Background
As the leading cause of gynecologic cancer-related deaths worldwide, almost 75% ovarian cancer patients are diagnosed in late stage, losing the best time for operation. Thus, it is urgent to find new prognosis or therapeutic targets for ovarian cancer.
In humans, S100 family is consisted of 21 acidic-Ca
2+ binding proteins, which are highly conserved and functioned as both intracellular Ca
2+ sensors and extracellular factors [
1]. Among 21 members of S100 family, four are found dispersed throughout the genome (including S100B-21q22, S100G-Xp22, S100P-4p16 and S100Z-5q14); and the rest ones (from S100A1 to A100A14, S100A7 and S100A16) are encoded in two series clusters within a 2 Mb region on chromosome locus 1q21 [
1]. Exhibiting high consistence in sequence and structure, S100 family members are interchangeable in function and they show a wide spectrum of biological processes, including proliferation, apoptosis,migration,inflammation and differentiation and the like [
2].
Dysregulated expression of multiple S100 proteins have been found to be involved in the progression of numerous tumors. In addition, the protein expression of S100 family members has been used to facilitate diagnosis, predict prognosis, determine treatment options and monitor therapeutic efficacy [
2,
3].
The expression and function of S100 protein vary across different malignant tumors or tumor subtypes. For instance, in oral carcinoma, S100A2 acts as a tumor-suppressor, while tumor-promoter in lung carcinoma [
4,
5]. In ovarian cancer, many S100 family members have been reported, such as S100A1, S100A4, S100A6, S100B and S100P. DeRycke MS’s report indicated that S100A1 mRNA and protein expression elevated with increasing Silverberg grade but not stage in serous ovarian cancer, while in endometrial subtype of ovarian cancer, there was no correlation between S100A1 expression and clinical stage or grade, but it was related with decreased relapse-free survival (RFS) [
6]. Serum S100A6 concentration has been found that it predicts peritoneal tumor burden and is associated with advanced stage in ovarian cancer [
7]. S100A4, S100B, S100A14, S100P expressions are associated with poor prognosis [
8‐
12], moreover,high cytoplasmic S100A10 staining is significantly associated with reduced overall survival (OS) and progression-free survival (PFS) in serous ovarian cancer [
13,
14]. Whereas, some other S100 family members, including S100A3, S100A5, S100A7, S100A7A, S100A8, S100A16 and S100G are less reported in ovarian carcinoma. The prognosis value of each S100 member, particularly at the transcriptional level in ovarian cancer patients needs further to be explored. In this report, we evaluated the prognosis value of each S100 member mRNA level in ovarian cancer patients by Kaplan Meier plotter platform.
Discussion
Until now, a vast majority of previous researches has focused on overall survival of ovarian cancer patients, and transcriptome analysis has been used to hunt for molecular signatures correlated with prognosis of ovarian cancer patients. However, the small sample sizes in some researches raised doubts about creditability and generalizability of the results. In our work, we identified the relationship between the prognosis and transcriptional level of the S100 family members in a large scale of patients by use of Kaplan Meier plotter database.
Variation in S100A1 expression have been reported in various diseases, including malignant tumor, heart failure, diabetes, ischemia, and chronic pulmonary hypertension [
16,
17]. It has been reported that S100A1 expression was significantly related with favorable OS in breast cancer patients [
3], which is in accordance with all ovarian cancer patients, regardless of specific subtypes. However, DeRycke M. S et al. have reported that S100A1 expression was a prognostic indicator of relapse-free survival in endometrioid subtypes of ovarian cancer [
6]. In addition, although we found no significant OS prognostic value of S100A1 expression, it signifies poor PFS (progressive free survival) for endometrioid ovarian cancer patients (Additional file
5: Figure S1). Thus, we suppose that S100A1 may play a more important role in disease relapse or progression in endometrioid subtypes of ovarian cancer patients. The specific mechanisms behind this situation need further investigation.
S100A2 has been identified as a markedly downregulated gene in tumor-derived mammary epithelial cell lines and was assumed as a tumor suppressor gene [
18]. Meanwhile,a down-regulation of S100A2 can be verified in many human cancers and certified as a prognostic marker, such as melanoma, prostate and breast cancer [
19‐
22]. On the other hand, some other type of tumors, including pancreas, colorectal, non-small-cell lung carcinoma (NSCLC) exhibit an upregulation of S100A2, which are associated with poor patient survival [
5,
23,
24]. The role of S100A2 in ovarian cancer has rarely been explored. Hough, C. D., et al. have shown the expression of S100A2 is elevated in ovarian cancer [
25]. In accordance, in our study, for all ovarian cancer patients, S100A2 mRNA expression was correlated with poor prognosis. However, for wild-TP53-type ovarian cancer patients, the higher level of S100A2 predicted better OS, but showed no prognostic value for mutant-p53-type ovarian cancer patients. Buckley NE has reported that S100A2 modulated binding of mutant TP53 to HSP90, and loss of S100A2 leaded to an HSP90-dependent stabilization of mutant TP53 [
26]. It was also reported S100A2 protein co-localized with the tumor suppressor wild type TP53, and increased TP53 transcription activity through a Ca
2+-dependent interaction [
27]. We guess that the interaction between S100A2 and TP53 may mediate the tumor suppressor effect of S100A2 in wild-TP53-type ovarian cancer patients. As to its poor prognosis for all patients, Buckley NE also demonstrated that, as downstream of the BRCA1/ΔNp63 signaling axis, S100A2 played an important role in regulating transcriptional responses and enhancing growth control mechanisms through destabilization of mutant TP53 [
26]. Thus, we speculated that, as the downstream of the BRCA1/ΔNp63 signaling axis, S100A2 might quite possibly be upregulated in modulating transcriptional responses in those patients with genomic instability, including BRCA1 or TP53 mutation, but not limited to those, who had worse survival than those without genomic instability.
S100A3 is a matricellular protein, which has been identified in various tissues, such as diaphragm, heart, liver, placenta, skeletal muscle and so on [
28,
29]. S100A3 promotes tumor progression in a variety of tumors. In colorectal cancer, the high level of S100A3 were related to the tumor occurrence and progression [
30]. In castration-resistant prostate cancer cells, S100A3 inhibition reduced invasion and tumor growth [
31]. It was also reported that S100A3 was relatively highly expressed in poorly differentiated and advanced gastric cancer [
28]. While, in breast cancer, the loss of S100A3 expression was associated with malignant development [
29], which was in accordance with its favorable prognosis in ovarian cancer in our study. Up to now, the prognostic role of S100A3 in ovarian cancer has not been reported.
S100A5 is a novel member of S100 protein family which is characterized by its interaction with Ca
2+, Zn
2+, and Cu
[2+.[32] It was detected that S100A5 expressed in very restricted region of adult brain [
32,
33]. Hancq, S et al. showed an inverse relationship between the S100A5 expression and the risk of recurrence of totally resected WHO grade I meningiomas [
34]. In consistent with this result, this study indicated that S100A5 predicted favorable OS in all EOC patients.
As a member of S100 protein family, S100A6 is highly expressed in epithelial cells, fibroblasts and in several types of tumor cells, which can regulate the cellular function of proliferation, apoptosis, cytoskeleton dynamics, tumorigenesis [
35,
36]. In several human carcinomas, the expression of S100A6 appears to be associated with aggressive disease and high levels of S100A6 was an independent marker for poor prognostic, such as colorectal, gastric, thyroid, hepatocellular carcinoma, and so on [
37‐
40]. Serum S100A6 concentration has been found that it predicts peritoneal tumor burden and is associated with advanced stage in ovarian cancer [
7]. Nevertheless, in our study, high expression of S100A6 predicted favorable OS in ovarian cancer patients, especially in serous type ovarian cancer. Interestingly, we observed a positive relationship between S100A6 mRNA level and OS in stage II patients, but negative in stage IV patients. Thus, we suggest that S100A6 may play different roles in early and late stage ovarian cancer patients. It also reported that S100A6 could interact with p53 and affect its biological activity [
41]. While, S100A6 expression did not show any correlation with overall survival in patients with wild or mutant TP53 status.
S100A8 and S100A9 originally existed as homodimers similar with many other S100 family members, and preferentially expressed in myeloid cells, such as granulocytes, monocytes, osteoclasts, and so on [
42‐
44]. It has been longer recognized the relationship between the S100A8/S100A9 expression and inflammatory disorders, including rheumatoid arthritis, inflammatory bowel disease, multiple sclerosis [
42,
45,
46]. As is well known, acute and chronic inflammation has been confirmed to increase the risk of tumorigenesis. Strong up-regulation of S100A8/S100A9 has also been detected in various human cancers and promotes tumors progression, including gastric, colon, breast, and live cancer [
46‐
50]. While, several other studies have indicated that extracellular S100A8/S100A9 exhibits powerful anti-tumor responses by promoting cytotoxicity and apoptosis [
51]. Until now, little was known about the relationship between the S100A8/S100A9 and prognosis in ovarian cancer. In our research, the higher expression of S100A8 predicted favorable OS in serous, as well as grade III and stage III + IV ovarian cancer patients. However, higher expression of S100A9 was not correlated with prognostic value in ovarian cancer patients.
S100A10 is known to form (annexinA2)
2-(S100A10)
2 heterotetramer (AIIt), which results in the translocation of S100A10 to the plasma membrane [
52]. Extracellularly, the S100A10/annexin II complex functions as a plasminogen receptor, and regulates the tissue plasminogen activator (tPA)-dependent plasminogen activation and the plasmin formation on cancer cells. Intracellularly, S100A10 implicates in various cell processes, including cell growth, cell cycle, transcription, and cell differentiation [
53].
Furthermore, these processes mentioned above were thought to be associated with tumor invasion, metastasis and angiogenesis [
53,
54]. In addition, S100 plays a role in the regulation of cell proliferation by negatively affecting BAD-induced apoptosis. In colorectal cancer cells, S100A10 expression are associated with resistance to oxaliplatin (L-OHP) [
55]. It has been also reported that S100A10 expression contributed to the aggressive characteristics of thyroid anaplastic carcinoma [
56]. Recently, Lokman, N. A., et al. reported high S100A10 mRNA levels powerfully predicted poor outcome in serous ovarian cancer [
13]. Here, our research also supported that the high level of S100A10 may indicated reduced OS in serous, grade III, or stage III + IV ovarian cancer patients.
S100A13 is a novel member of S100 family that characterized by its specificity in various form of cancer. Previously, S100A13 has been indicated to be a powerfully angiogenic biomarker for melanoma and astrocytic gliomas [
57‐
59]. Recently, it has been reported that S100A13 expression contributes to more aggressive invasive phenotype in lung cancer cells [
60]. S100A13 also influences chemoresistance by regulating the secretion of FGF1 and IL1A, or functions in several key signaling pathways such as cytokine and NF-kB signaling. Contrary to these functions, our research showed that upregulated S100A13 was associated better OS in all ovarian cancer, grade II, stage I + II, and mutant-p53-type ovarian cancer patients.
S100A16 was a recent addition to the S100 family that isolated from astrocytoma [
61]. The expression of S100A16 was upregulated in various tumors, such as bladder, lung, thyroid gland, pancreas, and so on [
62,
63]. In lung adenocarcinoma, S100A16 expression was associated with vessel invasion and poor outcome and confirmed as an independent prognostic marker [
62]. Similarity,higher expression of S100A16 predicted worse OS in breast cancer by promoting epithelial-mesenchymal transition [
64,
65]. However, high membrane S100A16 predicted better OS in colorectal cancer as well as oral squamous cell carcinoma [
66,
67]. Our results indicated that higher level of S100A16 predicted worse OS in ovarian cancer, especially in grade II, grade III, and stage III EOC patients.
S100P was first discovered in placental tissue. Accumulating evidences have suggested a link between S100P expression and progression of various types of cancers [
68‐
72]. In previous studies, statistical analysis has shown S100P positively correlated with serum level of CA125, worse survival and progressive disease in EOC patients [
8,
73]. While, Umezaki, Y., et al. reported that S100P low expression was associated with bad prognosis in clear cell adenocarcinoma of the ovary [
74]. In our study, we find that high expression of S100P predicted worse OS in grade III, stage I + II and mutant-p53-type of ovarian cancer patients.
As intracellular Ca2+ sensors and as extracellular factors, this family of proteins modulates a wide spectrum of other cellular functions in the normal or non-cancerous conditions. We had reviewed some recently published literatures, and found that S100A12 has pro-inflammatory properties that are likely to be stable in an oxidative environment. Conversely, S100A8 and S100A9 may have important protective mechanisms in inflammation [
75]. In endometrial stromal cells, S100A6 over-expression promoted beta-catenin expression at the RNA and protein levels [
76]. Additionally, the expression of RAGE and EN-RAGE, the downstream effectors of S100A12, was significantly increased, as evidenced by the significantly greater mRNA and protein expression in the cells of the endometriosis patients [
77]. As to hepatic or renal illnesses and diabetes, S100A4 promotes liver fibrosis by activating HSCs (hepatic stellate cell) [
78]; serum concentrations of S100B are higher in patients with cirrhosis than in healthy volunteers; silencing S100A9 clearly alleviated G-MDSCs (Granuloid medullary sources inhibit cells) expansion and hepatic steatosis caused by ATF3 deficiency or GC (glucocorticoid) treatment [
79]; the sunitinib may protect against renal damage from diabetes mellitus through regulating the levels of vimentin, E-cadherin and S100 [
80]; there is coexistence of increased expression of the S100B and the type 2 diabetes mellitus gene in patients [
81]; patients with type 2 diabetes have reticulated thrombocytosis that correlates with glycated hemoglobin as well as increased plasma S100A8/A9 levels [
82]; S100A8/A9 regulates renal damage and fibrosis, probably through loss of tubular epithelial cell contacts and irreversible damage [
83]; higher urine S100 levels are associated with increased lupus nephritis activity in childhood-onset systemic lupus erythematosus, whereas serum S100 (S100A4, S100A6, S100A8/9, and S100A12) levels do not correlate with disease activity [
84]; high S100A12 levels are associated with the presence and severity of coronary artery disease in patients with T2DM (type 2 diabetes mellitus) [
85].
Our research has some limitations. Firstly, all clinical data were collected by Kaplan-Meier database where the Kaplan-Meier survival curves and the log-rank
P value were fully analyzed, and the multivariate analysis by Logistical or COX regression could not be achieved for the reason of data limitation. Thus, the survival according to expression level of S100 family members might be cross- influenced by variable factors such as grade, stage, surgery and so on. Secondly, as for grade, secondary classification system about ovarian serous cancer, namely the low-grade serous carcinoma (LGSC) and high-grade serous carcinoma (HGSC), were recommended by 2014 WHO for the superiority that can be simple and easy to use, high repeatability, better able to guide the clinical treatment. While, the Kaplan-Meier database still used 2003 edition of the WHO classification method, namely, Silverberg grading system, which were divided into tumor Grade 1 (high differentiation), Grade 2 (moderate differentiation), Grade 3 (poorly differentiation). Thirdly, we can only obtain the differential expression information of each S100 family member between normal and ovarian cancer tissues, which provides clues to enhance the points indirectly. (Additional file
1: Table S6).