This study presents data on the protein expression of the activin homodimer inhibin beta-A subunit in a comprehensive European cohort of AEG/AS tumors.
Our cohort studied encompasses 277 patients treated at a single center primarily by surgery, we observed high activin homodimer subunit inhibin beta A expression in roughly a quarter of the cases (71/277, 26,0%). Activin expression correlated inversely with tumor stage, and as therefore expected, we observed a statistically significant higher overall survival in patients with tumors exhibiting high activin expression (72.1 +/− 8.3 versus 51.2 +/− 4.6 months, p < 0.01). This effect was predominantly found in tumors of UICC stage III (35.5 +/− 9.2 versus 15.6 +/− 2.4 months, p < 0.01). Additionally, higher activin expression scores were seen in tumors of the expansive type rather than tumors of the infiltrative type following the Ming-Classification (p = 0.04). A multivariate regression did not show activin subunit protein expression as independent predictor of survival, which is most probably due to the strong inverse correlation of activin with tumor stage in the cohort studied.
These data are complementing prior in vitro data demonstrating a growth inhibiting effect of activin on gastric cancer cell lines [22, 23]. Studies investigating the mRNA expression of activin in AEG/AS cohorts reported a shorter overall survival [25, 27]. This discrepancy in findings could be explained by post transcriptional control of activin levels, as well as genetic differences in the cohorts, as the cohort studied herein is the first predominantly European cohort investigated for activin protein expression in AEG/ASs. Arguably tumor protein activin expression is a more accurate measurement of activin signal activation as compared to activin mRNA. In our cohort, the correlation between overall survival and activin subunit protein expression was most pronounced in UICC stage III cancers. This observation might be due to a stronger pro-proliferative effect through other pathways in these tumors when compared to UICC stage I and II cancers which is mitigated in cases of higher activing protein expression, but further studies are needed. The higher activin expression in tumors of the expansive type rather than tumors of the infiltrative type is most likely due to the correlation of both expansive type  as well as activin expression with lower tumor stage.
No correlation was found between activin expression in metastases or lymph nodes and overall survival, albeit a smaller cohort size reduces the power to detect a difference in these cases.
Interestingly, a correlation of activin protein expression and CD4+ T- helper cell infiltration was found, but not overall CD3+ T-cell infiltration or CD8+ cytotoxic T-cell infiltration.
We previously reported on longer survival of patients with AEG/ASs exhibiting high CD4+ cell infiltration . Even though a non-statistical significant trend towards correlation between activin protein expression and overall survival was seen in tumors exhibiting a low CD4+ T-cell infiltration (high 57.4 +/− 11.0 months versus low 47.1 +/− 5.7 months, p = n.s.), the effect was more pronounced in tumors with high CD4 infiltration score (88.8 +/− 12.1 months versus 55.9 +/− 8.6, p < 0.05). Interestingly, the longer overall survival from high CD4+ T-cell infiltration and high activin expression were reliant on each other. This finding strongly hints towards a tumor suppressive effect, activin exerts through a yet unspecified effect on CD4+ T-cells infiltrating the tumor. Whether this presumed effect is a direct effect on lymphocytes or an indirect effect through activins effect on tumor epithelium or fibroblasts and the induction of other signaling molecules cannot be said at this point. Due to the correlative nature of data presented here, it is also conceivable that combined high activin expression and CD4+ T-cell infiltration a marker for a yet to specify tumor subtype. CD4+ T-cells are a heterogeneous group of cells including pro- and anti-inflammatory T-helper cells (Th1 and Th2), IL-17 producing Th17 cells, and regulatory T-cells, and further subclassification of CD4+ T-cells should be performed in future studies to possible discern mechanistic interaction between activin and CD4+ T-cells in AEG/AS tumors, especially given the recent data on activin driving differentiation towards Th17 cells .
Several limitations should be kept in mind when interpreting the data presented here. TMAs are a valuable tool to assay protein expression in large cohorts such as ours, but some concern exists due to the incomplete reflection of tumor heterogeneity, especially with regards to tumor infiltrating lymphocytes. The heterogeneity in between TMA cores, at least for activin staining was low (difference > 1 point in either intensity or positive tumor cell sub-scores in 2,4% of cases), and the tumor infiltrating lymphocyte working group sees TMAs as adequate tool for rapid evaluation of large clinical cohorts . Nevertheless, full reflection of tumor heterogeneity was not possible due to our approach, a limitation that should be kept in mind. In our study, we investigated protein expression of the activin homodimer subunit inhibin beta A. Inhibin beta A dimerizes with either the inhibin alpha subunit or Inhibin beta subunit, for which several isoforms exist. A cross-reactivity with other homo- and heterodimers of inhibin beta A besides activin A cannot be ruled out by immunohistochemistry, an important limitation when interpreting the results of this study.
In conclusion, the data presented herein show a correlation of higher activin tumor protein expression and longer overall survival in a cohort of patients with AEG/AS tumors. The data imply a net anti-tumorigenic effect of activin signaling in this tumor entity, albeit due to the context specific nature of activin signaling observed in other tumor entities a more distinguished function with a net pro tumorigenic effect in a subset of tumors is conceivable. Therefore, further mechanistic in vitro and in vivo studies of activin in the context of AEG/ASs, especially with regards to its effect on CD4+ tumor infiltrating lymphocytes, are warranted.