Associations of circulating mediators of inflammation, cell regulation and immune response with esophageal squamous cell carcinoma

Abstract

Background

Esophageal squamous cell carcinoma (ESCC) is the most common histologic subtype of esophageal cancer globally. The development of squamous cell carcinoma has important inflammatory influences and effects. We, therefore, examined circulating levels of inflammation- and immune-related proteins for associations with ESCC.

Methods

We used pre-treatment EDTA plasma from 80 ESCC patients (44% clinical stages I and II) and 80 cancer-free control individuals within the Hospital-based Epidemiologic Research Program at Aichi Cancer Center. Levels of 184 biomarkers were measured by high-throughput multiplexed proximity extension assays using Olink’s Proseek Cell Regulation and Immuno-Oncology Panels. ESCC odds ratios (OR) per quantile (based on two to four categories) of each biomarker were calculated by unconditional logistic regression models, adjusted for age, sex, cigarette smoking and alcohol consumption. Correlations among continuous biomarker levels were assessed by Spearman’s rank correlation. All statistical tests were two-sided with p values < 0.05 considered as significant. Given the exploratory nature of the study, we did not adjust for multiple comparisons.

Results

Seven proteins were undetectable in nearly all samples. Of the remaining 177 evaluable biomarkers, levels of cluster of differentiation 40 (CD40, per quartile OR 1.64; p trend = 0.018), syntaxin 16 (STX16, per quartile OR 1.63; p trend = 0.008), heme oxygenase 1 (per quartile OR 1.59; p trend = 0.014), and γ-secretase activating protein (GSAP, per quartile OR 1.47; p trend = 0.036) were significantly associated with ESCC. Amongst these significant markers, levels of CD40, STX16, and GSPA were moderately correlated (Rho coefficients 0.46–0.55; p < 0.05).

Conclusion

Our case–control study expands the range of inflammation and immune molecules associated with ESCC. These novel findings warrant replication and functional characterization.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00432-021-03687-3.

Introduction

Patients with esophageal cancer are frequently diagnosed at advanced clinical stages, making this malignancy the sixth deadliest cancer worldwide in 2020 (Sung et al. 2021). Esophageal squamous cell carcinoma (ESCC) is the most common histologic subtype of esophageal cancer, accounting for ~ 90% globally (Arnold et al. 2017). The causes of ESCC vary among geographic regions. Cigarette smoking and alcohol drinking are the primary known risk factors in non-endemic areas (Murphy et al. 2017) such as Japan, where the incidence of ESCC is increasing in both males and females (Arnold et al. 2017).

Chronic inflammation plays an important role in the initiation and development of squamous cell carcinomas. Smoking- and alcohol-induced inflammation and immune modulation are potentially important mechanisms in cancer development. Mucosal regeneration of the esophagus after injury involves a complex interaction among different immune cells, extracellular matrix, and growth/differentiation factors. Assuming that locally produced molecules could move from the esophageal tissue microenvironment to circulation, blood levels could be informative either as markers of the organ-specific activity or systemic effects, which may have etiologic and clinical implications. Previous studies of inflammation biomarkers in ESCC are limited. To gain a better understanding of the role of inflammation in ESCC, our study aimed to examine associations between circulating levels of inflammation- and immune-related proteins and ESCC in a Japanese case–control study. As a secondary aim, we evaluated associations with overall disease survival.

Methods

Study population

Participants in the Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC) study represent first-visit outpatients who were invited to provide lifestyle data and pretreatment blood samples. Among the participants recruited between January 2001 and December 2005 (HERPACC-II), 80 patients with ESCC (ICD-O-3 codes C15) and 80 cancer-free control individuals were selected for this analysis. In ESCC patients, vital status was collected from medical records. Survival time was calculated in months from the date of enrollment to death or the last follow-up in 2008.

Informed consent was obtained from all participants. The HERPACC study was approved by the institutional review board of the Ethics Committees of Aichi Cancer Center and Nagoya University.

Laboratory methods

EDTA plasma samples from ESCC cases, cancer-free controls and laboratory controls were tested for levels of 184 proteins (Supplementary Table 1) by multiplex proximity extension assays on two panels (Proseek Multiplex Cell Regulation and Immuno-Oncology Panels, Olink Bioscience, Uppsala, Sweden). Blinded testing was performed on a Fluidigm Biomark reader (Fluidigm Corporation, USA) at the facility of the kit manufacturer. The two assay panels include unique sets of 92 proteins related to apoptosis, cell activation, adhesion, communication, cycle, differentiation, proliferation, localization, metabolic progress, and chemotaxis, autophagy, promotion and suppression of tumor immunity, MAPK cascade, phosphorylation, proteolysis, regulation of gene expression, and vascular/tissue remodeling. Relative levels were calculated from quantitative PCR cycle threshold values with corrections for assay variation and expressed as normalized protein expression on a logarithmic scale. Seven biomarkers (IFN-beta, IFN-gamma, IL1-alpha, IL35, IL13, IL2 and IL33) were undetectable in > 90% of samples and were, therefore, excluded from statistical analysis. To assess reproducibility, we tested a common pooled sample on each plate. Intra- and inter-assay coefficients of variations of linear normalized values were < 10% for both panels.

Statistical analysis

Statistical calculations were based on relative marker levels transformed to a linear representation of protein concentration. Normalized levels of the 177 biomarkers detected in > 10% of samples were grouped into quantiles based on their distributions among controls. A priori, the measurements were categorized as (i) quartiles for biomarkers with < 25% levels below the lower limit of detection (LLOD; n = 153); (ii) tertiles for biomarkers with 25–50% levels below the LLOD (n = 7); (iii) undetectable, < median or ≥ median for biomarkers with 50–75% levels below the LLOD (n = 9); and (iv) undetectable or detectable for biomarkers with 75–90% levels below the LLOD (n = 8). For biomarkers with three or four categories, the quantiles were tested for linear trends as ordinal variables.

Unconditional logistic regression analysis was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the association of each biomarker quantile with ESCC, adjusted for age, sex, cigarette smoking status, and alcohol drinking status. Correlations among biomarker levels were evaluated by pairwise Spearman’s rank correlation. We also modeled the association between ESCC and the significant biomarkers by stepwise selection (slentry = 0.25 slstay = 0.15). A stratified analysis by clinical stage (I–II vs. III–VI) was performed for the significant markers. P-heterogeneity for testing the strata was calculated using Wald test. Cox proportional hazard regression analysis was used to calculate hazard ratios (HRs) and 95% CIs for the association of each biomarker quantile with ESCC survival, adjusted for clinical stage, age, sex, cigarette smoking status, and alcohol drinking status. We also modeled the associations between ESCC survival and the significant biomarkers by stepwise selection (slentry = 0.25 slstay = 0.15). For all tests, p values < 0.05 were considered statistically significant. We did not adjust for multiple comparisons because of the exploratory nature of the study. Statistical analyses were conducted using SAS version 9.4 software (SAS Inc, Cary, NC).

Results

Selected characteristics of ESCC cases and controls are presented in Table 1. Compared with the controls, cases were more likely to be male, smokers and alcohol drinkers. During a median of 2 years of follow-up, 42 patients (54%) with ESCC died.

Table 1.

Baseline selected characteristics of esophageal squamous cell carcinoma (ESCC) patients and cancer-free control individuals within the HERPACC Study

	Controls n = 80	ESCC cases n = 80	p value
Age in years, mean (SD)	59 (11)	62 (8)	0.06
Male, n (%)	66 (82.5)	75 (93.8)	0.05*
Cigarette smoking status, n (%)
Never	25 (31.3)	3 (3.8)	< 0.0001
Former	31 (38.8)	25 (31.3)
Current	24 (30.0)	52 (65.0)
Alcohol drinking status, n (%)
Never	26 (32.5)	2 (2.5)	< 0.0001
Former	2 (2.5)	7 (8.8)
Current	52 (65.0)	71 (88.8)
TNM staging, n (%)
Stage I		21 (26.3)
Stage II		14 (17.5)
Stage III		20 (25.0)
Stage IV		24 (30.0)
Unspecified		1 (1.3)

SD standard deviation

*p value calculated by Fisher exact test

Figure 1 (Cell Regulation panel) and Fig. 2 (Immuno-oncology panel) show the multivariable-adjusted associations of biomarker quantile with ESCC diagnosis. Levels of syntaxin 16 (STX16; per quartile OR 1.63; p trend = 0.008), heme oxygenase 1 (HO-1; 1.59; 0.01), cluster of differentiation 40 (CD40; 1.64; 0.02), and γ-secretase activating protein (GSAP; 1.47; 0.04) were associated with ESCC. Quartile-specific ORs for the above nominally significant markers are presented in Table 2. The ORs for ESCC associations with STX16 and CD40 were more than fivefold for quartile 4 vs. quartile 1. The marker association did not differ by early (I–II) vs. late (III–IV) clinical stage (data not shown).

Fig. 1.

Estimated odds ratios (OR) and 95% confidence intervals (CIs) of esophageal squamous cell carcinoma per quantile of biomarker levels measured in the Cell Regulation panel. Logistic regression models were adjusted for age (continuous), sex, cigarette smoking status (never, former, current) and alcohol drinking status (never, former, current). Biomarker levels were analyzed by quartiles (filled circles), three categories (undetectable, < median vs. ≥ median; half-filled circles) and two categories (undetectable vs. detectable levels; open circles). *Indicates p trend < 0.05

Fig. 2.

Estimated odds ratios (OR) and 95% confidence intervals (CIs) of esophageal squamous cell carcinoma per quantile of biomarker levels measured in the Immune-Oncology panel. Logistic regression models were adjusted for age (continuous), sex, cigarette smoking status (never, former, current) and alcohol drinking status (never, former, current). Biomarker levels were analyzed by quartiles (filled circles), three categories (undetectable, < median vs. ≥ median; half-filled circles) and two categories (undetectable vs. detectable levels; open circles). *Indicates ptrend < 0.05

Table 2.

Circulating immune and inflammation-related proteins with statistically significant trends for association with esophageal squamous cell carcinoma

Proteins (other names)	Quartile	Adjusted odds ratio (95% CI)
Syntaxin-16 (STX16)*	1	1.0
	2	2.49 (0.76–8.19)
	3	2.94 (0.87–10.01)
	4	5.05 (1.59–16.06)
	p trend	0.008
Heme oxygenase 1 (HO-1)**	1	1.0
	2	0.40 (0.10–1.54)
	3	3.52 (1.12–11.05)
	4	2.39 (0.74–7.72)
	p trend	0.014
Cluster of differentiation 40 (CD40, tumor necrosis factor receptor superfamily member 5)**	1	1.0
	2	2.71 (0.75–9.79)
	3	3.29 (0.92–11.75)
	4	5.54 (1.42–21.57)
	p trend	0.018
Gamma-secretase-activating protein (GSAP)*	1	1.0
	2	0.42 (0.13–1.37)
	3	1.30 (0.41–4.06)
	4	2.59 (0.82–8.21)
	p trend	0.036

CI confidence interval

*Biomarker included in the Immuno-oncology panel

**Biomarker included in the Cell Regulation panel

Spearman’s correlations among the four associated biomarkers ranged from − 0.10 to 0.55. Three pairwise correlations had statistically significant Rho coefficients between 0.46 and 0.55: CD40 vs. GSAP (Rho = 0.55; p < 0.0001), CD40 vs. STX16 (0.53; p < 0.0001), and STX16 vs. GSAP (0.46; p < 0.0001). Consistent with the high degree of correlation among these three biomarkers, CD40 and GSAP were dropped by stepwise regression, with the final model retaining only STX16 (per quartile OR 1.74; p trend = 0.02) and HO-1 (1.50; 0.05).

Supplementary Figs. 1 (Cell Regulation panel) and 2 (Immuno-oncology panel) show the multivariable-adjusted associations of biomarker quantile with overall ESCC survival. Higher levels of tumor-associated calcium signal transducer 2 (TACSTD2; per quartile HR, 1.44; p trend = 0.005), neural leucine-rich repeat 1 (LRRN1; 1.38; 0.03), SLIT and NTRK like family member 2 (SLITRK2; 1.38; 0.04), CCL17 (1.45; 0.04), and collagen type IV α1 (COL4A1; 1.40; 0.04) were associated with increased mortality. On the contrary, higher levels of glucagon (GCG; per quartile HR, 0.59; p trend = 0.0008), T cell surface glycoprotein CD4 (0.60; 0.002), TNF-related apoptosis-inducing ligand (TRAIL; 0.59; 0.003), placenta growth factor (PGF; 0.62; 0.005), CX3CL1 (0.64; 0.006), CD40 (0.59; 0.002), nitric oxide synthase (NOS3; 0.70; 0.02), vascular endothelial growth factor receptor-2 (VEGFR2; 0.67; 0.02), monocyte chemotactic protein 1 (MCP1; 0.72; 0.02), hepatocyte growth factor (HGF; 0.69; 0.02), CXCL9 (0.69; 0.03), T cell surface glycoprotein CD5 (0.73; 0.03), VEGFA (0.66; 0.03), CXCL13 (0.66; 0.03), macrophage colony-stimulating factor 1 (CSF1; 0.74; 0.03), interleukin-12 receptor subunit beta-1 (IL12RB1; 0.72; 0.04), adenosine deaminase (ADA; 0.73; 0.05) were associated with decreased mortality. The final stepwise model only included three biomarkers: TACSTD2 (per quartile HR, 1.66; p trend = 0.002), CX3CL1 (0.61; 0.004), and GCG (0.62; 0.01).

Discussion

Our case–control study has comprehensively evaluated a wide range of circulating mediators of inflammation, cell regulation and immune response for potential association with non-endemic ESCC. We found four novel proteins (STX16, CD40, HO-1 and GSAP) nominally associated with this carcinoma. Although smoking and alcohol cause inflammation of the esophageal mucosa, our study adjusted for these factors arguing against confounding. Associations did not vary by clinical stage. Circulating levels of CD40, STX16 and GSAP were moderately correlated, suggestive of a potential common pathway. Although underlying biologic mechanisms for our findings remain to be established as we could not measure alterations occurring locally in the tissue of the esophagus, this study adds to the understanding of the complex role of chronic inflammation in esophageal cancer.

Tissue expression of CD40 and HO-1 has been previously investigated in ESCC. CD40 is a costimulatory protein that is expressed on antigen-presenting cells. Its ligand, CD40L (CD154), is a member of the tumor necrosis factor superfamily that is primarily expressed on activated T cells. The CD40–CD40L pathway is essential in mediating core processes of the adaptative immune response, including T cell-dependent immunoglobulin class switching, memory B cell development, and germinal center formation (Law and Grewal 2009). CD40 expression is frequently shown to be associated with prolonged cancer survival. However, ESCC patients with CD40-positive tumors have been reported to have a more advanced pathological stage, poorer differentiation, higher frequency of positive lymph node metastasis, and a greater tumor extension (Matsumura et al. 2016).

HO-1, an enzyme that catalyzes the degradation of heme, has important antioxidant, anti-inflammatory, anti-apoptotic, anti-proliferative, and immunomodulatory effects in vascular cells (Araujo et al. 2012). HO-1 is expressed in ESCC tumors (Ren et al. 2016). In addition, knockdown of HO-1 promotes apoptosis through activation of a ROS-mediated caspase pathway in ESCC (Ren et al. 2016). Candidate gene studies of HO-1 promoter variants have associated long (GT)n repeats with increased risk of ESCC (Hu et al. 2010).

Little is known about the functions of STX16 and GSAP in carcinogenesis. STX16, a member of the soluble N-ethylmaleimide sensitive factor attachment protein receptor family, is involved in several membrane-trafficking activities, particularly in the transport processes at the trans-Golgi network. Recent studies suggest that STX16 appears to be a key regulator of cytokinesis (Neto et al. 2013) and autophagy (Tang 2019). Failure of these cellular processes has been linked to neoplastic transformation (Sagona and Stenmark 2010, Mathiassen et al. 2017). GSAP has been proposed to promote amyloid-β production (Chu et al. 2015). Accordingly, GSAP has mostly been studied in the context of Alzheimer’s disease pathology (Hussain et al. 2013). Interestingly, TIAF1, a gene associated with the progression of Alzheimer's disease, might function as a tumor suppressor in ESCC cell proliferation, invasion and metastasis (Xing and Liu 2017).

Previous investigations have identified increased levels of several immune- and inflammation-related proteins in ESCC cases as compared to controls (Ren et al. 2005a, b; Ren et al. 2005a, b; Kozlowski et al. 2013; Chu et al. 2020). In the current study, we found no significant associations with nine of 11 markers in the previous studies (VEGF-A, VEGF-C, VEGF-D, IL-6, IL-8, HGF, CX3CL1, CXCL12 and CCL20) and did not examine the remaining two (MIF and MMP1).

A few prospective studies have addressed associations between baseline levels of immune- and inflammation-related biomarkers and subsequent ESCC risk. Using Milliplex Luminex bead-based assays and the Proseek Multiplex Inflammation panel in the Japan Public Health Center-based Prospective Study (Camargo et al. 2019; Aversa et al. 2020), we previously reported modest ESCC associations (nominal p trends < 0.05) with levels of 22 (AXIN1, CASP8, CD5, CD6, CDCP1, CCL4, CCL15, CCL23, CRP, CSF1, CXCL1, CXCL5, CXCL6, CXCL11, FGF19, NT3, OPG, OSM, SLAMF1, ST1A1, STAMBP and TGFA) of the ~ 100 biomarkers evaluated. The current case–control study included 8 (CASP8, CD5, CSF1, CCL4, CCL23, CXCL1, CXCL5 and CXCL11) of these 22 candidate biomarkers and found none significantly associated with ESCC. In a smaller study from an endemic area of Iran, Keeley et al. (Keeley et al. 2014) reported modest-to-strong associations of ESCC with 17 of 21 inflammatory biomarkers that were measured. Our case–control study evaluated five (FGF-2, IL12p70, IL17α, MCP3 and TNFα) of these candidates and found no significant associations. Considering the complexity of the immune system-tumor interplay, further research is required to identify underlying mechanisms of ESCC carcinogenesis.

Regarding overall ESCC survival, we found five proteins (TACSTD2, LRRN1, SLITRK2, CCL17, and COL4A1) positivity associated and 17 (GCG, CD4, TRAIL, PGF, CX3CL1, CD40, NOS3, VEGFR2, MCP1, HGF, CXCL9, CD5, VEGFA, CXCL13, CSF1, IL12RB1, and ADA) inversely associated with mortality. There are limited data on the relationship between these biomarkers and ESCC prognosis. Among the proteins associated with increased mortality, higher CCL17 expression within the tumor microenvironment is related to the accumulation of FoxP3 + T-regs in ESCC (Maruyama et al. 2010) that usually correlates with poor clinical outcomes (Saleh and Elkord 2020). Gene expression and survival analyses using data from The Cancer Genome Atlas showed that COL4A1 is expressed at a high level in ESCC and is associated with poor prognosis (Chen et al. 2019). In terms of proteins associated with reduced mortality, growing evidence suggests that TRAIL may be a promising anti-cancer agent for ESCC (Ma et al. 2016; Han et al. 2019). The prognostic value of the VEGF family in patients with ESCC is controversial (Liu et al. 2009; Luo et al. 2011; Luz et al. 2018). Our findings contradict previous reports suggesting that high tissue expression of MCP1 (Koide et al. 2004) and HGF (Ren et al. 2005a, b), as well as high serum levels of HGF (Xie et al. 2019) and CSF (Lukaszewicz-Zajac et al. 2010) are associated with poor prognosis. CD40 is the only marker that was significantly different between cases and controls and was also associated with the prognosis of ESCC cases. The effects of CD40 signaling are multifaceted, depending on the microenvironment (Vonderheide and Glennie 2013). Hypothetically, CD40 may mediate tumor regression through immune activation and/or cytotoxic effects.

In conclusion, our study assessed a broad range of plausible carcinogenic pathways and identified novel associations with ESCC diagnosis and prognosis. Our data indicate that local tissue inflammation may be reflected by alterations of specific proteins in circulation, which could have etiologic and clinical implications. These findings warrant replication and evaluation for functional characterization.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conception and design: MCC, MS, CSR. Acquisition of samples and data: HI, IO, YNK, YK, KM. Analysis and interpretation of data: MCC, MS, CSR, KM. Writing, review, and/or revision of the manuscript: all authors. Study supervision and funding: MCC, KM.

Funding

The HERPACC study was supported by the Ministry of Education, Science, Sports, Culture and Technology of Japan Grant-in-Aid for Scientific Research; Ministry of Health, Labour and Welfare of Japan Grant-in-Aid for Cancer Research and Grant-in-Aid for the Third Term Comprehensive 10-Year Strategy for Cancer Control; and Uehara Memorial Foundation research grant. This analysis was supported by the Intramural Research Program of the U.S. National Cancer Institute.

Data availability

The data sets generated and/or analyzed during the current study are available from the authors on reasonable request.

Declarations

Conflict of interest

The authors declare no conflicts of interest.