Combination of high anti-SKI and low anti-TMED5 antibody levels is preferable prognostic factor in esophageal carcinoma

Abstract

Given that esophageal cancer is highly malignant, the discovery of novel prognostic markers is eagerly awaited. We performed serological identification of antigens by recombinant cDNA expression cloning (SEREX) and identified SKI proto-oncogene protein and transmembrane p24 trafficking protein 5 (TMED5) as antigens recognized by serum IgG antibodies in patients with esophageal carcinoma. SKI and TMED5 proteins were expressed in Escherichia coli, purified by affinity chromatography, and used as antigens. The serum anti-SKI antibody (s-SKI-Ab) and anti-TMED5 antibody (s-TMED5-Ab) levels were significantly higher in 192 patients with esophageal carcinoma than in 96 healthy donors. The presence of s-SKI-Abs and s-TMED5-Abs in the patients' sera was confirmed by western blotting. Immunohistochemical staining showed that the TMED5 protein was highly expressed in the cytoplasm and nuclear compartments of the esophageal squamous cell carcinoma tissues, whereas the SKI protein was localized predominantly in the nuclei. Regarding the overall survival in 91 patients who underwent radical surgery, the s-SKI-Ab-positive and s-TMED5-Ab-negative statuses were significantly associated with a favorable prognosis. Additionally, the combination of s-SKI-Ab-positive and s-TMED5-Ab-negative cases showed an even clearer difference in overall survival as compared with that of s-SKI-Ab-negative and s-TMED5-Ab-positive cases. The s-SKI-Ab and s-TMED5-Ab biomarkers are useful for diagnosing esophageal carcinoma and distinguishing between favorable and poor prognoses.

INTRODUCTION

Esophageal carcinomas can grow expeditiously, and the prognosis for cases with aggressive invasion and metastasis is poor.¹ Regarding the therapeutic strategy, surgical invasion is overwhelming, and chemotherapy and/or radiotherapy often induce adverse events or side effects.^2,3 Therefore, the outcomes of esophageal carcinoma therapy were not fully satisfactory. Therefore, we aimed to diagnose esophageal carcinoma at an early stage to improve disease prognosis, using p53 antibodies.⁴ If there are useful biomarkers to predict prognosis, they will be effective for postoperative therapy and follow-up.

For several decades, we have performed serological identification of antigens using the recombinant cDNA expression cloning (SEREX) method. It is an effective screening method for detecting novel tumor markers.⁵ SEREX involves immune screening of phage cDNA libraries prepared from tumor specimens using the sera from the patients. The sequencing of isolated cDNA clones makes SEREX suitable for large-scale screening of tumor antigens. Our SEREX screening method has identified many antibody biomarkers or antigens, including trophoblast cell surface antigen 2 (TROP2),⁶ solute carrier family 2 member 1/glucose transporter 1 (SLC2A1/GLUT1),⁷ ZIC2,⁸ BAMBI,⁹ striatin,¹⁰ LDL receptor-related protein-associated protein 1,¹¹ proprotein convertase subtilisin/kexin type 9,¹² cofilin,¹³ and WD repeat-containing protein 1.¹⁴

In the present study, SEREX screening for esophageal carcinoma identified the SKI proto-oncogene protein and transmembrane p24 trafficking protein 5 (TMED5). The serum anti-SKI antibody (s-SKI-Ab) and anti-TMED5 antibody (s-TMED5-Ab) levels were compared between patients with esophageal carcinoma and healthy donors (HDs). We also analyzed the relationship between these two antibody levels and clinicopathological features of patients and prognosis.

MATERIALS AND METHODS Collection of serum samples

Altogether, 192 serum samples of esophageal carcinoma were collected. Of these, 91 were surgical cases, and their serum samples were collected before treatment at Toho University, Omori Medical Center, between June 2010 and February 2016. Among the 91 patients, 70 were men and 21 were women with a median age of 67 years. Surgical cases were included with or without preoperative chemoradiotherapy in the criteria, and 63 received neoadjuvant chemotherapy. All patients were followed up over 5 years after surgery or until death. The remaining 101 cases underwent chemotherapy, radiation therapy, non-curative treatment, or best supportive care. Duplicated and distant metastasis cases were also excluded. Data on the clinicopathological characteristics and prognoses were retrospectively obtained. The sera of 96 HDs were collected from the Port Square Kashiwado Clinic. This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Toho University, Graduate School of Medicine (No. A18103_A17052_A16035_A16001_26095_25024_24038_22047_22047). Retrospective analysis of the patients' medical records was approved by the Ethics Committee of Toho University Omori Medical Center (approval nos. M21038_20197_19213 and M21320_21039_20200_30196_19056_18002). The Ethics Committee of Chiba University Graduate School of Medicine (No. 2018–320) (Japan) and Port Square Kashiwado Clinic, Kashiwado Memorial Foundation (approval no. 2012–001) also approved the study protocol. We collected the sera from patients who had provided written informed consent.

Preparation and purification of antigenic SKI and TMED5 proteins

cDNA clones of SKI and TMED5 were isolated by SEREX screening using λZAP II phage cDNA library of the human esophageal squamous cell carcinoma cell line, T.Tn, and sera of patients with esophageal carcinoma.^6,7 The 1353–2259 region of SKI cDNA (Accession number: NM_003036; coding sequence: 73–2259) and full-length TMED5 cDNA (Accession number: NM_016040; coding sequence: 113–802) were isolated. The cDNAs were then recombined into the prokaryotic expression plasmid, pGEX-4 T-1, and expressed into proteins. ECOS™ competent Escherichia coli (E. coli) BL-21 cells (Nippon Gene; Tokyo, Japan) were transformed with pGEX-4 T-1, pGEX-4 T-1-SKI, and pGEX-4 T-1-TMED5, and then cultured for 3 h in 200-mL Luria broth containing 0.1 mM isopropyl β-d-thiogalactopyranoside (IPTG; Wako Pure Chemicals, Osaka, Japan). The cells were lysed by sonication in BugBuster Protein Extraction Reagent (Merck Millipore, Darmstadt, Germany), and GST, GST-SKI, and GST-TMED5 proteins were purified by affinity chromatography with a Glutathione-Sepharose column (Cytiva, Pittsburgh, PA) as previously described.¹⁵

Measurement of s-SKI-Ab and s-TMED5-Ab levels and conventional serum markers

Serum samples were obtained before treatment, centrifuged at 3000 g for 10 min, and stored at −80°C until use. The s-SKI-Ab and s-TMED5-Ab levels were measured using an amplified luminescence proximity homogeneous assay-linked immunosorbent assay (AlphaLISA). AlphaLISA was conducted using 384-well microtiter plates (white opaque OptiPlate™, Revvity, Waltham, MA) containing 2.5-μL 1/100-diluted sera and 2.5-μL GST, GST-SKI, or GST- TMED5 (10 μg/mL) in an AlphaLISA buffer (25-mM HEPES, pH 7.4, 0.1% casein, 0.5% Triton X-100, 1-mg/mL dextran-500, and 0.05% Proclin-300) according to the manufacturer's instructions (Revvity).¹⁶ The reaction mixture was incubated at room temperature for 6–8 h. Next, anti-human IgG-conjugated acceptor beads (2.5 μL of 40 μg/mL) and glutathione-conjugated donor beads (2.5 μL of 40 μg/mL) were added and incubated further for 7–28 days at room temperature in the dark. The chemical emission was read on an EnSpire Alpha microplate reader (Revvity) as previously described.¹⁷ Specific reactions were calculated by subtracting the Alpha photon counts of the GST control from those of the GST fusion proteins.

The s-p53-Ab and squamous cell carcinoma antigen (SCC-Ag) levels were measured using the STACIA® MEBLux TEST anti-p53 (Medical & Biological Laboratories, Tokyo, Japan)¹⁸ and Elecsys® SCC (Roche, Swiss),¹⁹ respectively. The cutoff values for s-p53-Abs and SCC-Ag were fixed at 1.3 IU/mL and 1.5 ng/mL, respectively.

Comparison of the clinicopathological parameters between high and low groups of s-SKI-Ab and s-TMED5-Ab for the surgically treated esophageal carcinoma patients

We divided clinicopathological information into two groups and compared the titers of s-SKI-Ab and s-TMED5-Ab among 91 surgical cases using Fisher's exact probability test. Specifically, the patients were divided into two groups by sex (male versus female), age (<64 years versus >65 years due to the median age of 67 years), tumor depth (T1 versus T2–T4 or T1–T2 versus T3–T4), presence and absence of lymph node metastasis, and lesion site (upper and lower carcinoma lesion sites). The outlier boundaries used in clinical medicine divided the serological cutoff levels.

Immunohistochemical staining

Formalin-fixed, paraffin-embedded esophageal cancer tissues were cut into 4-μm-thick sections. The sections were deparaffinized and blocked with 5% bovine serum albumin. Next, the sections were reacted with primary anti-SKI (GTX81524, rabbit polyclonal antibodies, GeneTex, CA) or anti-TMED5 antibodies at 2 μg/mL for 1 h at room temperature, incubated with horseradish peroxidase-conjugated anti-rabbit IgG (ab7090, Abcam, Cambridge, UK), and visualized using a diaminobenzidine chromogen substrate (MK210, Takara Bio, Kusatsu, Japan).^6,12

Western blotting

GST, GST-fused SKI, and TMED5 proteins (0.3 μg) were separated on sodium dodecyl sulfate-polyacrylamide gels. After transfer, the membranes were blocked with 0.5% dry milk in Tris-buffered saline [150-mM NaCl, 20-mM Tris–HCl (pH 7.6), and 0.1% Tween-20; TBS-T] and incubated with anti-GST antibodies (Rockland, Gilbertsville, PA) or sera from HDs or patients with esophageal carcinoma. After incubation with a horseradish peroxidase-conjugated secondary antibody, immunoreactivity was detected using Immobilon (Merck Millipore, Darmstadt, Germany) as described previously.^6,12,20,21

Statistical analysis

Fisher's exact test analyzed the correlations between the two variables. The continuous data between the HDs and patients with esophageal carcinoma for the non-parametric tests were analyzed using the Mann–Whitney U-test.

The cutoff value for detecting 192 esophageal carcinomas was calculated by performing a receiver operating characteristic (ROC) curve analysis. In particular, X-tile software (Yale University, New Haven, CT)²² was used to determine the cutoff level between survival and mortality cases among 91 operated esophageal carcinoma cases.

The survival curves were measured using the Kaplan–Meier method. The log-rank test compared the univariate results. The significant predictors were evaluated via a multivariate analysis using the Cox proportional hazards model. These statistical results were analyzed using EZR software²³ (Jichi Medical University, Saitama, Japan, version 1.41). Statistically significant levels were defined as p < 0.05.

RESULTS Identification of SKI and TMED5 using the SEREX screening method and comparison of the s-SKI-Ab and s-TMED5-Ab levels between HDs and patients with esophageal carcinoma

Large-scale SEREX screening was performed using the sera of patients with esophageal carcinoma. SKI and TMED5 were identified as antigens recognized by the serum IgG antibodies.

Next, the s-SKI-Ab and s-TMED5-Ab levels were evaluated by AlphaLISA using GST-SKI and GST-TMED5, respectively, as antigens. Both the s-SKI-Ab and s-TMED5-Ab levels were significantly higher in patients with esophageal carcinoma than in HDs (Figure 1A,B; p < 0.001). The average ± SD of the s-SKI-Ab levels (Alpha photon counts) in HDs and patients with esophageal carcinoma were 130,689 ± 52,961 and 176,701 ± 71,678, respectively, whereas the s-TMED5-Ab levels were 81,067 ± 48,801 and 109,137 ± 61,909, respectively.

FIGURE 1. Comparison of the serum anti-SKI antibody (s-SKI-Ab) and serum anti-TMED5 antibody (s-TMED5-Ab) levels and receiver operating characteristic (ROC) curve analysis between healthy donors (HDs) and surgically treated 91 esophageal carcinomas. The s-SKI-Ab (A) and s-TMED5-Ab (B) levels in patients with esophageal carcinoma and HDs were examined by AlphaLISA and shown in box–whisker plots. Box plots represent 25, 50, and 75 percentiles. The upper and lower horizontal lines represent the 90 percentile. ***p [less than] 0.001; evaluated using the Mann–Whitney U-test. The ROC curve analysis for SKI-Ab and TMED5-Ab are shown in panels (C) and (D), respectively. The numbers in (C) and (D) represent the area under the curve (AUC), 95% confidence interval (CI), cutoff level, specificity and sensitivity, and p-values.

The ROC curve indicated that the area under the ROC curve (AUC) of s-SKI-Ab was 0.678. When the study participants were divided into positive and negative s-SKI-Ab groups using the Youden index, 122,712, the sensitivity and specificity were 59.47% and 68.82%, respectively (Figure 1C). The AUC value for s-TMED5-Ab was 0.716. When the Youden index of 91,902 was used as the cutoff value for s-TMED5-Abs, the sensitivity and specificity were 46.88% and 86.32%, respectively (Figure 1D).

Western blotting

The presence of anti-SKI and anti-TMED5 antibodies in the patients' sera was confirmed by western blotting. The GST-fused SKI and GST-TMED5 proteins were purified and used as antigens in the subsequent study. The purity exceeded 90% (Figure 2A, gel I). The GST-SKI, GST-TMED5, and GST proteins were detected as 26-, 62-, and 50-kDa proteins, respectively, using the anti-GST antibody (Figure 2A, gel II). GST-SKI and GST-TMED5 were recognized by the serum IgG antibodies of patients with esophageal carcinoma (EC#61, EC#62, EC#64) but not by those of an HD (HD#13) (Figure 2A, gels III–VI). The GST protein exhibited no obvious reactivity against the serum IgG antibodies, regardless of whether they were obtained from HDs or patients with esophageal carcinoma.

FIGURE 2. The result of western blotting and immunohistochemistry with esophageal carcinoma patients. (A) Representative results by western blotting are shown. GST, GST-SKI, and GST-TMED5 proteins were electrophoresed through SDS-PAGE gels followed by staining with Coomassie Brilliant Blue (CBB) (I), or western blotting using anti-GST (αGST) (II), sera of patients [EC#61 (III), EC#62 (VI), EC#64 (V)], or a healthy donor serum [HD#13 (VI)]. The arrows indicate the GST-SKI, GST-TMED5 and GST positions. The left vertical axis indicates the molecular weight. In the figure, the single asterisk ‘*’ represents the degradation products, whereas the double asterisks ‘**’ represent non-specific reaction for EC#61. The immunohistochemical staining for SKI (B) and TMED5 (C) of esophageal carcinoma is shown.

Immunohistochemical staining

We examined the SKI and TMED5 expressions by immunohistochemical staining. A high expression of the SKI protein around the nucleus in esophageal SCC tissues was observed²⁴ (Figure 2B). The nuclear and Golgi fields of esophageal SCC tissues, but not the surrounding normal tissues or interstitial lesions, were predominantly stained with the TMED5 antibody (Figure 2C).²⁵

Overall survival according to the s-SKI-Ab and s-TMED5-Ab levels among surgically treated esophageal carcinoma

To clarify the prognostic features of s-SKI-Abs and s-TMED5-Abs, the survival curves were drawn utilizing the Kaplan–Meier method using surgical cases. To evaluate the prognosis of surgical cases, the X-tile software²² was used to determine the optimal cutoff level between the survival and mortality cases among patients with operated esophageal carcinomas. The X-tile 3.6.1 software is a bioinformatics tool for assessing the biological relationships between the biomarkers and outcomes and the discovery of cutoff points based on the marker levels. Moreover, the number with the maximum chi-square test statistic values of the log-rank tests will be used as the final cutoff value.^26–28 Dividing the participants into positive and negative groups according to the s-SKI-Ab and s-TMED5-Ab levels (SKI cutoff: 102888, TMED5 cutoff: 179720), The s-SKI-Ab-positive patients showed significantly better overall survival than the s-SKI-Ab-negative patients (p = 0.003) (Figure 3A). Contrarily, the s-TMED5-Ab-negative patients showed significantly better overall survival than the s-TMED5-Ab-positive patients (p = 0.048) (Figure 3B).

FIGURE 3. Comparison of the overall survival between the positive and negative s-SKI-Ab and s-TMED5-Ab cases. The overall survival of surgically treated esophageal carcinoma is presented in the Kaplan–Meyer plots for s-SKI-Ab (A) and s-TMED5-Ab (B). The s-SKI-Ab and s-TMED5-Ab levels are divided into high and low groups. The cutoff level was assessed using ROC curve analysis between the survival and mortality cases of esophageal carcinoma using X-tile software (SKI cutoff: 102888; TMED5 cutoff: 179720). Statistical analyses were performed by using the log-rank test in each group. (C) The prognosis when using a combination of the s-SKI-Ab and s-TMED5-Ab levels. All s-TMED5-Ab-positive cases were included in the s-SKI-Ab positive group, and the combinations were categorized into three (C). A flow chart of s-SKI-Ab and s-TMED5-Ab level was created to allow easy understanding of the prognosis (D). First, the s-TMED5-Ab level of the esophageal carcinoma patients was measured before treatment. Only the s-TMED5-Ab-negative cases will undergo s-SKI-Ab measurement in the next step. Only the s-SKI-Ab-positive cases were expected to have a favorable prognosis.

A combined analysis of the s-SKI-Ab and s-TMED5-Ab levels for survival

Given that both s-SKI-Abs and s-TMED5-Abs were related to prognosis, we combined these antibody markers to examine their prognostic characteristics. Given that all of the s-TMED5-Ab-positive cases were included in the s-SKI-Ab-positive cases, we divided the participants into three groups based on their prognosis: s-SKI-Ab-negative and s-TMED5-Ab-negative (Group A), s-SKI-Ab-positive and s-TMED5-Ab-negative (Group B), and s-SKI-Ab-positive and s-TMED5-Ab-positive (Group C) groups. The survival rate was significantly better in the B group than in the A and C groups (p = 0.0007 and 0.0094, respectively) (Figure 3C). Thus, the flowchart of s-SKI-Abs and s-TMED5-Abs can help predict the pre-treatment prognosis of patients with esophageal carcinoma (Figure 3D).

Univariate and multivariate analyses of the clinicopathological parameters and serum antibody levels in patients with surgically treated esophageal carcinoma

The cutoff s-SKI-Ab and s-TMED5-Ab levels were set using the X-tile software described above, and the clinicopathological characteristics were evaluated using Fisher's exact probability test. The univariate results did not show any correlation with sex, age, tumor depth, lymph node metastasis, tumor location, white blood cell count, neutrophil count, hemoglobin level, C-reactive protein level, or albumin level. No significant correlation was observed between these antibody markers and other conventional tumor markers, including SCC-Ag or p53-Abs (Tables 1 and 2). However, in the multivariate analysis with logistic regression analysis, s-TMED5-Ab significantly influenced tumor depth and lymph node metastasis (Table 2).

TABLE 1 Comparison of the clinicopathological characteristics between the patients with esophageal carcinoma stratified according to the s-SKI-Ab level in the univariate and multivariate analyses.

Variables	Fisher's exact probability test			Logistic regression analysis
	s-SKI-Ab low	s-SKI-Ab high	p-value	Odds ratio	95% CI	p-value
Sex
Male	10	59	>0.999
Female	3	18
Age
>65	6	47	0.369
≤65	7	30
Tumor depth
T1	4	25	>0.999
T2 T3 T4	9	52
Tumor depth
T1 T2	4	32	0.551	0.418	0.092–1.91	0.261
T3 T4	9	45
Lymph node metastasis
N0	5	36	0.765	0.972	0.245–3.87	0.968
N1	8	41
Location
Upper	2	12	>0.999
Lower	11	65
WBC (/μL)
>8000	2	9	0.657
≤8000	11	68
Neutrophil (%)
>70	4	15	0.461
≤70	9	62
Lymphocyte (%)
>35	1	16	0.448
≤35	12	61
Hemoglobin (g/dL)
>12	7	53	0.345
≤12	6	24
CRP (mg/dL)
>0.3	4	25	>0.999
≤0.3	8	51
Albumin (g/dL)
>3.5	10	57	>0.999
≤3.5	3	20
SCC-Ag (ng/mL)
>1.5	3	28	0.368	2.810	0.647–12.2	0.168
≤1.5	10	47
p53-Abs (U/mL)
>1.30	3	14	0.710
≤1.30	10	61

Note: The total numbers for CRP, SCC-Ag and p53-Ab are less than the total of 91 cases because in some cases these parameters were not measured. The parameters, sex, age, tumor depth, lymph node metastasis, tumor location, WBC, neutrophil, lymphocyte, hemoglobin, CRP, albumin, SCC-Ag, and p53-Abs were divided into a binary category, and statistical evaluation for low and high s-SKI-Ab titers was performed using Fisher's exact probability test. Next, multivariate analysis was performed for tumor depth, lymph node metastasis, and SCC-Ag using logistic regression analysis.

Abbreviations: Ab, antibody; CRP, C-reactive protein; SCC-Ag, squamous cell carcinoma antigen; s-SKI-Ab, serum SKI antibody; WBC, white blood cell.

TABLE 2 Comparison of the clinicopathological characters between patients with esophageal carcinoma stratified according to s-TMED5-Ab included in the univariate and multivariate analyses.

Variables	Fisher's exact probability test			Logistic regression analysis
	s-TMED5-Ab low	s-TMED5-Ab high	p-value	odds ratio	95% CI	p-value
Sex
Male	61	9	>0.999
Female	18	3
Age
>65	47	6	0.547
≤65	32	6
Tumor depth
T1	27	2	0.325
T2 T3 T4	52	10
Tumor depth
T1 T2	35	2	0.113	8.500	1.270–56.6	0.027
T3 T4	44	10
Lymph node metastasis
N0	33	8	0.128	0.186	0.042–0.818	0.026
N1	46	4
Location
Upper	12	2	>0.999
Lower	67	10
WBC (/μL)
>8000	10	1	>0999
≤8000	69	11
Neutrophil (%)
>70	17	2	>0.999
≤70	62	10
Lymphocyte (%)
>35	14	3	0.690
≤35	65	9
Hemoglobin (g/dL)
>12	52	9	0.744
≤12	27	3
CRP (mg/dL)
>0.3	24	5	0.516
≤0.3	53	7
Albumin (g/dL)
>3.5	61	7	0.171
≤3.5	18	5
SCC-Ag (ng/mL)
>1.5	26	5	0.505	0.816	0.187–3.56	0.787
≤1.5	52	6
p53-Abs (U/mL)
>1.30	15	2	>0.999
≤1.30	62	10

Note: The total numbers for CRP, SCC-Ag, and p53-Ab are less than the total of 91 cases because in some cases these parameters were not measured. The parameters sex, age, tumor depth, lymph node metastasis, tumor location, WBC, neutrophil, lymphocyte, hemoglobin, CRP, albumin, SCC-Ag, and p53-Abs were divided into a binary category, and statistical evaluation for low and high s-SKI-Ab titers was performed using Fisher's exact probability test. Next, multivariate analysis was performed for tumor depth, lymph node metastasis, and SCC-Ag using logistic regression analysis.

Abbreviations: CRP, C-reactive protein; Ab, antibody; SCC-Ag, squamous cell carcinoma antigen; s-TMED5-Ab, serum TMED5 antibody; WBC, white blood cell.

The relationship between clinicopathological characteristics and each serum antibody titer in all the 192 cases was also analyzed using Fisher's exact test. The cutoff values of s-SKI-Ab and s-TMED5-Ab were determined as 122,712 and 91,902, respectively, according to the ROC curve analysis as described above. Although some of the characteristics were not obtained, no statistically significant correlation was observed between each serum antibody levels and clinicopathological factors (age, tumor location, white blood cell (WBC), neutrophil, lymphocyte, hemoglobin, CRP, albumin, SCC-Ag, and p53-Abs) (Table S1).

DISCUSSION

To explore the useful biomarkers, we identified SKI and TMED5 as antigens recognized by serum IgG antibodies in patients with esophageal carcinoma. Western blotting confirmed the presence of their antibodies in the sera (Figure 2A). Immunohistochemistry revealed that the SKI and TMED5 proteins were highly expressed in esophageal carcinoma tissues (Figure 2B,C). The results suggested that the development of serum autoantibodies was attributable to their high expression. The s-SKI-Ab and s-TMED5-Ab levels were significantly higher in patients with esophageal carcinoma than in HDs (Figure 1A,B). The s-SKI-Ab and s-TMED5-Ab levels showed an inverse relationship with overall survival (Figure 3A,B), and the s-SKI-Ab-positive/s-TMED5-Ab-negative patients had better survival than the s-SKI-Ab-negative/s-TMED5-Ab-positive patients (Figure 3C). This result suggested that the combination of these two antibody markers would be useful for predicting the prognosis.

In 192 esophageal carcinoma cases, there was no significant relationship between two antibody markers and clinicopathological factors (Table S1). In surgical cases, univariate analysis showed that none of the clinicopathological characteristics was significantly correlated with the s-SKI-Ab or s-TMED5-Ab levels (Tables 1 and 2). However, the multivariate analysis showed a statistically significant relationship between tumor depth and lymph node metastasis with s-TMED5-Abs. That is, the odds ratio was higher in cases with T1/T2 depth stage than in those with T3/T4 depth stage but it was lower in cases with N0 metastasis than in those with N1 metastasis. These results implied that the poor prognosis of the s-TMED5-Ab-positive group (Figure 3B) was attributable to the proliferating ability but not the metastatic ability of esophageal carcinomas, which may account for the unfavorable prognosis of the s-TMED5-Ab-positive group compared with the favorable prognosis of the s-TMED5-Ab-negative group (Figure 3B). We thought that these results were based on the TGF-β signaling pathway.

SKI is an evolutionarily conserved protein, and the SKI antigen is highly expressed as tumor invasion and lymph node metastasis of esophageal cancer progress.^29,30 Additionally, overexpression of the Ski/Sno family of proto-oncogenes interdicts TGF-β signaling and causes cells to become refractory to the antiproliferative activity of TGF-β.³¹ TMED5 is highly expressed in cervical cancer,³² bladder cancer,³³ and HCC³⁴ cell lines. The overall survival in cases with high TMED5 transcription levels was considered to be poor, and TMED5 may be positively correlated with the TGF-β signaling pathway.³⁵ ski/sno is a key negative regulator of TGF-β signaling.³⁶ However, a direct relationship between serum ski/sno and TMED5 levels has not been clarified.

TGF-β is a cytokine involved in diverse biological processes, including cell proliferation, differentiation, cell–cell interactions, cell migration, and apoptosis. TGF-β is the most important mediator of epithelial–mesenchymal transition (EMT) in human cancer.³⁷ EMT was also known to activate tumor initiation, invasion, metastasis, and resistance to therapy.³⁸ One of the poor prognostic reasons in the s-TMED5-Ab-positive and s-SKI-Ab-negative groups was thought to be TGF-β-mediated EMT activation. This is compatible with the results of our multivariate analysis showing that the s-TMED5-Ab levels were positively correlated with tumor depth-related proliferation and negatively correlated with lymph node metastasis of esophageal carcinoma (Table 2).

We reported the results of our large-scale SEREX screening of esophageal carcinoma markers. Interestingly, some, not a few, of the SEREX antigens are related to TGF-β. For example, TROP2⁶ and SLC2A1/GLUT1⁷ were strongly induced in response to TGF-β1.^39,40 Cofilin¹³ was translocated in the mitochondrial fraction after treatment with TGF-β1.⁴¹ ZIC2⁸ promoted colorectal cancer growth and metastasis through the TGF-β signaling pathway.⁴² BAMBI⁹ inhibited TGF-β signaling as a dominant-negative, non-signaling, pseudoreceptor for the TGF-β type I receptor family.⁴³ Thus, the progression of esophageal carcinoma can depend on the TGF-β signaling pathway. As described above, SKI inhibits TGF-β signaling,³² whereas TMED5 mediates and promotes TGF-β signaling,³⁶ indicating that SKI and TMED5 have opposite effects on the TGF-β signaling pathway. This is consistent with the result that s-SKI-Ab and s-TMED5-Ab levels showed opposite effects on the overall survival of patients with esophageal carcinoma (Figure 3A–C), if the serum antibody levels are associated with the protein expression levels. Some other reports have shown that the high autoantibody titers were accompanied by the high expression of their antigenic proteins in cancer tissues.^11,12,44,45

The prognostic prediction regarding the two antibody levels is summarized in Figure 3D. The s-SKI-Ab-positive and s-TMED5-Ab-negative groups showed favorable prognoses than the both s-SKI-Ab- and s-TMED5-Ab-negative and -positive groups. This simple flow chart can be applied to postoperative prognostic prediction and adjuvant chemotherapy.

The strength of our study was that it was easy to evaluate the prognosis, but the limitations were that it was a retrospective study and the number of cases was rather small.

In conclusion, the s-SKI-Ab and s-TMED5-Ab levels were significantly higher in patients with esophageal carcinoma than in HDs. These antibody markers are useful for predicting the prognosis of patients with esophageal carcinoma individually or in combination.

AUTHOR CONTRIBUTIONS

Masaaki Ito: Conceptualization; writing – original draft. Satoshi Yajima: Data curation. Takashi Suzuki: Data curation. Yoko Oshima: Data curation. Tatsuki Nanami: Data curation. Makoto Sumazaki: Data curation. Fumiaki Shiratori: Data curation. Hirotaka Takizawa: Data curation. Shu-Yang Li: Methodology; validation. Bo-Shi Zhang: Methodology. Yoichi Yoshida: Methodology. Tomoo Matsutani: Methodology. Takaki Hiwasa: Conceptualization; writing – review and editing. Hideaki Shimada: Conceptualization; funding acquisition; writing – review and editing.

ACKNOWLEDGMENTS

We thank Professor Kimihiko Funahashi, MD, PhD, and Yuichiro Otsuka, MD, PhD for supporting our study and Ms. Seiko Otsuka, Ms. Masae Suzuki, Ms. Chiho Kusaka, and Ms. Satoko Ishibashi for preparing the patient data.

FUNDING INFORMATION

This research was supported by the Project for Cancer Research and Therapeutic Evolution (P-CREATE) from the Japan Agency for Medical Research and Development (AMED grant no. 21cm0106403h0006), and Grants-in-Aid for Scientific Research (KAKENHI) from the Japan Society for the Promotion of Science (JSPS) grant nos. 16 K10520, 21 K08695, 20 K16396, 20 K17953, and 22 K07273.

CONFLICT OF INTEREST STATEMENT

Hideaki Shimada is an editorial board member of Cancer Science. The other authors declare no conflict of interest.

ETHICS STATEMENTS

Approval of the research protocol by an institutional review board: This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Toho University, Graduate School of Medicine (No. A18103_A17052_A16035_A16001_26095_25024_24038_22047_22047). Retrospective analysis of the patients' medical records was approved by the Ethics Committee of Toho University Omori Medical Center (approval nos. M21038_20197_19213 and M21320_21039_20200_30196_19056_18002). The Ethics Committee of Chiba University Graduate School of Medicine (No. 2018–320) (Japan) and Port Square Kashiwado Clinic, Kashiwado Memorial Foundation (approval no. 2012–001) also approved the study protocol.

Informed Consent: We collected the sera from patients who had provided written informed consent.

Registry and Registration No. of the study: N/A.

Animal Studies: N/A.