.1 | 16,381,425 | 17,709,418 | Gain | 9 | 26 |
21p11.1 | 10,013,263 | 10,117,957 | Gain | 6 | 26 |
21p11.1 | 10,013,263 | 10,117,957 | Loss | 6 | 26 |
Yp11.2 | 7,278,858 | 10,143,912 | Loss | 19 | 29 |
Yq11.21–q11.221 | 13,924,906 | 17,801,127 | Loss | 6 | 39 |
Yq11.222–q11.223 | 19,700,039 | 23,283,748 | Loss | 28 | 39 |
Genomic region . | Start (bp) . | End (bp) . | Event . | Genes . | Frequency (%) . |
---|---|---|---|---|---|
3p26.3–p25.1 | 993,748 | 14,191,316 | Loss | 143 | 26 |
3p24.3–p22.2 | 15,550,927 | 37,778,461 | Loss | 109 | 42 |
3p22.2 | 38,062,745 | 38,411,982 | Loss | 9 | 26 |
3p22.2–p21.32 | 38,941,951 | 44,629,118 | Loss | 74 | 34 |
3p21.1–p14.3 | 54,190,324 | 57,080,457 | Loss | 15 | 26 |
3p14.1 | 66,457,075 | 70,311,198 | Loss | 23 | 32 |
3p13–p11.1 | 72,491,521 | 90,264,177 | Loss | 44 | 34 |
3q12.3–q13.13 | 104,132,986 | 110,515,632 | Gain | 34 | 29 |
3q13.13–q13.31 | 111,130,268 | 115,247,042 | Gain | 46 | 34 |
3q13.31–q25.31 | 116,240,263 | 157,523,060 | Gain | 428 | 42 |
3q25.31–q29 | 157,878,836 | 199,501,827 | Gain | 382 | 42 |
5p13.3 | 31,077,746 | 32,825,265 | Gain | 13 | 26 |
5p15.33 | 0 | 1,305,409 | Gain | 33 | 26 |
8p23.3–p11.21 | 0 | 43,175,310 | Loss | 444 | 50 |
8q11.1–q11.23 | 47,062,121 | 55,033,387 | Gain | 28 | 34 |
8q12.1 | 55,682,239 | 56,316,934 | Gain | 2 | 26 |
8q12.1–q21.13 | 56,967,220 | 82,092,323 | Gain | 172 | 34 |
8q21.2–q24.3 | 87,183,119 | 146,274,826 | Gain | 461 | 66 |
8q22.3 | 101,799,120 | 103,732,868 | Gain | 16 | 61 |
9p21.3 | 21,905,405 | 22,067,827 | Loss | 6 | 26 |
9p12–p11.2 | 42,014,069 | 42,702,421 | Gain | 9 | 26 |
9q33.3–q34.11 | 129,284,384 | 130,899,653 | Gain | 79 | 29 |
20p12.1 | 16,381,425 | 17,709,418 | Gain | 9 | 26 |
21p11.1 | 10,013,263 | 10,117,957 | Gain | 6 | 26 |
21p11.1 | 10,013,263 | 10,117,957 | Loss | 6 | 26 |
Yp11.2 | 7,278,858 | 10,143,912 | Loss | 19 | 29 |
Yq11.21–q11.221 | 13,924,906 | 17,801,127 | Loss | 6 | 39 |
Yq11.222–q11.223 | 19,700,039 | 23,283,748 | Loss | 28 | 39 |
Abbreviation: bp, base pairs.
Genomic alterations mapped at 3p and 8p were significantly associated with advanced clinical and T stage, recurrence and death due to disease (P ≤ 0.05; Table 2). Significantly lower CSS and DFS were observed in cases with losses of 3p21.1–p14.3 (P = 0.006 and P = 0.023, respectively), and gains of 3q25.31–q29 (P = 0.017 and P = 0.042, respectively; Table 3, Fig. 1).
Recurrent copy-number alterations related to prognosis. Kaplan–Meier curves showing significantly reduced CSS and DFS for patients with penile cancer with losses on 3p21.1–p14.3 (A and B) and gains on 3q25.31–q29 (C and D). P values were determined using the log-rank test.
Recurrent copy-number alterations related to prognosis. Kaplan–Meier curves showing significantly reduced CSS and DFS for patients with penile cancer with losses on 3p21.1–p14.3 (A and B) and gains on 3q25.31–q29 (C and D). P values were determined using the log-rank test.
Genomic alterations associated with clinical and pathologic features in penile carcinomas
Genomic region . | Event . | Pa . | ||||
---|---|---|---|---|---|---|
. | . | Clinical stage III–IV . | Recurrence . | T stage T3-T4 . | Lymph node metastasis . | Death . |
aCGH | ||||||
3p26.3–p25.1 | Loss | 0.013 | 0.033 | ns | ns | ns |
3p24.3–p22.2 | Loss | 0.013 | 0.010 | 0.001 | ns | 0.021 |
3p22.2 | Loss | 0.013 | 0.033 | 0.032 | ns | ns |
3p22.2–p21.32 | Loss | 0.030 | ns | ns | ns | ns |
3p21.1–p14.3 | Loss | 0.001 | 0.033 | 0.032 | ns | 0.018 |
3p14.1 | Loss | 0.022 | ns | 0.020 | ns | 0.048 |
3p13–p11.1 | Loss | 0.004 | ns | 0.007 | ns | ns |
8p23.3–p11.21 | Loss | ns | 0.041 | 0.010 | ns | ns |
qPCR | ||||||
DLC1-3 | Loss | ns | ns | 0.047 | 0.046 | ns |
DLC1-7 | Loss | ns | ns | ns | 0.011 | ns |
PPARG-1 | Loss | 0.005 | ns | 0.019 | 0.003 | ns |
Genomic region . | Event . | Pa . | ||||
---|---|---|---|---|---|---|
. | . | Clinical stage III–IV . | Recurrence . | T stage T3-T4 . | Lymph node metastasis . | Death . |
aCGH | ||||||
3p26.3–p25.1 | Loss | 0.013 | 0.033 | ns | ns | ns |
3p24.3–p22.2 | Loss | 0.013 | 0.010 | 0.001 | ns | 0.021 |
3p22.2 | Loss | 0.013 | 0.033 | 0.032 | ns | ns |
3p22.2–p21.32 | Loss | 0.030 | ns | ns | ns | ns |
3p21.1–p14.3 | Loss | 0.001 | 0.033 | 0.032 | ns | 0.018 |
3p14.1 | Loss | 0.022 | ns | 0.020 | ns | 0.048 |
3p13–p11.1 | Loss | 0.004 | ns | 0.007 | ns | ns |
8p23.3–p11.21 | Loss | ns | 0.041 | 0.010 | ns | ns |
qPCR | ||||||
DLC1-3 | Loss | ns | ns | 0.047 | 0.046 | ns |
DLC1-7 | Loss | ns | ns | ns | 0.011 | ns |
PPARG-1 | Loss | 0.005 | ns | 0.019 | 0.003 | ns |
Abbreviation: ns, not significant.
aCalculated using Fisher exact with P ≤ 0.05 for significance.
Clinical features and genomic alterations associated with CSS and DFS by univariate and multivariate analysis in penile cancer
Clinical feature/genomic region . | Event . | Cases . | Death (%) . | Recurrence (%) . | Punivariatea . | Pmultivariateb/HR (95% CI) . |
---|---|---|---|---|---|---|
CSS | ||||||
Clinical stage | I–II | 21 | 0 (0) | — | 0.004 | ns |
III–IV | 16 | 6 (38) | — | |||
Histologic grade | I | 8 | 0 (0) | — | 0.085 | ns |
II–III | 26 | 6 (23) | — | |||
3p21.1–p14.3 | No alteration | 26 | 3 (11) | — | 0.006 | ns |
Loss | 8 | 4 (50) | — | |||
3q25.31–q29 | No alteration | 19 | 1 (5) | — | 0.017 | ns |
Gain | 15 | 6 (40) | — | |||
DLC1-4 | No alteration | 24 | 1 (4) | — | 0.048 | ns |
Loss | 10 | 3 (30) | — | |||
DLC1-7 | No alteration | 19 | 1 (5) | — | 0.034 | ns |
Loss | 16 | 4 (25) | — | |||
PPARG-1 | No alteration | 26 | 0 (0) | — | <0.001 | ns |
Loss | 8 | 5 (63) | — | |||
PPARG-2 | No alteration | 21 | 2 (10) | — | 0.025 | ns |
Loss | 13 | 3 (23) | — | |||
PPARG-3 | No alteration | 26 | 1 (4) | — | 0.002 | ns |
Loss | 9 | 4 (44) | — | |||
DFS | ||||||
Tumor stage | T1-T2 | 24 | — | 2 (8) | 0.011 | ns |
T3-T4 | 14 | — | 7 (50) | |||
3p21.1–p14.3 | No alteration | 26 | — | 3 (11) | 0.023 | ns |
Loss | 8 | — | 4 (50) | |||
3q25.31–q29 | No alteration | 19 | — | 2 (10) | 0.042 | ns |
Gain | 15 | — | 5 (33) | |||
DLC1-1 | No alteration | 8 | — | 0 (0) | 0.083 | ns |
Loss | 28 | — | 9 (32) | |||
DLC1-3 | No alteration | 15 | — | 1 (7) | 0.045 | ns |
Loss | 21 | — | 8 (38) | |||
DLC1-6 | No alteration | 23 | — | 3 (13) | 0.022 | 0.040/5.5 (1.1–27.8) |
Loss | 7 | — | 3 (43) |
Clinical feature/genomic region . | Event . | Cases . | Death (%) . | Recurrence (%) . | Punivariatea . | Pmultivariateb/HR (95% CI) . |
---|---|---|---|---|---|---|
CSS | ||||||
Clinical stage | I–II | 21 | 0 (0) | — | 0.004 | ns |
III–IV | 16 | 6 (38) | — | |||
Histologic grade | I | 8 | 0 (0) | — | 0.085 | ns |
II–III | 26 | 6 (23) | — | |||
3p21.1–p14.3 | No alteration | 26 | 3 (11) | — | 0.006 | ns |
Loss | 8 | 4 (50) | — | |||
3q25.31–q29 | No alteration | 19 | 1 (5) | — | 0.017 | ns |
Gain | 15 | 6 (40) | — | |||
DLC1-4 | No alteration | 24 | 1 (4) | — | 0.048 | ns |
Loss | 10 | 3 (30) | — | |||
DLC1-7 | No alteration | 19 | 1 (5) | — | 0.034 | ns |
Loss | 16 | 4 (25) | — | |||
PPARG-1 | No alteration | 26 | 0 (0) | — | <0.001 | ns |
Loss | 8 | 5 (63) | — | |||
PPARG-2 | No alteration | 21 | 2 (10) | — | 0.025 | ns |
Loss | 13 | 3 (23) | — | |||
PPARG-3 | No alteration | 26 | 1 (4) | — | 0.002 | ns |
Loss | 9 | 4 (44) | — | |||
DFS | ||||||
Tumor stage | T1-T2 | 24 | — | 2 (8) | 0.011 | ns |
T3-T4 | 14 | — | 7 (50) | |||
3p21.1–p14.3 | No alteration | 26 | — | 3 (11) | 0.023 | ns |
Loss | 8 | — | 4 (50) | |||
3q25.31–q29 | No alteration | 19 | — | 2 (10) | 0.042 | ns |
Gain | 15 | — | 5 (33) | |||
DLC1-1 | No alteration | 8 | — | 0 (0) | 0.083 | ns |
Loss | 28 | — | 9 (32) | |||
DLC1-3 | No alteration | 15 | — | 1 (7) | 0.045 | ns |
Loss | 21 | — | 8 (38) | |||
DLC1-6 | No alteration | 23 | — | 3 (13) | 0.022 | 0.040/5.5 (1.1–27.8) |
Loss | 7 | — | 3 (43) |
Abbreviations: CI, confidence interval; ns, not significant.
aCalculated using the log-rank test with P ≤ 0.05 for significance.
bCalculated using Cox regression with P ≤ 0.05 for significance.
Unsupervised hierarchical clustering analysis revealed four distinct groups nearly classified according to chromosome 3 and 8 genomic alterations (Fig. 2A; Supplementary Table S4). However, no significant difference was found between these clusters and the clinicopathologic data.
Genomic profile of 38 penile carcinoma cases evaluated by aCGH (chromosome 1 to Y). Four main clusters were identified on the basis of the genomic profiles for each group (A). Two genomic profiles were generated according to HPV infection (B). Nineteen regions significantly more common in HPV-positive samples are represented in line “S.” The top bars (blue) indicate genomic gains, whereas the lower bars (red) refer to chromosomal losses.
Genomic profile of 38 penile carcinoma cases evaluated by aCGH (chromosome 1 to Y). Four main clusters were identified on the basis of the genomic profiles for each group (A). Two genomic profiles were generated according to HPV infection (B). Nineteen regions significantly more common in HPV-positive samples are represented in line “S.” The top bars (blue) indicate genomic gains, whereas the lower bars (red) refer to chromosomal losses.
Copy-number alterations according to HPV status
Sixteen of 46 penile carcinoma (34.8%) were positive for HPV genotyping: HPV16 (12 cases), HPV53 (one case), HPV16/40 (one case), HPV16/62 (one case), and HPV18/40 (one case).
Nineteen genomic regions were more commonly detected in HPV-positive tumors, including losses of 2q33.2–q33.3, 2q35, 2q36.3–q37.1, 2q37.1, 2q37.3, 2q37.3, 2q37.3, 2q37.1, 3p21.1, 4p16.1–p15.2, 4p14–p13, 5q31.1, 17p13.1, 17p12–p11.2, 17q11.2 and gains of 8p12, 9p13.3, 16p13.3, and 19q13.32 (Fig. 2B; Supplementary Table S5). Assuming that high-risk HPV are mostly integrated into the genome, integration sites for HPV16 and HPV18 were investigated as previously described (18). Nine of 19 significant regions corresponded with HPV integration sites in cervical carcinomas or cell lines: 2q33.2–q33.3, 2q35, 2q36.3–q37.1, 3p21.1, 4p16.1–p15.2, 5q31.1, 8p12, 9p13.3, and 19q13.32; (Supplementary Table S5). Furthermore, three corresponded to genomic fragile sites: FRA2I (2q33.2–q33.3), FRA5C (5q31.1), and FRA19A (19q13.32; Supplementary Table S5).
Quantitative analysis of DNA copy number and gene expression
The qPCR analysis was first performed in the same sample set used in microarray (N = 35). Regions significantly altered were selected to be investigated in an independent set of samples (N = 6). Losses were confirmed for six of eight regions evaluated for DLC1 and four regions for PPARG; gains were confirmed for one region of TNFSF10 (Supplementary Table S6). Downexpression of DLC1 and PPARG genes (P ≤ 0.001) and overexpression of LAMP3 (P ≤ 0.001) and TNFSF10 (P ≤ 0.010) were detected (Supplementary Fig. S1).
Association was found between DLC1- and PPARG-specific alterations and advanced clinical and T stage and development of lymph node metastasis (P ≤ 0.05; Table 2). On univariate analysis, PPARG-1 (P ≤ 0.001), PPARG-2 (P = 0.025), PPARG-3 (P = 0.002), DLC1-4 (P = 0.048), and DLC1-7 (P = 0.034) losses influenced CSS; DLC1-3 (P = 0.045) and DLC1-6 (P = 0.022) losses were associated with reduced DFS (Table 3; Supplementary Fig. S2). Multivariate analysis showed DLC1-6 as an independent risk factor for recurrence (P = 0.040; HR, 5.5; 95% confidence interval, 1.1–27.8; Table 3). No association was observed between transcript profile and clinicopathologic features (data not shown).
FISH evaluation of genomic alterations
Recurrent Y chromosome losses and MYC gains detected by aCGH were confirmed in all tumors evaluated (N = 7). Five cases showed Y chromosome loss, whereas two presented no alterations for both methodologies. Four penile carcinoma showed gains of 8q24 (MYC gene), whereas three were unaltered. A summary of aCGH, qPCR, RT-qPCR, and FISH data is presented on Table 4.
Summary of the main alterations detected by aCGH, qPCR, RT-qPCR, and FISH in penile cancer
Chromosome region altered by aCGH (number of genes altered) . | Gene (location) . | aCGH (N = 38) . | qPCR (N = 41a) . | RT-qPCR (N = 36b) . | FISH (N = 7c) . |
---|---|---|---|---|---|
3p26.3–p25.1 (143) | PPARG (3p25) | Loss | Loss | Downexpression | — |
3q25.31–q29 (382) | LAMP3 (3q26.3–q27) | Gain | No alteration | Overexpression | — |
TNFSF10 (3q26) | Gain | Gain | Overexpression | — | |
8p23.3–p11.21 (444) | DLC1 (8p22) | Loss | Loss | Downexpression | — |
8q21.2–q24.3 (461) | MYC (8q24) | Gain | — | — | Gain |
Chromosome Y (53) | — | Loss | — | — | Loss |
Chromosome region altered by aCGH (number of genes altered) . | Gene (location) . | aCGH (N = 38) . | qPCR (N = 41a) . | RT-qPCR (N = 36b) . | FISH (N = 7c) . |
---|---|---|---|---|---|
3p26.3–p25.1 (143) | PPARG (3p25) | Loss | Loss | Downexpression | — |
3q25.31–q29 (382) | LAMP3 (3q26.3–q27) | Gain | No alteration | Overexpression | — |
TNFSF10 (3q26) | Gain | Gain | Overexpression | — | |
8p23.3–p11.21 (444) | DLC1 (8p22) | Loss | Loss | Downexpression | — |
8q21.2–q24.3 (461) | MYC (8q24) | Gain | — | — | Gain |
Chromosome Y (53) | — | Loss | — | — | Loss |
a35 aCGH-dependent and six array independent samples.
b34 aCGH-dependent and two array independent samples.
cThe samples tested are aCGH dependent.
Discussion
To the best of our knowledge, this is the first study using a large-scale approach to uncover the genomic alterations with prognostic value in penile carcinomas. Using chromosomal CGH, gains were reported for 8q24, 16p11-12, 20q11-13, 22q, 19q13, and 13q21, and losses for 5p15-22, 4q21-32, and X chromosome in penile carcinoma; however, no association was found between these alterations and clinicopathologic data (19).
The correlation between genomic alterations and clinical data showed interesting and novel results, especially for chromosomes 3 and 8. Copy-number alterations of 3p, 3q, and 8p were related to a worse prognosis, as well as a reduced CSS and DFS. Several studies have reported preliminary evidences that the same regions and genes mapped on chromosomes 3 and 8 may have an important role in the pathogenesis of epithelial tumors, including esophageal, non-small cell lung, head and neck, and bladder carcinomas (20–24). The association between genomic alterations and prognosis may be used to identify patients with penile carcinoma with poor outcome and highlights potential targets for therapy.
Four genes mapped on chromosome 3 and 8 were selected for data confirmation by qPCR due to the frequent association of specific chromosomal regions with prognosis. Using array-dependent and -independent samples sets, DLC1, PPARG, and TNFSF10 were confirmed as altered, and DLC1 and PPARG were associated with poor prognosis. Furthermore, copy-number alterations of these genes were translated for altered expression levels by RT-qPCR assays. Although LAMP3 gains have not been confirmed by qPCR, the transcript evaluation showed gene overexpression, indicating a possible genomic alteration covering a region not evaluated by qPCR or other mechanisms of transcript alteration. In addition to report new putative molecular markers of prognosis in penile carcinoma, our analysis also detected genomic alterations in two relevant therapeutic targets, PPARG (25, 26) and TNFSF10 (27), which may be further studied to design new treatment strategies for penile tumors.
MYC gene gains and Y chromosome losses detected by aCGH were also submitted to validation. The FISH analysis was in agreement with large-scale data and showed the reliability of the gains and losses detected by our microarray approach. The most frequent alteration detected by aCGH was 8q21.2–q24.3 gain (66% of cases), which contains the MYC gene (8q24). MYC is a transcription factor responsible for the regulation of approximately 15% of the human gene set, acting in processes such as cell growth and proliferation, cell-cycle progression, transcription, differentiation, apoptosis, and cell motility (28). Previously, MYC gains or amplification, as well as increased gene and protein expression related and unrelated to HPV positivity, were described in penile carcinomas (5, 29). It was reported that MYC activation was mediated by viral integration, which may be an important event in penile oncogenesis (5). In the present study, MYC amplification was detected in both HPV-positive (N = 7) and -negative cases (N = 18), with no significant correlation between these parameters. The high frequency of this alteration independent of HPV status suggests that MYC is a potential therapeutic target in penile carcinoma. Drugs targeting the MYC or associated products in related pathways have been extensively evaluated for cancer treatment (30). The FISH genomic data were not submitted to clinical and pathologic features association due to the low sample size (N = 7).
Frequent Y chromosome losses detected by aCGH were also confirmed by FISH, in a subset of cases. Despite the significance of this alteration in tumor biology having been controversial, a recent head and neck tumor case–control study revealed that Y chromosome loss is independent of patient age and was significantly associated with cancer cells (31).
HPV is considered an important etiologic factor in penile carcinoma. In this study, 34.8% of penile carcinoma samples were positive for HPV infection; 15 of 16 positive cases showed at least one oncogenic HPV genotype (HPV16 or 18). Three cases presented multiple infections (HPV16/40, HPV16/62, and HPV18/40) and, although its significance remains unknown, the presence of at least one oncogenic HPV genotype in each case supports their relevance in tumor biology.
Comparing tumors according to HPV status, it was revealed that 19 regions were able to differentiate between the groups, nine involving breakpoints previously described as associated with HPV integration sites in cervical carcinomas and cell lineages (18). HPV integration into cellular genome has been associated with the progression of preneoplastic lesions, mainly due to structural changes of the viral genome that allow deregulated expression of the viral oncogenes and confer neoplastic selective pressure (18). It is also speculated that critical cellular genes are affected by integration of viral genome (18). The detection of HPV integration sites described herein is in agreement with the importance of these regions in the development and progression of tumors, including penile carcinoma.
The distinct genomic profile between HPV-positive and -negative samples gives additional support for the hypothesis that these tumors have two distinct etiologies: one dependent and the other independent of HPV infection (11, 32, 33). The regions frequently detected as altered in HPV-positive cases must be further explored, aiming to better understand the involvement of HPV in penile carcinoma and its potential application in clinical practice.
There are currently no targets for therapy for clinical use in penile cancer. Using aCGH, we provided insights into novel therapeutic targets for penile carcinoma and identified putative prognostic molecular markers. Distinct genomic profiles were found related to HPV status. In particular, the presence of breakpoints regions previously described in cervical carcinomas, revealed similarities between these two HPV associated diseases. These findings provide novel insights into penile carcinogenesis.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: S.R. Rogatto
Development of methodology: A.F. Busso-Lopes, F.A. Marchi, H. Kuasne
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.F. Busso-Lopes, C. Scapulatempo-Neto, J.C.S. Trindade-Filho, C.M.N. de Jesus, A. Lopes, G.C. Guimarães
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.F. Busso-Lopes, F.A. Marchi, H. Kuasne, S.R. Rogatto
Writing, review, and/or revision of the manuscript: A.F. Busso-Lopes, F.A. Marchi, C. Scapulatempo-Neto, J.C.S. Trindade-Filho, A. Lopes, G.C. Guimarães, H. Kuasne, S.R. Rogatto
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Scapulatempo-Neto, G.C. Guimarães
Study supervision: S.R. Rogatto
Acknowledgments
The authors specially thank the staff of the Department of Pathology and the Macromolecules Laboratory at the A.C. Camargo Cancer Center, São Paulo, and Barretos Cancer Hospital, Barretos, São Paulo, Brazil.
Grant Support
This study was supported by CNPq (National Council of Technological and Scientific Development), and FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo). S.R. Rogatto received research grants from FAPESP (grant numbers 2009/52088-3 and 2010/51601-6) and CNPq. A.F. Busso-Lopes received scholarships from FAPESP (grant numbers 2009/06851-7 and 2011/03974-0).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.