1 Introduction

Assessment of several cancer-associated signature factors can provide critical and more accurate diagnostic and prognostic information than reliance on a single biomarker. Two such pathologically relevant markers are the cell cycle-related protein Septin9 (SEPT9) and the serine protease inhibitor SERPINE1, also known as plasminogen activator inhibitor-1 (PAI-1). SEPT9 is a scaffold protein that interacts with cytoskeletal proteins and plasma membranes [1, 2] and is involved in various fundamental cellular processes, including cytokinesis, vesicle trafficking, adhesion, and motility [3, 4]. Abnormal levels of SEPT9 mRNA or protein are evident across multiple cancers [5, 6], namely breast [7, 8], ovarian [9, 10], colorectal [11], and hepatocellular carcinoma (HCC) [12]. PAI-1 was initially identified as an inhibitor of urokinase-type plasminogen activator (uPA) [13]. It has been demonstrated that PAI-1 is not merely a fibrinolysis inhibitor but acts as a key driver in various human malignancies [14] by stimulating angiogenesis [15], inducing cell migration [16], and evading apoptosis [17].

Our previous immunohistochemistry (IHC) study illustrated that malignant hepatic lesions are more likely to be SEPT9 positive than benign or precursor hepatic lesions, and SEPT9 expression correlates with higher tumor grade in HCC [12]. These findings suggest a possible role of SEPT9 in HCC formation and progression. However, the mechanism by which SEPT9 promotes HCC is not fully understood. Similarly, PAI-1 and its substrate uPA are significantly increased in HCC compared to the adjacent uninvolved liver, and their expression correlates with advanced tumor grade, invasion, and metastasis [18]. Indeed, HCC patients with higher PAI-1 expression have worse overall survival (OS) and recurrence-free survival (RFS) [19]. Notably, recent studies also highlighted PAI-1 as a regulator of the tumor immune microenvironment [20,21,22]. This is particularly relevant since multiple immune cell lineages, including neutrophils and macrophages, express CXCR2 [23, 24]. CXCR2 ligands stimulate HCC cell proliferation and migration in vitro [25] and likely do so in the tumor microenvironment since CXCL8, a ligand for CXCR2, is upregulated as HCC develops in cirrhotic liver [26]. In murine models, small molecule inhibitors of CXCR2 or CXCR2 knockout diminished tumor growth and progression [27, 28]. Indeed, human studies confirmed that CXCR2+ cells are important contributors to HCC progression and bear prognostic significance [29, 30]. Collectively, these findings suggest a possible link between PAI-1 and CXCR2+ immune cells in the HCC immune microenvironment.

The study is the first to evaluate the relationship between SEPT9, PAI-1, and the tumor immune microenvironment in HCC. To this end, SEPT9, PAI-1, and immune marker expression in HCC and in the background benign liver were assessed by IHC. Special attention was paid to CXCR2+ immune cells in the HCC microenvironment. The potential prognostic implications of SEPT9 and PAI-1 in HCC were also evaluated.

2 Methods

This study was approved by the Institutional Review Board (IRB) at Albany Medical College (Albany, NY, USA) (protocol #5718, approval date: 06/18/2020) The IRB granted a waiver of informed patient consent, determining that the study qualified as secondary research for which informed patient consent is not required. All methods were carried out in accordance with relevant guidelines and regulations. The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

2.1 Study population

Archived partial hepatectomies (n = 76; 2003–2019) for HCC were retrieved from Albany Medical Center. Neither explants nor biopsies were included in our cohort. Fibrolamellar type HCC and combined hepatocellular-cholangiocarcinoma were excluded.

Electronic medical records and tumor registries were reviewed to obtain demographic features (age, sex), tumor characteristics (tumor size, AFP level at diagnosis, presence of multifocal hepatic lesions, T stage), HCC risk factors (family history of HCC, hepatitis B, hepatitis C, cirrhosis in background liver, past and current alcohol use, smoking history), metabolic dysfunction-associated liver disease (MASLD) risk factors (hypertension, dyslipidemia, diabetes, obesity), other medical conditions, and survival outcomes.

2.2 Histologic review

Archived hematoxylin & eosin (H&E)-stained slides were reviewed. Tumor grades were assigned per the 2019 WHO grading system [31]. For samples containing areas with different grades, the highest grade was assigned. For SEPT9 IHC, representative tissue blocks showing tumor-benign liver interface were selected for each case. For tissue microarray (TMA) construction, slides showing tumor and/or benign liver tissue were selected and the areas to be sampled were circled. Antibody clones, titers, and detection kits used in this study are summarized in Table 1.

Table 1 Antibody specification and detection kits
Full size table

2.3 Tissue microarray construction

TMAs (core diameter 1 mm, Galileo TMA CK3600 Computer Driven; Integrated Systems Engineering, Brugherio, Italy) were constructed from corresponding formalin-fixed paraffin-embedded (FFPE) tissue blocks. From each HCC tissue and benign liver, up to 10 and 5 cores were randomly selected, respectively. The constructed TMAs were used for PAI-1, CXCR2, CD3, CD15, CD68, and CD163 IHC.

2.4 SEPT9 immunohistochemical assays

5-micron thick whole sections representative of the tumor-benign liver interface were prepared for SEPT9 immunostaining. Native interlobular bile ducts served as positive internal controls. SEPT9 staining results were dichotomized into positive (≥ 5% of lesion staining) and negative (< 5% of lesion staining). In our previous study, SEPT9 staining was considered positive when > 5% of tumor cells exhibited distinct membranous/cytoplasmic staining with membranous accentuation compared with the benign liver [12]. In this study, any staining within the tumor, including weak cytoplasmic granular pattern, was considered positive if it spanned ≥ 5% of the tissue area.

2.5 PAI-1 immunohistochemical assays

5-micron thick TMA sections were prepared for PAI-1 immunostaining. Fibroblasts within desmoplasia were used as positive controls [32]. Any granular cytoplasmic and/or membranous staining within hepatocytes or tumor cells was considered positive.

2.6 Immune marker immunohistochemical assays

CXCR2, CD15, CD3, CD68, and CD163 IHC were performed using 5-micron thick sections from the TMAs. CXCR2 and CD15 staining were localized in the nucleus and/or cytoplasm of inflammatory cells [33]. The number of CXCR2+ and CD15+ inflammatory cells was counted at a hotspot in a ×400 high power field. CD3+, CD68+, and CD163+ inflammatory cells were considered uncountable since numerous cells were stained by the antibodies. Therefore, their staining extents, instead of cell counts, were assessed and scored as + 1, + 2, and + 3. The means of CXCR2+ and CD15+ cell counts in HCC and benign liver cores were selected as representative values. For CD3, CD68, and CD163 expression levels, the median staining extents in HCC and benign liver cores were used for the analyses.

2.7 Statistical analysis

Statistical analysis was performed using R version 4.3.3, with a p-value < 0.05 as a threshold for statistical significance. Continuous variables were reported as range (mean, median), and categorical variables were presented as frequencies and proportions (%). Normality and equality of variances for continuous variables were evaluated using Shapiro–Wilk test and Levene’s test. Depending on distribution, either T-test or Wilcoxon rank sum test was applied to identify their relationship with PAI-1 and SEPT9 IHC results. PAI-1/SEPT9 expression with respect to categorical variables were analyzed by Fisher’s exact test and if needed, post-hoc analysis by false discovery rate (FDR) adjusted p-value. Ordinal variables (grade and T stage) were further assessed by Cochran-Armitage trend test.

Survival analysis was conducted with Kaplan–Meier estimation and Cox proportional hazards model. Log-rank test was performed to evaluate Kaplan–Meier estimation. Simple Cox regression was performed for each variable, and those with p < 0.05 were then included in multiple Cox regression to identify independent prognostic factors. The statistical significance of the multiple Cox regression model was tested using the Wald test. Clinical features and selected IHC markers (CXCR2, CD15, SEPT9, and PAI-1) were analyzed as potential prognostic factors. The proportional hazard assumption was evaluated for each predictor following simple Cox regression, and variables that do not meet the assumption were excluded from the analysis.

3 Results

3.1 Study population

Demographics and clinical information of the cohort are outlined in Table 2. The mean age at HCC diagnosis was 65 (range 37–88) years; 48 (63.2%) patients were male. The mean tumor size was 6.4 (range 0.8–20) cm. AFP levels at diagnosis were available in 51 patients and AFP levels exceed 400 ng/ml in 10 (19.2%) of them. 9.9% (7) had more than one hepatic lesion. There were 17 (22.4%) grade 1, 38 (50%) grade 2, and 21 (27.6%) grade 3 cases. The mean follow-up was 61.2 (median: 46, range: 0–210) months.

Table 2 Demographics and clinicopathologic characteristics of the patients
Full size table

3.2 SEPT9 and PAI-1 are related to each other and associated with higher tumor grade

SEPT9 IHC results were available for 68 cases, with 34 (50%) being SEPT9(+) within the tumor. Representative images of SEPT9 IHC are shown in Fig. 1. 22 (29%) cases showed weak cytoplasmic granular staining in neighboring benign liver. SEPT9 staining in the benign region was strongly associated with SEPT9 positivity in HCC (p = 0.02). However, SEPT9 staining in the benign liver was not associated with any of the patient characteristics or immune marker expression (CD3, CD15, CD68, CD163, CXCR2, and PAI-1), thus staining in the benign liver tissue was disregarded. PAI-1 IHC results were available for all 76 cases, with 17 (22%) cases being PAI-1(+) within the tumor (Fig. 2). None of the adjacent benign livers was PAI-1 positive. SEPT9 and PAI-1 staining positivity were associated with each other (p = 0.02).

Fig. 1
figure 1

Hepatocellular carcinoma (R45) with SEPT9 positivity; a hematoxylin and eosin, ×50; b SEPT9, ×50

Full size image
Fig. 2
figure 2

Two tissue microarray cores of hepatocellular carcinoma (R4 and R49) with focal PAI-1 positivity; a, c hematoxylin and eosin, ×100; b, d, PAI-1, ×100

Full size image

Table 2 outlines clinical characteristics and their association with SEPT9 and PAI-1. SEPT9 and PAI-1 staining in HCC showed a significant difference according to tumor grade (SEPT9: p = 0.04, PAI-1: p < 0.01). Post-hoc analysis revealed that SEPT9 and PAI-1 expression were more frequently observed in grade 3 and grade 2 HCC patients compared to those with grade 1 (Table 2). Additionally, there was a significant trend indicating that patients with higher tumor grades were more likely to be SEPT9(+) (p = 0.02) and PAI-1(+) (p < 0.01). Other clinical or pathologic features were not significantly associated with SEPT9 and PAI-1 staining.

3.3 SEPT9 and PAI-1 positive HCC have distinct intratumoral CXCR2 + immune cell makeup

Next, we investigated whether SEPT9 or PAI-1 expression is related to CXCR2+ immune cell makeup in HCC. Specific CXCR2+ cell types, CD15 (neutrophils), CD68 (pan-macrophage), CD163 (M2 macrophage), and CD3 (T lymphocytes) were identified by IHC (See methods). Representative images for CD3, CD68, and CD163 IHC with staining extent scores are shown in Fig. 3. Results are summarized in Table 3. SEPT9(+) HCCs had a higher CXCR2+ (p < 0.01) and CD15+ (p = 0.02) cell count within the tumor. Intratumoral PAI-1 staining showed a positive correlation with CXCR2+ cell count (p = 0.01), CD3 expression (p = 0.04), CD15+ cell count (p < 0.01), CD68 expression (p = 0.03), and CD163 expression (p = 0.03) within the tumor.

Fig. 3
figure 3

Representative images of 1+ to 3+ staining for CD3 (left), CD68 (middle) and CD163 (right) immunohistochemistry on hepatocellular carcinoma tissue microarrays, ×100

Full size image
Table 3 PAI-1, SEPT9 and immune markers
Full size table

3.4 SEPT9 and PAI-1 expression in HCC and benign liver CXCR2+ immune cell makeup

Neither SEPT9 nor PAI-1 expression was significantly associated with CXCR2+ cell counts or specific CXCR2+ cell marker expression (CD3, CD15, CD68, and CD163) in the adjacent benign liver (Table 3, Online Resource 1). However, SEPT9(+) HCC had marginally increased CXCR2+ [p = 0.15, mean in SEPT9(−): 21.8 vs. mean in SEPT9(+): 31.4] and CD15+ [p = 0.08, mean in SEPT9(−): 42.3 vs. mean in SEPT9(+): 56.7] cell count in the uninvolved benign liver compared to SEPT9(−) HCC.

3.5 SEPT9 and PAI-1 influence overall survival in HCC patients only when the other is absent

Previous studies suggested that CXCR2+ and CD15+ cells have prognostic value in HCC [29, 30]. SEPT9, PAI-1, CXCR2, and CD15 IHC results were examined for possible associations with oncologic outcomes (Fig. 4, Table 4). Kaplan–Meier analysis revealed that SEPT9(+) HCC patients had shorter OS compared to SEPT9(−) HCC patients (p = 0.01) (Fig. 4a). No significant association was found between PAI-1 staining and OS (p = 0.07) (Fig. 4d). Interestingly, worse OS was associated with SEPT9 or PAI-1 expression only when the other marker was absent (Fig. 4b, c, e, f). Neither SEPT9 nor PAI-1 was related to RFS or 5-year survival rate.

Fig. 4
figure 4

SEPT9, PAI-1, and overall survival. a SEPT9 and overall survival, b SEPT9 and overall survival in PAI-1(−) patients, c SEPT9 and overall survival in PAI-1(+) patients. d PAI-1 and overall survival, e PAI-1 and overall survival in SEPT9(−) patients, f PAI-1 and overall survival in SEPT9(+) patients

Full size image
Table 4 Survival analysis
Full size table

Simple Cox regression for OS identified tumor size (p = 0.03), CD15+ cell counts in the benign liver (p < 0.01), and SEPT9 expression (p = 0.02) as significant prognostic factors. The multiple Cox regression with tumor size, benign liver CD15+ cells, and SEPT9 was performed. Multiple regression model was statistically significant (Wald test, p < 0.01). SEPT9 was not an independent predictor of worse prognosis (p = 0.29), while tumor size [adjusted HR = 1.13, 95% CI (1.03–1.24), p = 0.01] and benign CD15+ cell counts [adjusted HR = 1.02, 95% CI (1.00–1.03), p = 0.02] retained their significance.

4 Discussion

Previous studies demonstrated that SEPT9 interacts with various tumorigenic proteins, such as HIF-1alpha [34], JNK, cyclin D [7], and Rho [35]. In this study, we showed that PAI-1, another pro-tumorigenic protein, is associated with SEPT9 in HCC. Moreover, PAI-1 and SEPT9 protein expression correlated with higher HCC grade, consistent with findings from other cancers [12, 36,37,38].

Growing evidence has suggested that PAI-1 modulates the tumor immune microenvironment, especially in relation to CXCR2+ immune cells. PAI-1 induced PD-L1 expression, and a PAI-1 inhibitor increased the number of CD8+ T cells in the tumor microenvironment while decreasing that of regulatory T cells [21]. Gene expression array analysis demonstrated that PAI-1 is positively correlated with CD163 (M2 macrophage marker) in various cancers [20]. Further, PAI-1 stimulates macrophages to infiltrate into tumors and adopt a pro-tumoral M2 phenotype [20]. PAI-1-uPA dimer has a similar effect on neutrophils [22]. Importantly, CXCL8, a CXCR2 ligand, expression in colon cancer is induced by PAI-1 [39]. Our findings that PAI-1(+) HCCs had higher intratumoral CD3 (T lymphocytes), CD15 (neutrophils), CD68 (pan-macrophage), and CD163 (M2 macrophage) expression, are in line with the previous studies.

On the other hand, the role of SEPT9 in the tumor immune microenvironment is not fully understood, although several studies reported a potential link between SEPT9 and immunological processes [3, 40, 41]. Jiao et al., showed that monocytes incubated with conditioned media from irradiated, SEPT9-overexpressing HeLa cells differentiated into M2 macrophages [42]. Our results showed that SEPT9(+) HCCs have a higher number of intratumoral CXCR2+ and CD15+ cells compared to SEPT9(−) HCC. CXCR2+ and CD15+ cells were highlighted as important mediators of HCC tumorigenesis and progression [29, 30]. Thus, CXCR2+ cells in the tumor microenvironment might serve as a link between SEPT9 and HCC progression.

Interestingly, although patients with SEPT9(+) HCCs tend to also express SEPT9 in neighboring benign liver (p = 0.02), SEPT9 staining in benign regions was not associated with any of the immune markers (CXCR2, CD3, CD15, CD68, and CD163). It is possible that aberrant SEPT9 protein expression initiates from background benign liver tissue in a subset of SEPT9(+) HCCs. Alternatively, SEPT9 expression may be an early event in HCC tumorigenesis, extending to adjacent benign tissue as HCC progresses. Further studies are required to determine the significance of SEPT9 in benign tissue.

It is worth noting that SEPT9(+) and PAI-1(+) HCCs have distinct immune landscapes. Li et al., showed that neutrophils, macrophages, and T lymphocytes are the major subpopulations of CXCR2+ cells in HCC [29]. Intratumoral PAI-1 expression showed a positive correlation with CXCR2+ cell count and specific CXCR2+ cell marker (CD3, CD15, CD68, and CD168) expressions. However, only CXCR2+ and CD15+ cell counts were significantly associated with SEPT9. SEPT9(+) HCC did not exhibit higher CD3, CD68 and CD163 expression levels compared to SEPT9(−) HCC (Table 3). This implies SEPT9 IHC can identify HCCs with distinct intratumoral CXCR2+ immune cell composition, which in turn may harbor clinical significance.

In our HCC cohort, intratumoral SEPT9 expression was associated with shorter OS. Similarly, Stanbery et al., demonstrated that in head and neck cancer, higher SEPT9 IHC intensity correlated with worse oncologic outcomes [43]. However, SEPT9 was not an independent prognostic factor in multiple Cox regression, while CD15+ cell count in benign tissue, which was marginally associated with SEPT9 expression, continued to show significance. Previous studies identified peritumoral CD15+ cell count as an independent predictor for shorter OS and RFS in HCC patients [30], while the exact definition of ‘peritumoral’ was not available. Since our benign liver tissue cores were sampled adjacent to the HCC, the impact of SEPT9 expression on OS might be partly due to higher CD15+ cell count in ‘peritumoral’ benign tissue.

PAI-1 expression is an indicator of worse prognosis in various cancers, including breast [44], colorectal [45], gastric [46], ovarian [47], esophageal cancer [48], and HCC [19]. However, in our cohort, although patients with PAI-1(+) HCC tended to show shorter overall survival (p = 0.07), such an association was not statistically significant. Profound regional differences in HCC risk factors [49, 50] or small cohort size might be responsible for this discrepancy.

Considering that SEPT9 and PAI-1 expression are related, and their common association with CXCR2+ and CD15+ cell count within HCC, it is possible that they interact via modulating the HCC tumor microenvironment. Interestingly, despite their strong correlation, SEPT9 and PAI-1 affect overall survival only when the other is absent (Fig. 4). Their influences on survival outcomes and existence of a possible interaction should be investigated in future studies.

One limitation of our study is that we were unable to differentiate specific isoforms of SEPT9. SEPT9 undergoes complex alternative splicing, with each isoform having distinct effects on tumorigenesis [7, 11]. Additionally, we could not distinguish specific subtypes of CD3+ T cells and CD68+ macrophages, although M2 macrophage was identified by CD163 IHC. Distinct T cell subpopulations, such as CD4+, CD8+, and regulatory T cells, differentially influence HCC development and prognosis [51]. Likewise, M1 macrophages are known to have anti-tumor effects, while anti-inflammatory M2 macrophages promote tumor progression [52]. Kupffer cells, the resident macrophages of the liver, play a crucial role in hepatic fibrosis [53] and non-alcoholic fatty liver disease [54], both of which are well-established risk factors of HCC. Therefore, identifying specific T cell and macrophage subtypes is necessary for the comprehensive understanding of the HCC immune microenvironment. Finally, small cohort size might have led to limited statistical power, especially in subgroup analysis.

To the best of our knowledge, this is the first study to elucidate the relationship between SEPT9, PAI-1, and the HCC immune microenvironment. SEPT9(+) HCCs had a distinct immune microenvironment, with potential clinical implications. Further research is required to validate the existence and mechanism of PAI-1/SEPT9 interaction. Additionally, the potential clinical utility of SEPT9 IHC in HCC should be explored. Of note, small molecule inhibitors of CXCR2 and PAI-1 have been developed and are under clinical investigation [23, 32, 55]. In the future, SEPT9 IHC could provide additional insights regarding these novel therapeutics.