Mesothelin Expression in Triple Negative Breast Carcinomas Correlates Significantly with Basal-Like Phenotype, Distant Metastases and Decreased Survival

Mesothelin is a cell surface associated antigen expressed on mesothelial cells and in some malignant neoplasms. Mesothelin-targeted therapies are in phase I/II clinical trials. The clinicopathologic and prognostic significance of mesothelin expression in triple negative breast carcinomas (TNBC) has not been fully assessed. We evaluated the expression of mesothelin and of basal markers in tissue microarrays of 226 TNBC and 88 non-TNBC and assessed the clinicopathologic features of mesothelin-expressing breast carcinomas. Furthermore, we investigated the impact of mesothelin expression on the disease-free and overall survival of patients with TNBC. We found that mesothelin expression is significantly more frequent in TNBC than in non-TNBC (36% vs 16%, respectively; p = 0.0006), and is significantly correlated with immunoreactivity for basal keratins, but not for EGFR. Mesothelin-positive and mesothelin-negative TNBC were not significantly different by patients’ race, tumor size, histologic grade, tumor subtype, lymphovascular invasion and lymph node metastases. Patients with mesothelin-positive TNBC were older than patients with mesothelin-negative TNBC, developed more distant metastases with a shorter interval, and had significantly lower overall and disease-free survival. Based on our results, patients with mesothelin-positive TNBC could benefit from mesothelin-targeted therapies.


Introduction
Mesothelin (MSLN) is a 40-kDa glycosylphosphatidylinositol-linked cell surface antigen present in normal mesothelial cells and overexpressed in several human malignancies, including mesothelioma, pancreatobiliriary, ovarian and lung adenocarcinomas [1][2][3][4][5][6][7][8]. In mesothelioma MSLN promotes tumor cell invasion by increased MMP-9 secretion [9]. MSLN also binds CA-125/MUC16 with very high affinity and may contribute to the adhesion of tumor cells in peritoneal metastasis [10,11]. Mesothelin expression increases resistance to TNFa-induced apoptosis through Akt/PI3K/NF-kB activation and IL-6/Mcl-1 expression in pancreatic carcinoma cell lines [12]. MSLN-overexpressing pancreatic cancer cell lines showed increased cyclin E and cyclin dependent kinase 2 expression, resulting in increased cell proliferation and cell cycle progression [13]. Membrane-bound MSLN is also released into body fluids and its use as a potential serum tumor marker is currently under investigation [14,15]. MSLN is an attractive target for targeted therapy due to its limited distribution in normal tissues, high immunogenicity, and elevated expression in several human malignancies [16]. Several ongoing clinical trials in patients with ovarian cancer, with pancreatic cancer or with mesothelioma suggest that MSLN-specific T-cell responses have a beneficial effect [16][17][18][19][20][21][22].
In this study, we assessed the expression of MSLN in a large cohort of TNBC and non-TNBC. We also correlated MSLN overexpression with clinicopathologic features and basal-like immunophenotype of TNBC [39,43]. Furthermore, we evaluated MSLN as a potential prognostic marker in TNBC by correlating its expression with clinical outcome.

Tissue microarrays
Tissue microarrays (TMAs) containing 226 TNBC and 88 non-TNBC were used in this study. A breast carcinoma was defined as TNBC if nuclear staining for ER and PR was detected in less than 1% of the tumor cells, and HER2 was negative (0 or 1+) by immunohistochemistry (IHC) or equivocal (2+) by IHC and showed no HER2 gene amplification by fluorescence in situ hybridization (FISH) [44,45]. The TNBC cases were obtained from consecutive patients who underwent surgical excision of the primary breast carcinoma at our center between 2002 and 2006 and for which slides and blocks were available for the study. A TMA of non-TNBC from consecutive patients treated at our institution in 2004 was used for reference. Triplicate 0.6-mm diameter cores from formalin-fixed, paraffin-embedded blocks were used to construct the TMAs. Only carcinomas spanning 0.5 cm or larger were used for the TMAs, to ensure the availability of residual carcinoma for possible future clinical use. Tumor size, grade and the presence or absence of lymphovascular invasion (LVI) were extracted from the original pathology reports. Collection or study of existing data -application for exemption from IRB/ PB review (including waiver of HIPAA authorization and informed consent) was reviewed and approved by the Institutional Review Board (IRB) and Human Biospecimen Utilization Committee.

Statistical analysis
The relationship between MSLN staining, basal-like phenotype, and clinicopathologic features was assessed using Fisher's exact test. Five-year estimates of overall survival (OS) and disease-free survival (DFS) by MSLN positivity, basallike phenotype and clinicopathologic features were calculated using Kaplan-Meier methods. Differences between the Kaplan-Meier curves were tested using log-rank test. A p-value ,0.05 was considered as statistically significant.
Among non-TNBC, 67 cases were ER and/or PR positive and HER2 negative, 21 were ER and/or PR positive and HER2 positive. MSLN was positive in 13 out of 76 IDC-NOS and one mixed ductal and lobular carcinoma. No statistically difference in the prevalence of MSLN expression was observed between ER/PR positive, HER2 negative and ER/PR positive, HER2 positive tumors (Table 3).

Correlation with clinical outcome
The median follow-up time was 5.3 years (range 0.7-8.2). We observed a general trend towards increased frequency of distant metastases in patients with MSLNpositive TNBC, compared to patients with MSLN-negative TNBC and non-TNBC. Patients with MSLN-positive TNBC also had shorter interval to metastases and showed a greater propensity to develop brain metastasis ( Table 6).

Discussion
Most TNBC have an aggressive clinical course and do not respond to current therapies targeting ER, PR, and HER2. MSLN is a cell-surface antigen overexpressed in several human malignancies and constitutes a promising immunotherapy target [1][2][3][4][5][6][7][8], which could provide a much needed therapeutic option for patients with TNBC. Several published studies have evaluated MSLN expression in different tumor types, using various scoring systems to quantify the immunohistochemical expression of MSLN. Argani et al [3] categorized any tumor with $1% staining of any intensity as ''positive'' for MSLN and any staining between 1%-25% as ''focal''. Swierczyncki et al [4] also used similar cut-off percentage, but required ''moderate to strong'' labeling intensity. In their study, cases with ''focal'' staining and ''positive'' cases were combined together for statistical analysis [4]. Ho et al [6] did not explicitly specify an immunohistochemical cut-off score in their  [46]. Using these criteria, the authors reported MSLN overexpression in 67% (29/43) of TNBC, compared to only 3% and 4% of ERpositive and HER2-positive tumors respectively [46]. Table 7 summarizes studies evaluating the prognostic significance of mesothelin expression by immunohistochemistry in a variety of adenocarcinomas  [47][48][49][50][51][52][53][54][55]. Although there is no consensus on the scoring criteria of immunohistochemical staining for mesothelin, ''high level'' of mesothelin expression was significantly associated with worse outcome [48,50,52,53,55]. In our study of 226 TNBC treated at our institution we found MSLN to be overexpressed in 36% of cases. The rate of MSLN-positive TNBC in our study may appear relatively low compared to that reported by Tchou et al [46], because we regarded as MSLN-positive in only cases that showed substantial MSLN reactivity, including at least moderate staining intensity or if weak intensity in more than 50% of the tumor cells. Our scoring criteria is similar to that used in previous studies [48,52,53] demonstrating prognostic significance of mesothelin expression by immunohistochemistry in pancreatobiliary and gastric carcinomas. The use of a high cutoff of MSLN positivity, albeit arbitrarily selected, identifies cases that are more likely to be targetable to treatment with anti-MSLN. Interestingly, in our series, the use of strict criteria of MSLN-positivity identified carcinomas with significantly worse clinical behavior.  A recent study by Parinyanitikul and colleagues showed no correlation between MSLN expression and survival outcomes in triple negative breast carcinomas [54]. MSLN staining was quantified using an H-score that combined the percentage (0-100%) and intensity (1+, 2+, 3+). The H-score was calculated by multiplying the  percentage of positive cells by a factor representing the intensity of immunoreactivity, with final score ranging between 0 and 300. An H-score of 10 was chosen as the threshold for MSLN positivity [54]. Using this scoring system, 37 (34%) of 109 TNBC were deemed MSLN positive [54], but MSLN expression in TNBC did not show prognostic significance. In contrast, we found that MSLN positive TNBC had significant worse prognosis. The difference between the study by Parinyanitikul et al and ours probably stems from the different criteria used for MSLN positivity. In addition, in the study by Parinyanitikul et al [54], TNBC was defined as ER and PR #5%, HER2 negative by IHC and or FISH. The different criteria used to assign ER and/or PR positivity in the study by Parinyanitikul et al [54] (,5%) and in our own (,1%) [44] may also account in part for the different results.
Over two-thirds of TNBC in our series demonstrated at least focal MSLN staining in at least 1% of tumor cells (Table 2), a percentage similar to that reported by Tchou et al [46], but we documented higher rates of MSLN reactivity in non-TNBC (45% showing at least focal staining, and 16% with substantial MSLN expression in our study versus only 3% in a prior study [46]). However, in our series significantly more TNBC than non-TNBC were strongly MSLN-positive (82/226, 36% vs 14/88, 16%; p50.0006). Differences in proportion and intensity of staining could potentially be explained by differences in the choice of MSLN antibody dilution utilized, and definitive characterization of MSLN expression in non-TNBC requires further evaluation.
To the best of our knowledge, our series is the largest to date to assess MLSN expression in TNBC, allowing to further evaluate its correlation, or lack thereof, with basal immunophenotype. In our study, MSLN expression in TNBC correlated with basal cytokeratin expression but not with that of EGFR. Furthermore, MSLN expression was a predictor of worse outcome independent of basal immunophenotype.
In conclusion, MSLN, a cell surface antigen overexpressed in several malignancies, shows substantial expression in TNBC. Among TNBC, MSLN appears to be an independent prognostic marker associated with distant metastasis and worse survival. Patients with MSLN-positive TNBC could benefit from MSLN-targeted immuno therapies currently in development.