Bae GE, Kim SM, Lee JK, Lee SY. First large-scale prospective evaluation of serum S100A6 as a complementary biomarker for early pancreatic cancer detection. World J Gastroenterol 2026; 32(21): 116581 [DOI: 10.3748/wjg.v32.i21.116581]
Corresponding Author of This Article
Soo-Youn Lee, MD, PhD, Professor, Department of Laboratory Medicine and Genetics, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, South Korea. suddenbz@skku.edu
Research Domain of This Article
Gastroenterology & Hepatology
Article-Type of This Article
research-article
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Bae GE, Kim SM, Lee JK, Lee SY. First large-scale prospective evaluation of serum S100A6 as a complementary biomarker for early pancreatic cancer detection. World J Gastroenterol 2026; 32(21): 116581 [DOI: 10.3748/wjg.v32.i21.116581]
Go Eun Bae, Sang-Mi Kim, Soo-Youn Lee, Department of Laboratory Medicine and Genetics, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul 06351, South Korea
Sang-Mi Kim, Department of Laboratory Medicine and Genetics, Chosun University Hospital, Chosun University School of Medicine, Gwangju 61453, South Korea
Jong Kyun Lee, Division of Gastroenterology, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul 06351, South Korea
Author contributions: Bae GE, Kim SM, and Lee SY conceived and designed the analysis; Kim SM and Lee SY collected the data; Bae GE and Lee JK contributed data or analysis tools; Bae GE performed the analysis; Bae GE and Kim SM wrote the paper, they contributed equally to this manuscript and are co-first authors.
Supported by the National Research Foundation of Korea Grant funded by the Korea government, No. 2021R1A2C1006409.
Institutional review board statement: The study was approved by the Institutional Review Board of Samsung Medical Center (No. SMC IRB 2021-02-153).
Informed consent statement: The Institutional Review Board of Samsung Medical Center waived the requirement for informed consent.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
STROBE statement: The authors have read the STROBE Statement-checklist of items, and the manuscript was prepared and revised according to the STROBE Statement-checklist of items.
Data sharing statement: The data that support the findings of this study are available from the corresponding author (Lee SY), upon reasonable request.
Corresponding author: Soo-Youn Lee, MD, PhD, Professor, Department of Laboratory Medicine and Genetics, Samsung Medical Center, Sungkyunkwan University School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 06351, South Korea. suddenbz@skku.edu
Received: November 18, 2025 Revised: January 27, 2026 Accepted: March 5, 2026 Published online: June 7, 2026 Processing time: 192 Days and 23.9 Hours
Abstract
BACKGROUND
Pancreatic cancer (PC) remains one of the most lethal malignancies due to late diagnosis and lack of early detection strategies. Carbohydrate antigen 19-9 (CA19-9), the only Food and Drug Administration-approved biomarker, demonstrates limited sensitivity for early-stage disease and poor specificity in distinguishing PC from benign pancreatic conditions. S100A6, a calcium-binding protein implicated in cell proliferation and apoptosis, has been identified as a potential diagnostic biomarker in various malignancies. While overexpression of S100A6 has been reported in PC tissues, its diagnostic value as a serum biomarker for PC remains unexplored.
AIM
To assess the diagnostic value of serum S100A6, alone and combined with CA19-9 and carcinoembryonic antigen (CEA), for detecting PC, particularly early-stage disease.
METHODS
Serum samples from 414 subjects were analyzed: 301 PC patients (149 early-stage, 152 advanced-stage), 52 with chronic pancreatitis (CP) patients, and 61 healthy controls (HCs). Serum S100A6 was measured by enzyme-linked immunosorbent assay, while CA19-9 and CEA by electrochemiluminescence immunoassay. Logistic regression was used to adjust for covariates, and diagnostic performance was assessed using receiver operating characteristic curves. Internal validation was conducted using 2000 bootstrap resamples.
RESULTS
Serum S100A6 concentrations were significantly higher in both early- and late-stage PC compared with CP and HCs (P < 0.001). In multivariate logistic regression, only S100A6 [odds ratio (OR) = 1.82, 95% confidence interval (CI): 1.40-2.37, P < 0.001] and CA19-9 (OR = 2.21, 95%CI: 1.80-2.72, P < 0.001) remained independent predictors of PC. CA19-9 showed the highest area under the curve (AUC) among single markers (0.839, 95%CI: 0.802-0.877), but combined panel of S100A6, CA19-9, and CEA achieved a higher AUCs of 0.868 (95%CI: 0.834-0.902, P = 0.007). Critically, for distinguishing early-stage PC from CP, the triple panel of S100A6, CA19-9, and CEA achieved an AUC of 0.821 (95%CI: 0.763-0.879) and was the only panel to show a statistically significant improvement over CA19-9 alone (P = 0.017), underscoring its potential for more timely identification of surgically resectable PC.
CONCLUSION
This large cohort study is the first to demonstrate the diagnostic potential of serum S100A6 in PC. When combined with CA19-9, serum S100A6 improves diagnostic performance for both overall and early-stage PC, supporting its role as a complementary biomarker in clinical practice.
Core Tip: This large-scale case-control study is the first to evaluate the clinical potential of serum S100A6 as a biomarker for pancreatic cancer (PC). When used together with carbohydrate antigen 19-9 (CA19-9), serum S100A6 significantly improved diagnostic accuracy over CA19-9 alone. Importantly, the triple panel including S100A6, CA19-9, and carcinoembryonic antigen significantly improved the differentiation of early-stage PC from chronic pancreatitis. These findings suggest that serum S100A6 may contribute to earlier and more accurate diagnosis of PC. Such improvement could enable timely identification of patients eligible for curative resection. Additional validation in independent and diverse patient populations is warranted.
Citation: Bae GE, Kim SM, Lee JK, Lee SY. First large-scale prospective evaluation of serum S100A6 as a complementary biomarker for early pancreatic cancer detection. World J Gastroenterol 2026; 32(21): 116581
Pancreatic cancer (PC) remains one of the most lethal malignancies worldwide, with stage at diagnosis being the most important determinant of survival[1,2]. Early-stage PC typically presents with non-specific symptoms or remains asymptomatic, underscoring the critical need for effective early screening methods[3,4]. Currently, carbohydrate antigen 19-9 (CA19-9) is the only Food and Drug Administration-approved biomarker for PC; however, it shows limited sensitivity for early-stage disease and yields false-positive results in benign conditions and non-pancreatic malignancies[5-7]. Advanced imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and endoscopic ultrasound are used for surveillance in high-risk individuals but remain impractical for population-wide screening due to high cost and limited sensitivity for early-stage PC[3,8,9].
The S100A6 protein, a Ca2+-binding protein in the S100 family, has shown elevated expression in PC tissues and cytology samples compared with non-cancerous tissues, with expression progressively increasing during carcinogenesis[10-14]. Despite extensive tissue-based studies, its diagnostic potential in blood - a more clinically accessible sample - remains unexplored[15]. To address this gap, we evaluated serum S100A6 as a diagnostic biomarker for PC in a large clinical cohort, both as a standalone marker and in combination with CA19-9 and carcinoembryonic antigen (CEA), with particular focus on early-stage PC detection. To our knowledge, this is the first study to investigate the diagnostic value of serum S100A6 in PC.
MATERIALS AND METHODS
Study design and population
This was a case-control study utilizing prospectively collected samples (2008-2015) analyzed retrospectively for biomarker evaluation in 2023. From June 2008 to April 2015, samples from patients diagnosed with PC or chronic pancreatitis (CP) and from healthy controls (HCs) were prospectively collected at Samsung Medical Center (Seoul, South Korea). The flow chart of the study population is shown in Supplementary Figure 1.
Among 437 subjects initially enrolled, 23 were excluded due to indeterminate or missing diagnoses or tumor-node-metastasis (TNM) classifications. The final cohort included 414 subjects: 149 with early-stage PC (9 stage I and 140 stage II), 152 with advanced-stage PC (53 stage III and 99 stage IV), and 113 controls (52 with CP and 61 HCs).
In this study, early-stage PC was defined as resectable disease, corresponding to stages I and II according to the TNM classification, whereas advanced-stage disease included stages III and IV tumors. Diagnoses of PC were based on histopathological examination, fine-needle aspiration cytology, and/or endoscopic findings (including endoscopic retrograde cholangiopancreatography and/or endoscopic ultrasound), together with relevant clinical information. Cases were staged according to the TNM classification: Stage I, tumor confined to the gland; stage II, local extension without nodal involvement; stage III, regional lymph node involvement; and stage IV, distant metastasis. CP was diagnosed based on characteristic findings on CT and MRI, in conjunction with pancreatic function tests, including serum amylase and lipase concentrations. HCs were recruited from visitors to the preventive healthcare center at Samsung Medical Center without clinical or biochemical evidence of disease, and were matched to PC patients by age and sex. Demographic and clinical data, including age, sex, histology, tumor stage, smoking status, and comorbidities, were obtained from medical records. CA19-9 and CEA concentrations measured at the time of sample collection were also retrieved from medical records.
Ethical approval
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Samsung Medical Center (No. SMC IRB 2021-02-153), with informed consent waived.
Sample collection and storage
Samples from PC and CP patients were obtained prior to any therapeutic interventions, including surgery, chemotherapy, and radiotherapy. All serum samples were centrifuged immediately after collection at 2330 × g for 5 minutes, aliquoted, and stored at -70 °C until analysis.
Laboratory analysis
For S100A6 measurements, samples underwent a single thaw cycle at room temperature immediately before analysis. Serum S100A6 concentrations were measured by a sandwich enzyme-linked immunosorbent assay using a human S100A6 ELISA kit (Cat. No. NBP3-06952, Novus Biologicals, LLC, Centennial, CO, United States) according to the manufacturer’s instructions. All analyses were done between June and September 2023 in three batches. CA19-9 and CEA had been measured on the day of sample collection by electrochemiluminescence immunoassay using Elecsys® CA19-9 and CEA (Roche Diagnostics, Mannheim, Germany) on the Roche Cobas e 801 analyzer (Roche Diagnostics) at Samsung Medical Center, a College of American Pathologists-accredited laboratory. Assay platforms and protocols remained unchanged throughout the sample collection period, with acceptable internal quality control and external proficiency test results throughout. The manufacturer’s cutoff values for CA19-9 and CEA were 37 U/mL and 4.7 ng/mL, respectively.
Statistical analysis
Categorical variables were compared using χ2 or Fisher’s exact tests, and non-normally distributed numerical variables were compared using the Kruskal-Wallis test. Associations between serum markers and clinical outcomes (PC vs non-PC) were assessed with univariate and multivariate logistic regression, and results were reported as regression coefficients with 95% confidence intervals (CIs). Differences in serum S100A6 levels among the four groups were first assessed using the Kruskal-Wallis test. When a significant overall difference was observed, post hoc pairwise comparisons were performed using Scheffe’s test to identify which specific groups differed. For binary comparisons (PC vs non-PC), Wilcoxon rank-sum tests were applied. To assess potential temporal effects, biomarker concentrations were compared across three collection periods within each diagnosis group (HC, CP, early-stage PC, and late-stage PC) using the Kruskal-Wallis test. Additionally, the association between storage duration and serum S100A6 concentrations was evaluated using Spearman’s rank correlation test. Diagnostic performance was assessed with receiver operating characteristic (ROC) curves. Area under the curves (AUCs) were compared using DeLong’s method. For each biomarker panel, a logistic regression-based signature was derived and predictive performance was evaluated by AUC. Internal validation was performed using 2000 bootstrap resamples of equal size to the original dataset. A P value < 0.05 was considered statistically significant. Analyses were performed using SAS version 9.4 and R version 4.1.3.
RESULTS
Baseline clinical and laboratory characteristics
Baseline characteristics of the four groups are summarized in Table 1. Age, sex, smoking status, and the presence of other malignancies did not differ significantly across groups, and cancer histology was similar between early and late PC. In contrast, diabetes mellitus (DM) was more frequent in PC patients, particularly in the early PC group, while absent in HCs (P < 0.001). Median serum levels of S100A6, CA19-9, and CEA were significantly different across the four groups (all P < 0.001). To address potential temporal effects from the extended collection period, we compared the biomarker concentrations across three collection periods within each diagnostic group. While no significant temporal variation was observed for S100A6 and CA19-9 in any diagnostic group (all P > 0.05; Supplementary Figure 2), a statistically significant difference was noted for CEA among HC (median 1.05 ng/mL vs 2.00 ng/mL vs 1.27 ng/mL, P = 0.02); however, all values remained well below the cutoff (4.7 ng/mL), indicating no clinically meaningful temporal variation. Additionally, storage duration did not correlate with S100A6 concentration (Spearman’s rho = -0.083, P = 0.078).
Table 1 Baseline demographic and laboratory characteristics, n (%)/medians (interquartile range).
Median serum S100A6 concentrations were significantly higher in both early PC [61.4 ng/mL, interquartile range (IQR): 18.8-114.2] and late PC (43.6 ng/mL, IQR: 19.7-101.3) compared with CP (22.9 ng/mL, IQR: 11.9-41.9) and HC (21.2 ng/mL, IQR: 13.3-37.0) (all P < 0.001) (Figure 1). In contrast, there was no significant difference between HC and CP (P = 0.914) or between early and late PC (P = 0.930).
Figure 1 Serum S100A6 levels were compared among healthy controls, patients with chronic pancreatitis, and patients with pancreatic cancer stratified by stage (stage I/II and stage III/IV).
Data are presented as box plots showing the median and interquartile range, with whiskers indicating the range. Overall differences among the four groups were assessed using the Kruskal-Wallis test. When a significant overall difference was observed, post hoc pairwise comparisons were performed using Scheffé’s test based on ranks. For binary comparisons between pancreatic cancer and non-pancreatic cancer groups, Wilcoxon rank-sum tests were applied. Median values and interquartile ranges are shown below the plot. IQR: Interquartile range.
Logistic regression analysis for distinction of PC from control groups
The results of binary logistic regression distinguishing PC patients from CP patients and HCs are shown in Table 2. Known risk factors for PC were included as covariates: Age, sex, underlying DM, smoking status, and presence of other malignancies[16,17]. In univariate analysis, serum S100A6, CA19-9, CEA, and DM were significantly associated with PC. However, in multivariate analysis adjusting for covariates, only serum S100A6 [odds ratio (OR) = 1.82, 95%CI: 1.40-2.37, P < 0.001] and CA19-9 (OR = 2.21, 95%CI: 1.80-2.72, P < 0.001) remained independent determinants. This indicates that the association between S100A6 and PC was not confounded by DM. Although DM (OR = 1.85, 95%CI: 0.94-3.63, P = 0.075) and CEA (OR = 1.52, 95%CI: 0.51-4.55, P = 0.454) showed elevated odds ratios, their associations were not statistically significant.
Table 2 Univariate and multivariate binary logistic regression analysis for distinguishing pancreatic cancer from chronic pancreatitis and healthy controls.
The ROC curves of S100A6 for distinguishing early and/or late PC from various combinations of non-PC states are shown in Figure 2. The highest AUC (0.697; 95%CI: 0.634-0.760) was observed for detecting early PC from the entire control group (HCs + CP patients), with a comparable AUC (0.692; 95%CI: 0.640-0.744) for detecting all stages of PC from the same control group (HCs + CP patients). Comparisons of the ROC curves of S100A6 with those of other single biomarkers (CA19-9 and CEA) were performed to distinguish all stages of PC from HC or CP status (Figure 3A). The AUCs of S100A6, CA19-9, and CEA were 0.692 (95%CI: 0.640-0.744), 0.839 (95%CI: 0.802-0.877), and 0.637 (95%CI: 0.581-0.692), respectively, with CA19-9 showing a significantly higher AUC than S100A6 (P < 0.001) and S100A6 demonstrating a higher AUC than CEA (P = 0.150) for detecting PC.
Figure 2 Receiver operating characteristic curves for distinguishing pancreatic cancer from control groups.
Areas under the curves with 95% confidence intervals are shown for the original dataset and internal validation, and areas under the curve comparisons were performed using DeLong’s test. EPC: Early pancreatic cancer; LPC: Late pancreatic cancer; H: Healthy control; CP: Chronic pancreatitis.
Figure 3 Receiver operating characteristic curves for single biomarkers (S100A6, carbohydrate antigen 19-9, and carcinoembryonic antigen) in the diagnosis of pancreatic cancer.
A: Receiver operating characteristic curves comparing early and late pancreatic cancer (early pancreatic cancer + late pancreatic cancer) with all control subjects, including healthy controls and patients with chronic pancreatitis (healthy control + chronic pancreatitis); B: Receiver operating characteristic curves comparing early pancreatic cancer with chronic pancreatitis. Areas under the curve with 95% confidence intervals are shown for both the original dataset and internal validation. Comparisons of areas under the curves between S100A6 and the other biomarkers were performed using DeLong’s test. aP < 0.05. CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen; EPC: Early pancreatic cancer; LPC: Late pancreatic cancer; H: Healthy control; CP: Chronic pancreatitis; AUC: Areas under the curve.
Notably, we compared the AUCs in detecting early-stage PC in patients with CP. For individual markers, the AUCs were 0.670 (95%CI: 0.591-0.750) for S100A6, 0.770 (95%CI: 0.705-0.835) for CA19-9, and 0.561 (95%CI: 0.473-0.648) for CEA, respectively (Figure 3B). Although the AUC of CA19-9 (0.770) was numerically higher than that of S100A6 (0.670), this difference did not reach statistical significance (P = 0.053). We analyzed the sensitivity of serum S100A6 compared to serum CA19-9 and CEA using ROC curve analysis (Supplementary Table 1). At fixed specificities of 90% and 80%, serum CA19-9 consistently demonstrated the highest sensitivities for both early and overall PC, confirming its role as the most reliable single marker. S100A6 showed lower sensitivity than CA19-9 but consistently outperformed CEA. In particular, for early PC vs CP at 80% specificity, S100A6 achieved a sensitivity of 56.4%, which was slightly lower than CA19-9 (62.4%) but clearly higher than CEA (29.5%). These results indicate that although S100A6 alone cannot surpass CA19-9, it retains modest diagnostic value and may serve as a complementary biomarker, especially for early-stage disease.
To further assess the clinical utility of S100A6, we compared its diagnostic performance at its optimized cutoff based on the maximum Youden index, with those of CA19-9 and CEA at their respective conventional cutoff values (Supplementary Table 2). The positive likelihood ratios of S100A6 ranged from 2.75 to 5.72 across comparisons, with positive predictive values of 88.2% to 94.1%. Notably, in early-stage PC comparisons (early-stage PC vs H + CP and early-stage PC vs CP), the positive likelihood ratios of S100A6 (5.72 and 3.18, respectively) were comparable to or greater than those of CA19-9 (4.89 and 2.42, respectively) and were substantially greater than those of CEA (1.76 and 1.04, respectively). Nevertheless, the negative likelihood ratios of S100A6 ranged from 0.54 to 0.62, with negative predictive values of 22.8% to 58.2%, which were consistently inferior to those of CA19-9 (negative likelihood ratios: 0.37-0.48) but superior to those of CEA (0.81-0.99) in all comparisons. Collectively, these findings suggest that S100A6 may have greater clinical utility for ruling in early-stage PC rather than ruling out disease, supporting its role as a complementary diagnostic marker.
Diagnostic performance of combined biomarker sets including S100A6
The ROC curves of the combined biomarker panels are shown in Figure 4. Among these, both the dual panel of S100A6 plus CA19-9 and the triple panel of S100A6, CA19-9, and CEA demonstrated the highest AUC of 0.868 (95%CI: 0.834-0.902), whereas the dual panel of CA19-9 and CEA showed an AUC of 0.842 (95%CI: 0.804-0.879) (Figure 4A). Compared with CA19-9 alone, which had the highest AUC among single markers, both panels including S100A6 achieved significantly higher AUCs (P = 0.007). In contrast, the dual panel of CA19-9 and CEA did not differ significantly from CA19-9 alone (P = 0.582). Notably, there was no significant difference between the triple panel (S100A6, CA19-9, and CEA) and the dual panel (S100A6 and CA19-9) (P = 0.820). For detecting early-stage PC from the patients with CP, the triple panel achieved the highest AUC of 0.821 (95%CI: 0.763-0.879) and was the only panel that performed significantly better than CA19-9 alone (Figure 4B).
Figure 4 Receiver operating characteristic curves for combined biomarker sets in the diagnosis of pancreatic cancer.
A: Receiver operating characteristic curves comparing early and late pancreatic cancer (early pancreatic cancer + late pancreatic cancer) with all control subjects, including healthy controls and patients with chronic pancreatitis (healthy control + chronic pancreatitis); B: Receiver operating characteristic curves comparing early pancreatic cancer with chronic pancreatitis. Areas under the curve with 95% confidence intervals are shown for both the original dataset and internal validation. Comparisons of areas under the curves between single carbohydrate antigen 19-9 and the combined biomarker sets were performed using DeLong’s test. aP < 0.05. CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen; EPC: Early pancreatic cancer; LPC: Late pancreatic cancer; H: Healthy control; CP: Chronic pancreatitis; AUC: Areas under the curve.
Internal validation
Internal validation was performed using bootstrap resampling with the same sample size as the original dataset. The validated models demonstrated predictive performances that were comparable to those obtained from the original data. As shown in Figures 2, 3 and 4, the AUC values of both single and combined biomarker signatures remained stable after validation, indicating the robustness and reproducibility of our findings.
DISCUSSION
S100A6, also known as calcyclin, is a member of the S100 protein family and functions as a calcium-binding protein. The expression of S100A6 mRNA is increased in PC tissue compared to non-cancerous tissue[11]. Using GEPIA2 analysis of The Cancer Genome Atlas and Genotype-Tissue Expression datasets, the median transcripts per million of S100A6 was markedly higher in pancreatic adenocarcinoma (5221.69) compared with normal pancreatic tissue (53.89)[18]. Considering this, S100A6 has been studied as a biomarker for the diagnosis and prognosis of PC in tissue or pancreatic juice[19]. However, to the best of our knowledge, no study has assessed serum S100A6 as biomarker for PC[13]. To date, serum S100A6 has only been reported as a diagnostic marker for bladder cancer[20], non-small-cell lung cancer[21], and epithelial ovarian cancer[22], as well as a prognostic marker for cholangiocarcinoma[23] and gastric cancer[24]. This is the first study that demonstrated the potential of serum S100A6 as a diagnostic marker for PC.
PC is one of the cancers in which symptoms appear late and are vague, making early diagnosis extremely challenging[25]. Therefore, the development of tumor markers for PC screening is crucial for the favorable prognosis of patients. In particular, if early detection through blood tests before adopting invasive diagnostic methods such as tissue biopsy is possible, it would be very useful in real world practice. Furthermore, tissue biopsy poses risks of misinterpretation due to variations according to biopsy site and factors such as desmoplastic reactions, including inflammation and fibrosis[26]. Several studies have reported various novel serum biomarkers as candidate markers for diagnosing PC, but there is no known better biomarker than serum CA19-9[27,28]. Unfortunately, serum S100A6 also showed inferior overall diagnostic performance compared to CA19-9 as a single marker in this study. However, when used in combination with CA19-9, its performance improved. Our findings showed that a combined panel of serum S100A6 and CA19-9 had superior predictive performance compared to using CA19-9 alone and even outperformed the combination of CA19-9 and CEA commonly used in current clinical practice[29]. Therefore, this study highlights the utility of S100A6 as a promising complementary biomarker to CA19-9 for diagnosis of PC.
In addition, we demonstrated that serum S100A6 can be especially useful for the early diagnosis of PC in patients with CP. CA19-9 is less effective for distinguishing early PC from CP because elevated CA19-9 can be present in benign conditions such as CP and non-malignant jaundice[30-33]. In this study, we found that CA19-9 alone is not significantly superior to S100A6, particularly for detecting early PC in CP patients. Instead, the triple panel of S100A6, CA19-9, and CEA showed the best predictive performance compared to any other panel. This result suggests that S100A6 has the potential to serve as a complementary marker to overcome the limitations of CA19-9 in the early diagnosis of PC. However, it should be noted that the “early-stage” group in this study primarily represents resectable disease (stage I-II), with a limited number of stage I cases (n = 9). Therefore, the observed complementary role of S100A6 should be interpreted mainly in the context of distinguishing resectable PC from CP.
Obviously, CA19-9 is currently the best single marker for diagnosing PC; however, its low sensitivity still poses challenges in PC diagnosis[6,7]. In line with previous reports, our study also confirmed that, as individual markers, CA19-9 demonstrated superior sensitivity compared to S100A6, while S100A6 showed better sensitivity than CEA. A well-known limitation of CA19-9 is its dependence on Lewis antigen status, with approximately 5%-10% of the population is Lewis antigen-negative and therefore at the risk of false-negative CA19-9 results[34]. Because S100A6 expression is not related to glycan synthesis or Lewis antigen-dependent pathways[11], it could provide complementary diagnostic information in patients with low or undetectable CA19-9 levels. However, since the Lewis antigen status was not evaluated in this study, stratified analysis according to Lewis antigen status was not possible. Thus, we cannot attribute the complementary role of serum S100A6 to compensation for Lewis antigen-negative cases in the false negative CA19-9 results. Nevertheless, our results did reveal significant patterns. In our cohort, 32% (96/301) of PC patients had CA19-9 concentrations below the conventional diagnostic cutoff (< 37 U/mL) and thus were not identified by CA19-9 alone. Of these CA19-9 false-negative cases, 41% (39/96) were positive for serum S100A6 using the study-defined cutoff by Youden index (≥ 60.2 ng/mL), indicating that S100A6 may partially compensate for this limitation of CA19-9 by Lewis antigen status. To determine whether S100A6-positive/CA19-9-negative PC patients represent a distinct subgroup of patients, we compared their clinicopathological features with those of CA19-9-positive patients. However, no statistically significant differences were found in terms of age, sex, histological type, TNM stage, presence of DM or other malignancies, or smoking status (Supplementary Table 3), indicating that the complementary diagnostic role of S100A6 is unlikely to be driven by overt clinicopathological differences. Future studies incorporating stratification based on Lewis antigen status will be required to clarify the complementary role of S100A6.
Correlation analysis between CA19-9, CEA, and S100A6 showed a weak correlation between CA19-9 and S100A6 (r = 0.18, P < 0.001) (Supplementary Table 4), suggesting that S100A6 may reflect a biological mechanism distinct from that of CA19-9. S100A6 and CA19-9 are both overexpressed in PC, yet they represent fundamentally different classes of biomarkers with distinct biological roles. CA19-9 is an extracellular carbohydrate antigen that primarily reflects tumor-associated aberrant glycosylation and mucin expression, and therefore is a passive indicator of tumor burden. However, its serum levels are substantially influenced by biliary obstruction and Lewis antigen status, limiting its diagnostic reliability[35]. In contrast, S100A6 is an intracellular calcium-binding oncoprotein that actively participates in pancreatic carcinogenesis rather than merely reflecting disease presence. S100A6 expression progressively increases from early pancreatic intraepithelial neoplasia to invasive PC, driven in part by transcriptional activation through the Hedgehog-GLI1 signaling pathway[36]. S100A6 also forms a positive feedback loop with the Wnt/β-catenin pathway, one of the most important signaling pathways involved in epithelial-mesenchymal transition by upregulating the expression of β-catenin, N-cadherin, and vimentin while downregulating E-cadherin expression, thereby promoting epithelial-mesenchymal transition[37,38]. This mechanistic difference between S100A6 and CA19-9 supports its potential to identify PC that may be missed by CA19-9 alone.
Although serum S100A6 concentrations were significantly higher in patients with PC than in control groups, no clear association was observed between S100A6 levels and advancing tumor stage (Supplementary Figure 3). This distinct pattern contrasts with CA19-9, which tends to rise predominantly in advanced-stage PC and therefore suffers from reduced sensitivity in early detection[6,29,31]. Regarding stage-independent expression, previous tissue-based studies in PC have reported findings that overlap with our serum results. In an analysis of publicly available transcriptomic data, S100A6 mRNA expression in PC tissue was not significantly associated with N or M stage, although higher levels were observed in T3-4 compared with T1-2 tumors, suggesting that S100A6 may increase with local tumor extent but does not show a linear relationship with overall TNM stage[39]. Similarly, immunohistochemical studies of PC tissue have found no significant association between nuclear S100A6 expression and TNM stage[10]. In contrast, in several other malignancies, S100A6 expression has been reported to correlate with advancing stages. In colorectal cancer, S100A6 expression was significantly associated with Duke’s stage and lymphatic invasion[40,41], and in gastric cancer[42], high S100A6 expression was linked to deeper wall invasion, lymph node metastasis, liver metastasis, higher TNM stage, and poorer clinical outcomes. These findings suggest that the relationship between S100A6 and tumor stage is cancer-type specific and may depend on distinct biological contexts. The absence of a stage-dependent increase suggests that S100A6 may provide complementary value by capturing early-stage disease that CA19-9 often misses. Indeed, in our cohort, CA19-9 did not significantly outperform S100A6 in differentiating early PC from CP, and the triple combination of S100A6, CA19-9, and CEA was the only panel that showed statistically significant improvement compared with CA19-9 alone. These findings imply that while S100A6 may have limited utility as a prognostic marker of tumor burden or progression, its stage-independent expression profile could enhance early diagnostic strategies, particularly when incorporated into combined biomarker panels.
Regarding clinical implications, serum S100A6 is best positioned as a complementary, rule-in biomarker rather than a stand-alone screening test for PC. Given its favorable positive likelihood ratios and higher positive predictive values, S100A6 may help identify patients with CP who warrant further evaluation when used alongside CA19-9. Importantly, S100A6 provides additive diagnostic value in patients with low or undetectable CA19-9 levels, such as those that are potentially Lewis antigen-negative. Consistent with this, a combined panel of S100A6 and CA19-9 (with or without CEA) significantly improved diagnostic performance compared with CA19-9 alone, particularly for distinguishing resectable PC from CP.
However, these findings should be interpreted within the appropriate clinical context. S100A6 is not specific to PC and has been reported in other malignancies, including bladder, lung, ovarian, cholangiocarcinoma, gastric cancers, and even thyroid cancer[20-24,39]. Therefore, its clinical utility in this study should be interpreted as that of a complementary biomarker to CA19-9 rather than a pancreas-specific marker.
Several limitations should be considered when interpreting our results. First, despite internal validation, the diagnostic performance of serum S100A6 was not confirmed in an independent external cohort, thereby limiting the generalizability of our findings. As a result, the applicability of our results to other populations and clinical settings remains uncertain. However, our results provide a strong rationale for future external validation studies, and validation in multi-center cohorts may facilitate the clinical translation of serum S100A6 as a complementary diagnostic biomarker. Second, the study was conducted in a single center, and longitudinal follow-up data were lacking to determine whether S100A6 could serve as a prognostic or monitoring biomarker. Moreover, as a tertiary hospital, our institution may have preferentially enrolled patients with more symptomatic disease, higher comorbidity burden, or more complex diagnostic profiles, which could have introduced selection bias and may limit the generalizability of our findings to broader screening or community-based populations. Third, Lewis antigen phenotyping or genotyping could not be performed because no residual specimens were available, which prevented us from determining whether the CA19-9 false-negative results were attributable to Lewis antigen negativity. Consequently, some CA19-9 negative cases in our cohort may have been attributable to Lewis-negative status. Although S100A6 identified a substantial proportion of CA19-9 negative PC cases, the absence of Lewis antigen data precludes definitive interpretation of its complementary role. Fourth, cost-benefit analysis of combined S100A6 and CA19-9 testing was not performed. Future multicenter studies with larger, independently validated cohorts, Lewis antigen assessment, longitudinal follow-up, and cost-benefit analyses are warranted to confirm the clinical utility and economic feasibility of adding serum S100A6 as a complementary biomarker to CA19-9 in PC diagnosis.
CONCLUSION
This study provides the first evaluation of serum S100A6 as a biomarker for PC. While CA19-9 remains the most reliable single marker, the combination of serum S100A6 and CA19-9 demonstrated superior predictive performance compared with CA19-9 alone and with the CA19-9 and CEA combination. This combined approach was especially effective for distinguishing early-stage PC from CP. Further validation in larger, independent cohorts is required to confirm its diagnostic utility and define patient subgroups most likely to benefit from inclusion of S100A6 as a complementary biomarker.
Fahrmann JF, Schmidt CM, Mao X, Irajizad E, Loftus M, Zhang J, Patel N, Vykoukal J, Dennison JB, Long JP, Do KA, Zhang J, Chabot JA, Kluger MD, Kastrinos F, Brais L, Babic A, Jajoo K, Lee LS, Clancy TE, Ng K, Bullock A, Genkinger J, Yip-Schneider MT, Maitra A, Wolpin BM, Hanash S. Lead-Time Trajectory of CA19-9 as an Anchor Marker for Pancreatic Cancer Early Detection.Gastroenterology. 2021;160:1373-1383.e6.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 175][Cited by in RCA: 166][Article Influence: 33.2][Reference Citation Analysis (7)]
Vimalachandran D, Greenhalf W, Thompson C, Lüttges J, Prime W, Campbell F, Dodson A, Watson R, Crnogorac-Jurcevic T, Lemoine N, Neoptolemos J, Costello E. High nuclear S100A6 (Calcyclin) is significantly associated with poor survival in pancreatic cancer patients.Cancer Res. 2005;65:3218-3225.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 93][Cited by in RCA: 97][Article Influence: 4.6][Reference Citation Analysis (0)]
Wei BR, Hoover SB, Ross MM, Zhou W, Meani F, Edwards JB, Spehalski EI, Risinger JI, Alvord WG, Quiñones OA, Belluco C, Martella L, Campagnutta E, Ravaggi A, Dai RM, Goldsmith PK, Woolard KD, Pecorelli S, Liotta LA, Petricoin EF, Simpson RM. Serum S100A6 concentration predicts peritoneal tumor burden in mice with epithelial ovarian cancer and is associated with advanced stage in patients.PLoS One. 2009;4:e7670.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 26][Cited by in RCA: 42][Article Influence: 2.5][Reference Citation Analysis (0)]
Chen B, Zheng D, Liu C, Bhandari A, Hirachan S, Shen C, Mainali S, Li H, Jiang W, Xu J, Zhang X, Tang K, Zhang W. S100A6 promotes the development of thyroid cancer and inhibits apoptosis of thyroid cancer cells through the PI3K/AKT/mTOR pathway.Pathol Res Pract. 2023;242:154325.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 9][Reference Citation Analysis (0)]
Komatsu K, Andoh A, Ishiguro S, Suzuki N, Hunai H, Kobune-Fujiwara Y, Kameyama M, Miyoshi J, Akedo H, Nakamura H. Increased expression of S100A6 (Calcyclin), a calcium-binding protein of the S100 family, in human colorectal adenocarcinomas.Clin Cancer Res. 2000;6:172-177.
[PubMed] [DOI]
