Prostate cancer is the most commonly diagnosed malignancy, and the second leading cause of cancer-related death in men in the United States.1 Radical retropubic prostatectomy (RRP) is the most common and effective treatment for clinically localized prostate cancer. However, ∼35% of prostate cancer patients submitting to RRP will have biochemical-recurrence of the disease within 10 years after surgery, as defined by an increase in Prostate Specific Antigen (PSA) level to >0.2 ng/mL.2 Subsequently, these patients will develop clinical relapse of the disease, typically with the development of bone metastases.
Biochemical-recurrence after RRP has been associated with multiple factors including preoperative PSA level and velocity, clinical stage, Gleason score, level of extracapsular extension, seminal vesicle invasion, pelvic lymph node status and surgical margin status and at present there is no cure for metastatic prostate cancer.3, 4 However, due to the significant clinical heterogeneity of the disease, these factors are far from accurate in identifying patients at high risk for cancer progression and the characterization of novel biomarkers to predict biochemical-recurrence, as well as the development of targeted therapies for prostate cancer are, thus, of considerable interest.5–8
Cancer/testis (CT) antigens are proteins expressed in normal gametogenic tissues and in different types of tumors.9, 10 Although it seems to exist an association between CT antigen expression and more aggressive tumors, this association varies according to antigen and tumor type and the potential use of CT antigens as tumor markers remains poorly explored.11 However, most CT antigens are highly immunogenic, eliciting both humoral and cellular immune responses in cancer patients. These antigens are not immunogenic in normal tissues, as their expression is restricted to MHC class I-negative germ cells. Thus, CT antigens are considered ideal targets for cancer immunotherapy.9, 10, 12
We have recently described a novel CT antigen, named CTSP-1, which is aberrantly expressed in 58% of prostate tumors and capable of eliciting a humoral immune response in ∼20% of prostate cancer patients.13 In the present study, we analyzed CTSP-1 protein expression and the presence of antibodies against CTSP-1 in a larger population of prostate cancer patients with clinically localized disease. Our results suggest that CTSP-1 can be considered a promising target for prostate cancer immunotherapy, since it is frequently expressed in localized prostate tumors and is capable of eliciting humoral immune response in a significant fraction of prostate cancer patients. We also demonstrated that the presence of anti-CTSP-1 antibodies in the plasma of prostate cancer patients is associated with a better prognosis and could be used as a prognostic marker for biochemical-recurrence in these patients.
Patients and methods
The study included 147 consecutive prostate cancer patients who underwent RRP with simultaneous bilateral pelvic lymphadenectomy at the Urology Division of the Pelvic Surgery Department of the Hospital do Cancer, São Paulo, Brazil, between November 1998 and July 2005. All cases had clinically localized disease as suggested by elevated PSA serum levels or palpable nodules on digital rectal examination, and then confirmed by transrectal ultrasound-guided needle biopsy of the prostate. Patients with a second nonprostate primary tumor, or who were submitted to radiotherapy before the surgery, were excluded from this study. Twenty-two patients received neoadjuvant androgen suppression. After RRP, patients were followed up with serum PSA tests every 3 months during the first year, twice a year between the second and the fifth year and then annually thereafter. Patients with an increase in PSA levels after RRP were submitted to rectal digital examination, pelvic and abdominal computerized tomography and transrectal ultrasound-guided needle biopsy to confirm local recurrence and bone scan to detect distant metastases. Surgical specimens were staged according to the 1997 American Joint Committee on Cancer System and graded using the Gleason system. Specimens were evaluated for surgical margin status and perineural, angiolymphatic, extracapsular and seminal vesicle invasion, as well as for the presence of metastasis in all removed lymph nodes. Clinicopathological data and follow-up information were obtained retrospectively from patient's medical records. The pathological information for these patients is summarized in Table I. The median age of the patients at diagnosis was 62 years (range 41–75) and the median preoperative PSA level was 8.6 ng/mL (range 1.2–44, percentiles 5.7, 8.7, 14.4). Complete clinical follow-up, including disease progression and mortality, was available on 143 patients included in this study. Patients lost to follow-up were followed for at least 2 years (range 29–64 months). The median follow up time of this cohort was 54.7 months (range 2.9–96.2) and the median follow-up time of censored cases was 74.6 months (range 35.1–96.2). Biochemical-recurrence was defined as a serum PSA level > 0.2 ng/mL on 2 consecutive PSA tests after RRP.
Table I. Pathological Characteristics of the Prostate Cancer Patients
Tumor volume has eight patient data missing and perineural and angiolymphatic invasion have 1 patient data missing.
Plasma samples were collected at the time of the surgery and after explicit informed consent. The study was reviewed and approved by the institution's ethical committee. Patient identifiers were removed and samples were coded to assure confidentiality. Plasma antibody responses against the recombinant CTSP-1 protein were tested by Western blot analysis. Five hundred nanograms of his-taged CTSP-1 recombinant protein (∼32 KDa) were fractioned on 12% SDS/PAGE gel electrophoresis and transferred to Hybond-P PVDF membranes (Amersham Biosciences, Piscataway, NJ). After blocking with PBS solution containing 5% low-fat milk, the membranes were incubated for 90 min at room temperature with plasma from cancer patients at a 1:25 dilution. Plasma antibodies binding to CTSP-1 protein were detected by incubation with goat anti-human IgG HRP-conjugate (Amersham Biosciences, Piscataway, NJ) and visualized with ECL™ Western Blotting Detection Reagents (Amersham Biosciences, Piscataway, NJ). Results were analyzed qualitatively, since no reactivity against CTSP-1 protein was detected among plasma samples from 50 healthy blood donors collected at the same Institution.13 Immunoblots were carried out in duplicates, and triplicates when necessary. A plasma sample was considered positive when reactivity against the recombinant CTSP-1 protein was observed in 2 independent experiments. Positive plasma samples were then tested against an irrelevant protein from sugar cane expressed in the same expression system as the CTSP-1 recombinant protein and used to correct for unspecific binding.
Prostate tumors were fixed with buffered formalin and embedded in paraffin. Briefly, sections were heated at 60°C for 20 min and deparaffinized in xylene, followed by a graded series of alcohols (100–75%) and rehydrated in water. Slides were placed 3 times in 3% hydrogen peroxide for 5 min and washed in water rinses for 5 min prior to incubation for 24 hr with an anti-CTSP-1 polyclonal antibody13 diluted 1:100. Slides were washed in PBS, incubated with biotinylated goat anti-rabbit IgG for 20 min and then with streptavidin-biotin peroxidase LSAB kit (Dako, Glostrup, Denmark). The immunostaining was performed by incubating slides in diaminobenzidine (Dako, Glostrup, Denmark) solution containing 1 μL of chromogen for 50 μL of buffer substrate during 5 min. After chromogen development, slides were washed, dehydrate with alcohol and xylene, and mounted with coverslips using a permanent mounting medium. Normal testis and prostate specimens were used as positive and negative controls of the reactions, respectively.
Fisher exact test or chi-square tests were used to evaluate the association between plasma antibody response against CTSP-1 and categorical variables. For continuous variables, Mann-Whitney U test was used. The preoperative PSA level was considered a continuous variable, the pathological Gleason score was grouped in 2 categorical variables (4–6 and 7–8) and tumor volume was grouped in 4 categorical variables (1–10, 11–30, 31–50 and 51–100%). Biochemical-recurrence free survival (from the date of the RRP until the date of the first PSA measurement >0.2 ng/mL or to the date of the last follow up) and clinical-recurrence free survival (from the date of the RRP until the date of detection of local or distant disease or until the date of the last follow up) curves were calculated with Kaplan-Meier method. A log-rank test was used to assess statistical differences between the curves. Multivariate analysis was carried out using Cox proportional hazard regression model. All variables presenting p-value < 0.20 on the univariate analysis were selected for building a multiple model (stepwise forward method), since this level of significance eliminates many insignificant variables and ensures that all potential explanatory variables are included. For all tests, type 1 error (alpha) was established as 0.05 and results were considered statistically significant when p < 0.05. Patients with missing values for tumor volume, perineural and angiolymphatic invasion were excluded from the analysis of that particular variable. Statistical analyses were performed using SPSS Software 15.0 (SPSS, Chicago, IL).
Antibodies against CTSP-1 are frequently found in prostate cancer patients but are not associated with established pathological variables
To evaluate the presence of antibodies against CTSP-1 in the plasma from cancer patients, we established an immunoblotting assay using a his-tagged CTSP-1 recombinant protein.13 Antibodies against CTSP-1 were detected in 37 of the 147 (25.2%) prostate cancer patients (Fig. 1). We then investigated the association between the presence of anti-CTSP-1 antibodies and well-established pathological parameters for prostate cancer. The distribution of the pathological parameters according to the presence of humoral response against CTSP-1 is listed in Table II. No statistically significant association was found between the presence of anti-CTSP-1 antibodies and any of the pathological parameters analyzed.
Table II. Patient's Distribution According to the Existence of Humoral Response Against CTSP-1
For 49 of these 147 patients, tumor samples embedded in paraffin were available for protein expression analysis by immunohistochemistry (IHC). CTSP-1 protein expression was detected in 61.2% (30/49) of these tumors (Fig. 2), confirming our previous results on the high frequency of CTSP-1 expression in primary prostate tumors.13 Eleven out of 30 (36.7%) patients bearing CTSP-1 positive tumors presented anti-CTSP-1 antibodies in their plasma and only 1 out of 19 (5.3%) patients with negative CTSP-1 expression presented humoral response against this antigen. Although a significant correlation (p = 0.017) between CTSP-1 protein expression in the primary tumor and the presence of humoral response against this antigen was observed, no statistically significant association was found between CTSP-1 protein expression and any of the pathological parameters analyzed [Supplementary material (table)].
Humoral response against CTSP-1 as an independent prognostic factor for biochemical-recurrence
Kaplan-Meier method was used to estimate the relationship between the presence of humoral response against CTSP-1 and the risk of biochemical-recurrence of the disease after surgery. The probability of 5-year biochemical-recurrence among all 147 patients evaluated was 61.8%. Patients who did not produce antibodies against CTSP-1 had a higher probability of biochemical-recurrence than did those producing anti-CTSP-1 antibodies, although the difference between the groups was not statistically significant (57.2 vs. 75.6%, p = 0.075; Fig. 3a). Interestingly, none of the patients producing antibodies against CTSP-1 developed clinical symptoms of the disease progression (Fig. 3b). Five-year biochemical-recurrence free survival in our patient population was also influenced by the following parameters: preoperative PSA level, tumor stage, Gleason score, tumor volume, seminal vesicle invasion and surgical margins (Table III).
Table III. Univariate Analysis of Biochemical-Recurrence Free Survival (BRFS)
A multivariate analysis was then performed to determine whether the presence of a natural humoral immune response against CTSP-1 was an independent factor in predicting biochemical-recurrence. All variables presenting p-value < 0.20 on the univariate analysis (pathologic T staging, Gleason score, tumor volume, perineural invasion, angiolymphatic invasion, extracapsular invasion, tumor margins, seminal vesicle invasion and preoperative PSA level) were selected for building a multiple model using a stepwise forward method.
The preoperative PSA level, the Gleason score and the presence of antibodies against CTSP-1 were considered as independent prognostic factors for biochemical-recurrence free survival (Table IV) after adjusting the analysis for the use of neoadjuvant androgen suppression. Patients producing antibodies against CTSP-1 had a lower risk for biochemical-recurrence than patients without anti-CTSP-1 (HR 0.41, 95%CI 0.18–0.90, p = 0.039).
Table IV. Multivariate Analysis of Biochemical-Recurrence Free Survival
Cox regression model adjusted for the use of neoadjuvant androgen suppression.
During the multivariate analysis, we observed that Gleason score was the most important variable changing the significance of the presence of antibodies against CTSP-1 in predicting biochemical-recurrence of the disease. The Gleason score is the most widespread method used for prostate tumor grading and is based upon the degree of loss of the normal glandular tissue architecture.14 The higher the Gleason score, the more aggressive the tumor is. Indeed, the Gleason score is the single most important prognostic factor in prostate cancer influencing decisions regarding options for therapy.15
Although no statistically significant association was observed between anti-CTSP-1 antibodies and Gleason score, antibodies against CTSP-1 tend to be more frequently observed among patients with a higher Gleason score as compared to those with lower Gleason score (32.7 vs. 21.4%, p = 0.139, Table II). We then decide to perform a stratified survival analysis according to Gleason score (Figs. 3c and 3d). The presence of antibodies against CTSP-1 was significantly associated with a better prognosis in patients with higher Gleason score (36.4 vs. 80.0%, p = 0.028; Fig. 3d) as compared to those with a lower Gleason score (67.9 vs. 72.2%, p = 0.292; Fig. 3c), explaining the absence of statistical significance in the univariate analysis when all patients were analyzed as a single group (Fig. 3a).
Among various types of epithelial cancers, prostate cancer appears to be one of the best targets for specific immunotherapy since it is generally recognized that the human immune system is capable of mounting a significant immune response against the prostate tissue,16 and that immunity against prostate tissue can be induced by a variety of immunotherapies, including peptide, DNA, virus-based, dendritic cell and genetically modified cancer vaccines.7, 8
The results of several cancer vaccine trials indeed suggest a clinical benefit to patients with prostate cancer and phase III trials are current underway in patients with androgen-independent prostate cancer. However, the antigens included in these studies are mainly proteins with known prostate-restricted expression (e.g., PSA, PSMA, PAP) and most of them are not natural targets of an immune response in patients with cancer.7, 8 Identifying immunogenic proteins in patients with prostate cancer could, thus, lead to the development of other potential vaccine target antigens.
Several CT antigens are expressed in prostate tumors, including PAGE-4, NY-ESO-1, LAGE-1, XAGE-1 and members of the MAGE-A family17–21; however, little is known about their expression pattern during the natural history of the disease and their immunogenicity in prostate cancer patients. In one of the more comprehensive studies, Nakada et al. examined NY-ESO-1 mRNA expression and the presence of NY-ESO-1 antibodies in the sera of patients with localized and advanced prostate cancer.17 A higher frequency of NY-ESO-1 mRNA expression was observed in prostate cancer patients with systemic metastasis (43%), compared to those with local tumors (27%) or regional lymph node metastasis (33%). These differences, however, were not statistically significant. Anti-NY-ESO-1 antibodies were found in patients with metastatic disease (7.1%) but not in patients with localized tumors. Out of 20 patients with NY-ESO-1 mRNA positive tumors, 2 (10%) produced anti-NY-ESO-1 antibodies and seropositivity was observed for the serum from 1 patient with NY-ESO-1 mRNA negative tumor.17
In another study on prostate tumors, NY-ESO-1 protein was predominantly detected in hormone-refractory tumors (14.6%) compared to localized tumors (3%) and anti-NY-ESO-1 antibodies were also more frequently found in the serum of patients with advanced disease (18.9%) as compared to localized tumors (0.9%).18 Matched serum samples and biopsies were available for 66 and 25 patients with localized and advanced disease, respectively. None of the 2 patients with NY-ESO-1 positive tumor and localized disease was positive for anti-NY-ESO-1 antibodies and 2 out of 4 patients with NY-ESO-1 mRNA positive tumors and advanced disease produced anti-NY-ESO-1 antibodies. Seropositivity was observed for the serum from 1 patient with NY-ESO-1 mRNA negative tumor.18
Correlation between CT antigen expression and disease progression seems to be controversial mainly due to the small sample size analyzed in individual studies, variations in immunohistochemical staining and interpretation of the results. These correlations can also vary between different antigens and tumor types.22 In one of the most comprehensive studies addressing this issue, Barrow et al. assessed the expression of MAGE-A1, MAGE-A4 and NY-ESO-1 in a series of 586 primary and metastatic melanoma samples.22 Although MAGE-A1 and MAGE-A4 expression was acquired with advancing disease, NY-ESO-1 expression remained constant during disease progression and was indeed commonly lost in patients for which consecutive biopsies were available.22
The fact that CT antigen expression may be lost or acquired during disease progression is very interesting, since this could be related to antigen immunogenicity and the immunoediting process during the course of the disease.22 Immunoediting is the concept that the immune system can “sculpt” the tumor phenotype, selecting tumor variants that can evade immune defences.23 A tumor that develops in the presence of an active immune response against one specific antigen can evolve to escape the control of the immune system by loosing antigen expression. On the other hand, in the absence of an effective immune response and of a selective pressure applied to the evolving tumor, there may be little change or even increase in the tumor antigen expression, during disease progression. The same reasoning can be used to explain differences in the correlation between CT antigen expression and disease progression observed between different tumor types, since persistence of antigen expression would be expected in tumors derived from more “immunologically privileged” tissues.
We demonstrated that CTSP-1 protein, as opposed to NY-ESO-1, is already frequently expressed (61.2%) in localized prostate tumors. Moreover, a significant correlation between CTSP-1 protein expression in the primary tumor and the presence of humoral response was observed, since only 1 out of 19 (5.3%) patients bearing CTSP-1 negative tumors presented humoral response against this antigen. In addition, CTSP-1 seems to be highly immunogenic, since a significant fraction (36.7%) of patients bearing CTSP-1 positive tumors presented spontaneous anti-CTSP-1 antibodies in their plasma. To our knowledge CTSP-1 is the most immunogenic CT antigen described to date for prostate cancer and taken together our results suggest that CTSP-1 could be considered a promising candidate for prostate cancer immunotherapy. Although patients with advanced disease were not included in the present study, we expect CTSP-1 expression not to be correlated with disease progression due to its high immunogenicity.
We have also demonstrated that the presence of antibodies against CTSP-1 in the plasma of patients with clinically localized prostate cancer is associated with a better disease outcome, especially in patients with higher Gleason score, and is an independent predictor of biochemical-recurrence after RRP. Reports on the association between CT antigen expression and/or spontaneous serological responses with disease outcome are limited to a few studies involving a small number of prostate cancer patients. Nakada et al. observed that spontaneous serological response against NY-ESO-1 was associated with poor survival in hormone-refractory prostate cancer patients (HRPC).17 The median survival of HRPC patients with anti-NY-ESO-1 antibodies was significantly less than the medial survival of HRPC patients without antibodies (7 vs. 13 months, p = 0.03).17 Our results, together with the ones presented by others,17, 18 suggest that the presence of a natural immune response against CT antigens could also be used as prognostic factors in prostate cancer and deserves further evaluation.
We thank Dr. Sarah White and Dr. Erney Camargo for critically reading this manuscript. Mrs. Valéria Paixão, Mr. Carlos Nascimento, Mrs. Miyuki da Silva and Mr. Severino Ferreira provided outstanding technical assistance. R.B.P. and F.B. were sponsored by fellowships from FAPESP.