Involvement of Membranous S100A10 Expression in the Tumor Budding of Colorectal Cancer: An Immunohistochemical Study.

Introduction
Colorectal cancer (CRC) is the second leading cause of cancer death worldwide [1]. In CRC, tumor budding (TB) is a histologic prognostic parameter independent of the conventional tumor, node, and metastasis (TNM) staging system [2]. In fact, studies have shown that CRCs with prominent TB have a significantly higher incidence of metastasis and vessel invasion and a poorer prognosis [2–4]. Therefore, TB may influence therapeutic management [4]. TB is defined as a single tumor cell or a cell cluster of up to four tumor cells, and it has been mainly observed at the invasive front (IF) of a cancer [5]. Recently, a three-dimensional analysis of TB has shown that the majority of the tumor buds are not individual islands but are cord-like branches or protrusions that continue from the origin tumor mass [6, 7]. TB, a unique form of invasion, has been considered one type of the epithelial–mesenchymal transition (EMT) at the leading edge of collective invasion [6–9]. The mechanisms by which EMT occurs remain exceedingly complex [10, 11]. Therefore, further discovery of reliable molecular markers is needed to elucidate the mechanisms and to establish effective therapeutic strategies for EMT [10–12]. We believe that a better understanding of the proteins involved in TB will help suppress the aggressiveness of CRC.
We had previously reported that S100A10 may be involved in TB based on two patients with CRC [13]. S100A10 is one of the 25 members of the S100 protein family [14] and is involved in various cellular processes, such as differentiation and motility/invasion [15]. S100A10 forms a heterocomplex with most of the annexin A2 (ANX A2) in the cytoplasm and translocates to the cell membrane in response to various stimuli [14, 16]. The S100A10–ANX A2 heterocomplex functions as a plasminogen receptor [14, 16] in cooperation with RAS, including KRAS [17]. Furthermore, the heterocomplex can contribute to the degradation of the extracellular matrix (ECM) and promote cancer cell migration/invasion and metastasis [14, 16, 17]. Membranous S100A10 has been considered an activation modulator of the heterocomplex [14, 16]. Moreover, it regulates the remodeling of cytoskeletal actin, which has been associated with cell migration, and the morphological changes of cells [18, 19]. We had previously reported that in breast cancer, overexpression of membranous S100A10 was more frequently observed in thin trabecular components [20]. Recent studies have also suggested the involvement of S100A10 in EMT [21, 22]. Therefore, the current study immunohistochemically examined S100A10 expression in CRC tumor buds and its relationship with the various pathological parameters associated with worse prognosis.

Methods

Study Design
This study was designed as an exploratory analysis to evaluate the fundamental association between S100A10 and TB in CRC. Tissue sections from low-grade (well and moderately differentiated) adenocarcinomas not otherwise specified (LGAC-NOS) [23] were used given that adenocarcinoma (AC) of this type was most prevalent [23]. TB was assessed at the IF of the tumors using the hotspot method [5] recommended by the International Tumor Budding Consensus Conference. Immunohistochemical analysis of S100A10 expression focused on both the tumor buds at the hotspot and tumor glands (TGs) serving as the direct background for TB. S100A10 expression at the stroma-facing membranes (SFMs) of constitutive tumor cells was evaluated, given that previous studies have demonstrated a direct link between the SFM of cancer cells and stromal invasion [10, 11, 14, 16–19, 24]. Thereafter, the relationship between S100A10 immunoreactivity and various pathological parameters was investigated.

Samples
Tissue samples were obtained from 339 patients with LGAC-NOS (mean diameter: 4.3 cm; range, 1.5–13.5) who had not received preoperative therapy and from 992 patients with primary CRC who underwent surgical resection at Shizuoka General Hospital between January 2001 and December 2020. Among the 339 patients, 209 and 130 were men and women, respectively, with a mean age of 68.3 years (range, 38–92 years). This study was approved by the Ethics Committee of Shizuoka General Hospital (approval number: SGHIRB# 2018091/4). All tumor samples were fixed in 20% neutral-buffered formalin, and paraffin-embedded tissue sections were routinely stained with hematoxylin and eosin (H&E). The histological type was classified according to the criteria of the WHO [23]. High-grade (poorly differentiated) ACs and special subtypes of AC [23], such as micropapillary AC and mucinous carcinoma, were excluded.

TB Assessment
TB was evaluated on H&E-stained slides (3–13 slides, average 6 slides per patient). An IF containing high amounts of inflammatory cell infiltrate was excluded from the areas evaluated given that TG fragments from inflammatory destruction can obscure the identification of tumor buds [5]. We selected one tissue block that contained the TB hotspot per patient and performed new serial sections. One tissue section was then stained with H&E, after which the maximum number of tumor buds per patient was counted. Given our use of a ×20 objective lens with a field area of 1.227 mm2, the tumor bud counts were normalized to the recommended field of 0.785 mm2 [5]. Using a two-tier system, we classified the TB for each patient into low grade (< 10 buds) and high grade (≥ 10 buds) based on the tumor bud count.

Immunohistochemistry

Immunostaining
Immunohistochemistry for S100A10 and pan-cytokeratin (pan-CK), AE1/AE3, was performed with the above serial tissue sections using Leica Bond-Max (Leica Biosystems, Melbourne, Victoria, Australia). The characterization and staining protocol for the S100A10 antibody have been described elsewhere [13, 20, 25], including the staining protocol for the AE1/AE3 antibody [13]. Although pan-CK immunostaining can also detect non-tumor buds, including cellular debris [5], tumor buds have better visibility with this stain [5, 13].

Evaluation of S100A10 Immunostaining
Semiquantitative assessment was performed in a single field under ×20 objective lens. Unlike TB assessment, no correction of the field of view area was performed. Immunoreactivity at the SFMs was evaluated regardless of whether they were found at the intercellular borders, although samples with faint membranous positivity were excluded.

Tumor Buds
The percentage of S100A10-positive tumor buds at the TB hotspot was evaluated. Considering that tumor buds are composed of a fairly small number of tumor cells, we classified the tumor bud as S100A10-positive even if one of the component cells showed positivity. In samples with a small number of tumor buds, even the immunoreactivity of a single tumor bud had a significant impact. To minimize this problem, we evaluated a 50% cutoff value, which we and other researchers often employ as an indicator of high expression [20, 26, 27]. Thus, immunopositivity rate was classified into two broad categories: low score, < 50% positive tumor buds (including negative cases); high score, ≥ 50% positive tumor buds. If more than two hotspots were present in a tissue section, the field with the highest S100A10-positive percentage was selected.

Background TGs
The maximum number of positive tumor cells per 100 tumor cells facing the stroma was semiquantitatively calculated in the area adjacent to the TB hotspot. Immunopositivity was evaluated with the cutoff value of 10%, which we and other researchers often employ as an indicator of significant expression [20, 24, 26, 27]. Thus, the counts were classified into two categories: low positivity, < 10 positive tumor cells (including negative cases) and high positivity, ≥ 10 positive tumor cells. Even in tumors without TB, the number of S100A10-positive tumor cells was similarly evaluated at the IF.

Statistical Analysis
All categorical data were analyzed using chi-squared test and Fisher’s exact test. Data analysis was performed using js-STAR XR+ (version 2.1.3) (https://www.kisnet.or.jp/nappa/software/star8/index.htm) and R (version 4.3.1) (https://www.R-project.org/). If necessary, P values were adjusted using Bonferroni correction. In all analyses, a P value < 0.05 indicated statistical significance.

Results

TB Grading
Among the 339 patients with LGAC-NOS, 190 (56.0%) were confirmed to have TB. The maximum number of tumor buds was 21 per field. Moreover, among the 339 patients analyzed, 281 had (82.9%) had low-grade TB, consisting of 132 cases with TB and 149 cases without TB, whereas 58 (17.1%) had high-grade TB (Table 1).

Immunohistochemical Evaluation of S100A10 at the SFMs

Tumor Buds
Among the 190 patients with TB, 153 (80.5%) exhibited S100A10 positivity (Figs. 1, 2, and 3), including all high-grade TB patients. S100A10 immunoreactivity was heterogenous, with variations in the staining intensity of individual tumor buds (Figs. 1, 2, and 3). Among the 153 patients, 120 (78.4%) were classified as having a high score, including 47 (81.0%) of the 58 patients with high-grade TB (Table 2; Figs. 1 and 2). Among the 132 patients with low-grade TB, 59 (44.7%) were classified as having a low score (Table 2; Fig. 3). All 37 patients who tested negative for immunoreactivity had low-grade TB.

Background TGs
Among the 339 patients, 211 (62.2%) showed S100A10 positivity, including 165 (86.8%) of 190 patients with TB. Among the 211 patients who exhibited S100A10 positivity, 102 (48.3%) were classified as having high positivity (Table 3; Figs. 1 and 2). Among the 237 patients classified as having low positivity, 227 (96%) were had low-grade TB (Table 3; Fig. 3). Furthermore, among the 149 patients without TB, 103 (69.1%) were negative for S100A10 (Fig. 4). S100A10 positivity was mainly observed in the tumor cells with an irregular border with the stroma, especially those with cord-like or nest-like protrusions toward the stroma (Figs. 1, 2, and 3). In contrast, TGs without conspicuous protrusions into the stroma (Fig. 2) tested negative for S100A10, and even TGs with TB were S100A10-negative in portions where the border with the stroma was smooth (Fig. 3), which was more noticeable in well-differentiated ACs. Non-inflammatory surface TGs without TB were also negative for S100A10, including normal crypts.

Relationship Between TB Grade and Pathological Parameters
Correlations between TB grade and pathological parameters are presented in Table 1. Accordingly, TB grade was significantly correlated with advanced pT categories (P = 0.039), node metastasis (P < 0.001), poor pStage (P < 0.001), and lymphatic permeation (P = 0.046) using Fisher’s exact test but not with vascular invasion (P = 0.29, Fisher’s exact test) and tumor location (P = 1.00, chi-squared test after Bonferroni correction).

Relationship Between S100A10 Immunoreactivity at the SFMs and Pathological Parameters

Tumor Buds
Correlations between S100A10 immunoreactivity and pathological parameters are presented in Table 2. Notably, S100A10 immunoreactivity was significantly correlated with high-grade TB (P < 0.001) and background TGs with high positivity for S100A10 (P < 0.001) using Fisher’s exact test but not with advanced pT categories, node metastasis, poor pStage, vascular invasion, lymphatic permeation (P = 0.27, P = 0.76, P = 0.76, P = 0.75, P = 1.00, respectively, Fisher’s exact test), and tumor location (P = 1.00, chi-squared test after Bonferroni correction).

Background TGs
Correlations between S100A10 immunoreactivity and pathological parameters are presented in Table 3. Accordingly, background TGs were significantly correlated with high-grade TB (P < 0.001), node metastasis (P = 0.005), poor pStage (P = 0.005), and lymphatic permeation (P = 0.016) using Fisher’s exact test but not with advanced pT categories, vascular invasion (P = 0.89 and P = 0.90, respectively, Fisher’s exact test), and tumor location (P = 1.00, chi-squared test after Bonferroni correction).

Discussion
Consistent with previous research, the current study found that TB grade was correlated with advanced TNM stage [2–4]. Notably, we found that S100A10 expression at the SFMs of background TGs was significantly more common in high-grade TB cases and was correlated with advanced TNM stage. Furthermore, S100A10 expression at the SFMs of TGs was localized to the tumor cell population that forms tumor buds, similar to our previous study [13]. S100A10 at the SFMs has been suggested to promote the formation of tumor buds from the earliest stage of TB. High-grade TB reflects the complexity of branching or elongated protrusions from the origin tumor mass [6, 7, 9, 28]. Therefore, S100A10 expression is also considered to be maintained during the branching and elongation of protrusions. The increase in the number of S100A10-expressing tumor cells in the background TGs may induce more adverse biological behavior in CRC.
Cancer cells have been known to undergo EMT at the leading edge of collective invasion [10, 11]. EMT involves the activation of several signal transduction pathways, such as RAS and TGF-β, in cancer cells [10, 11]. Current evidence suggests that the EMT process is not binary, albeit uneven, but rather gradient and plastic [10, 11, 29, 30]. The hybrid EMT status reflects the plasticity and heterogeneity of cancer cells undergoing the EMT process [29, 30]. Hybrid EMT has been noted to play a critical role in tumor aggressiveness, migration/invasion, metastasis, and therapeutic resistance [29, 30]. Moreover, changes in cell function have a profound effect on cell morphology and motility [28, 29]. TB has also been regarded as one manifestation of hybrid EMT [31].
S100A10 is suggested to be a key regulator of the plasminogen activation system during TGF-β-induced EMT [21]. Recent studies on hepatocellular carcinoma have also suggested that S100A10 promotes EMT and metastasis [22]. Another study showed that S100A10 expression is upregulated by KRAS activation [17], and the KRAS mutation has been known to be associated with a high frequency of TB [32]. The mentioned studies [17, 32] support the association between TB and S100A10 expression. We did not demonstrate a direct association between S100A10 and EMT. However, a relationship between the two could be expected, through the association with TB [6–11, 29–32]. S100A10 expression at the SFMs was heterogeneous among tumor buds. We speculate that this finding reflects the diversity of tumor cells under a hybrid EMT state [29, 30]. In TGs, the portions with a smooth border around the stroma showed no S100A10 expression, suggesting that epithelial differentiation of tumor cells was robustly maintained [10, 11, 29, 30].
S100A10 expression has also been noted in many other cancers [33]. However, the relationship between S100A10 and TB has not yet been investigated. Aberrant S100A10 expression correlates with poor prognosis not only in CRC but also in lung cancer, gastric cancer, ovarian cancer, and adult malignant glioma [24, 26, 33, 34]. In hepatocellular carcinoma and ovarian cancer, S100A10 has also been reported to enhance chemotherapy resistance [22, 26]. These reports are interesting given that chemotherapy resistance is deeply associated with EMT [29, 30].
ANX A2 has also been suggested to be associated with the EMT of CRC and with TB [35, 36]. However, the cited studies have shown a discrepancy in the cellular localization of ANX A2, with neither study mentioning the presence or absence of S100A10 effects [35, 36]. In our previous study, we found that the immunolocalization of ANX A2 was similar to that of S100A10 [13]. Further investigations are therefore needed to determine whether such ANX A2 functions are independent of S100A10.
The present study has several limitations worth noting. First, this study employed a retrospective design. Second, no matching or regression to adjust for potential confounders and mediators in between-group characteristics (e.g., missing results for the presence of KRAS mutation in many cases) was performed, which may limit the generalizability of our results. Third, we did not perform immunohistochemical examination of S100A10-expressing tumor buds with known EMT markers. Fourth, we did not perform multivariate analysis. Furthermore, we were unable to analyze prognostic data due to its insufficiency. Although under these limited conditions, we believe that our study provides important implications for the association between S100A10 and TB. However, whether S100A10 can serve as an independent determinant requires future studies involving large cohorts. Fifth, although this study also observed S100A10 positivity at the intercellular borders (Figs. 2 and 3), especially in the TB of well-differentiated ACs, it did not investigate its association with TB. In addition, poorly differentiated clusters, which could be a possible histological finding in a sequence of TB [37], will be the subject of further examination.

Conclusions
S100A10 expression at the SFMs of TGs likely promotes TB from its earliest stage and remains active during TB. Moreover, the increase in the number of S100A10-expressing tumor cells confers worse biological behavior in CRC. Overall, our findings suggest that S100A10 could be a potential biomarker for TB.