For this study, a total of 246 patients with ESCC from Anyang Tumor Hospital (ATH; Anyang, Henan, China) were enrolled after radical resection. Tumor tissue specimens and corresponding paracancerous tissue specimens (esophageal mucosa more than 5 cm from the edge of the cancer tissue) were collected from each patient for formalin fixation and paraffin embedding. The inclusion criteria were as follows: 1) patients did not receive radiotherapy, chemotherapy, or immunotherapy before the operation; 2) no antibiotics were used within four weeks prior to operation; 3) specimens were taken after therapeutic resection of ESCC; 4) postoperative pathological diagnosis of ESCC was performed; 5) comprehensive case information was provided; and 6) the follow-up period was 60 months (5 years). The exclusion criteria were as follows: 1) preoperative radiotherapy, chemotherapy, or immunotherapy; 2) postoperative pathological diagnosis that was not ESCC or complicated with other tumors; 3) received immunotherapy after the operation; 4) incomplete case information; and 5) death not caused by ESCC.

The institutional review boards of ATH reviewed and approved this study (No. 2020-05-B005), and all participants provided written informed consent. Patient information was anonymized and deidentified prior to analysis. Basic patient information, including age, sex, and tobacco and alcohol histories (Smokers were defined as those who smoked at least one cigarette per day and smoked regularly for at least 6 months continuously or cumulatively and were considered smoking positive; others were considered "never smokers" and were considered smoking negative [22]. Alcohol drinkers were defined as those who consumed at least 25 g of alcohol per day on average and were considered positive for alcohol consumption; others were considered "occasional drinkers or never drinkers" and were considered negative for alcohol consumption [23].), and clinical information, including the histological type, degree of tumor invasion and differentiation, lymph node metastasis, tumor-node-metastasis (TNM) stage, and clinical stage, were collected from hospital records. The clinical stage and histological type were based on the 2017 Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) TNM classification system (eighth edition) for esophageal cancer. ATH has a standard follow-up center, and the patients or their families were contacted every 3 months until the end of follow-up. The end of follow-up was postoperative death, and the surviving patients were followed up to 60 months after the operation. If patients could not be contacted by telephone, they were visited at their homes in the countryside or were instructed to go to the local police station for confirmation.

The RNAscope® 2.5 High Definition (HD)-Red Assay is based on Advanced Cell Diagnostics, Inc. (ACD)’s patented signal amplification and background suppression technology. The assay uses a novel and proprietary method of in situ hybridization (ISH) to visualize single RNA molecules per cell in a multitude of sample types mounted on slides. The RNAscope® 2.5 HD (ACD) was used in accordance with the manufacturer’s instructions [22], with minor modifications (see below). The RNAscope assay was performed on formalin-fixed paraffin-embedded (FFPE) sections using the RNAscope 2.5 HD Assay-RED (322,350, ACD, United States). The paraffin-embedded specimens of ESCC tissues and corresponding paracancerous tissues from each patient were sectioned continuously at 3 μm. The specimens were immediately placed into xylene for dewaxing at 60°C for 1.5 h (2 × 5 min) and subjected to ethanol dehydration (2 × 1 min). RNAscope hydrogen peroxide (322,335, ACD, United States) was added and incubated for 10 min, the slides were placed in boiling (98–102°C) RNAscope target repair reagent (322,000, ACD, United States) for 15 min and immediately rinsed in distilled water (3 times), and hydrophobic circles were drawn when the slides were completely dry (room temperature). RNAscope Protease Plus (322,331, ACD, United States) was added and incubated for 30 min at 40°C in a HybEZ Hybridization furnace (220 VAC, 310,013, ACD, United States). The slides were then hybridized with an F. nucleatum probe (RNAscope® Probe—B-Fusobacterium-23S-3zz, Cat. No. 486411, ACD, United States) for 2 h at 40°C. After hybridization, slides were subjected to signal amplification using the RNAscope® 2.5 HD Detection Kit-RED (322,360, ACD, United States). The tissue sections that had been incubated with the probe were incubated with Amp 1 (preamplifier) for 30 min at 40°C, Amp 2 (background reducer) for 15 min at 40°C, Amp 3 (amplifier) at 40°C, Amp 4 (label probe) for 15 min at 40°C, Amp 5 for 30 min at room temperature, and finally Amp 6 for 15 min at room temperature. After each of these steps, the sections were rinsed in wash buffer (310,091, ACD, United States). Then, the hybridization signal was detected using a mixture of Fast-RED solutions A and B (1:60). After counterstaining with Gill's hematoxylin, slides were dried in a 60°C dry oven for 15 min and sealed with VectaMount (321,584, ACD, United States). The F. nucleatum signal was observed in tumor cells as red granules. As recommended by ACD, a semi-quantitative scoring approach was applied to evaluate the staining results [24]. RNAscope results were examined under a standard bright-field microscope at 20-40× magnification. We used the scoring system provided by the vendor, as follows: 0: negative, 0–1 dots/10 tumor cells at 40×; 1+: 1−3 dots/cell visible at 20−40× magnification; 2+: 4–10 dots/cell visible at 20−40× magnification; 3+: >10 dots/cell and <10% positive cells with dot clusters visible at 20× magnification; and 4+: >10 dots/cell and >10% positive cells with dot clusters visible at 20× magnification. For each slide, five areas containing the highest number of positive cells or dot clusters were selected. All tumor cells within each field were counted, and then the percentage of positive F. nucleatum mRNA signals was used as the final score. Image acquisition and analysis were performed using 3DHISTECH software (automatic digital slide scanner, Pannoramic® MIDI I, 3DHISTECH, HU) and Lucia G software (Laboratory Imaging, Prague, Czech Republic) for microscopic image analysis [25-26].

The paraffin-embedded ESCC tissues and corresponding paracancerous tissues from the same patients were fixed with formalin, embedded in paraffin, and serially sectioned at a thickness of 3 μm. After the wax was dissolved in the oven at 60°C for 1.5 h, the samples were immediately placed into xylene for dewaxing (3 times in total, 10 min each time). The samples were successively hydrated in 100%, 95%, 85%, and 70% ethanol for 5 min, followed by continuous rinsing with distilled water for 5 min. Antigen repair was performed at 94–98°C with citrate buffer solution (C1032, Solarbio, China) for 15 min to fully expose antigens. Then, peroxidase blocker was added for 10 min to reduce endogenous peroxidase activity. After rinsing with phosphate-buffered saline (PBS, P1002, Solarbio, China), goat serum was added to reduce non-specific binding, and background staining was performed for 30 min. Serial sections were incubated with CD4 ⁻ , CD25 ⁻ , and FoxP3-specific primary antibodies (PBS dilution ratio 1:100, Abcam, United Kingdom) overnight at 4°C. The next day, after the samples were reheated for 1 h and rinsed with PBS (3 times × 5 min), biotin-labelled secondary antibodies were added and incubated at room temperature for 1 h, and the samples were washed with PBS 3 times × 5 min. The SP-9000 SPlink Detection Kit (Biotin-Streptavidin HRP Detection Systems, SP-9000, ZSGB-BIO, China) was used according to the manufacturer’s instructions. Then, peroxidase-conjugated streptavidin, a biotin-binding protein, was added for sealing at room temperature for 15 min. The samples were washed with PBS (3 times × 5 min) and stained with diaminobenzidine tetrahydrochloride (DAB, DA1010, Solarbio, China) for color detection of antigen-antibody binding. The color was observed under a binocular stereomicroscope (Eclipse 80i, Nikon, Japan), and then the reaction was terminated with distilled water; then, hematoxylin redyeing was performed. The samples were subjected to gradient ethanol dehydration (70%, 85%, 95%, 100%) for 1 min each and then cleared with xylene (3 times × 3 min). After air drying, the samples were sealed with neutral gum. The expression of CD4, CD25, and FoxP3 in ESCC tissue and corresponding paracancerous tissue sections was observed by optical microscopy in five randomly selected fields at 400x magnification (Eclipse 80i, Nikon, Japan). The positive sections jointly determined by two senior pathologists were taken as positive controls, and the negative sections (PBS was used instead of the primary antibody) were taken as negative controls. Conflicting results were resolved with a multihead microscope (Eclipse 80i, Nikon, Japan). Samples with light yellow, tan, or brown coloring in the same location of the lymphocyte membrane (CD4 and CD25) or nucleus (FoxP3) in successive sections were considered positive for Tregs infiltration. A semi-quantitative scoring scheme based upon the average density of positive-staining cells for the section was utilized for assessment of expression of each immune marker. Five high-power microscopic fields (400 times) in the tissue sections were randomly selected for observation. Each tissue slide was evaluated by two senior pathologists. Staining was scored based on the number of positive-staining cells per 400x field with 1-5 cells designated as 1+, 6–30 cells as 2+, >30 cells as 3+, and no positive-staining cells as 0 [27]. A final score of ≥1 on all three slides of each patient was defined as positive for Tregs infiltration.

All statistical analyses were performed with the SPSS statistical package, version 26.0 (SPSS Inc., Chicago, IL, United States): 1) the correlation analysis was performed with the chi-square test, 2) the survival curve was drawn with the Kaplan-Meier method, 3) the difference between survival times was analyzed by the log-rank test, 4) the effect of factors on prognosis was analyzed using Cox regression, and 5) a p value < 0.05 was considered statistically significant.

We used RNAscope and IHC to detect F. nucleatum infection and Treg enrichment in ESCC tissues and corresponding paracancerous tissues (Figure 1) and analyzed the correlation between the two factors. Red granules can be seen in the cytoplasm of ESCC tissues, indicating F. nucleatum infection (Figure 1A). In the serial sections, light yellow, tan, or brown coloring was seen on the membrane (CD4 and CD25) or nucleus (FoxP3) of lymphocytes at the same location, indicating the positive infiltration of Tregs (Figure 1C,D). F. nucleatum infection was significantly correlated with Tregs infiltration (Table 1, p < 0.05). However, most cells in the corresponding paracancerous tissues were negative (Figure 1I,J,K,L). F. nucleatum infection in ESCC tissues was significantly higher than that in the corresponding paracancerous tissues (p < 0.05, Table 2). Tregs infiltration in ESCC tissues was significantly higher than that in the corresponding paracancerous tissues (p < 0.05, Table 3).

Among the patients with F. nucleatum infection (Table 4), there were significantly more males, smokers, and drinkers (p < 0.05), and most of the tumor cells in the positive group were hypodifferentiated, with deeper tumor tissue infiltration, significantly more lymph node metastases, and more frequent occurrences of clinical stages III/IV (p < 0.05).

The 5 years survival rate and median survival time of the 246 patients with ESCC after surgery were 30.08% and 36.0 months, respectively. The 5 years survival rate and median survival time in the F. nucleatum infection-positive group (12.94% and 47.0 months, respectively) were significantly lower (p < 0.05) than those in the negative group (39.13% and 24.0 months, respectively), as presented in Table 5 and Figure 2.

A total of 246 patients with ESCC successfully completed postoperative follow-up, with the death due to ESCC as the endpoint event, and univariate Cox regression analysis was conducted, F. nucleatum infection (positive = 1/negative = 0), sex (male = 1/female = 0), age (≥60 = 1/<60 = 0), smoking (positive = 1/negative = 0), alcohol consumption (positive = 1/negative = 0), degree of tumor differentiation (poor = 1/moderate/high = 0), depth of infiltration (≥adventitia = 1/<adventitia = 0), lymph node metastasis (positive = 1/negative = 0), and clinical stage (III/IV = 1/I/II = 0) were used as independent variables. The data showed that sex, smoking, alcohol consumption, degree of tumor differentiation, depth of infiltration, lymph node metastasis, clinical stage, and F. nucleatum infection were associated with overall survival (p < 0.05). Confounding factors (sex and age) were included, and variables that were significantly associated (p < 0.05) with ESCC in the univariate analysis were entered into multivariate Cox regression analysis, which showed that the degree of tumor differentiation, depth of infiltration, lymph node metastasis, clinical stage, and F. nucleatum infection are independent risk factors that affect the prognosis of ESCC (p < 0.05), as presented in Table 6.