To the Editor:

Mesothelin (MSLN) is a 40-kDa cell surface glycoprotein and expressed not only in normal mesothelial cells slightly [1, 2], but also in various types of cancers [3,4,5,6]. Previously, we demonstrated that high MSLN expression was correlated with poor prognosis in breast cancer [7]. CA125/MUC16 (CA125) is one of the binding partners for MSLN [8,9,10,11]. Heterotypic adhesion between MSLN and CA125 may cause intracavitary tumor metastasis [8, 10]. We showed co-expression of MSLN and CA125 (Co-expression) were correlated with poor prognosis in pancreatic cancer [11]. However, there have not been any studies regarding Co-expression in breast cancer. Therefore, we investigated CA125 expression in addition to MSLN in breast cancer by immunohistochemistry and examined its association between their co-expression and clinicopathological factors.

Subjects comprised 478 patients who underwent surgical resection for primary breast cancer from January 2002 and December 2013. The clinicopathological parameters of these cases were summarized in Table S1. The immunohistochemical staining and evaluation of mesothelin and CA125 were performed as previously described [11] (Methods S1). The expression of MSLN and CA125 was positive when immunoreactivity was observed in 1 % or more of tumor cells, and negative when immunoreactivity was detected in less than 1 % of cancer cells or was absent. Co-expression was positive when the expression of both MSLN and CA125 was detected, and was negative when the expression of MSLN, CA125, or both was absent (Fig. 1 A).

Fig. 1
figure 1

A A representative cases of breast cancer is that mesothelin (A) and CA125 (B) is diffusely positive in triple-negative breast cancer. Immunoperoxidase stain, original magnification ×400. B Relapse-free survival curves for 478 patients with breast cancer after surgery classified with the status of the expressions of mesothelin and CA125. Their co-expression group shows the worst prognosis. C-D The result of Cox’s univariate analysis is shown in forest plots, and by the Cox’s multivariate analysis, NSLN and CA125 co-expression remains as an independent prognostic factor. E Relapse-free survival curves for 333 luminal-type breast cancer patients after surgery classified with the status of the expressions of mesothelin and CA125. Their co-expression group shows the worst prognosis. F-G The result of Cox’s univariate analysis was shown in forest plots, and by the Cox’s multivariate analysis, NSLN and CA125 co-expression remains as an independent prognostic factor.

Full size image

The expression of MSLN was positive in carcinoma cells in 75 (15.7 %) out of 478 breast cancer specimens, while the expression of CA125 was positive in 217 (45.4 %) out of 478 specimens and in 48 (64.0 %) out of 75 MSLN-positive specimens. The positive expression of MSLN correlated with the pathological T factor, triple-negative subtype, Grade 3, a higher Ki-67 labeling index (LI), and higher relapse rate. The positive expression of CA125 also correlated with the subtype and a higher relapse rate. The Co-expression was observed in 48 cases (10.0 %) and correlated with the pathological T factor, triple-negative subtype, Grade 3, a higher Ki-67 LI, and higher relapse rate (Table 1).

Table 1 Clinicopathological parameters according to mesothelin and CA125 expression levels
Full size table

The relapse free survival (RFS) rate was significantly poorer in patients expressing MSLN or CA125 than in those not expressing MSLN or CA125 Moreover, the prognosis of the group showing the Co-expression was the worst (Fig. 1 B). Cox’s univariate proportional hazards model analyses identified the pathological T factor, NG, lymphatic invasion, Ki-67 LI, and pathological N factor as significant risk factors for recurrence. Both the expressions of MSLN and CA125 were identified as significant risk factors for recurrence: [hazard ratio (HR) 1.89, 95 % confidence interval (CI) 1.06-3.18, P = 0.0313 for MSLN; HR 1.67, 95 %CI=1.04-2.68, P = 0.0319 for CA125], while Co-expression was a much stronger risk factor (HR 2.94, 95 %CI 1.60-5.06, P = 0.0009) (Fig. 1 C). In Cox’s multivariate analyses, Co-expression was an independent predictor of RFS in breast cancer patients (HR =1 0.92, 95 %CI 1.01-3.46, P = 0.0483) as well as the pathological T factor (HR = 2.26, 95 %CI 1.31-4.08, P = 0.0032) and pathological N factor (HR = 2.45, 95 %CI 1.43-4.28, P = 0.0009) (Fig. 1 D, including MSLN and CA125 analysis Table S2).

In 333 patients with hormone receptor-positive (luminal type) breast cancer, the RFS rate was significantly poorer in patients expressing MSLN than in those not expressing MSLN (P = 0.0021). The RFS rate also tended to be lower in patients expressing CA125 than in those not expressing CA125 (P = 0.057). The prognosis of the group with Co-expression was the poorest (Fig. 1 E). Cox’s univariate and multivariate analyses were performed on 333 luminal-type cases (Fig. 1 F). The expression of MSLN was identified as a significant risk factor for recurrence (HR 3.16, 95 %CI 1.36-6.54, P = 0.010). In luminal-type patients, the expression of CA125 was a marginal risk factor for recurrence (HR = 1.80, 95 %CI 0.97-3.37, P = 0.0606); however, Co-expression was identified as a significant risk factor (HR = 5.00, 95 %CI 1.87-11.2, P = 0.0027). In the multivariate analysis, Co-expression was independent predictors of RFS in luminal-type breast cancer patients (Fig. 1 G, including MSLN and CA125 analysis Table S3).

In conclusion, we herein reported the clinicopathological significance of the co-expression of MSLN and CA125 in breast cancer, particularly in the luminal type, as an independent prognostic factor.