A Comprehensive Expression Analysis of Cancer Testis Antigens in Head and Neck Squamous Cell Carcinoma Revels MAGEA3/6 as a Marker for Recurrence

Head and neck squamous cell carcinoma (HNSCC) affects 600,000 new patients worldwide each year (1). Despite significant advances in therapy, the survival rate for these patients has not improved markedly over the past 30 years (2). Late diagnosis and frequent locoregional recurrences are the most frequent causes of treatment failure. Therefore, the development of new therapeutic approaches and their integration into current forms of treatment, which include surgery, radiotherapy, and chemotherapy, are pursued to improve this prognosis.

The recent discovery of tumor antigens has opened new doors for specific tumor-targeted treatments using passive or active immunotherapeutic strategies. In particular, cancer–testis antigens (CTA) represent promising targets due to their high immunogenicity and specific expression pattern (3–5). To date, more than 200 genes encoding CTAs have been identified and cataloged in a dedicated database (CTDatabase— http://www.cta.lncc.br/index.php; ref. 6). As spontaneous antitumor immune responses can be elicited by CTAs, they are promising candidates for cancer immunotherapy and, in recent years, gained importance in the development of cancer vaccine–based clinical trials (7). The expression of these antigens has been detected in various types of malignant tumors of different histological types, but rarely in normal somatic tissues, with the exception of immunoprivileged gametogenic tissues. CTAs are described as frequently expressed in melanomas, multiple myeloma, glioblastomas, and carcinomas of the bladder, lung, ovarian, and liver (8–14). A moderate expression could be detected in breast and prostate tumors (5, 15), and they are rarely expressed in renal cell carcinoma, colon cancer, lymphomas, and leukemias (9, 15–17). Few studies with small sample sizes have detected CTA expression in HNSCCs (18–26).

In light of these considerations, the current study provides a comprehensive evaluation of CTA gene expression in a large number of HNSCC cases to assess their potential as targets for immunotherapy and to evaluate their prognostic significance.

This retrospective study involved tissue specimens from 89 patients with HNSCC who underwent tumor resection between 2007 and 2010 at the Barretos Cancer Hospital (Sao Paulo, Brazil). These samples were available at the Tissue Bank of the Hospital. Only patients diagnosed with primary HNSCC, not previously treated, presenting with tumors at oral cavity, pharynx, or larynx and treated with curative intent were included in the study. All samples were checked microscopically for the presence of neoplastic tissue and the absence of contaminating normal mucosa. In addition, 20 oral mucosa tissue samples from healthy donors undergoing odontological and preprosthetic surgeries were used as normal controls. All tissue samples were snap-frozen in liquid nitrogen within 30 minutes after resection and stored at −80°C until the RNA extraction.

The medical records of patients were reviewed for standard demographic data, clinical and pathologic information, and outcome of the disease. Smoking was defined as use of tobacco, chewable or smoked, for at least 1 year continuously. Alcohol use was defined as intake of more than 2 alcoholic drinks per day, for at least 1 year continuously. Tissue sampling and study design were approved by the Institutional Review Board of the Barretos Cancer Hospital.

Total RNA was isolated using Trizol reagent (Life Technologies), following the manufacturer's recommendations. After extraction, RNA samples were quantified in a spectrophotometer (NanoDrop ND1000; Thermo Fisher Scientific), and the quality of the RNA was checked by electrophoresis on 1% agarose gels. Two micrograms of each RNA sample was subjected to cDNA synthesis using the SuperScript III First-Strand Synthesis System (Invitrogen). The cDNA obtained was diluted 10-fold before use.

The mRNA expression of the selected CTA in tumor and control samples was determined by RT-PCR using primer sequences and amplification conditions described on the CTDatabase (http://www.cta.lncc.br/index.php), except for PRAME (ref. 27; Supplementary Table S1). Testicular tissue was used as a positive control, and the ACTB gene was used as loading control. MAGEA3 and MAGEA6 share more than 98% similarity in their sequences and, for this reason, the primers used in this study were not able to discriminate the expression of these two genes. The same is valid for SPANX-C and SPANX-D genes. Electrophoresis gels were visualized by SYBR Safe (Invitrogen) staining using the EP AlphaImager (Alpha Innotech). Each RT-PCR experiment was performed in duplicate using the same cDNA sample. All cases with a detectable band on both replicates were considered positive. The intensities of the PCR products were heterogeneous, and some specimens yielded only faint bands. These were scored positive only if the result could be reproduced by a repeated RT-PCR. Cases with very low transcript levels that were not reproducibly positive were regarded as negative.

We defined a strategy to select, among the CTAs cataloged in the CTDatabase, the most promising ones to have their mRNA positivity evaluated in HNSCC samples. Firstly, we mined the Oncomine (http://www.oncomine.org) and SAGE Anatomic Viewer (http://cgap.nci.nih.gov/SAGE/AnatomicViewer) databases to search for CTA genes expressed in HNSCCs. After that, we evaluated high-throughput expression data (EST, MPSS, CAGE, and RT-PCR experiments) previously provided by Hofmann and colleagues (28). In the end, we also accessed published data of CTA mRNA expression in cancer cell lines (CTDatabase), as well as, information retrieved from the literature.

Statistical analysis was performed using the statistical software SPSS 19.0 for Windows. To assess the association between the anatomical and pathologic variables of patients and the molecular findings in the tumor samples, the χ² or Fisher exact tests were performed, as appropriate. Survival curves were calculated by the Kaplan–Meier method, and differences between groups were compared using the log-rank test. Overall survival (OS) was measured as the time interval between the date of the initial treatment for the primary tumor and the date of the last follow-up or death, whereas the disease-free survival (DFS) was defined as the time interval between the date of initial treatment and the date of the diagnosis of the first recurrence. For evaluation of the independent contribution of significant clinical and molecular variables on DFS or OS, all factors with significance in the Kaplan–Meier analysis (P ≤ 0.2) were tested in the multivariate analyses using the Cox proportional hazard model. Results were calculated with 95% confidence intervals (CI). For all analyses, we considered statistical significance when P value <0.05.

The clinical and histological features of the 89 patients with HNSCC enrolled in this study are presented in Table 1. They were mainly males (86.5%), with age ranging from 32 to 82 years (median, 58.9 years). Tobacco or alcohol consumption (current or former) was reported by 90.7% and 73.5%, respectively. Primary tumor sites included oral cavity (68.5%), oropharynx/hypopharynx (15.7%), and larynx (15.7%). Most tumors were presented at advanced stages (T3/T4, 81.8%). Among 78 patients with data available, metastatic cells could be detected in lymph nodes of 39 patients (50%) and 21 (28.4%) of them presented extracapsular spread. Perineural invasion was present in 21 cases (24.7%) among 85 patients with available data. Lymph vascular invasion was observed in 14 (17.1%) cases.

To select CTA-candidate genes with high probability to be expressed in HNSCC, an in silico screening was performed using all the 139 CTA genes cataloged on the CTDatabase to interrogate the NCBI-CGAP EST and SAGE databases, Oncomine microarray data collection, published data of CTA mRNA expression in cancer cell lines (CTDatabase), and high-throughput expression data provided by Hofmann and colleagues (28). By the end, we were able to select 36 CTA genes for the assessment of gene expression in a context of high expression in HNSCC samples and absence in normal mucosa (Supplementary Table S2). Of note, previous studies have already reported the expression of 11 of those selected CTAs (CTAGE1, LIPI, MAGEA1, MAGEA12, MAGEA2, MAGEA3/6, MAGEA4, MAGEC1, MAGEC2, PRAME, and SSX2) in HNSCCs, whereas the presence of transcripts from the remaining 25 CTAs (ACTL8, CABYR, CASC5, CEP290, CRISP2, CSAG2, CXORF48, HSPB9, MAGEA9, MAGEB2, MAGEB6, NXF2, OBP2B, OIP5, OTOA, PAGE5, PBK, SPANX-CD, TAF7L, TDRD1, TFDP3, TSSK6, VENTXP1, XAGE2, and XAGE3) has never been described in head and neck cancers.

Based on the above results, 36 CTA genes were selected to be evaluated in HNSCC and normal mucosa samples using an RT-PCR analysis. Because of the scarcity of RNA quantity of many samples and the high number of genes selected, it would be virtually impossible to evaluate all possible candidate-genes in all samples. So, we firstly decided to conduct a discovery study, and then a more limited set of “best” CTA genes would be used in the prevalence set of samples. The first step was to verify the expression pattern of the 36 selected CTA genes in mucosa samples collected from healthy individuals (controls, n = 10). This analysis showed that 17 CTAs (ACTL8, CABYR, CASC5, CEP290, CSAG2, CTAGE1, HSPB9, MAGEA2, NXF2, OBP2B, OTOA, PBK, TAF7L, TDRD1, TFDP3, TSSK6, and VENTPX1) were expressed in more than 20% of the normal controls, indicating low specificity. Therefore, these CTAs were excluded from the study.

Next, we evaluated the mRNA positivity of the remaining 19 CTA genes in 17 HNSCC samples. LIPI (0%), PAGE5 (5.9%), SSX2 (5.9%), XAGE2 (5.9%), OIP5 (11.8%), and XAGE3 (11.8%) were rarely expressed in HNSCC and were set aside. By the end, 13 CTA genes that could better distinguish HNSCC from control samples were selected to be tested in the prevalence set (n = 89) and control subjects (n = 20).

MAGEC1 and CRISP2 were found rarely expressed in the HNSCC specimens (15.7% and 11.2%, respectively). However, we were able to identify 11 CTA genes whose expression was highly associated with HNSCC cases (PRAME 79.8%, SPANX-CD 71.9%, MAGEA4 60.7%, MAGEA3/6 56.2%, MAGEA12 49.2%, MAGEB2 44.9%, MAGEA1 44.9%, MAGEA9 40.5% MAGEC2 39.3%, MAGEB6 32.6%, and CXORF48 27.0%; Table 2).

According to the CTDatabase, the expression of MAGEB2, MAGEA1, MAGEB6, SPANX-CD, and CXORF48 can be detected in several tumors, whereas it is absent in normal tissues. For this reason, they are described as “testis-restricted” CT antigens. Notably, in our study, 85.4% of the examined HNSCC cases showed expression of at least one of these five CTAs. Among them, 20.2% of the cases expressed only one of these CTA genes, whereas coexpression of two, three, four, or five antigens was detected in 23.5%, 18.1%, 18.1%, and 5.5% of the HNSCC cases, respectively. Evaluating different panels formed by combinations of these five “testis-restricted” CT antigens expressed in HNSCCs, we observed that sensitivity of the panels can range from 57.3% to 85.4%, whereas the specificity observed can be as high as 100% in some combinations of genes (Table 3).

Significant associations between demographical and clinical characteristics (age, gender, tobacco consumption, alcohol consumption, tumor site, T stage, N stage, perineural invasion, lymphovascular invasion, and extracapsular spread) of the 89 patients with HSNCC enrolled in this study and the mRNA expression of 11 selected CTA genes were evaluated. This analysis showed MAGEA12 expression associated with nonsmokers (P = 0.028) and SPANX-CD mRNA positivity correlated with age <55 years (P = 0.005). The lack of MAGEB6 expression was associated with the absence of extracapsular spread (P = 0.005), whereas CXORF48 expression was associated with perineural (P = 0.023) and lymphovascular invasions (P = 0.003). The detection of CRISP2 transcripts was also associated with lymphovascular invasion (P = 0.006; Supplementary Table S3). No other significant correlation could be observed between the expression of the selected CTA genes and the patient's characteristics.

In addition, we investigated the correlation of the OS and DFS with clinical and molecular variables. As expected, the OS was better for those patients with early T stage (86.6% T1/T2 vs. 41.4% T3/T4, P = 0.002), without perineural invasion (65.4% absent vs. 31.7% present, P = 0.038) and with a tendency to without metastasis (64.1% N0 vs. 43.7 N1/N2/N3, P = 0.065; Table 4).

The analyses of OS were not able to identify any significant association with the mRNA expression of the 11 selected CTA genes. However, the 5-year DFS analyses showed that 23.8% of the patients with HNSCC expressing MAGEA3/6 presented recurrences, whereas relapses were detected in 65.9% of the patients with no expression of this CTA, and this difference was significant (P = 0.014; Fig. 1).

To detect independent predictors of survival, an analysis of prognostic variables based on the Cox proportional hazards model was performed involving significant clinical features (age, gender, alcohol consumption, tumor site, T stage, N stage, perineural invasion, lymphovascular invasion, and extracapsular spread) and molecular variables (MAGEA3/6 and MAGEA12 mRNA positives) associated with survival probability. In 5-year DFS, this multivariate analysis revealed that gender (P = 0.032; HR = 3.21; 95% CI, 1.11–9.30) and MAGEA3/6 expression (P = 0.008; HR = 0.30; 95% CI, 0.128–0.732) remained as independent predictors of recurrence (Table 5).

Over the last years, diagnosis and management of patients with HNSCC have improved through combined efforts in surgery, radiotherapy, and chemotherapy, but long-term survival rate is still around 50% (29). Therefore, novel forms of treatment are urgently needed, and immunotherapy represents an approach that has yet to be fully explored in HNSCC. Immunotherapy is a promising cancer treatment due to the potential for persistence of the antitumor effect caused by immunologic memory (30). In this way, immunotherapeutic intervention could help in the control of the HNSCC progression and relapse. There is a growing interest in applying immunotherapy approaches to different tumors, encouraged by the recent FDA approval for Sipuleucel-T in prostate cancer and Ipilimumab for metastatic melanoma (31, 32).

Although there are few studies in this field, evidences show that the weakening of the immune response may play an important role in the installation, progression, and recurrence of HNSCC. For instance, solid organ transplant recipients who are diagnosed with HNSCC while receiving high doses of immunosuppressive progress extremely poorly, with advanced diagnosis of the tumor and short survival (33). Besides, studies have associated lower rates of survival with immune system depression in patients with HNSCC (34). These facts point up an involvement of the immune response in the control of HNSCC and highlight the potential value of cancer immunotherapy for these patients (34, 35).

In the present study, we surveyed the landscape of CTA expression in HNSCC by conducting a deep screening of the CTA listed in the CTDatabase. A small number of studies have reported a relatively frequent, but sometimes conflicting, expression of selected CTA genes in HNSCC (18–25). According to them, MAGEA3 expression could be detected in 43% to 72% of HNSCC cases evaluated in five studies, whereas MAGEA4 gene product was detected in 27% to 60% of tumors examined by four independent research groups. MAGEA1 is expressed in a considerable proportion of HNSCCs (30%–40%), as well as PRAME (42%–49%), NY-ESO-1 (6%–33%), MAGEC1 (28%–48%), MAGEC2 (10%–33%), MAGEA12 (49%), SSX1 (45%), and SSX4 (30%). Other CTAs reported as occasionally expressed in HNSCC are MAGEA2, MAGEA10, BAGE, SCP1, GAGE, and SSX2. Consistent with this, we also found MAGEA1 (44.9%), MAGEA4 (60.7%), MAGEA3 (56.2%), MAGEC1 (15%), MAGEC2 (39.3%), and MAGEA12 (49.4%) frequently expressed in HNSCC samples. Further, we were able to detect PRAME transcripts in 79.8% of the HNSCC samples, a frequency higher than observed previously. Worth mentioning, this is the first study to identify a frequent expression of CXORF48, MAGEA9, MAGEB2, MAGEB6, CRISP2, and SPANX-CD in HNSCC.

Several clinical trials have evaluated different CTAs as targets for anticancer vaccines in patients with lung cancer, prostate, ovarian, and melanoma (12, 36–42). In HNSCC, a clinical trial conducted by the University of Maryland School of Medicine, Baltimore, United States, is evaluating a vaccine targeting MAGEA3 and HPV-16 antigens (NCT00257738, http://clinicaltrials.gov/). Our results show a large proportion of patients with HNSCC expressing at least one of five CTAs, and this expression seems to be restricted to tumor tissues. Taken collectively, these data suggest that patients with HNSCC expressing CTAs might be eligible for future immunotherapy approaches targeting multiple antigens; however, the demonstration of immunogenicity in the human host is crucial for a CTA to be considered as a potential cancer vaccine target. Hence, further studies are necessary to evaluate whether some of the CTAs highlighted in this study are able to elicit coordinated humoral and cell-mediated responses in patients with HNSCC. It will also be important to verify the presence of IgG antibodies against these antigens in sera of these patients.

The identification of human tumor antigens could be important not only for the analysis of antitumor immune responses and the development of immunotherapy, but also for the identification of molecular markers useful for diagnosis or prognosis determination (43). Probably, due to the small sample sizes analyzed, all except one of the previous studies failed in establishing a clear correlation between CTA expression and HNSCC prognosis. Cuffel and colleagues (21) showed that patients with tumors expressing MAGEA4 or multiple CTA genes had a poorer OS, and MAGEA4 mRNA positivity was associated with poor outcome independent of clinical parameters. Our results reveal, for the first time, a significant association between MAGEB6 expression and the tumor extracapsular spread. The biologic significance of the extracapsular spread in the outcome of the patients is referred as a manifestation of a more aggressive neoplasm providing higher rates of recurrence (44–47). In addition, CRISP2 and CXORF48 expression was correlated with the presence of lymphovascular invasion, a recognized clinical factor of poor prognosis. Based on these findings, we can speculate an association between the expression of these CT antigens and a more malignant phenotype, suggesting their usefulness as prognostic markers in HNSCC.

Interestingly, our study showed the positive correlation between MAGEA3/6 expression and better disease outcome. Although most of previous studies have reported an association of CTA expression with poor outcome in multiple myeloma, ovarian, lung, head and neck, and gastric cancers (9, 11, 21, 48, 49), Sharma and colleagues (10) described a positive association between expression of CT10 and improved survival for patients with urothelial carcinoma. Along the same line, Freitas and colleagues presented the mRNA positivity of CT antigens as an independent predictor of better OS for patients with glioblastoma (14). Additional prospective studies with independent cohorts of patients with HNSCC are needed to confirm the positive correlation between MAGEA3/6 expression and better disease outcome, but, based on our results, we can speculate that the expression of these CTA in head and neck tumors may have elicited a spontaneous immune response and this could impact favorably the prognosis.

The biologic function of MAGEA3 remains poorly understood, but this antigen is found frequently expressed in different cancer types. Yang and colleagues proposed that the expression of MAGEA3 could protect cells from programmed cell death and contribute to tumor development by stimulating the cell survival and proliferation (50).

Some points in this report warrant emphasis. First, this study identifies 11 CTA genes frequently expressed in HNSCC, and five of them are reported as “testis-restricted” CTAs. Second, different panels formed by combinations of these “testis-restricted” CTAs can show up to 85.4% of positivity for at least one member of the panel, making them potential targets for immunotherapy approaches. Third, MAGEB6, CRISP2, and CXORF48 expressions are significantly correlated with well-known poor outcome clinical features. Fourth, the MAGEA3/6 expression is an independent marker for tumor recurrence. On the other hand, there are also some limitations to the present study, such as the limited patient number evaluated and the fact of these patients were not consecutively and prospectively collected for the study. Furthermore, we used a frozen tissue cohort in which a selection bias existed as a result of the selection of specimens with large volume tumors appropriate for frozen tissue collection.

In conclusion, this study, one of the largest studies reported to date evaluating CTA expression in HNSCC, identified 11 CTA genes frequently expressed in HNSCC. Five of them are considered as “testis-restricted” CTAs and, in the future, should be evaluated as immunotherapy targets for this neoplasm. Furthermore, on the basis of our results, the expression of MAGEA3/6 is an independent predictor of DFS in patients with HNSCC, and the mRNA positivity of MAGEB6, CRISP2, and CXORF48 is associated with bad prognosis for these patients.

No potential conflicts of interest were disclosed.

Conception and design: A.L. Carvalho, A.L. Vettore

Development of methodology: F.T. Zamunér, C.Z. de Oliveira

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F.T. Zamunér, B.T.R. Karia, C.R. dos Santos

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F.T. Zamunér, B.T.R. Karia, C.Z. de Oliveira, A.L. Carvalho

Writing, review, and/or revision of the manuscript: F.T. Zamunér, C.R. dos Santos, A.L. Carvalho, A.L. Vettore

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.T.R. Karia

Study supervision: C.R. dos Santos, A.L. Carvalho, A.L. Vettore

This work was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo. F.T. Zamunér was recipient of scholarship from Fundação de Amparo à Pesquisa do Estado de São Paulo—FAPESP, and B.T.R. Karia was recipient of scholarship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—CAPES. A.L. Carvalho has a National Counsel of Technological and Scientific Development (CNPq) scholarship.

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.