Investigating the prognostic impact of NY-ESO-1 expression and HLA subtypes in metastatic synovial sarcoma

Background To better understand the importance of New York esophageal squamous cell carcinoma 1 (NY-ESO-1) and human leukocyte antigen (HLA) subtype in treatment decision making, further investigation of their prevalence and prognostic impact among patients with metastatic synovial sarcoma (mSS) is needed. Patients and methods This was a retrospective clinico-biological cohort study of adults with mSS. Patient data were collected from the French Sarcoma Group NetSARC database and supplemented by electronic medical records. Primary tumor samples were collected and analyzed for NY-ESO-1 expression by immunohistochemistry (IHC) and HLA-A∗02 status by RNA sequencing (RNA-seq). The primary cohort included patients with available primary tumor samples; the impact of a larger sample size was explored by including patients who had either a primary or metastatic sample (termed the exploratory cohort). P values are provided for descriptive purposes. Results In 92 patients with primary tumor samples, ∼25% (n = 23) were positive for NY-ESO-1 and HLA-A∗02 expression (dual positive). Among 106 patients with IHC data, 61% (n = 65) were NY-ESO-1 positive, and among 94 patients with RNA-seq data, 45% (n = 42) were HLA-A∗02 positive. The median overall survival (OS) for positive versus negative NY-ESO-1 status was 35.3 and 21.7 months, respectively (unadjusted P = 0.0428). We observed no difference in median OS for HLA-A∗02-positive versus -negative and dual-positive patients versus others (both unadjusted P > 0.05). Multivariate analyses of OS showed no prognostic impact for NY-ESO-1 among primary tumor samples and in the exploratory cohort. However, in the latter, we observed an association between NY-ESO-1 expression and OS in the first-line (P = 0.0041) but not in the second-line setting. Conclusions The primary tumor cohort showed no association between NY-ESO-1 expression and OS (including stratification by HLA-A∗02 subtype and treatment line), when adjusting for important prognostic factors, possibly due to small sample sizes.


INTRODUCTION
Synovial sarcomas (SSs) comprise w5%-10% of all soft tissue sarcomas (STSs). 1 SS is characterized by translocation t(X;18)(p11.2;q11.2),which results in the fusion of genes SSX1, SSX2, or SSX4, and SYT (SS18). 2,3SS usually presents in the lower extremities of adolescents and young adults. 1 Although most patients are diagnosed with localized disease, around 50% of them will develop metastatic disease. 4,5hemotherapy is the standard treatment in metastatic or unresectable disease and typically comprises anthracycline and/or ifosfamide as first line (L1); however, this approach has been shown to have limited efficacy. 1Treatment options in the second-line (L2) setting for advanced SS include trabectedin and the oral angiogenesis inhibitor pazopanib. 6azopanib has been reported to provide a progression-free survival (PFS) rate of 49% at 12 weeks in patients with advanced SS; nevertheless, no complete responses have been observed. 7The need for novel treatments with improved efficacy was highlighted by a recent meta-analysis of 16 studies with pazopanib or trabectedin in previously treated patients with metastatic SS (mSS), in which the overall estimate of pooled objective response rate was 14.7% and median overall survival (OS) was <12 months for both medications. 8ew York esophageal squamous cell carcinoma 1 (NY-ESO-1) cancer testis antigens are tumor-associated proteins that have been found specifically expressed in several tumor types, including a sizable proportion of SS. 9,10 NY-ESO-1 has been investigated as a therapeutic target of adoptive cell therapy in several clinical trials, and T-cell receptor gene therapies have been exploring specific human leukocyte antigen (HLA) alleles for presentation of the NY-ESO-1 antigen.][13][14][15] The literature shows variation in the relationship between NY-ESO-1 expression and OS in STS.One study associated NY-ESO-1 expression with favorable survival prognosis, 11 one study reported worse survival prognosis, 13 and three studies have shown that OS is not affected by NY-ESO-1 expression. 12,15,16However, in these studies, interpretation of the impact of NY-ESO-1 expression on OS is limited by small sample sizes 12,15,16 and/or the absence of SS-specific analyses. 11,13here remains a need for a better understanding of the prevalence of HLA subtype A*02 (HLA-A*02) and NY-ESO-1 expression among a broad population of patients with mSS, as well as the prognostic impact of these markers for this patient population.The objectives of this study were to (i) describe the prevalence and levels of expression of NY-ESO-1 and HLA-A*02 subtypes, and clinical characteristics of adult patients with mSS; and to (ii) describe and compare clinical outcomes among patients with mSS who express NY-ESO-1 and relevant HLA-A*02 subtypes against those who do not (independently, and together), and by lines of systemic treatment.

Study design
This was a retrospective clinico-biological cohort study in which data were obtained from the NetSARC database of the French Sarcoma Group.The study included adult patients with a histologically confirmed diagnosis of mSS diagnosed since 2000 and with available formalin-fixed paraffin-embedded archival primary tumor samples (and metastatic sample if available) with a corresponding pathological report.The patients were diagnosed and managed in 1 of the 27 national reference centers of the NetSARC network, across France.None of the patients were previously treated with T-cell therapy.
Patient information and tumor characteristics were collected from the NetSARC database, and data from any missing fields and updates were collected from patient electronic medical records from visiting treatment centers.Further information on patient data protection can be found in the Supplementary Methods, available at https:// doi.org/10.1016/j.esmoop.2024.103645.

Biomarker detection
Patients with mSS were identified from the French Sarcoma Group database and each case was reviewed by a sarcoma expert pathologist.NY-ESO-1 antigen expression on primary tumor sample was assessed using immunohistochemistry (IHC).Determination of NY-ESO-1 positivity was based on the diagnostic enrolment criteria of !30% tumor cell staining at 2þ/3þ intensity.HLA-A*02 positivity was determined by RNA sequencing of the primary tumor sample and included patients with HLA-A*02:01, HLA-A*02:05, or HLA-A*02:06 status.Further information on tissue collection, validation, and concordance assessment of the Centre Léon Bérard NY-ESO-1 IHC assay can be found in the Supplementary Methods, available at https://doi.org/10.1016/j.esmoop.2024.103645.

Study cohorts
While the pre-planned primary analysis was conducted among patients with primary tumor samples available (termed the primary cohort), ad hoc exploratory analyses were conducted in patients with either a primary or metastatic sample (termed the exploratory cohort; for patients with both primary and metastatic samples, the metastatic sample was selected) to explore the effect of an increase in sample size on results.

Endpoints
The study endpoints included OS, PFS, and objective response rate, measured as per best RECIST 1.1 criteria, and time to next treatment among patients with mSS with NY-ESO-1 and HLA-A*02 subtype positivity against those who do not (independently, and together), and by lines of systemic treatment.This study focuses on OS analyses of patients with mSS according to NY-ESO-1 expression, HLA-A*02 subtypes, and lines of systemic treatment.Patients who were NY-ESO-1 positive and HLA*A02 positive were termed 'dual positive'.

Statistical analysis
Clinical covariates assessed included age at diagnosis of metastatic disease, sex, histology, size, grade, and location of primary tumor, time between original and metastasis diagnosis, year of metastasis diagnosis, site of first metastasis, number of metastatic sites, number of chemotherapy lines, and drugs used in L1 systemic therapy.
For the 'overall' patient population, which included treated and untreated patients, irrespective of the line of systemic treatment, OS was measured from date of metastatic diagnosis to date of death, or date of last news for patients still alive, at the time of database lock (1 August 2021).For patients receiving L1 and L2 systemic therapy, OS was calculated from systemic therapy initiation until date of death, or date of last news for patients still alive, at database lock.
As the final sample size was determined by the availability of tissue and completeness of data, and the a priori hypothesis was that HLA subtype and NY-ESO-1 positivity have no effect on prognosis, a sample size or power calculation would have little benefit.Instead, the calculation of the minimum detectable effect (MDE) was implemented.At a given significance level, this estimate could yield a significant result that could distinguish survival between the different subtypes.
The MDE is a function of total sample size, the proportion of events expected across the sample, the proportion of the sample allocated to the positive subgroup, and the type 1 and type 2 errors.The proportion of events expected across the sample was expected to be 70% across all scenarios.Type 1 and type 2 errors were assumed to be 5% and 20%, respectively.The proportion of sample allocated to the positive subgroup was likely to be 50% for patients who were HLA-A*02 positive and 70% for patients who were NY-ESO-1 positive.The scenario in which 70% of patients are NY-ESO-1 positive also applies to patients who are HLA-A*02/NY-ESO-1 positive, which is expected to represent w35% of patients.Based on a sample size of 100, the maximum detectable hazard ratio (HR) yielding a significant difference in survival between positive and negative subgroups was 0.63 for HLA-A*02 and 0.60 for NY-ESO-1.
Statistical analyses of patient demographics and characteristics of patients in the dual-positive cohort versus all metastatic patients were carried out using Fisher's exact test.Several tests were conducted without adjustments for multiple testing; alpha adjustments and a predefined testing hierarchy were not included.
Survival distributions were estimated using the Kaplane Meier method and compared between biomarker expression groups using the log-rank test.Univariate analyses were used to evaluate the prognostic value of each covariate and biomarker on OS.Covariates found to be significant at 0.1 alpha level in the univariate analysis were then included in the multivariate model.All multivariate analyses P values were calculated by Wald chi-square test.
All analyses were provided with corresponding P values for descriptive purposes and were evaluated against a nominal P value of 0.05 for exploratory purposes.All statistical analyses were carried out using SAS® software, v9.4 (SAS Institute Inc, Cary, NC).
At the time of data analysis, 80.7% (n ¼ 88) of patients had died and the median follow-up was 69 months (range 1.1-138 months).Out of 109 patients, 45% (n ¼ 49) were female and 45.9% (n ¼ 50) were 45 years of age at metastatic diagnosis.Table 2 reports the distribution of tumor, patient, and management characteristics in the primary cohort comparing dual-positive patients with others.The proportion of patients with location of the primary tumor in the limbs [n ¼ 20 (87%), P ¼ 0.003] and location of metastases only in the lungs [n ¼ 21 (91.3%),P ¼ 0.029] was higher for the dual-positive group versus the total population (Table 2).Nine patients (8.3%) did not receive any chemotherapy, whereas 20 (18.3%) and 80 (73.4%) patients received L1 and L2þ chemotherapy, respectively.

Primary and metastatic paired sample summary
Fourteen patients had paired primary and metastatic samples available (Supplementary Table S1, available at https:// doi.org/10.1016/j.esmoop.2024.103645).Although there was high concordance in NY-ESO-1 positivity between patients' metastatic and primary tumor samples (10/14), it was not fully concordant.In three cases (21.4%),NY-ESO-1 expression was positive in the metastatic sample but negative in the primary sample.In one case (7.1%), NY-ESO-1 expression was negative in the metastatic sample but positive in the primary sample.In these four cases, patients received systemic chemotherapy between primary and metastatic sampling.Therefore, both primary and exploratory cohort results are reported.

DISCUSSION
Assessment of the prevalence of NY-ESO-1 expression and HLA-A*02 subtypes is crucial for the further clinical development of agents targeting these biomarkers in mSS.This is especially relevant with adoptive T-cell therapies, in which assessment of efficacy through randomized trials is challenging.In this context, it is essential to confirm that  expression of NY-ESO-1 and HLA-A*02 subtypes does not modify the natural history of the disease.To our knowledge, this is the largest clinico-biological cohort to examine the prevalence and the prognostic impact of both NY-ESO-1 and HLA-A subtypes in patients with mSS, and the first study in this population using a combination of real-world data and archival tissue samples.In this study, 61% (n ¼ 65) of patients were NY-ESO-1 positive and 45% (n ¼ 42) were HLA-A*02 positive; 25% were dual positive, with comparable prevalence in L1 and L2 settings.][19] High concordance in NY-ESO-1 positivity between patients' primary and metastatic tumor samples was observed, but a lack of total concordance highlighted the importance of including both primary and exploratory cohort results.
The median OS in the primary tumor cohort was short (26.4 months) in the mSS population, irrespective of NY-ESO-1 or HLA-A*02 positivity or treatment line.Our findings on OS are comparable with recent data from the French Sarcoma Group using the same database in patients with mSS (median OS of 22.3 months). 20Unadjusted results reported a longer OS in patients whose tumors expressed NY-ESO-1, although no prognostic impact was observed in the multivariate analysis.In the exploratory cohort, there was an association between NY-ESO-1 positivity and OS since L1 treatment initiation, despite adjusting for certain parameters such as the site of primary tumor, localization of first metastases, and treatment for L1.This may be explained by a subgroup of patients with very good prognostic factors who benefited from surgical approach alone for oligometastatic disease and did not receive any systemic chemotherapy.Therefore, this subgroup was not included in the L1 analysis.There were no differences in survival outcomes based on HLA-A*02 positivity or on dual positivity, including in the L2 setting.
Several studies have attempted to elucidate the association between NY-ESO-1 expression and OS in STS, with varying conclusions, and each with low sample sizes and/or lack of specificity to an SS population.In a recent study of high-risk patients with STS with long-term follow-up, high NY-ESO-1 expression correlated with favorable survival prognosis (>120 versus 58 months; P ¼ 0.0370) and lower tumor grade (grade 2 versus grade 3; P ¼ 0.0290); however, this study did not focus on SS specifically. 11In a metaanalysis, Wang et al. reported an association between NY-ESO-1 expression and worse OS and PFS in patients with solid tumors (including SS); nevertheless, no SS-specific analyses were conducted. 13Finally, three studies have shown no impact on OS.In a Cancer Genome Atlas-based study, OS in patients with SS was not affected by NY-ESO-1 expression; however, this analysis was only based on 10 patients with SS. 12 In a study investigating the expression of NY-ESO-1 by IHC and real-time polymerase chain reaction in SS, Iura et al. 15 reported a strong correlation between high NY-ESO-1 expression and the presence of necrosis and advanced clinical stage; nevertheless, the study was limited by a lack of survival data and the small number of SS samples (n ¼ 11) used for gene expression profiling.Our findings align with the recent study from Gyurdieva et al., in which OS was not affected by NY-ESO-1 expression in 45 patients with SS. 16 To our knowledge, the present study is one of the largest clinico-biological cohort studies specifically focusing on the prognostic value of NY-ESO-1 in patients with mSS, a rare subtype of STS.
HLA-A*02 positivity had no prognostic effect on OS in the current study, consistent with a recent study in which patients with mSS underwent HLA typing to determine eligibility for a clinical trial of NY-ESO-1-specific engineered T cells, and reported that HLA-A*02:01, HLA-A*02:05, and HLA-A*02:06 genotypes did not have a prognostic effect on OS in patients with mSS. 14etrospective studies are generally associated with certain limitations, including possible bias from missing data in medical records, possible inconsistencies within and across physician assessments and radiological reports, underpowered estimates that may be unreliable due to the rarity of SS, and age of samples.In our study, both the primary and exploratory cohorts had very few missing IHC patient data; any data missing from the NetSARC database were obtained by contacting treatment centers and accessing individual patient electronic medical records.The possible bias occurring from the low sample size due to the rarity of the disease and available metastatic tissue samples was minimized with the inclusion of the exploratory cohort analysis.Moreover, based on a sample size of 100, it was estimated that the maximum HR yielding a significant difference in survival between positive and negative subgroups for NY-ESO-1 was 0.60.This threshold was surpassed in our study, ensuring the possible bias of underpowered estimates due to the rarity of SS was minimized.
Overall, our findings provide further clarity on the lack of prognostic impact of NY-ESO-1 expression in a broad population of patients with mSS and expected clinical outcomes in this population.The results of this study can inform trial design in this rare population, for novel treatments that target NY-ESO-1 with or without HLA restriction.

Conclusion
To our knowledge, this is the first study to examine the prognostic impact of both NY-ESO-1 and HLA-A subtype in patients with mSS, and the first study in this population using both real-world data and archival tissue samples.This study provides much-needed information on outcomes in mSS, a distinct and rare subset of the broader STS population.NY-ESO-1 and HLA-A type, independently and together, show no significant impact on OS.Prognosis remains poor for this population, regardless of NY-ESO-1 expression and HLA-A subtype, and there is a need for new treatments that improve OS.NY-ESO-1 remains an important target in mSS, and this study can inform the selection of relevant populations of patients for clinical trials and trial design.

Figure 3 .Volume 9 -
Figure3.OS according to dual positivity in the primary (A) and exploratory (B) cohorts.Data show median OS for patients who were positive for NY-ESO-1 and had HLA-A subtype A*02 versus those who were negative for NY-ESO-1 and had other HLA-A subtypes (with unadjusted HR and P values).CI, confidence interval; HLA, human leukocyte antigen; HLA-A, human leukocyte antigen-A; HR, hazard ratio; IHC, immunohistochemistry; NY-ESO-1, New York esophageal squamous cell carcinoma 1; OS, overall survival.

Table 1
HLAþ refers to HLA-A*02.All values are n (%).HLA, human leukocyte antigen; HLA-A, human leukocyte antigen-A; IHC, immunohistochemistry; NY-ESO-1, New York esophageal squamous cell carcinoma 1. a One of the two patients with missing data for HLA had available IHC data but their NY-ESO-1 result was not informative.This patient was included in overall survival analyses of the dual-positive group versus others.

Multivariate analysis of OS using the Cox model a
Figure 2. OS according to NY-ESO-1 positivity by IHC in the primary cohort.All multivariate analysis P values were calculated by Wald chi-square test.CI, confidence interval; HR, hazard ratio; IHC, immunohistochemistry; NE, not evaluable; NY-ESO-1, New York esophageal squamous cell carcinoma 1; OS, overall survival; SS, synovial sarcoma.aFinal model is estimated on 105 patients and 85 deaths due to missing values.dualpositivity and OS, including by line of treatment.Similar results were reported in the exploratory cohort, in which the median OS for patients with dual-positive expression was 35.3 months (95% CI 16.2-42.8months) versus 26.6 months (95% CI 20.2-34.5 months) for others [unadjusted HR 0.77 (95% CI 0.47-1.25),unadjusted P ¼ 0.2584] (Figure