Cancer/testis (CT) antigens are potential targets for cancer immunotherapy, with NY-ESO-1 being among the most immunogenic. In several clinical trials in malignant melanoma (MM) patients, NY-ESO-1 protein/peptides showed clear evidence of inducing specific immunity. However, little is known about NY-ESO-1 expression in primary and metastatic MM and its relationship to disease progression. We analyzed NY-ESO-1 expression immunohistochemically in a series of primary and metastatic MMs and its relation to prognostic parameters and survival. We studied 61 primary and 63 metastatic MM specimens (from 61 and 56 patients, respectively). The prevalence of NY-ESO-1 expression was significantly higher in metastatic versus primary tumors [18/56 (32%) versus 8/61 (13%), P = 0.015]. There was a significant association between initial stage at presentation and NY-ESO-1 expression [stage I (3.45%), stage II (9.52%) and stage III (45.45%), P = 0.0014]. Primary MMs expressing NY-ESO-1 were significantly thicker than NY-ESO-1 negative cases (median thickness 4.7 mm versus 1.53 mm respectively, P = 0.03). No significant difference was seen in overall survival. In conclusion, NY-ESO-1 is more frequently expressed in metastatic than in primary MM and its expression is associated with thicker primary lesions and a higher frequency of metastatic disease, indicative of a worse prognosis. Our study suggests that patients with metastatic MM who express NY-ESO-1 may benefit from NY-ESO-1-based immunotherapy.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
The identification of tumor-associated antigens recognized by cellular or humoral effectors of the immune system has opened new perspectives for cancer immunotherapy (1, 2). Current melanoma vaccine studies focus on two main antigenic targets: melanocyte differentiation antigens (MDAs) and cancer/testis (CT) antigens. As their name implies, CT antigens are expressed in a variety of cancers and in germ cells of the adult testis, as well as occasionally in placenta and fetal ovary (3). More than 50 CT genes or gene families, such as MAGE, GAGE, NY-ESO-1/CTAG1B and others, are known today. The concept of immunotherapy is based on the assumption that antigenic proteins expressed in tumors can be used as targets in therapeutic approaches employing the autologous immune system (1, 3). Although serological reagents have become available, current knowledge is still mainly based on mRNA data and little is known about CT antigen expression at the protein level, especially with regard to correlation with clinical data such as autologous serological responses and prognosis. Consequently, there is a great need to increase knowledge of the actual antigen expression and its potential clinical implications. In various tumors, humoral immune responses to NY-ESO-1 are more frequently observed than for most other CT antigens and hence NY-ESO-1 is regarded as one of the most immunogenic CT antigens (3, 4).
In the present study, we attempted to determine the clinical relevance of NY-ESO-1 by an immunohistochemical analysis of NY-ESO-1 protein expression in a series of primary and metastatic malignant melanomas from a well-characterized cohort of melanoma patients with extended follow-up in order to correlate its presence with morphologic and pathologic prognostic factors.
For the 61 primary malignant melanomas analyzed, the median tumor thickness was 1.8 mm (mean 3.12 mm, range 0.3-24.0 mm). Of the 61 tumors, 13 tumors were less than or 1.0 mm thick (defined as thin MM), 37 tumors were 1.01-4.0 mm thick (intermediate MM), and 11 tumors were >4.0 mm in Breslow thickness (thick MM). Thirty-one (50.8%) tumors were axial and 30 (49.1%) were on the extremities. Histologically, 27 (44.3%) of these primary MMs were superficial spreading, 23 (37.7%) were nodular, 6 (9.8%) were acral, 3 (4.9%) were lentigo maligna and 2 (3.3%) were of the desmoplastic type.
For the metastatic melanomas, 63 lesions from 56 patients were available. Sixteen samples from 15 patients were metastases in sentinel or regional lymph nodes derived from the primary melanoma patient group while 47 specimens were derived from the 41 patients of metastatic malignant melanoma cohort without available primary lesions. Nineteen of these 63 metastases were cutaneous, 36 from the lymph nodes and 8 visceral. Five patients had two metastatic tumor sites (1 case each with metastases in: skin and lymph node, skin and lung, lung and lymph node, colon and lymph node, two lymph nodes) and one patient had three metastatic tumor sites (skin and two lymph nodes) which were included in the analysis.
In summary, 45 patients provided 45 primary MM specimens exclusively; 16 patients provided 16 primary as well as 16 metastatic MM specimens (of these 1 metastatic MM specimen was too small for immunohistochemical analysis), and 41 patients provided 47 metastatic MM specimens but no primary lesion. Consequently, a total of 102 patients provided 61 primary and 63 metastatic melanoma specimens, of which 62 metastases were suitable for immunohistochemical analysis.
NY-ESO-1 protein expression in primary melanoma
NY-ESO-1 expression was assayed immunohistochemically using mAb E978. Eight of the 61 primary melanomas were positive for NY-ESO-1 (13%): 4/8 cases were focally positive (50%), 1/8 cases moderately positive (12.5%) and 3/8 cases showed diffuse immunopositivity (37.5%). Two of the 4 focally positive cases showed only few isolated positive cells. The moderately positive case showed a heterogeneous mosaic-like immunostaining pattern with intermixed positive and negative tumor cells. Even in the cases with diffuse immunopositivity, single cells or small clusters of cells without immunostaining were present (Figure 1). In all of the positive cases, the staining was predominantly cytoplasmic, while in a few cases nuclear staining was also present. Primary malignant melanomas expressing NY-ESO-1 were significantly thicker than NY-ESO-1 negative cases (median thickness 4.7 mm versus 1.53 mm; P = 0.03, Wilcoxon rank-sum test). There was also a correlation between NY-ESO-1 expression and tumor thickness when divided into thin, intermediate and thick melanomas. All 13 thin melanomas were negative for NY-ESO-1 expression, whereas 4 of the 37 intermediate tumors (10.8%) and 4 of the 11 thick melanomas (36.4%) expressed NY-ESO-1 (P = 0.01, trend test; Figure 2). In addition, NY-ESO-1 positive lesions were more frequently nodular (62.5%) and ulcerated (50%) compared to NY-ESO-1 negative cases (34% and 26% respectively); however, these differences were not statistically significant. The presence and severity of tumor infiltrating lymphocytes were not significantly different in primary melanomas expressing NY-ESO-1 versus the NY-ESO- negative cases in this series. No significant difference in gender distribution or patient age was found among primary lesions which did or did not express NY-ESO-1. NY-ESO-1 was expressed in 39% of the female and 33% of the male melanoma cases studied. The mean patient age for NY-ESO-1 positive cases was 58.3 years compared to 58.9 years for NY-ESO-1 negative cases. Twenty-nine of these 61 patients (47.5%) whose primary melanoma tissue was available for immunohistochemical studies were stage I, 21 (34.4%) were stage II and 11 (18.04%) were stage III at initial presentation according to the American Joint Committee on Cancer (AJCC) 2002 staging guidelines. Interestingly, there was a significant association between disease stage at the time of diagnosis and NY-ESO-1 expression [stage I (3.45%), stage II (9.52%) and stage III (45.45%); P = 0.0014, Cochran-Armitage Trend test; Figure 3].
NY-ESO-1 protein expression in metastatic melanoma
Sixty-three melanoma metastases from 56 patients were available. In the six patients with more than one metastasis, the immunohistochemical analysis rendered congruent expression patterns in all except one case in which a cutaneous metastasis was negative; however, a lymph node metastasis 3 years later showed moderate NY-ESO-1 expression. No primary melanoma tissue was available for that case. Only one NY-ESO-1 positive metastatic lesion per patient was used to calculate the prevalence of NY-ESO-1 expression in metastatic melanoma (the patient with both a positive and negative metastasis for NY-ESO-1 was counted as positive). Eighteen of the 56 metastatic melanoma patients (32%) were positive for NY-ESO-1 by immunohistochemistry. This prevalence was significantly higher than that in primary patients [8/61 (13%); P = 0.015, Fisher's exact test]. Fifty percent of the visceral metastases expressed NY-ESO-1 compared to 28% of the lymph node and 26% of the cutaneous metastases (P = 0.4220, Cochran-Armitage Trend test). Nine of the 18 NY-ESO-1-expressing metastatic cases (50%) were focally positive (Figure 4, panel 4), 3 (17%) were moderately positive and 6 (33%) showed diffuse immunopositivity. Among the focally positive cases, 1 case showed only a few isolated NY-ESO-1 positive cells. Similar to what was observed in the primary tumors, the moderately positive cases often showed a heterogeneous mosaic-like immunostaining pattern (Figure 4, panel 5) and, in cases with diffuse immunopositivity, single cells or small clusters of cells without immunostaining were present. One case showed diffuse immunostaining without the characteristic heterogeneous pattern (Figure 4, panel 6).
Sixteen of the 61 primary melanomas had metastatic disease; 11/61 had positive sentinel lymph node at initial presentation and 4 eventually developed metastases in different sites. One of the sentinel lymph node metastases was too small and insufficient for immunohistochemical studies. However, in 15 of these 16 patients, the primary and metastatic lesions were available for this study; 6/15 primary and 4/15 metastatic melanoma lesions were NY-ESO-1 positive. In four cases, both the primary and corresponding metastatic lesions showed NY-ESO-1 expression (in two cases the primary lesions showed focal immunopositivity while their corresponding metastases showed moderate expression; one case showed focal positivity in the primary and metastatic lesion while the remaining case was diffusely immunopositive in the primary and metastatic site). In two cases, only the primary tumor was immunopositive while the corresponding metastases were negative for NY-ESO-1. However, one of these primary lesions showed very focal expression of NY-ESO-1 in isolated cells; the other primary MM was diffusely immunopositive but the corresponding sentinel lymph node showed only a micrometastatic focus that was negative for NY-ESO-1.
NY-ESO-1 protein expression: correlation with survival endpoints
Median overall survival time was 2.0 years shorter for patients with a melanoma metastasis expressing NY-ESO-1 [10.7 years, 95% confidence interval (CI) 6.17-15.54 years] compared to patients whose metastases were negative for NY-ESO-1 (12.7 years, 95% CI 9.5 years-not reached; P = 0.49, log-rank test). Patients with metastatic melanoma demonstrating NY-ESO-1 expression had a 9 months shorter median disease-free survival time (6.2 years, 95% CI 1.81-10.12 years) than those with NY-ESO-1 negative lesions (7.1 years, 95% CI 2.24-10.44 years; P = 0.59, log-rank test). Finally, 2-year post-recurrence survival rates were 69.8% and 71.4% for patients expressing NY-ESO-1 and those who did not, respectively (P = 0.35, log-rank test; Figure 5).
We analyzed the clinicopathological relevance of NY-ESO-1 expression in a well-characterized cohort of melanoma patients who were enrolled in the Interdisciplinary Melanoma Cooperative Group (IMCG) program at New York University (NYU) Medical Center. The IMCG's program's primary focus is to study the clinical relevance of molecular alterations in melanoma progression, in order to identify better prognostic markers and new treatments for this deadly disease. Clinical, pathological and follow-up data were available for the patients. We chose to study NY-ESO-1 expression in melanoma since CT antigens in general, and NY-ESO-1 in particular, appear to be ideal targets for active immunotherapy in patients with advanced melanoma. The identification of tumor antigens capable of inducing an immune response in cancer patients and the development of cancer vaccines utilizing these antigens are important goals in tumor immunology and cancer immunotherapy (4, 5). CT antigens are a distinct and unique class of tumor-associated antigens, first identified by autologous T-cell epitope typing in melanoma patients (6, 7). They were named after their typical pattern of expression since they are present in various tumor types and in germ cells of the adult testis (8, 9). Their presence in various tumor types, their limited expression in normal tissue (germ cells and occasionally placental trophoblastic tissue), and their ability to elicit autologous immune responses make CT antigens almost ideal targets for vaccine-based immunotherapy of cancer (9). More than 50 CT genes or gene families have been isolated to date (5). NY-ESO-1 is the most immunogenic, eliciting both humoral and cellular immune responses in a large percentage of patients with advanced NY-ESO-1 expressing tumors (10). The NY-ESO-1/CTAG1B gene was identified by the serological analysis of recombinant tumor cDNA expression libraries (SEREX) in a patient with esophageal squamous cell carcinoma (8). Subsequently, monoclonal antibodies to detect NY-ESO-1 at the protein level were generated (11, 12). Using mAb E978 which was previously developed by our group for use in formalin-fixed, paraffin-embedded tissues, NY-ESO-1 expression in primary melanoma has recently been reported in 45% of the cases in an Australian series (13); in a prior study, Goydos et al. had observed NY-ESO-1 mRNA expression in 10% of the primary MMs studied (14). We found NY-ESO-1 expression in 13% of the primary melanomas, a rate similar to the latter RT-PCR based study. In concordance with the Australian series in which a very low frequency of NY-ESO-1 expression was reported in primary tumors measuring up to 1 mm in thickness (13), all thin primary melanomas in our series were negative for NY-ESO-1. We observed a significant progressive increase in NY-ESO-1 positivity with increasing thickness of the primary lesion. NY-ESO-1 expression was three times as prevalent in thick as opposed to intermediate primary melanomas. Curiously, Barrow et al. found that NY-ESO-1 immunopositivity predominated in lesions measuring between 1.0-4.0 mm in thickness, and decreased in lesions measuring more than 4.0 mm (13). In our series, there was a consistent trend towards increased NY-ESO-1 expression in thicker primary lesions, with NY-ESO-1 being most prevalent in MMs measuring more than 4 mm in thickness. Although the differences were not statistically significant, we also found increased expression of NY-ESO-1 in nodular and ulcerated primary melanomas compared to other patterns of growth and non-ulcerated lesions. Altogether our findings support the hypothesis that NY-ESO-1 expression correlates with a worse prognosis and in this respect are in concordance with findings for other CT antigens, such as MAGE-1 and MAGE-4, that expression increased with disease progression (13). A correlation between CT antigen expression and negative disease parameters, such as level of invasion, has previously been reported on the mRNA level for several CT antigens (15).
Several morphological parameters, such as the presence of lymphocytic infiltrates, ulceration, and growth pattern are prognostic markers for cutaneous melanomas. Interestingly, no study has previously analyzed the relationship between NY-ESO-1 expression and the presence and severity of tumor lymphocytic infiltrate. A prior pathological study analyzing immunoreactivity with the anti-MAGE reagent mAb 57B in a series of 65 primary invasive melanomas found a significant association between immunopositivity and the presence of a brisk tumor lymphocytic infiltrate (16). We could not find any significant correlation between the presence of a brisk tumor lymphocytic infiltrate in primary melanomas and NY-ESO-1 expression in this series.
Few studies have looked at patterns of NY-ESO-1 expression in primary and metastatic melanomas and the results appear to be controversial. NY-ESO-1 expression in metastatic melanoma has previously been reported to be 46.8 % in a previous RT-PCR analysis (14) and 45% in an immunohistochemical study (13). Other studies found variable NY-ESO-1 expression in melanomas ranging from 26% (17), 31.6 % (18), 36% (12) to 45% (19); however no distinction between primary and metastatic lesions was made in those analyses. We found that 32% of the metastatic lesions in this series expressed NY-ESO-1 protein and this rate was significantly higher than that of primary lesions (13%). Interestingly, in their recent immunohistochemical study Barrow et al. found that NY-ESO-1 expression did not increase with disease progression and/or stage (13). Furthermore, that study showed that NY-ESO-1 expression was most commonly lost (in 31% of the patients) on serial biopsies over time suggesting that this antigen may be susceptible to modifications by the autologous immune system, in what is commonly referred to as "immunoediting" (13). Our findings parallel those of Goydos et al. (14) who demonstrated increased NY-ESO-1 mRNA expression in metastases compared to primary samples, i.e. 46.8% and 10% respectively. We did not have sufficient numbers of serial biopsies to draw a conclusion with regard to NY-ESO-1 expression in the same patient over time. Although statistically not significant, we found that there is a trend towards increased NY-ESO-1 expression in visceral metastases compared to cutaneous and lymph node metastases, suggesting that NY-ESO-1 expression may be related to biological events during tumor spread. Our ability to further explore this relationship is limited by the availability of tumor tissue, as many patients with visceral metastases do not undergo biopsies of these tumors. Of interest, Barrow et al. reported that NY-ESO-1 expression tended to be most common in brain metastases (13).
Information regarding NY-ESO-1 expression and correlation with disease stage and survival is very limited. Our results show a significant association between initial stage and NY-ESO-1 expression in primary tumors [stage I (3.45%), stage II (9.52%) and stage III (45.45%)] suggesting that NY-ESO-1 expression in primary melanomas may be a marker of worse prognosis and a predictor of positive sentinel lymph nodes. We also examined whether NY-ESO-1 expression correlated with patient survival by analyzing three measures of survival: the overall survival time from initial melanoma diagnosis, the time from initial diagnosis to first recurrence and the survival time after the first recurrence. We observed a 2-year shorter overall survival for metastatic patients whose tumors were NY-ESO-1 immunopositive. The 2-year post-recurrence survival rates were 67% and 73% for NY-ESO-1 positive and negative patients respectively. These differences however, did not reach statistical significance. The disease-free survival was 9 months shorter in patients whose tumors were NY-ESO-1 positive versus patients with NY-ESO-1 negative lesions. A larger sample size with extended follow-up is needed to validate these preliminary findings.
The frequent 'heterogeneous' pattern of expression of NY-ESO-1 in primary as well as metastatic tissues was noted in prior studies with several other CT antigens (5, 20). A 'mosaic-like' pattern, in which immunopositive cells were intermixed with negative cells, was a frequent finding in positive lesions. Even in diffusely positive cases, single cells or small clusters of cells without immunostaining were present. The significance of this peculiar pattern, which appears to be a constant finding with CT antigens, as well as the significance of cases with only a few scattered positive cells, needs to be further analyzed since they may have implications for immunotherapeutic approaches. A prior study suggested that several CT genes and their products, including NY-ESO-1, were expressed in undifferentiated mesenchymal stem cells and down-regulated after differentiation (21). Classical CT antigens, such as MAGE and NY-ESO-1, map to chromosome X (CT-X antigens) and they are expressed in the spermatogonia, the most immature testicular germ cells. This could be indicative of a potential association of CT antigens with stem cells and current on-going studies are analyzing their potential as stem cell markers (5).
In conclusion, we find an increased expression of NY-ESO-1 in thicker primary melanomas (especially melanomas measuring more than 4.0 mm in thickness), and an increased prevalence of NY-ESO-1 expression in metastatic, compared to primary, lesions (32% versus 13%, respectively), as well as an association of NY-ESO-1 expression with a more advanced clinical stage at the moment of initial diagnosis. This is partially in contradiction with recent data generated by the Australian series and further studies are necessary to clarify these discordances which are particularly important given the high immunogenicity and potential use of NY-ESO-1 for immunotherapy (13). Our findings suggest that in cutaneous melanoma, NY-ESO-1 is related to the progression of the disease and a marker of worse prognosis.
Materials and methods
The study cohort consisted of 102 melanoma patients. All patients were identified through the Interdisciplinary Melanoma Cooperative Group database at the NYU School of Medicine (42 females and 60 males, mean age 58.7 years). The study was performed under valid legal regulations and approved by the Institutional Review Board of NYU.
Primary lesions were available from 61 patients; 29 of these 61 patients (47.5%) were stage I, 21/61 (34.4%) were stage II and 11/61 (18.04%) were stage III at time of the diagnosis, according to the AJCC 2002 staging guidelines (22). In order to ensure the validity of stage-specific analyses, only primary melanoma specimens from patients who underwent sentinel lymph node biopsy were included so that cases with sentinel node biopsy were selected (without intention) from consecutive melanoma specimens. Specimens from metastatic melanoma sites were available from another 41 patients.
All specimens were standard, formalin-fixed, paraffin-embedded tissue blocks for morphological analysis. Hematoxylin-eosin stained sections of each lesion were reviewed to verify the diagnosis and assess the presence of representative lesional material.
The cases included in this study were separated into two groups:
For the primary malignant melanomas, the following clinicopathologic and demographic information was recorded: patient's age, sex, disease stage, location of primary tumor, histologic type (23, 24), Breslow thickness (25), presence of ulceration, presence and intensity of tumor lymphocytic infiltrate (26), location and timing of recurrence and metastasis, treatment and survival data.
For metastatic melanomas, recorded data included patient's age, sex, disease stage at time of primary diagnosis, location and timing of recurrence and metastasis, treatment and survival data.
Immunohistochemical analysis and scoring
Expression of NY-ESO-1 was assessed by immunohistochemistry using mAb E978 as previously described (12). Briefly, tissue sections were deparaffinized and rehydrated in xylene and a series of graded alcohols. Heat-based antigen retrieval using DAKO hipH solution (DAKOCytomation, Carpinteria, CA) in a household vegetable steamer was performed for 30 minutes. Primary antibody incubation was carried out overnight at 4˚C. As a secondary reagent, the Powervision system (Immunovision Labs, Brisbane, CA) was used. Diaminobenzidine served as a chromogen. Counterstaining was performed with Gill's hematoxylin. The extent of NY-ESO-1 expression in tumor cells was estimated microscopically by two pathologists (E.F.V. and A.A.J.) who were blinded to the patient's data and graded as follows: negative (0%), focal (<20%), moderate (20-60%) and diffuse (>60%). A total of 124 specimens from 102 patients were available for the immunohistochemical analysis of NY-ESO-1 expression.
Descriptive statistics were calculated for baseline demographic and clinicopathologic characteristics. Associations between NY-ES0-1 immunoreactivity and clinicopathologic features were assessed by Fisher's exact test, the chi-square test, or the Cochran-Armitage trend test, as appropriate. Three survival outcome measures, including overall survival (time from initial diagnosis of melanoma to death), disease-free survival (time from initial diagnosis of melanoma to first recurrence), and survival after first recurrence (time from first recurrence to death), were analyzed using the Kaplan-Meier method. All P values are two-sided, with statistical significance evaluated at the 0.05 alpha level. Ninety-five percent confidence intervals (95% CI) were calculated to assess the precision of the estimates obtained. All analyses were performed using SAS Version 9.1 (SAS Institute, Inc., Cary, North Carolina) and Stata Version 8.0 (Stata Corporation, College Station, Texas).
We thank Dr. Lloyd Old (Ludwig Institute for Cancer Research) and Dr. Brian West (AmeriPath New York) for their invaluable guidance and advice.
- Received March 13, 2007.
- Accepted April 23, 2007.
- Copyright © 2007 by Nina Bhardwaj