Category: 3. Business

Introduction

Placenta accreta spectrum (PAS) is a serious obstetric complication usually associated with abnormal attachment or invasion of the placenta into the myometrium,¹ which may lead to life-threatening postpartum hemorrhage and hysterectomy.² Indeed, placental implantation spectrum disorders include several types: placental adhesion (anchoring villi attached to the superficial myometrium without insertion into the meconium), placental implantation (infiltration of placental villi into the myometrium), and penetrative placental implantation (anchoring villi tissue penetrating the entire uterine wall and even reaching surrounding organs).³ In recent years, the incidence of PAS has risen significantly with the rising rate of cesarean section and the increasing age of pregnant women.⁴ In order to accurately predict PAS and its associated risks in the prenatal period, researchers have developed various magnetic resonance imaging (MRI)-based prediction models. For example, Maurea et al proposed a model that utilizes ultrasound and MRI features (eg, loss of posterior interstitial space, intra-placental dark bands, and focal interruption of myometrial borders) to predict the risk of PAS in patients with placenta previa.⁵ Andrea et al developed an MRI-based scoring system that, by evaluating indicators such as T2 dark bands, uterine myometrial thinning, and abnormal vascular distribution, has further improving the diagnostic accuracy of PAS.⁶

However, most of these models do not adequately consider the morphological characteristics of the cervix and placenta. In recent years, it has been shown that placental volume and cervical length are closely associated with the occurrence of PAS and the risk of postpartum hemorrhage. In a study by Yue et al it was found that patients with smaller cervical areas and shorter cervical lengths had significantly more intraoperative hemorrhage and a higher incidence of adverse pregnancy outcomes in the presence of a complete placenta previa.⁷ And a study by Yue et al further confirmed that increased placental volume and T2 dark band volume were positively associated with the risk of major bleeding in patients with PAS.⁸ These findings suggest that morphological characteristics of the placenta and cervix are valuable in risk assessment of PAS.

Based on these studies, this study innovatively combined cervical volume, placental volume, and cervical length to construct a novel PAS risk prediction model. By quantifying these morphological indicators, we aimed to provide clinicians with a more accurate prenatal prediction tool to help them better assess the likelihood of PAS in patients with complete placenta previa who have a history of cesarean delivery. The establishment of this model not only fills the gap of previous studies in cervical and placental morphology, but also provides new ideas for early diagnosis and risk stratification of PAS, which is of great clinical significance.

Materials and Methods

This study was reviewed by the Ethics Committee of Suzhou Hospital of Nanjing Medical University (Approval K-2022-015-K01). Informed consent was waived because this anonymously selected study was retrospective and no new interventions were performed on patients. We reviewed the clinical data of pregnant women with complete placenta previa from January 2018 to August 2024 who had a history of cesarean delivery. Inclusion criteria were: pregnant women with MRI-confirmed diagnosis of complete placenta previa. Exclusion criteria were: (1) twin or multiple pregnancies, (2) no previous history of cesarean section, (3) no pelvic MRI, (4) incomplete clinical or surgical data, and (5) poor image quality affecting observation. The flow chart of the study design is shown in Figure 1. The reason for this design is that MRI is a key tool in the diagnosis of complete placenta previa and placenta implantation spectrum disorders (PAS), providing clearer images than ultrasound and ensuring diagnostic accuracy. By requiring all study subjects to undergo MRI, bias introduced by inconsistent diagnostic tools can be avoided, ensuring homogeneity and reliability of the study data. Ultimately, a total of 157 patients were included in the study, and the diagnosis of PAS was based on the 2019 FIGO criteria.

Figure 1 Flow chart of PAS patients with complete placenta previa included in the study.

MRI scans were performed at 3t (Siemens Medical Solutions, Erlangen, Germany) without gadolinium and keeping the bladder partially filled for optimal evaluation of the bladder-plasma membrane interface. Most patients were examined in the supine position, and a few patients who could not tolerate the supine position were examined in the left lateral position. MRI image acquisition employs a T2-weighted half-Fourier single-pass turbo spin-echo sequence to obtain axial, sagittal, and coronal images covering the entire uterus.

Cervical volume, placental volume, and cervical length were measured by three radiologists with over 20 years of experience in obstetric and gynecologic MRI. Radiologists were unaware of all clinical information and other radiologists’ impressions. First, MRI images from patients with complete placenta previa were imported into 3D Slicer software (version 5.2.1, www.slicer.org) to create placental and cervical contours and measure placental and cervical volumes with coronal, sagittal, and axial views of the placenta as shown (Figure 2A–C). Subsequently, MRI images from patients with complete placenta previa were imported into ImageJ software version 1.50 (National Institutes of Health, Bethesda, USA) to measure cervical length. The measurement method involved lines a and b passing through the internal and external cervical canals, respectively, and perpendicular to the cervical canal. The shortest distance between these two lines represented the cervical length (Figure 2D). After segmentation in 3D Slicer software, its 3D reconstruction function generated three-dimensional models of the placenta and cervix. The main steps were as follows: (1) Import the patient’s original MRI images in DICOM format; (2) Run the Editor module in the 2D window; (3) Perform segmentation by manually tracing the external contours of the placenta and cervix on each slice; (4) Utilize the program’s 3D segmentation function to calculate the volumes of all placental and cervical voxels (Figure 2E and F).

Figure 2 Magnetic resonance imaging (MRI) and three-dimensional (3D) reconstruction of the placenta and cervix in a patient with placenta accreta spectrum (PAS). (A) Coronal T2-weighted MRI view of the placenta.(B) Sagittal T2-weighted MRI view of the placenta.(C) Axial T2-weighted MRI view of the placenta.(D) Measurement of cervical length (2.62 cm) on a sagittal T2-weighted MRI. Line a passes through the internal cervical os, line b passes through the external cervical os, and both are drawn perpendicular to the cervical canal. The double-headed arrow indicates the cervical length, defined as the shortest distance between the two lines.(E) 3D reconstruction model of the cervix, generated using 3D Slicer software.(F) 3D reconstruction model of the placenta, generated using 3D Slicer software.

Statistical Analysis

The normality of continuous variables was assessed using the Kolmogorov–Smirnov test. Normally distributed variables were expressed as mean ± standard deviation, and intergroup comparisons were performed using the independent samples t-test. Non-normally distributed variables were expressed as median (interquartile range), and intergroup comparisons were performed using the Mann–Whitney U-test. Categorical variables were expressed as case numbers (percentages). Inter-observer agreement in MRI feature assessment was measured using the Kappa coefficient. Pearson correlation analysis explored associations between cervical volume, placental volume, cervical length, and PAS. Receiver operating characteristic (ROC) curve analysis determined optimal cutoff values for MRI features predicting PAS, with corresponding sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) calculated. Sample size estimation was performed using PASS software (version 11.0.7). The current sample size of 157 cases achieves over 90% test power at an α=0.05 level. All statistical analyses were conducted using IBM SPSS Statistics software (version 23.0). P < 0.05 was considered statistically significant.

Results

In the current study, we performed transabdominal ultrasound and MRI on 157 women who had undergone at least one cesarean section and were confirmed to have complete placenta previa, all of whom had delivered at our healthcare facility. Of these, 72 had placenta accreta (PAS), while the remaining 85 did not. Table 1 details the clinical characteristics of these patients. The mean differences in continuous variables such as age and weight between the two groups were assessed using t-tests. Z-scores were employed to compare the distribution differences of continuous variables between the two groups. For categorical variables (eg, parity, surgical history), the differences in proportions between the two groups were evaluated using chi-square (χ²) tests. By comparing the general clinical data, we found that maternal age, body mass index (BMI), gestational age, number of deliveries, history of dilatation and curettage, history of cesarean section, gestational age as revealed by MRI, and perioperative hemorrhage did not differ significantly between the two groups (P > 0.05). This finding may be related to multiple factors. All patients in the study had a history of cesarean delivery, which means that the two groups were already similar in the baseline characteristic of history of cesarean delivery, and thus the number of cesarean deliveries may no longer be a key factor in distinguishing PAS from non-PAS in this particular population. Second, although the number of cesarean deliveries is an important risk factor for PAS, other factors such as placental volume, cervical volume, and cervical length may have played a more significant role in the development of PAS in this study. In addition, the limitations of the sample size may have led to a statistical failure to detect differences in the number of cesarean deliveries. Therefore, although a history of cesarean delivery is a known risk factor for PAS, other factors may have been more critical in the specific population of this study, resulting in a history of cesarean delivery that was not significantly different between the two groups. In addition, we performed a three-dimensional reconstruction using 3D Slicer software and calculated placental volume, cervical volume, and cervical length. Interobserver variability in MRI images was nearly consistent, with Kappa values for cervical length, cervical volume, and placental volume all exceeding 0.900 (see Table 2). Of particular importance, we found that the placental volume of PAS pregnant women was significantly greater than that of non-PAS pregnant women, while their cervical volume and cervical length were significantly smaller than those of non-PAS pregnant women, and these differences were statistically significant (P < 0.001), as detailed in Table 3.

Table 1 Clinical Characteristics of Study Participants

Table 2 Interobserver Reliability of Magnetic Resonance Imaging (MRI) in the Measurement of MRI Features

Table 3 MRI Signs Related to PAS in Caesarean Section

To explore in more depth the risk factors for the development of placental implantation disease (PAS) in patients with complete placenta previa with a history of cesarean delivery, we plotted two subject operating characteristic (ROC) curves (Figure 3). In this case, the dashed line demonstrates the role of three indicators, placental volume, cervical volume, and cervical length, in predicting the occurrence of PAS. From the ROC curve analysis, we identified optimal threshold values for placental volume (880.0 cm³), cervical volume (20.0 cm³), and cervical length (3.0 cm) (Figure 4), which are important for distinguishing between pregnant women who are likely to develop PAS and those who are unlikely to do so. Subsequently, we created a novel predictive model based on these thresholds. The model was designed to explore the complex relationship between cervical volume, placental volume, and cervical length and the likelihood of PAS in pregnancies with complete placenta previa with a history of cesarean delivery. To assess this relationship more visually, we further developed a scoring system (shown in Table 4). The design of the scoring system was based on the Odds Ratio (OR) value of each indicator, ie, their independent predictive value for the occurrence of PAS. The OR for cervical volume was 4.132, for cervical length was 2.875, and for placental volume was 2.076. These ORs reflect the strength of the association between each indicator and the occurrence of PAS, with a higher OR indicating a stronger prediction of PAS by that indicator. Thus, cervical volume was assigned a score of 2, while cervical length and placental volume had relatively low OR values and were assigned a score of 1 each. Based on the scores, patients were categorized into low-risk (0–1), intermediate-risk (2) and high-risk (3–4) groups, with 76 cases in the low-risk group and 13 cases (17.1%) with combined PAS disease, 33 cases in the intermediate-risk group with combined PAS (21 cases (63.6%)) and 47 cases in the high-risk group with combined PAS (38 cases (80.9%)), which shows that with the increase in the scores, the incidence of the probability of PAS disease increases (Figure 5). In order to verify the accuracy of the scoring system, we again plotted a ROC curve (solid line). This curve demonstrates the predictive effect of the score after scoring assignment on the occurrence of PAS in those with complete placenta previa with a history of cesarean delivery. The results showed that the AUC value of the dashed line (raw index) was 0.891, while the AUC value of the solid line (scoring system) was 0.902. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for predicting PAS based on individual signs are shown in Table 5. When combined with cervical length, cervical volume, and placental volume, the sensitivity, specificity, PPV, and NPV for predicting PAS were 87.305%, 83.946%, 82.163%, and 88.645%, respectively. When applying the scoring system to predict PAS, the sensitivity, specificity, PPV, and NPV were 88.723%, 84.209%, 82.663%, and 89.817%, respectively. The high agreement between the two curves indicates that the scoring system we developed has a high accuracy in predicting the occurrence of PAS. This provides a powerful tool for assessing the risk of PAS in patients with complete placenta previa who have a history of cesarean delivery.

Table 4 PAS Index in Complete Placenta Previa with Prior Cesarean (PIPs)

Table 5 Receiver Operating Characteristic Analysis for Prediction of PAS Based on Cervical Volume, Cervical Length, and Placental Volume

Figure 3 Receiver operator characteristic curves for the regression model with cervical length, cervical volume, and placental volume, and for prediction score. Dashed line AUC= 0.891. solid line AUC= 0.902.

Figure 4 Multivariate logistic regression analysis of risk factors for patients with PAS.

Figure 5 Simple scoring model evaluated for correspondence with PAS.

Discussion

Previous studies have shown that the risk of placenta previa (PP) is usually associated with previous cesarean section, increases with the number of previous CS, and is an independent risk factor for PAS.⁹ In addition, PP combined with scarred uterus significantly increases the risk of developing PAS.^10,11 Therefore it is crucial to accurately predict the likelihood of PAS in women with placenta previa with a history of cesarean section in the antenatal period, which enables adequate preoperative preparation, for example, by planning the delivery in a referral labor ward, where more necessary means (blood transfusion, interventional radiology, resuscitation and surgical skills, etc.) are available.¹² When women with PAS deliver in a tertiary referral center with an experienced multidisciplinary team, maternal mortality and complication rates associated with PAS are significantly reduced.^13–16

In recent years, several researchers have also created different ultrasound scoring systems to help clinicians effectively predict PAS and adverse clinical outcomes before delivery. Zou¹⁷ et al scored the number of previous cesarean deliveries, placental position, placental/uterine augmentation, placental heterogeneity, placental T2 dark bands, intra-placental vascular anomalies, placental bed vascular anomalies, loss of the T2 low-signal interface, interruption of the bladder wall, penetrating PI and muscle thinning and interruption to investigate whether magnetic resonance imaging can effectively predict the diagnosis of malignant placenta previa with or without PAS and adverse clinical outcomes. In addition, Tovbin.¹⁸ created a new ultrasound scoring system that scored the number and size of placental sockets; occlusion of the boundary between the uterus and the placenta; placental position; color Doppler signals within the placental sockets; vascular richness of the placenta-bladder and/or utero-placental interface area; and number of previous cesarean deliveries. It was found that all ultrasound criteria of the scoring system were significantly associated with pathologically adherent placenta (MAP) (P<0.001). This provides additional and stronger imaging evidence for the diagnosis of MAP. Correct prenatal diagnosis buys more time for the multidisciplinary team to plan the delivery, which will help to reduce surgical complications, maternal blood loss, and length of stay in the intensive care unit.^14,19,20

In contrast to previous studies, the scoring system created in this study incorporated cervical morphology, and we chose placental volume, cervical volume, and length as key predictors of PAS based primarily on their pathophysiologic relevance. Increased placental volume usually reflects hyperplasia of placental tissue, especially in pregnant women with a history of cesarean section, where damage to the endometrium and myometrium may lead to abnormal invasion of placental villi into the myometrium, which may in turn increase the risk of PAS. Yue et al showed that the of PAS in patients with a placental volume greater than 887 cm³ was 85.531%, and the specificity was 83.907%, which indicating that placental volume is an important predictor of PAS.²¹ Shorter cervical length and volume were negatively associated with the severity of PAS and the risk of hemorrhage. Shorter cervical length usually means that the placenta may have invaded the cervical region, leading to disruption of cervical structures and abnormal vascular distribution. Yue et al found that patients with cervical length less than 30 mm had a significantly increased risk of hemorrhage, and cervical length was negatively correlated with hemorrhage volume.⁷ Based on these studies, this study innovatively combined cervical volume, placental volume and cervical length to construct a new PAS risk prediction model.

Notably, the results of this study indicate that the proportion of anterior placenta was significantly higher in the PAS group than in the non-PAS group (62.5% vs 41.2%, P = 0.008). This difference may be related to the fact that uterine scars after cesarean section are often located on the anterior wall, making the placenta more likely to attach to this area. Despite differences in placental location distribution between groups, this model’s advantage lies in its independence from placental location as a single variable. Instead, it achieves precise disease risk assessment by directly capturing PAS-induced morphological end changes in the cervix and placenta—such as reduced cervical volume and increased placental volume. Consequently, this model demonstrates greater universality. Regardless of whether the placenta is located in the anterior or posterior wall, characteristic morphological alterations can be effectively identified, achieving high predictive performance (AUC = 0.902).

The International Society for Abnormal Invasive Placenta (IS-AIP) has proposed several criteria for MRI signs of PAS, including Focal exophytic mass, Myometrial thinning, Bladder wall interruption, Abnormal vascularization of the placental bed, etc.¹² These MRI signs suggested by IS-AIP are high risk signals for PAS. In contrast, the present study identified PAS by measuring placental volume, cervical volume, and cervical length, which provides new MRI signs for prenatal prediction of PAS. It is noteworthy that the scoring system achieves quantitative evaluation by assigning corresponding points to each parameter: cervical length (CL) < 3.0 cm receives 1 point, cervical volume (CV) < 20.0 cm³ receives 2 points, and placental volume ≥ 880.0 cm³ receives 1 point. Thus, pregnant women meeting all these criteria achieve the maximum score of 4 points. Based on this scoring, patients with 3–4 points belong to the high-risk group for PAS disease. For example, PA patients with a history of CS who have CL < 3.0, CV < 20.0, and PV ≥ 880.0 strongly suggest concomitant PAS disease. This approach reduces diagnostic subjectivity to some extent. Thus, this scoring system offers clinicians a novel approach for prenatal assessment of PAS occurrence probability.

Furthermore, numerous previous studies have demonstrated associations between molecular biomarkers such as Cripto-1, AFP, and PAPP-A with PAS and placenta previa. Serum Cripto-1 levels in patients with PAS were significantly higher than in those with PP but without pregnancy complications. Elevated AFP levels during mid-pregnancy independently predicted PAS requiring hysterectomy, while elevated PAPP-A correlated with PAS and postpartum hemorrhage volume. In the future, integrating biomarkers like Cripto-1, AFP, and PAPP-A with imaging morphological parameters into predictive models may further enhance the comprehensiveness and accuracy of PAS prediction.

This study has several limitations. First, its retrospective design carries a risk of selection bias, and the limited sample size prevented complete matching between the PAS and non-PAS groups. Second, discrepancies between specimen collection times and disease diagnosis times in some laboratories may have affected the accuracy of certain indicators. Third, while inter-observer agreement was assessed and demonstrated good reproducibility (Kappa > 0.9), intra-observer variability was not analyzed. Although all measurements were performed by uniformly trained radiologists, future studies may incorporate automated segmentation techniques to further enhance measurement efficiency and objectivity. Finally, placental position, as a potential confounding factor, requires further clarification regarding its independent contribution to PAS occurrence alongside cervical-placental morphological parameters. Subsequent prospective studies with larger samples should employ multivariate or stratified analyses to determine the independent predictive value of each parameter, thereby optimizing model structure and diagnostic performance.

Conclusions

Cervical length, cervical volume and placental volume are independent risk factors for having PAS disease in patients with complete placenta previa with a history of cesarean delivery. According to the scoring system in this study, the higher the score, the higher the risk of having PAS disease in patients with complete placenta previa with a history of cesarean delivery. This scoring system has the potential to be applied to predict the likelihood of having PAS in patients with complete placenta previa with a history of cesarean delivery, thus contributing to the prenatal selection of rational treatment.

Abbreviations

PAS, placental implantation spectrum; MRI, magnetic resonance imaging; FIGO, International Federation of Gynecology and Obstetrics; DICOM, digital imaging and communications in medicine; PPV, positive predictive value; NPV, negative predictive value; ROC, receiver operating characteristic; BMI, body mass index; NICU, neonatal intensive care unit; OR, odds ratio; PP, placenta previa; CS, caesarean section; MAP, morbidly adherent placenta; IS-AIP, International Society for Abnormal Invasive Placenta; CL, cervical length; CV, cervical volume; PV, placental volume; PA, placenta accreta.

Data Sharing Statement

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics Approval and Consent to Participate

This study was reviewed by the Ethics Committee of Suzhou Hospital of Nanjing Medical University (Approval K-2022-015-K01). Informed consent was waived because this anonymously selected study was retrospective and no new interventions were performed on patients.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This study was supported by SuZhou Gusu Medical Youth Talent (grant number GSWS2023055), Suzhou Science and Technology Development Plan (grant number SYW2025052) and Suqian Science and Technology Plan Research Project (grant number SY202210).

Disclosure

The authors declare no competing interests in this work.

References

1. Jauniaux E, Aplin JD, Fox KA, et al. Placenta accreta spectrum. Nature Reviews Disease Primers. 2025;11(1):40. doi:10.1038/s41572-025-00624-3

2. Liu X, Wang Y, Wu Y, et al. What we know about placenta accreta spectrum (PAS). Eur J Obstet Gynecol Reprod Biol. 2021;259:81–89. doi:10.1016/j.ejogrb.2021.02.001

3. Silver RM, Branch DW. Placenta accreta spectrum. N Engl J Med. 2018;378(16):1529–1536. doi:10.1056/NEJMcp1709324

4. Wu X, Yang H, Yu X, et al. The prenatal diagnostic indicators of placenta accreta spectrum disorders. Heliyon. 2023;9(5):e16241. doi:10.1016/j.heliyon.2023.e16241

5. Maurea S, Verde F, Romeo V, et al. Prediction of placenta accreta spectrum in patients with placenta previa using a clinical, US and MRI combined model: a retrospective study with external validation. Eur J Radiol. 2023;168:111116. doi:10.1016/j.ejrad.2023.111116

6. Delli Pizzi A, Tavoletta A, Narciso R, et al. Prenatal planning of placenta previa: diagnostic accuracy of a novel MRI-based prediction model for placenta accreta spectrum (PAS) and clinical outcome. Abdom Radiol. 2019;44(5):1873–1882. doi:10.1007/s00261-018-1882-8

7. Yue Y, Zhu L, Liu C, et al. The relationship between cervical length and area measurements evaluated by MRI and the amount of hemorrhage in PAS cases. BMC Pregnancy Childbirth. 2024;24(1):293. doi:10.1186/s12884-024-06472-5

8. Yue Y, Song Y, Zhu L, et al. The MRI estimations of placental volume, T2 dark band volume, and cervical length correlate with massive hemorrhage in patients with placenta accreta spectrum disorders. Abdom Radiol. 2024;49(7):2525–2533. doi:10.1007/s00261-024-04272-1

9. Thurn L, Lindqvist PG, Jakobsson M, et al. Abnormally invasive placenta-prevalence, risk factors and antenatal suspicion: results from a large population-based pregnancy cohort study in the Nordic countries. BJOG. 2016;123(8):1348–1355. doi:10.1111/1471-0528.13547

10. Saxena U, Rana M, Tripathi S, et al. Prediction of placenta accreta spectrum by prenatal ultrasound staging system in women with placenta previa with scarred uterus. J Obstet Gynaecol India. 2023;73(Suppl 2):191–198. doi:10.1007/s13224-023-01830-3

11. Marshall NE, Fu R, Guise JM. Impact of multiple cesarean deliveries on maternal morbidity: a systematic review. Am J Obstet Gynecol. 2011;205(3):262.e1–8. doi:10.1016/j.ajog.2011.06.035

12. Morel O, Collins SL, Uzan-Augui J, et al. A proposal for standardized magnetic resonance imaging (MRI) descriptors of abnormally invasive placenta (AIP) – from the international society for AIP. Diagn Interv Imaging. 2019;100(6):319–325. doi:10.1016/j.diii.2019.02.004

13. Carusi DA, Duzyj CM, Hecht JL, et al. Knowledge gaps in placenta accreta spectrum. Am J Perinatol. 2023;40(9):962–969. doi:10.1055/s-0043-1761635

14. Erfani H, Fox KA, Clark SL, et al. Maternal outcomes in unexpected placenta accreta spectrum disorders: single-center experience with a multidisciplinary team. Am J Clin Exp Obstet Gynecol. 2019;221(4):337.e1–337.e5. doi:10.1016/j.ajog.2019.05.035

15. Yu FNY, Leung KY. Antenatal diagnosis of placenta accreta spectrum (PAS) disorders. Best Pract Res Clin Obstetrics Gynaecol. 2021;72:13–24.

16. Silver RM, Fox KA, Barton JR, et al. Center of excellence for placenta accreta. Am J Obstet Gynecol. 2015;212(5):561–568. doi:10.1016/j.ajog.2014.11.018

17. Zou L, Wang P, Song Z, et al. Effectiveness of a fetal magnetic resonance imaging scoring system for predicting the prognosis of pernicious placenta previa: a retrospective study. Front Physiol. 2022;13:921273. doi:10.3389/fphys.2022.921273

18. Tovbin J, Melcer Y, Shor S, et al. Prediction of morbidly adherent placenta using a scoring system. Ultrasound Obstet Gynecol. 2016;48(4):504–510. doi:10.1002/uog.15813

19. Kingdom JC, Hobson SR, Murji A, et al. Minimizing surgical blood loss at cesarean hysterectomy for placenta previa with evidence of placenta increta or placenta percreta: the state of play in 2020. Am J Clin Exp Obstet Gynecol. 2020;223(3):322–329. doi:10.1016/j.ajog.2020.01.044

20. Etori Y, Nagai R, Shimomoto Y, et al. Successful management of first-trimester uterine rupture and placenta previa: a case report. Cureus. 2025;17(1):e77857. doi:10.7759/cureus.77857

21. Yue Y, Wang X, Zhu L, et al. Placental volume as a novel sign for identifying placenta accreta spectrum in pregnancies with complete placenta previa. BMC Pregnancy Childbirth. 2024;24(1):52. doi:10.1186/s12884-024-06247-y

The Communications Regulatory Authority of Namibia (CRAN) has granted authorisation to Blue Telecommunications (Pty) Ltd, a subsidiary operating under Radio Electronic (Pty) Ltd’s license, to roll out Eutelsat OneWeb connectivity solutions nationwide. Operating from Low Earth Orbit, Eutelsat OneWeb’s satellite constellation provides high-speed, low-latency internet connectivity tailored for regions with inadequate or nonexistent ground-based network infrastructure. The service primarily serves sectors including enterprises, governmental institutions, maritime operations, aviation, and essential infrastructure providers.

Francois du Toit, Chief Executive Officer of Radio Electronic, said the approval marked a significant milestone for both the company and the country.

“This is a proud moment for Radio Electronic and for Namibia. Securing regulatory approval enables us to introduce Eutelsat OneWeb’s advanced LEO connectivity to the country for the first time,” du Toit said.

Du Toit stressed that LEO technology should be viewed as complementary rather than disruptive, stating it’s not a replacement for fibre, mobile or fixed wireless networks, but as an additional layer of infrastructure aimed at improving coverage, resilience and performance.

Laying the Groundwork for LEO Connectivity in Namibia

Namibia’s LEO connectivity landscape is at an early but consequential stage, shaped by a regulatory approval that reflects the strength of the country’s telecommunications framework. The authorisation granted to Radio Electronic positions it as Namibia’s first locally licensed provider of LEO satellite services, enabling advanced global satellite technologies to enter the market through a regulated, locally anchored operating model.

While the approval represents a historic milestone, full service activation remains subject to the completion of outstanding administrative steps, notably the formal issuance of the physical licence by CRAN, despite type approvals for the required hardware already being in place. Strategically, the rollout aligns closely with CRAN’s national objectives to deliver secure, reliable, and compliant connectivity solutions, particularly for enterprise, government, and other mission-critical users, reinforcing LEO satellite services as an increasingly integral component of Namibia’s digital infrastructure landscape.

Alongside this initial market entry, recent developments involving Starlink are further defining Namibia’s LEO connectivity environment, with the company entering a public consultation phase as part of its regulatory engagement process. During the consultation, Starlink received public support for its proposed operations in the country, reflecting interest in expanded satellite-based connectivity and broader awareness of LEO services among local stakeholders.

Empowering industrial teams to connect legacy equipment, collect actionable data, and visualize insights in real time. 

HOFFMAN ESTATES, Ill., Jan. 19, 2026 /PRNewswire/ — Omron Automation has released the DX100 Data Flow Edge Device, an industrial edge solution designed to connect directly to existing PLCs, sensors, and other automation devices. DX100 enables manufacturers to securely connects, collects, formats, and shares data intuitively. Without modifying current control logic or requiring extensive programming experience. 

DX100 is a single, industrial data edge device that supports a broad range of industrial protocols. It standardizes data at the source and delivers it to higher-level systems in a consistent, repeatable way, accelerating digital transformation initiative across the factory floor. 

• Industrial connectivity to existing machines: Collect data from installed equipment without modifying or replacing current control systems.
• SpeeDBee Synapse low-code environment: Low-code flow chart design with prebuilt collectors enables faster deployment by users with varying skill levels.
• Flexible development platform for custom logic: An easy-to-use environment supporting Python that allows teams to build and deploy tailored data processing logic.
• Built-in support for production dashboards: With Grafana built-in, the single unit can host a wide range of dashboards, delivering visualizations as unique as the data being collected.
• Standard northbound interfaces to cloud systems: Collect data over MQTT, and SQL, then share the consolidated data over MQTT to maximize network bandwidth.
• Predictable licensing for scalable deployments: Transparent, device-based licensing helps control costs as systems expand across multiple lines or facilities. 

Learn More: https://omron.pub/4pGJDY0

About Omron Automation 

Omron Automation is an industrial automation partner that creates, sells, and services fully integrated automation solutions that include sensing, control, safety, vision, motion, and more. Established in 1933, OMRON helps businesses solve problems with creativity in more than 110 countries. Visit https://automation.omron.com/en/us/

SOURCE Omron Automation Americas

A private debt collector has paid zero corporate tax since securing contracts worth tens of millions of dollars from the Australian Taxation Office to pursue arrears payments, including from welfare recipients.

The parent entity of the outsource operator, Recoveriescorp, has recorded large income streams in its two most recent annual accounts, according to Guardian Australia analysis, with revenue surpassing $100m during 2025.

At the same time, business expenses inflated by high consulting fees and elevated interest repayments to unnamed financiers has resulted in a series of annual losses, with no corporate tax payable.

Mark Zirnsak, secretariat at the Tax Justice Network Australia, said the accounts of the parent company, Symbos Bidco, raised questions.

_{Sign up: AU Breaking News email}

“Why is it a loss-making entity? It’s almost like it’s a not-for-profit, and what private business runs as a not-for-profit?,” Zirnsak said.

“As for its loan facilities, it’s a private company and there’s no obligation for them to explain why they have them, so we are left in the dark.”

There is no suggestion the company or its directors have acted illegally.

The ATO referred more than 355,000 taxpayers to Recoveriescorp between January 2024 and October 2025. It has awarded the firm $42.8m worth of contracts since 2022, according to the government tender portal.

A spokesperson for Recoveriescorp said the business and its parent were “fully compliant with their tax and regulatory obligations”.

“The business is in a growth phase so has been reinvesting in systems, frontline people and processes to continually improve the quality and expand the services offered,” the spokesperson said.

The private equity-backed Recoveriescorp is the second company with major government contracts identified by Guardian Australia as running at a financial loss, with no tax payable.

It comes after Guardian Australia revealed an unrelated outsource call centre operator for Centrelink, Telco Services Australia, also paid no corporate tax for several years after winning a large government agency contract.

Procurement rules

Tax Justice Network Australia, which advocates for reform on behalf of dozens of aid agencies, unions and governance groups, has said the tax compliance threshold to apply for government contracts in Australia was low and should be strengthened.

Recoveriescorp is one of the largest subsidiaries of Symbos Bidco, and a main driver of its revenue. All profits, losses and tax liabilities from its operations flow to the controlling entity.

The debt collector is ultimately controlled by the private equity firm Allegro, which bought the company in 2024.

Recoveriescorp staff have worked inside the ATO offices for several years, with its operations recently expanding to include offsite services, whereby the private collector chases debts under its own name on the agency’s behalf.

Symbos Bidco has paid hundreds of thousands of dollars for advisory services to the same firm that audits its accounts; a lawful practice but one that a recent parliamentary committee recommended against.

It has also paid an interest repayment rate in excess of 7% on a near $58m loan facility, although it is unclear why it needs access to such a large sum. It also has access to an $86m loan from a related party.

The Recoveriescorp spokesperson did not respond to questions about why it required the loans or why it doesn’t use different firms for audit and advisory services.

An ATO spokesperson said the agency was unable to comment on the tax affairs of Recoveriescorp or its parent company due to confidentiality laws.

“The ATO undertakes procurement processes in accordance with the commonwealth procurement rules,” the spokesperson said.

The ATO has been particularly reliant on outsource operators, and has awarded contracts to three private call centre operators, the US private equity-owned Probe Operations, the Nasdaq-listed Concentrix Services and the British multinational Serco, in addition to its contract with Recoveriescorp.

Staff at outsource call centres for agencies including the ATO and Centrelink have said training is poor, morale is low and the attrition rate is high.

The tax ombudsman has reported a spike in complaints over the ATO’s use of a third-party collector to chase tax debts and warned the agency to be considerate of a person’s circumstances.

Workday Inc. (NASDAQ:WDAY) is one of the best future stocks to buy for the long term. On January 12, Goldman Sachs assumed coverage of Workday with a Neutral rating and $238 price target. This decision was made as Goldman Sachs initiated coverage on 12 software stocks, highlighting AI as a major long-term growth driver that will expand the sector’s TAM over the next decade. However, the firm offered a more cautious outlook on Workday and noted that its next stage of capturing market share may prove more challenging than previous cycles.

A day before that, Goldman Sachs initiated coverage of Workday with a Neutral rating and a price target of $238. This assessment was part of a broader sector analysis in which the firm assumed coverage of 12 different software stocks. While the firm has a bullish view of the long-term adoption of AI, Goldman Sachs expressed caution regarding Workday’s specific growth trajectory as the company may face increasing difficulty in securing further market share gains during its next phase of expansion.

Additionally, on January 5, RBC Capital adjusted its price target for Workday Inc. (NASDAQ:WDAY) from $320 down to $300, while maintaining an Outperform rating. The firm suggested that 2026 will be a pivotal year where companies prepared for enterprise AI adoption begin to see clear benefits, while those less prepared may struggle against the narrative that AI threatens the traditional software model.

Workday Inc. (NASDAQ:WDAY) provides enterprise cloud applications in the US and internationally.

While we acknowledge the potential of WDAY as an investment, we believe certain AI stocks offer greater upside potential and carry less downside risk. If you’re looking for an extremely undervalued AI stock that also stands to benefit significantly from Trump-era tariffs and the onshoring trend, see our free report on the best short-term AI stock.

READ NEXT: 30 Stocks That Should Double in 3 Years and 11 Hidden AI Stocks to Buy Right Now.

Disclosure: None. This article is originally published at Insider Monkey.

The Decree transposes into Italy Directive (EU) 2024/1619 (CRD VI) and aligns the domestic framework with Regulation (EU) 2024/1624 (i.e., CRR III), amending the Italian Consolidated Banking Act (ICBA) and the Italian Consolidated Financial Act (ICFA).

The Decree reshapes the framework applicable to third-country branches/subsidiaries, group consolidated supervision, and the execution of certain corporate transactions, including but not limited to mergers and demergers involving banks (and/or (M)FHCs, as defined below) incorporated in Italy.

We set out below a brief description of the main amendments introduced by the Decree and the rollout of the transitional regimes. This publication will be followed by a series of insights relating to each of the main amendments outlined below.

Main amendments of the Decree

Third-country branches (TCB)

The main (and, from an in-scope entities’ perspective, the most burdensome) change brought by the Decree is the introduction of the so-called “branch establishment requirement”, which, upon expiration of the transitory regime, would apply to non-European undertakings engaging in certain “core banking” activities and services with Italy-based clients.

Notably, the new Article 14-bis of the ICBA provides that non-EU undertakings that intend to perform the following core banking services: (i) taking deposits and other repayable funds; (ii) lending; and (iii) granting guarantees and commitments (jointly, the Core BK Services) are required to establish a local branch (or, in certain situations, a subsidiary) within the Italian territory, unless an exemption applies.

Although the domestic regime envisaged by the Decree substantially aligns with that set forth in Articles 21c and 47 and the following of the CRD VI, the Italian legislator has nonetheless exercised certain national discretions, partly deviating from the harmonized European framework.

In particular, it is worth preliminarily mentioning that:

(i) reliance on the intra-group and intra-bank exemptions is nonetheless subject to a prior communication requirement to the Bank of Italy. Although this is not a strict prior approval procedure, the Italian regulator has the authority to prevent the notifying entity from commencing or continuing the performance of Core BK Services if certain requirements are not met (note, these requirements will have to be identified by the Bank of Italy in its secondary-level regulations, yet to be published).

(ii) the Decree seems to have not formally transposed the MiFID exemption, since Article 14-bis of the ICBA only includes a cross-reference to the ICFA (which sets out the existing regime for the provision of MiFID services in Italy by non-EU firms, among others). It is therefore not totally clear how this exemption will effectively apply to any non-EU undertakings intending to render Core BK Services that qualify as an activity ancillary to MiFID investment services. Based on the literal wording of the Decree, it seems that an authorization under the ICFA to provide investment services and activities (and relevant eligibility conditions) in Italy would nonetheless be required (subject to a case-by-case assessment).

(iii) the provision on a cross-border basis of banking services and activities other than those qualifying as Core BK Services by non-EU undertakings remains subject to a requirement for the prior notification of the Bank of Italy, though formal prior authorization is no longer required. This notwithstanding, the Bank of Italy has the authority to prevent the notifying entity from commencing or continuing the provision of the relevant services if certain conditions and requirements are not met (note, these requirements will have to be identified by the Bank of Italy in its secondary-level regulations, yet to be published).

(iv) the Decree finally codifies (under Article 14-bis of the ICBA) the reverse solicitation exemption but does not identify any (additional) eligibility criteria to define the permitted scope of the exemption.

The Bank of Italy has been mandated to enact secondary-level regulations and acts to further implement the provisions laid down by Article 14-bis of the ICBA to determine, inter alia, the requirements non-EU undertakings should comply with to rely on the intra-bank and intra-group exemptions, and also to operate in Italy on a cross-border basis (where feasible), and the criteria for assessing the applicability of the reverse solicitation exemption.

The Decree provides a “grandfather” regime benefiting pre-existing contracts (i.e., entered into before July 11, 2026) concerning Core BK Services but clarifies that non-EU undertakings may not novate or renew such contracts. Contrary to CRD VI, clients’ rights are not expressly addressed, and additional restrictions apply to open-ended pre-existing agreements, which must be terminated or transferred to other authorized intermediaries by January 10, 2028 (unless the reverse solicitation exemption applies).

For existing Italian branches of non-EU undertakings, these “grandfather” rights are conditional upon the relevant Italian branch filing a “refreshed” authorization application with the Bank of Italy by January 11, 2027.

As a final comment, it is also important to note that, according to the position historically taken by the Bank of Italy, the issuance and placement of debt securities in Italy by credit institutions is considered deposit-taking activity and is therefore currently subject to the requirement for prior authorization from the Bank of Italy (or passporting requirements, as the case may be). The Decree is silent on this and does not include a specific exemption the issuance of debt securities issuance may benefit from to fall outside the scope of the local branch requirement.

Fit and proper requirements

The Decree materially reshapes the fit and proper requirements applicable to banks (and other regulated intermediaries), as currently laid down by Article 26 of the ICBA.

Notably, to ensure full alignment with the changes brought by CRD VI at a European level, the Decree:

(i) introduces a reference to the “independence of mind” among the requirements for members of management bodies

(ii) exercises the discretion granted by CRD VI and, in the event of the replacement of the majority of members of the management bodies, allows suitability assessments to take place after newly appointed members have taken up their position

(iii) grants the Bank of Italy authority to independently assess the suitability of members of management bodies and key function holders for larger institutions. The relevant materiality thresholds for this assessment shall be identified by a secondary-level decree to be adopted by the Ministry of Economy and Finance.

The Decree also extends the application of the above requirements to regulated non-banking intermediaries, including Italian investment firms (e.g., SIMs) and Italian asset management companies, and, to certain extent, financial holding companies and mixed financial holding companies (the (M)FHCs).

Mergers and demergers

Irrespective of any materiality thresholds or the size of the relevant transactions, mergers and demergers are subject to the prior approval of the Bank of Italy in the following cases:

(i) In the case of a merger, the incorporating entity is a bank or an (M)FHC with its registered office in Italy.

(ii) In the case of a demerger, the demerged entity is a bank or an (M)FHC with its registered office in Italy.

Article 57 of the ICBA, as amended by the Decree, sets out the evaluation criteria for granting (or denying) authorization in accordance with the requirements laid down at a European level by CRD VI. Contrary to the approach followed by the European legislator, the Decree does not directly identify circumstances in which prior approval is not required, or the regulatory term for completing the assessment or filing further information upon the regulator’s request. The regulatory procedure that banks and/or (M)FHCs should follow will be governed by the secondary-level acts and regulations that the Bank of Italy has been mandated to enact.

Acquisition or sale of material holdings

The Decree introduces a new prior authorization or notification requirement for acquisition and sales of “material holdings” by banks and (M)FHCs, respectively. As opposed to CRD VI, the Decree (notably, the newly introduced Article 57-bis ICBA) does not define the meaning of “material holding”, nor set the thresholds for triggering a prior approval (or notification) requirement.

Among other things, the Bank of Italy is in charge of enacting a secondary-level regulation identifying the materiality thresholds for the application of the regime in question, governing the main terms and the timing of the relevant regulatory procedure in-scope entities should follow, and ensuring an adequate level of coordination with other regulatory regimes already in place (e.g., qualified holdings).

The thresholds triggering a prior approval (or notification) requirement may be exceeded on an individual or consolidated basis. In the latter scenario, the Bank of Italy shall consult and decide whether to grant the prior approval in coordination and cooperation with the European competent authority exercising the consolidated supervision (if different).

Similar to the approach followed for the qualified holdings regime, voting (or similar) rights attached to material holdings acquired/held in breach of the prior approval requirements cannot be exercised and the Bank of Italy may order the disposal of the relevant material holdings within a predetermined time frame.

Material transfer of assets and liabilities

Italian banks and (M)FHCs are required to notify the Bank of Italy, in writing, in advance of any material transfer of assets or liabilities they intend to execute. The newly introduced Article 58-bis of the ICBA does not identify the criteria for assessing whether a transfer of assets or liabilities is considered “material” or carry over the exemptions from this notification requirement laid down at a European level into Italian law.

As per the approach taken in relation to mergers and demergers and material holdings, the comprehensive regime concerning the material transfer will be envisaged by the secondary-level regulation the Bank of Italy is mandated to enact in this respect.

This notwithstanding, to avoid any regulatory misalignment and unlevel playing fields, the Decree amends the regulatory treatment applicable to bulk transfers of legal relationships (cessione di rapporti giuridici in blocco). Notably, the Decree:

(i) repeals the prior authorization procedure previously provided by Article 58 of the ICBA in relation to certain (material) bulk transfers of legal relationships

(ii) clearly envisages that in the case that a bulk transfer of legal relationships amounts to a material transfer of assets and liabilities, the relevant transaction will be subject to both regulatory regimes, as set out by Articles 58 and 58-bis of the ICBA, respectively.

Group supervision and consolidation perimeter

In terms of group consolidated supervision, the new Article 60-ter of the ICBA enables (M)FHCs that have been exempted from acting as a parent company to be excluded from the prudential consolidation perimeter of the group, provided that certain conditions and requirements are met. Upon the granting of the relevant authorization, the prudential requirements will be then calculated at the level of the controlled entity designated as a parent.

The Decrees also clarifies that an intermediated (M)FHC can be designated as responsible for ensuring compliance with the applicable legal and regulatory requirements on a consolidated basis.

The transitional regimes

The Decree entered into force on January 9, 2026, and provides for the following transitional regimes:

(i) The amendments introduced in relation to: (i) mergers and demergers; (ii) the acquisition and sale of material holdings; (iii) the material transfer of assets and liabilities; (iv) the regulatory treatment of (M)FHCs; and (v) the “fit and proper” requirements, would enter into force upon the adoption of the relevant secondary-level acts and regulations by the Bank of Italy. Therefore, as at the date of this publication, it is not possible to foresee when the transitory regimes for these matters would ultimately end.

(ii) The amendments introduced in terms of third-country branch regimes would enter into force on January 11, 2027. As from such date, non-EU undertakings may continue to provide, on a cross-border basis, any activities that are strictly necessary for the execution/performance of contracts pertaining to Core BK Services entered into before July 11, 2026 (with no ability to novate or renew such agreements).

Introduction

Chronic Atrophic Gastritis (CAG) is a chronic inflammatory disease characterised by gastric mucosal epithelium degradation, resulting in the reduction or absence of gastric mucosal glands, with or without intestinal epithelial or pyloric glandular metaplasia.¹ Recurrent or chronic inflammation has been linked with the onset and progression of various human cancers. In this regard, it is noteworthy that CAG is often the first step in gastric mucosal changes that could progress to irreversible gastric carcinogenesis.² Although immune-mediated inflammation has been significantly implicated in CAG’s complex pathogenesis, the precise mechanisms remain unclear. Complex interactions between immune cells and gastric mucosal epithelial cells could promote chronic inflammation in the gastric microenvironment. Notably, a complex network of signalling molecules, involving cytokines and chemokines, which facilitate immune cell recruitment and activation, thus influencing inflammation at local tissue sites, was reported to mediate the aforementioned dynamic interaction. Macrophages, as part of the immune system, help remove pathogens and damaged cells under normal circumstances, but can also, depending on the surrounding environment, become overactive and attack healthy tissues, leading to inflammation and damage.³ Owing to their remarkable plasticity, macrophages could polarise into classically activated (M1) or bypass-activated (M2) phenotypes in response to different factors and microenvironments. Notably, M1 macrophages are highly expressive of pro-inflammatory factors such as CD86, Major Histocompatibility Complex-II (MHC-II), and Inducible Nitric Oxide Synthase (iNOS) in response to Lipopolysaccharide (LPS), Interferon-gamma (IFN-γ), and TNF-α stimulation—a process mediated through the activation of pathways such as TLR- and NF-κB. These pro-inflammatory factors could exert tumour-suppressive and pro-inflammatory effects.⁴ Conversely, M2 macrophages, after treatment with TGF⁃β, IL-10, and Th2 cytokines (IL-4, IL-13), could exert anti-inflammatory effects and tissue repair functions—a process mediated through pathways such as JAK-STAT6 and PI3K-AKT.⁵ Following inflammation onset, numerous macrophages often infiltrate the gastric mucosal tissues in CAG;⁶ hence, CAG progression could be attributed to macrophage infiltration and polarisation. Moreover, macrophage conversion from the M1 to M2 phenotype could result in the upregulation of Transforming Growth Factor β (TGF-β) and Interleukin-10 (IL-10), thus alleviating inflammation.⁷ Therefore, proper macrophage phenotype modulation could precisely regulate the tissue and intracellular microenvironment, presenting a promising therapeutic strategy for CAG. Exploring this hypothesis could enhance our understanding of the pharmacological effects and potential mechanisms of CAG from a macrophage polarisation perspective.

The human CXC Chemokine Ligand 16 (CXCL16), a member of the chemokine family, is primarily expressed in monocytes, macrophages, dendritic cells, and endothelial cells, among other immune cells. Following inflammation occurrence, CXCL16, a vital inflammation transmitter, promotes immune cell chemotaxis to the inflammation site, facilitating inflammatory factor phagocytosis and release—a phenomenon that further aggravates the inflammatory response.^8,9 According to research, CXCL16 overexpression in the gastric mucosa could induce local CD8⁺ T lymphocyte infiltration, weakening the body’s defence function and ultimately promoting CAG development.¹⁰ Furthermore, CXCL16 could promote CXCL16/CXCR6 axis activation, exacerbating the gastric mucosa’s inflammatory response, ultimately disrupting local immune homeostasis and increasing the risk of gastric carcinogenesis.¹¹ Despite CXCL16 playing a key role in gastric mucosal injury, its precise mechanisms, especially in modulating the dysregulation of CAG immune response, remain to be elucidated. In addition to inflammatory responses, CXCL16 might also regulate macrophage migration, accelerating inflammatory mediator release and amplifying local and systemic inflammatory responses.¹² Based on these insights, we sought to identify the potential targets for CAG treatment from a macrophage polarisation perspective.

Presently, CAG treatment is primarily based on acid-suppressing drugs. Nonetheless, due to the heterogeneity and resistance to acid-suppressing drugs in CAG patients, these standardised therapeutic approaches may not be universally applicable. Therefore, developing alternative therapeutic strategies and targets would be imperative for improved clinical outcomes across different patient groups. In this study, in order to preliminarily investigate the mechanism of the interaction between CAG progression and macrophage polarization and to find out whether there is some kind of cytokine in CAG that acts on macrophages to promote their polarization or can attract the aggregation of M1-associated macrophages.we leverage a comprehensive analytical approach to identify potential biomarkers and elucidate the immune-mediated mechanisms in CAG. To establish whether CXCL16 regulates macrophage polarisation, we initially employed a comprehensive analytical approach to elucidate the immune-related biological functions of CAG and to identify potential biomarkers. Subsequently, data analysis utilizing the database revealed significant differences in the expression levels of CD86, CD163, and CXCL16 between the CAG and CNAG groups. Verification through multiplex immunohistochemistry (mIHC) technology demonstrated that M1 macrophages accumulate abundantly within the inflammatory microenvironment. Notably, CD86 and CXCL16 expression levels were significantly elevated in CAG patients, while CD163 protein was upregulated in CNAG patients; however, CXCL16 exhibited reduced expression in both CNAG and CAG-E patients. Further co-localization and correlation analyses indicated that CXCL16 is predominantly co-expressed with the M1 macrophage marker CD86 in CAG patients, suggesting its role in regulating M1 macrophage polarization. Additionally, in vitro experiments showed that stimulation with varying concentrations of CXCL16 resulted in a significant increase in CD86 mRNA expression alongside a marked decrease in CD163 mRNA expression. This further corroborates that CXCL16 can promote macrophage polarization towards an M1 phenotype. These findings provide novel insights and evidence for understanding the pathology of CAG as well as potential targeted therapeutic strategies.

Materials

Access to Public Databases

Based on the CAG disease type and human species, the gene expression datasets GSE153224 and GSE27411, containing CAG and Chronic Non-Atrophic Gastritis (CNAG) whole blood samples were collected from the US Centre for Biotechnology Information’s Gene Expression Omnibus (GEO) database (https://www.ncbi. nlm.nih.gov/geo/). Immune Genes (IGs) were also downloaded from the Import database (https://www.immport.org/).

Clinical Sample Collection and Patient Screening

This study included 20 cases each of gastric tissue wax blocks mainly from the gastric sinus area, extracted from patients with CNAG, CAG and Chronic atrophic gastritis with erosion (CAG-E) diagnosed via endoscopy and histopathological examination at Gansu Provincial Hospital’s Department of Gastroenterology from October 2023 to September 2024. The enrolled patients were predominantly aged 18–65 years. Diagnosis adhered to the (1) endoscopic diagnostic criteria outlined in the Chinese Guidelines for the Diagnosis and Treatment of Chronic Gastritis (2022) and (2) the pathohistological diagnostic criteria outlined in the Consensus on Pathological Diagnosis of Gastric Mucosal Biopsy for Chronic Gastritis and Epithelial Tumours (2017). Compliance procedures were informed by the ethical standards set by the Committee in Charge of Human Trials.

The inclusion criteria were: (1) Patients aged 18–65 years who met the diagnostic criteria and whose diagnosis was confirmed via gastroscopy and pathological examination; and (2) Patients with complete clinical history data who signed the informed consent form. On the other hand, the exclusion criteria were: (1) Patients who did not meet the inclusion criteria; (2) Patients who were treated with proton pump inhibitors, Non-Steroidal Anti-Inflammatory Drugs (NSAIDs), Traditional Chinese Medicine (TCM) herbs, antiplatelet drugs, and anticoagulants in the last one month; (3) Patients with a combination of severe cardiac, cerebral, renal, and pulmonary comorbidities, psychiatric disorders, or those who were pregnant/lactating women; (3) Patients who presented with reflux oesophagitis, peptic ulceration, polyp, and hypertrophic gastritis in the last month of endoscopic examination, Gastric Cancer (GC), and other malignancies intervened with surgery, radiotherapy, or chemotherapy within the last 5 years; and (4) Patients with autoimmune atrophic gastritis diagnosed using the anti-mural cell antibody test.

Reagents and materials

The key laboratory instruments and equipment included: An orthostatic fluorescence photomicrograph microscope (Nikon-eclipse ti2, Japan); an inverted white light/fluorescence photomicrograph microscope (Olympus, Japan); a water purifier, digital pendulum shaker, vortex mixer, and magnetic stirrer (ServiceBio, China); an electrothermal incubator (Shanghai Yiheng, China); a palm centrifuge (Scilogex, USA); a benchtop centrifuge (Shanghai Anting Scientific Instrument Factory, China); an ice maker (Changshu Xueke Electrical Appliance Co., China); refrigerators (4°C and −20°C; XINGX, China); pipettes (100–1000 μL, 20–200 μL, 10–100 μL, 0.5–10 μL, 0.1–2.5 μL; Eppendorf, German); a decolourisation shaker (Beijing Liuyi Instrument Factory, China); a PCR instrument (Hangzhou Bori Technology, China), and a fluorescence quantitative PCR instrument (Life technologies, USA).

The other laboratory reagents and consumables included: Triton X-100 and Bovine Serum Albumin (BSA; Beijing Solebo Technology Co., Ltd., China); an anti-fluorescence quenching sealer (SouthernBiotech, USA); pipette tips (1000 μL, 200 μL, and 10 μL), NaCl, Na2HPO4-12H2O, NaH2PO4-2H2O, disodium EDTA, NaOH, citrate buffer, xylene, anhydrous ethanol, Paraformaldehyde (PFA), Hydrochloric acid (HCl), glycerol, trichloromethane, and isopropanol (Sinopharm Chemical Reagent Co., Ltd., China); Phosphate Buffered Saline (PBS) solution, TSA-480, 570, 520, and 670, group paintbrush, coverslips, slides, and citric acid restoration solution 20* (Wuhan Snowpigeon Biotech Co., Ltd., China); DAPI staining solution and Tris base (Sigma, USA); Horseradish Peroxidase (HRP)-goat anti-rabbit IgG and HRP-goat anti-mouse IgG antibodies (KPL, USA); triple pure total RNA extraction reagent, EntiLink™ first strand cDNA synthesis kit, and EnTurbo™ SYBR Green PCR SuperMix kit (KPL, China).

Methods

Bioinformatics Analysis methods

Differentially Expressed Genes (DEGs) Analysis Using the GEO Database

DEGs from the 2 datasets with CAG samples were analysed using the online database GEO2R (https://wwwncb.i.nlm.nih.gov/geo/geo2r/) and volcano plots generated for visualisation. The conditions for screening were adj.P.Val<0.05 and an absolute value of the multiplicity of differences > 1 (|log2FC|>1).¹³ The Pheatmap package was used to generate differential gene heatmap displays for the top 15 genes. The online intersection analysis software (http://bioinformatics.psb.ugent.be/webtools/Venn/) venn graph was used to find the DEGs that are common to the above GEO data chips.

Acquisition of CAG-Related Immune Genes and Their Enrichment Analysis

CAG-associated immune genes were obtained by screening co-expressed genes between co-DEGs and Immune Genes (IGs) using Venn diagrams. These genes were then subjected to Gene Ontology (GO) functional enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses using the DAVID (https://david.ncifcrf.gov/home.jsp) online database to identify related functions and pathways. Specifically, KEGG pathway enrichment analysis was performed to identify the biological functions in which these IGs were involved, while GO analysis examined the key Biological Processes (BP), Cell Components (CCs), and Molecular Functions (MFs). Finally, the top-ranked results were visualised and analysed using a P-value Cutoff of 0.05 and a Q-value Cutoff of 0.05.

Target Core Proteins Screening via Protein-Protein Interaction (PPI) Network Analysis

The obtained proteins encoding CAG-associated immunity genes were introduced into STRING11.5 (https:// cn.string-db.org/) for visualisation, with the species set to “Homo sapiens”. A PPI network was then constructed and visualised. The results were further imported into Cytoscape 3.10.3 software in the xlsx format to construct a PPI network diagram. Topology analysis was performed using the cytoHubba plug-in of Cytoscape 3.10.3 based on the betweenness calculation method to filter out key proteins that could influence disease pathogenesis. To evaluate the target core proteins that most closely correlated with CAG, the key proteins were imported into Cytoscape again for interaction analysis.

Analyzing Differential Expression of Target Target Factors and Their Correlation Using Data From Databases

First, GSE153224-normalised microarray datasets were downloaded and screened for the expression of the target macrophage markers CD86 and CD163, as well as the impact factor CXCL16, across different gastritis tissues, respectively. Comparisons and Pearson’s correlation analyses were carried out by using GraphPad Prism 10.4 software, and the results were visualized.

Experimental Validation

Multiple Fluorescent Immunohistochemical Staining (the mIHC-Tetrachromatic Method)

The differential expressions of CD86, CD163, and CXCL16 proteins in CNAG, CAG, and CAG-E patients were verified using mIHC. Briefly, paraffin-embedded tissues were sectioned and baked in a 60°C oven for 1 h. They were then deparaffinized in xylene and hydrated in gradient alcohol (xylene I for 15 min, xylene II for 15 min, 100% anhydrous ethanol I for 5 min, 100% anhydrous ethanol II for 5 min, 95% ethanol I for 5 min, 85% ethanol for 5 min, and 75% ethanol for 5 min). After rinsing in tap water for 10 min, the sections were soaked in distilled water for 5 min. They were then subjected to high-temperature and high-pressure antigen repair in an autoclave. An appropriate amount of 0.01 M sodium citrate buffer (pH 6.0) repair solution was first added before turning on the autoclave and timing the heating to the point of jetting for 2 min, with the sections ultimately allowed to cool naturally after the thermal repair was completed. After rinsing in running tap water for 5 min, the sections were immersed in distilled water for 5 min and then placed in 3% H2O2 solution to seal the endogenous peroxidase. The sections were then incubated at Room Temperature (RT) for 20 min. After washing with distilled water, the non-specific binding site was sealed with 5% BSA before incubating at RT for an additional 30 min. Following that, the sealing solution was discarded, and the unwashed sections were incubated with antibody dilutions based on predetermined optimal concentrations. Specifically, the sections were subjected to primary antibody incubation at 37°C for 2 h, followed by a PBST-prepared, HRP enzyme-labelled secondary antibody incubation at RT for 1 h after washing three times in PBS (for 5 min each time). After washing away the secondary antibody with PBST three times (for 5 min each time), the sections were incubated with a Tyramine Signal Amplification (TSA) dye—added dropwise—at 37°C for 30 min. The tissues were then washed three times with PBS (for 5 min) and subjected to secondary antigenic repair (repeat the above steps without adding 3% hydrogen peroxide dropwise). After sealing, a second antibody was added, and the steps were repeated until the third antibody staining was completed. Finally, DAPI staining of nuclei was performed. Specifically, the DAPI working solution was added dropwise and incubated at RT for 5 min. The tissues were then washed with PBS three times (for 5 min each time). Subsequently, the liquid was shaken dry before adding an anti-fluorescence sealer dropwise and sealing with cover slips. The complete fluorescent sheets were stored at 4°C in the dark. Table 1 lists the antibodies and Table 2 shows the fluorescein information.

Table 1 Antibodies Used

Table 2 Fluorescein Information

Image Acquisition and Outcome Evaluation of mIHC

Stained images were first captured using a Leica DMi8 microscope equipped with high-efficiency fluorescent dye-specific filters for DAPI, FITC, Cy3, and Cy5. Scans of the Immunofluorescence (IF) multilabel were then examined for the number of positive cells, positive density, and co-localisation using the Indica Labs (U.S.A) digital image analysis software (HALO V2.0). Briefly, two pathologists blinded to the patient’s condition annotated the inflammatory borders. The software then circled the area to be assessed along the tissue to be tested, selecting either the number or the area of positives for the analysis module. Subsequently, fluorescent signals for each channel destination were selected manually and identified using the software across several iterations to ensure all positive signals were selected, while saving the initial colour selection criterion. The same colour selection criteria were applied to similar indexes within the same batch of sections. Following that, the software identified and located all nuclei with DAPI blue fluorescence and extended the cytoplasmic range, calculating different parameters such as the number of positive cells, the positive area, the positive intensity (grey value of fluorescence signals), and so on. Other parameters, such as the ratio of positive cells, the density of positive cells, the intensity of positive cells, and so on, were also calculated. Relevant parameters were then analysed through co-localisation to evaluate the strength of positivity. The area to be measured was calculated step by step under high magnification. The results of the analysis were then exported per predetermined requirements, and a report was generated. In cases where the positive cell ratio equalled the number of positive cells/total number of cells,^8,14 the percentage was calculated with the number of all nucleated cells (DAPI+) as the denominator. For statistical analysis, the following distinctions were made based on the ratio of positive cells in the sections: Non-expression group (no positive cell staining); low expression group (1–40% positive cell staining); and high expression group (>40% positive cell staining). Positive cell density, which evaluates the distribution and number of certain types of positive cells in the tissue, was calculated as follows: Positive cell density = number of positive cells/areas of tissue to be tested.¹⁵ This parameter was appropriately combined with mIHC staining intensity to assess the expression of each index in the tissue. On the other hand, positive intensity (staining intensity), which reflects the average depth of the positive signal, was evaluated based on the depth of positivity, with larger values indicating a greater brightness of the fluorescent signal. Notably, this parameter is particularly suitable when the positivity is patchy and widely expressed.^16,17 Herein, the positive intensity scoring criteria were as follows: Non-expression group (number of positive cells <10%), low expression group (10~40% positivity), and high expression group (>40% positivity).

Cell Culture

The human myeloid leukemia mononuclear cell line THP-1 utilized in this study was obtained from Saibaikang Biotechnology Co. Ltd. The cell line has the research resource identifier (RRID) CVCL_0006, as recorded in the Cellosaurus database. The use of this commercially available cell line was approved by the Ethics Committee of Gansu Provincial Hospital (Approval number: 2025–410). The cells were cultured in a Lymphocyte Medium (LM) supplemented with 10% Fetal Bovine Serum (FBS), 1% Penicillin-Streptomycin (P-S), and RPMI 164 basal medium at 37°C and 5% CO2.

Cellular IF Staining

First, THP-1 cells were cultured in LM containing 160 nmol/L phorbol ester (phorbol 12-myristate 13-acetate, PMA) for 24 h to induce their differentiation into macrophages. After discarding the old medium, the cells were washed three times (for 5 min each) with 200 μL PBS. The cells were then fixed in 4% PFA for 20 min and washed 3 times (for 5 min each) with PBS. To prevent the incubation solution from draining away in the later stages, circles were drawn with a histochemical pen. The cells were then sealed with 5% BSA to reduce non-specific staining. Following that, the cells were subjected to primary antibody incubation at 4°C overnight with mouse anti-CXCL16 (1:200) and rabbit anti-CXCR6 (1:200) polyclonal antibodies, followed by secondary antibody incubation with CoraLite488 and FITC-labelled secondary antibodies (Table 1) added dropwise at 37°C for 40 min in a water bath in the dark. The cells were then washed with PBS three times (for 5 min each time). Subsequently, the nuclei of the cells were restrained with DAPI in the dark at RT for 20 min, washed with anti-BSA, and sealed with an anti-quenching sealer for 20 min. Finally, the cells were observed under a fluorescence microscope, with the fluorescence images of CXCL16 and CXCR6 captured.

RT-qPCR Detection of the CD86 and CD163 mRNA Expression Levels in Macrophages

After inoculation into 6-well plates at a density of 1×10⁵ cells/well, the induced macrophages were cultivated in a cell culture incubator for 24 h. After wall attachment, the original medium was aspirated before washing the adherent cells twice with 200 µL PBS solution. Subsequently, different concentrations of the CXCL16 factor (0, 50, and 100 μg/mL, respectively) were added to stimulate the macrophages and incubated for an additional 48 h. Cells were collected from each group, and total RNA was extracted using TRIzol (Invitrogen, Carlsbad, USA) per the manufacturer’s instructions. Following that, 1µg of mRNA from each sample was reverse-transcribed into complementary DNA (cDNA) using the EntiLink™ 1st Strand cDNA Synthesis Kit. The reverse transcription products were then collected in the StepOne™ 1st Strand cDNA Synthesis Kit and subjected to Polymerase Chain Reaction (PCR) on a StepOne™ Real-Time PCR instrument under the following conditions: Pre-denaturation at 95°C for 3 min, 40 cycles (95°C 10s→58°C 30s→72°C 30s). The mRNA content was calculated using the 2^−ΔΔCT method, and GAPDH was used as the internal reference gene. Table 3 shows the primer sequences used.

Table 3 Primer Sequences

Statistical methods

The CAG transcriptomics results from the GEO database were mapped and analysed using the R4.0 software package. Meanwhile, the datas were analysed and graphed using SPSS 26.0 and GraphPad Prism 10.4 software. Statistical values of the samples from each group were assessed for normality using the Shapira-Wilkinson test, with P>0.05 indicating normal distribution. Metrological data between two groups were compared using the t-test, while those between multiple groups with the same variance were compared using one-way Analysis of Variance (ANOVA). Pairwise comparisons were performed using Bonferroni’s method. Pearson’s correlation analysis was used to analyse data on the relationship between macrophage-associated markers (CD86, CD163) and the chemokine CXCL16 obtained from mIHC. All tests were two-sided, and results with P<0.05 were considered statistically significant.

Results

Bioinformatics Analysis results

DEGs Identification

Combining the data from the 2 GEO database microarray sequences, we first analyzed the differential expression of genes in CAG using the GEO2R analysis system.It was found that 1096 genes were significantly up-regulated and 714 genes were significantly down-regulated in the GSE153224 dataset compared with the control CNAG. 220 genes were significantly up-regulated and 386 genes were down-regulated in the GSE27411 dataset (Figure 1A). Heatmaps of the top 15 DGEs of these two datasets were also shown separately (Figure 1B). And the intersection of the 2 data sets was taken to find the shared differential genes, and finally 239 shared DGEs were obtained (Figure 1C).

Figure 1 Differentially Expressed Genes Screening (A) Volcano plot of DEGs; (B) Heatmap of DEGs; (C) Venn diagram of 239 co-DEGs.

Screening and Biological Functions of CAG-Related Immune Genes

The intersection of differential genes with 459 IGs using Venn plots yielded 24 common IRGs in CAG (Figure 2). These genes were then subjected to GO annotation and KEGG enrichment analyses using the DAVID database to gain a deeper understanding of their biological roles. According to the GO-BP analysis results, these genes were involved in the positive regulation of BPs such as cell migration, signal transduction, cell chemotaxis, and immune responses (Figure 3A). On the other hand, GO-CC analysis revealed significant enrichment mainly in extracellular regions and the extracellular space (Figure 3B). Finally, the MF mainly included chemokine, growth factor, cytokine, and receptor ligand activities (Figure 3C). Additionally, cytokine-cytokine receptor interactions and chemokine signalling pathways were detected in KEGG enrichment analysis (Figure 3D). These findings collectively suggest that intrinsic immunity, adaptive immunomodulation, and chemokine activity are crucially involved in CAG pathogenesis—a phenomenon that aligns with the basic pathological features of macrophages and chemokines.

Figure 2 CAG immunity-related genes.

Figure 3 Functional and Pathway Enrichment Analysis Results for CAG-Related Immune Genes (A) Biological processes analysed by GO; (B) Cellular components analysed by GO; (C) Molecular functions analysed by GO; (D) KEGG analysis.

PPI Network Analysis and Screening of Key Markers

The 24 common IGs were subjected to protein interaction analysis and visualisation using the STRING database (Figure 4A), with CXCL16, CX3CL1, IL1RN, CD86, CD163, CCL28, and CCL15 emerging as the top seven key proteins. Since CD86 and CD163 are markers of different macrophage phenotypes, they were re-analysed in terms of interactions, revealing a close association between them and other proteins in CAG (Figure 4B). However, these proteins exhibited significant tissue specificity, among which CX3CL1 is distributed in dorsal root ganglia and spinal cord neurons, which are mainly involved in neurological disorders.¹⁸ IL1RN is highly expressed in cancer-related diseases, such as breast and gastric cancers,¹⁹ and very few literature reports on the correlation between IL1RN and the risk of CAG.CCL28 is mainly distributed in the mammary glands, small bowel, and colon, among others. CCL15 is expressed in the intestine and liver, and they have also been found to be associated with a variety of cancers.²⁰ While CXCL16 being highly expressed in antigen-presenting cells [macrophages and Dendritic Cells (DCs)] as a chemotactic agent for monocytes and macrophages, and correlating with inflammatory regulation. Nobly, this phenomenon aligns with our study goal. Consequently, we considered CD86 and CD163 as key markers for different macrophage phenotypes, and CXCL16 as a candidate for regulating macrophage polarisation in CAG.

Figure 4 PPI analysis network diagram (A) CAG-Related Immune Gene PPI Network Diagrams; (B) Interactions between the seven key proteins.

Analyzing Differential Expression and Correlation of Candidate Markers Using Database Information

Based on the datas screened in the GEO database, we first examined the expression of macrophage-related markers CD86 and CD163, as well as that of their potential influencing factor CXCL16 in the CNAG and CAG groups (Figure 5A). According to the results, the CAG group exhibited a higher expression of the M1 macrophage marker CD86 and chemokine CXCL16 than the CNAG group (P = 0.017 and 0.001, respectively). Conversely, the CNAG group showed a higher expression of the M2 macrophage marker CD163 than the CAG group (P=0.011). To further explore the effect of CXCL16 on macrophage polarisation in CAG, we analysed the correlations among CD86, CD163, and CXCL16 expressions in CAG (Figure 5B). According to the results, the M1 macrophage marker CD86 correlated strongly with the chemokine CXCL16 in CAG (0.8≤|r|<1, P<0.05). Conversely, the correlation of CD163 expression with CXCL16 was not statistically significant (P>0.05). These findings collectively suggest that CXCL16 and CD86 could play a mutually synergistic role in CAG, with CXCL16 potentially influencing M1 macrophage polarisation.

Figure 5 Bioinformatics analysis of the expression and correlation of macrophage markers with CXCL16 (A) The differential expression of CD86, CD163 and CXCL16 in CAG and CNAG tissues; (B) Correlation analysis for the association of CD86 with CXCL16 in CAG and CD163. * indicates P<0.05, ** indicates P<0.01, P<0.05 is statistically significant.

Experimental Validation Results

Multiple Fluorescence Immunohistochemistry Analysis

Establishment of Multiple Fluorescence Immunostaining methods

The CD86, CD163, and CXCL16 monoclonal antibodies were detected in gastric tissue sections, with clear blue, red, green, and yellow fluorescence observed in local magnification images. Furthermore, although CXCL16 exhibited a lower expression compared to CD86 and CD163, all of them were predominantly expressed in the cell membranes (Figure 6).

Figure 6 Multiplex fluorescent immunohistochemical staining was used to detect the expression and localization of CD86, CD163, and CXCL16 molecules in gastric tissue.Fluorescent labelling was examined by CaseViewer 2.4 scanning immunofluorescence microarray at 40x field of view, DAPI blue, CD86 red, CD163 green, CXCL16 yellow, scale bar 20μm.

Expression and Distribution of Macrophage-Related Markers and CXCL16 in Different Pathological Stages of Gastritis

To validate the differential expression of the two macrophage-related markers and CXCL16, we stained 60 gastritis tissue samples with mIHC and performed cell notation under high magnification. The number of CD86+, CD163+ and CXCL16+ macrophages across different cell types was analysed semi-quantitatively, with the ratio of positive cells calculated and statistically analysed. The results of mIHC staining and semi-quantitative analysis revealed that M1(CD86+) macrophages exhibited a higher expression in gastric tissues of CAG and CAG-E patients compared to CNAG patients. Furthermore, CD86 was sparsely expressed in gastritis tissues, exhibiting a step-wise upregulation with inflammation progression (both P<0.05). Conversely, the expression of CD163+ cells was lower in CAG and CAG-E patients than in CNAG patients (P<0.05). Finally, although CXCL16+ cells were up-regulated in the CAG stage, they were largely absent in CNAG and CAG-E patients (P<0.001) (Figures 7A and B).

Figure 7 mIHC technology validates the expression of macrophage markers and CXCL16 in gastric tissue (A) mIHC images of CD86 (red), CD163 (green) and CXCL16 (yellow) expressed in CNAG, CAG and CAG-E, respectively. Scale bar: 50 μm; (B) Differential Analysis of CD86+, CD163+, and CXCL16+ Cell Percentages in Different Types of Gastritis Tissue; (C) Box line plot of CD163+, CD86+ and CXCL16+ cell density in the study cohort. * indicates P<0.05, ** indicates P<0.01, *** indicates P<0.001, and **** indicates P<0.0001. Total sample size n=60, with n=20 per group.

To evaluate the distribution of positive cells in tissues and their number per unit area, we further assessed CD86⁺, CD163⁺ and CXCL16⁺ cell densities across three types of gastritis. According to the results, the positive cell density and percentage of positive cells analysed were consistent, with a more concentrated distribution of CD163⁺ and CD86⁺ cells observed in the CNAG group, which also exhibited the highest density of CD163⁺ cells (median number of positive cells/mm², 2451.7 vs 2418.7, P = 0.030; and 2451.7 vs 56.7, P= 0.009, respectively), followed by CD86⁺ cells higher than CXCL16 (median number of positive cells/mm², 2418.7 vs 56.7, P=0.001). Conversely, CD86⁺ cell density was significantly higher than that of the other two in the CAG and CAG-E groups, with a statistically significant difference in both cases (P<0.05). Furthermore, the distribution of CXCL16⁺ cell was highly centralised, with a low positive cell density in the CAG group and an even lower (almost negligible) density in the CNAG and CAG-E groups (Figure 7C).

Positive Cell Intensity Further Validates the Expression of CD86, CD163 and CXCL16 at Different Stages of Gastritis Progression After

Determining the staining intensity of CD86, CD163 and CXCL16 in the gastric tissues of patients with CNAG, CAG and CAG-E through multiple immunofluorescence assay, we further tested the expression levels of these indexes in different progression periods of gastritis, and observed that CD86, CD163 and CXCL16 were expressed in the early stage of gastritis, suggesting that they participate in the entire process of gastritis evolution. CD163 likewise stained with the strongest intensity in CNAG, and CAG was similar to that in CAG-E, with insignificant changes, and the difference between the 2 groups was not statistically significant (P>0.05). In contrast, CD86-positive macrophages exhibited increasing fluorescence intensity during the development of gastritis, and were significantly upregulated expression in gastric tissues, with the expression level being twice higher compared with that of the other 2 molecules,with pairwise comparisons showing statistically significant differences (P<0.05). The CXCL16-positive cells showed strong staining intensity in CAG (Figure 8).

Figure 8 Staining intensity of CD86, CD163, and CXCL16 positive cells in CNAG, CAG, and CAG-E CD86: P < 0.05, compared among the three gastritis groups; P < 0.01, the CAG group compared with the CNAG group; P < 0.001, the CAG-E group compared with the CNAG group; P < 0.01, the CAG-E group compared with the CAG group. CD613: P < 0.05, compared among the three gastritis groups; P < 0.01, comparison between CAG and CNAG groups; P < 0.05, comparison between CAG-E and CNAG groups. CXCL16: P < 0.05, compared among the three gastritis groups; P < 0.05, the CAG group compared with the CNAG groups.

The differential expression analyses indicated that in the CNAG group, CD163 expression was notably elevated, suggesting that the early stages of gastritis are predominantly characterized by M2 macrophage polarization. In contrast, the CAG and CAG-E groups exhibited higher CD86 expression, indicating a shift toward M1 macrophage activation, which likely plays a critical role in the progression and pathogenesis of gastritis. This indicated a potential correlation between macrophage polarisation status and the degree of inflammation and pathological changes in gastritis, with M1-type macrophages being associated with severe inflammation and tissue damage, and the M2-type macrophages exerted a restorative effect in the early stages of inflammation, but its anti-inflammatory response was relatively insufficient in the later stages of the disease, allowing persistent inflammation making it difficult to reverse the disease. CXCL16, a chemokine, was also upregulated and may act a synergistic effect with M1 macrophages to promote CAG development.

Co-Localisation of Expression and Correlation Analysis of CD86, CD163 and CXCL16 Proteins

To investigate the impact of CXCL16 on macrophage polarisation, we conducted the positive cell co-localisation and correlation analyses of CD86 and CD163 with CXCL16 expression in gastritis tissues. Such analyses enabled us to demonstrate the relationship between macrophage polarisation and CXCL16 in gastritis. Results showed that in CNAG and CAG-E groups, the expression of CXCL16 was higher in CD86⁺ cells than in CD163⁺ cells, despite low expression of CXCL16 and sparse co-localised expression of CD86⁺CXCL16⁺ and CD163+CXCL16⁺ (P<0.05). In contrast, in the CAG group, CXCL16 expression was significantly higher in M1 (CD86⁺) macrophages than in M2 macrophages (CD163⁺) (Figure 9A and B, P<0.05). Pairwise comparisons of CXCL16 co-localization with CD86 and CD163 among the three gastritis groups revealed that the CAG group exhibited significantly higher CXCL16 expression in both CD86+ and CD163+ cells compared to the other groups (P<0.05). Next, co-localisation positive cell density assessment of CXCL16 expression on CD86⁺ and CD163⁺ cells in different degrees of inflammation was examined (Figure 9C), which indicated a similar association, with CD86⁺ CXCL16⁺ cell density [median density of 160.0/mm² (range 93.7–339.4)] being significantly higher in the CAG state than CD163⁺ CXCL16⁺ cells [median density of 90.4/mm² (range 0.0–128.8)]. In contrast, the CD86⁺ CXCL16⁺ cells were dispersed in CNAG and CAG-E with median densities and ranges of [2.4/mm² (0.0–4.0) and 2.4/mm² (0.0–2.4)], respectively. This suggested that the CXCL16 expression primarily affected M1 macrophage polarisation, but it did not affect the degree of inflammation.

Figure 9 Co-expression and Correlation of CXCL16 and macrophage markers in gastric tissue (A) mIHC images of CD86, CD163, and CXCL16 co expressed in CNAG, CAG, and CAG-E, respectively; (B) The percentage of CXCL16 expression observed on CD86+and CD163+cells at different levels of inflammation; (C) The density of positive cells showing co-localization of CD86, CD163, and CXCL16 expression; (D) Correlation analysis of CD86, CD163 and CXCL16 expression. * indicates P<0.05, **** indicates P<0.0001.

Next, we investigated the relationship between macrophage polarisation status and CXCL16 based on the CD86,CD163 and CXCL16 protein correlation analysis. The data indicatedthat there was no correlation between the expression of CD163 and CXCL16 in CAG (P>0.05). In contrast, CD86 was positively correlated with CXCL16 expression (Figure 9D, P<0.05). This analysis suggested that CXCL16 was primarily involved in the regulation of M1 macrophage polarisation and participates in M1 macrophage-mediated inflammatory environment of CAG.

CXCL16 Promotes Macrophage Polarization to M1 and Inhibits Its Polarization to M2

THP-1 is a human monocyte cell line that is often applied in research on inflammation and immune response. To determine the regulatory effect of CXCL16 on macrophage polarisation, the THP-1 cells were cultured in vitro and induced to form macrophages, after which the expression of CXCL16 and CXCR6 in cells was examined via immunofluorescence staining experiments. In these test, we observed strong fluorescent signals of CXCL16 and its sole receptor CXCR6 in macrophages (Figure 10A). The qRT-PCR results showed that the mRNA expression of the M1 macrophage marker CD86 was significantly up-regulated after macrophage treated with different concentrations of CXCL16 (0, 50, and 100ug/mL) compared with the A1 control group in a concentration-dependent manner. Notably, as the concentration of CXCL16 increased, the mRNA expression level of CD86 was also significantly up-regulated (P<0.05, Figure 10B). However, the mRNA expression level of the M2 macrophage marker CD163, gradually decreased, with the lowest expression levels observed in the CI group, indicating that the expression of CD163 was significantly down-regulated as the concentration of CXCL16 increased (P<0.05, Figure 10C). These results suggested that CXCL16 promoted macrophage polarisation to M1 type, and inhibited its polarisation to M2 type, and this effect is more pronounced as the concentration of CXCL16 increased.

Figure 10 The effects of CXCL16 on macrophage polarization (A) Immunofluorescence staining of CXCL16 and CXCR6 in macrophages; (B) mRNA expression of CD86 in macrophages incubated with CXCL16 at different concentrations; (C) mRNA expression level of CD163 in macrophages. * indicates P<0.05, ** indicates P<0.01, *** indicates P<0.001.

Taken together, these results provide important clues that will expand the current understanding the role of macrophage polarisation in different types of gastritis and the role of CXCL16 in regulating macrophage function.The expression pattern of CD86, CD163 and CXCL16 molecules at different stages of gastritis development suggests that these molecules may serve as potential early biomarkers for the onset of gastritis. Moreover, their sustained expression during the progression phase suggests a role in influencing the disease’s advancement.

Discussion

To the best of our knowledge, this is the first study to assess the expression of CXCL16 and macrophage markers in CAG through bioinformatics analysis and experimental validation. Specifically, we examined the impact of CXCL16 expression on macrophage polarisation and the potential role of immune responses in CAG formation. Our findings revealed that macrophage polarisation correlated closely with CXCL16 expression. We also found a correlation between chemokines and immune cells, of which both were linked to changes in the CAG microenvironment. These results highlight the potential regulatory mechanism of CXCL16 and its potential clinical utility as a novel target for CAG immunotherapy.

Besides affecting food digestion functions, CAG, a characteristic precancerous lesion, significantly increases the risk of Gastric Cancer (GC). Gastric mucosa atrophy could impair folate and vitamin B12 adsorption, increasing the risk of severe hematological illnesses such as pernicious anemia, along with various neurological, psychiatric, cognitive, and ischemic heart diseases.^21,22 Moreover, the global incidence rate of CAG has been increasing annually in recent years, particularly in younger patients, severely impacting their health and quality of life. Therefore, early screening and development of effective pharmacological interventions would be imperative for improved clinical outcomes. Macrophages—bone marrow-derived mononuclear phagocytic cells—can polarise into different functional subtypes under specific stimulation conditions, finely regulating and responding to various stimuli and ultimately impacting inflammation or disease pathogenesis.²³ They could also release numerous inflammatory cytokines, thus mediating innate immune responses. For instance, significant M1 M φ s activation could cause gastric mucosal damage and inhibit M2 M φ s polarisation, potentially resulting in a more severe gastric inflammation.²⁴ Furthermore, Zhou and Naqvi et al reported a significant upregulation of the M1/M2 ratio in the gingival tissue of patients with chronic periodontitis, a phenomenon that correlated with disease severity, whereas the expression of M1 macrophage markers were downregulated following periodontitis treatment.^25,26 Additionally, M1 macrophages were implicated in early lung inflammation and injury. Meanwhile, M2 macrophages promoted pulmonary fibrosis, with macrophage depletion exerting a contrary effect.²⁷ These studies link the dynamic changes in macrophage polarisation closely to chronic inflammation, although their specific regulatory mechanisms in CAG remain unclear. Given the pro- and anti-inflammatory effects of M1 and M2 macrophages, respectively, we hypothesised that the regulation of the M1/M2 ratio could be a promising therapeutic strategy for CAG.

Owing to recent advancements in pertinent technologies, bioinformatics has recently emerged as a vital clinical tool, particularly in exploring the diagnostic markers and biological processes of diseases. Herein, to explore the biological functions and potential target markers of CAG immunity, we first screened CAG-related datasets and immune genes using the GEO database, yielding 24 common CAG IGs. Enrichment analyses further suggested that these genes were mainly involved in immune cell differentiation, recruitment, and homing, intrinsic and adaptive immunity regulation, cytokine-cytokine receptor interactions, and chemokine signalling pathways. Therefore, the identified genes might regulate functional modules that are highly comparable to the physiological functions of chemokines and macrophages and correlate closely with CAG onset. The identified IGs were further subjected to protein interaction analyses, revealing that the target markers C86 and CD163, and the chemokine CXCL16, crucially influenced CAG onset.

Chemokines, key molecules that regulate immune cell migration and localisation, play an important role in immune and inflammatory responses. Notably, CXCL16, a member of the CXC chemokine family, exists in two forms: Transmembrane CXCL16 (mCXCL16) and soluble CXCL16 (sCXCL16). Whereas sCXCL16 is responsible for the chemotaxis of cells carrying the CXCR6 receptor,⁸ mCXCL16 mainly functions as an intercellular adhesion molecule and is often cleaved by ADAM10 to form sCXCL16, ultimately inducing CXCR6+ cell recruitment towards the lesion site^17,28—a phenomenon that has been established to regulate the pathological processes of several inflammatory illnesses. Aberrant CXCL16 expression was previously reported in various inflammatory tissues, serving as a marker and promoter of inflammation-related diseases. For instance, Rheumatoid Arthritis (RA) patients exhibited serum CXCL16 upregulation.²⁹ Additionally, CXCL16 overexpression was reported in patients with Acute Kidney Injury (AKI), with CXCL16 inhibition reducing pro-inflammatory factor production post-AKI.³⁰ Furthermore, M1 macrophage and CXCL16 upregulation was detected in blood samples from acute stone cholecystitis patients, promoting neutrophil migration and Neutrophil Extracellular Trap (NET) formation, with a reduction of CXCL16 levels and macrophage polarisation alleviating the disease.³¹ Moreover, CXCL16 was implicated in liver disease onset and progression, with CXCL16, CXCR6 and ADAM10 upregulation and significant CXCL16 upregulation observed in the liver during inflammatory responses and infectious shock, respectively.^32,33 These findings collectively suggest that CXCL16 is crucially involved in pro-inflammatory microenvironment Formation. Nonetheless, the biological functions and molecular mechanisms of CXCL16 in CAG remain unclear.

Herein, we examined different phenotypic macrophage markers, CD86 and CD163, as well as the chemokine CXCL16. Based on GEO data and mIHC staining, we hypothesised that macrophage polarisation in gastric mucosal tissues might correlate with CXCL16 expression (positive expression; and co-expression of CXCL16 and macrophage markers). We observed significant M1 macrophage upregulation in CAG with gawith a higher expression in CAG and CAG-E than in CNAG at the protein level. These findistritis progression. Furthermore, the expression of the M1 macrophage marker CD86 exhibited a stepwise increase, ngs align with a previous study, which reported that M1 macrophage activation aggravated gastric inflammation,²⁴ further confirming that M1 macrophages play a major role in CAG. Conversely, M2 macrophages were correspondingly upregulated in the early stage of inflammation, inhibiting inflammatory overreactions and protecting gastric mucosal tissues from further damage. Additionally, CXCL16 was upregulated in CAG, particularly in CD86+ macrophages. Correlation analysis further revealed that M1 macrophage polarisation correlated positively with CXCL16 expression, suggesting that CXCL16 might influence M1 macrophage polarisation and CAG onset. To test this hypothesis, we performed in vitro experiments, and the results confirmed that CXCL16 exerts a potent chemotactic effect on M1 macrophages, promoting M1 macrophage polarisation and inhibiting M2 polarisation in the CAG microenvironment. These findings suggest that inhibiting CXCL16 expression could suppress M1 macrophage polarisation, thus alleviating gastric mucosal injury—a phenomenon that crucially elucidates CAG pathogenesis. However, the mechanisms behind macrophage transformation in CAG and those through which CXCL16 regulates the M1 phenotype remain unknown. Macrophage-expressed CXCR6 could bind to CXCL16 through its specific motifs, thereby affecting leukocyte recruitment and adhesion processes.³⁴ therefore, CXCL16/CXCR6 axis activation might crucially regulate macrophage-mediated inflammatory responses. Herein, macrophages exhibited an elevated expression of CXCL16 and its specific receptor CXCR6, implying that the regulatory effect of CXCL16 on macrophage polarisation (M1/M2) observed in this study could have been achieved via CXCL16/CXCR6 signalling axis activation. This mechanism aligns with the function of CXCL16/CXCR6 in immunomodulation as reported in the previous literature.¹¹ Moreover, CXCL16 overexpression in gastric mucosal tissues correlated with CXCR6 and ADAM10 upregulation.¹¹ Additionally, pro-inflammatory cytokines were reported to increase sCXCL16 shedding through ADAM10.^17,28,35 Based on these studies, we inferred that CXCL16 upregulation in CAG might promote ADAM10 expression, which, in turn, could cleave CXCL16 to produce sCXCL16, thus initiating CXCR6 activation. After several cell signaling transductions, we found that CXCL16 induced M1 macrophage proliferation and migration to the lesion site, triggering inflammatory cytokine production and exacerbating gastric mucosal tissue inflammation. This process highlights the specific mechanism through which CXCL16 could regulate macrophage polarisation in CAG; the positive feedback loop established through the CXCL16/CXCR6 axis.

The significant correlation between the overexpression of inflammatory factors (such as TNF-α, IL-6, and IL-1 β) and CAG onset and progression is well-documented.^36,37 While TNF-α exerts the strongest destructive effect on gastric mucosa, IL-1 β might activate specific immune responses and regulate immune surveillance function.³⁸ In the present study, we found that macrophages were stimulated by CXCL16 with a significant increase in M1 macrophages, while TNF-α, IL-6 and IL-1β were mainly secreted by M1 macrophages. This suggests that CXCL16 induced MI macrophage polarisation in CAG, promoting the secretion of numerous pro-inflammatory factors and causing gastric mucosal damage.Pro-inflammatory factors such as TNF-α, IFN-γ, IL-1β and IL-6 can increase the expression of CXCL16, and our study found that the expression of CXCL16 and CXCR6 was up-regulated in macrophages, and CXCL16 has the property of an adhesion protein that can bind to its specific receptor CXCR6^39,40—a phenomenon that could increase M1 macrophage accumulation at the inflammation site. Therefore, the roles of M1 macrophages and CXCL16 in CAG onset and progression could be reciprocal. According to research, the nuclear factor-κB (NF-κB) pathway, a pro-inflammatory signalling pathway, could initiate and regulate the transcriptional expression of pro-inflammatory genes.⁴¹ Herein, we mined the GEO dataset and found that CAG-related IGs, including the chemokine CXCL16, were mainly enriched in the cytokine-cytokine receptor interaction and chemokine signalling pathways. The cytokine-cytokine receptor interaction pathway, in which NF-κB expression was up-regulated in CAG, intestinal metaplasia, and GC, regulates the apoptosis and proliferation of gastric epithelial cells, and also plays an important role in “inflammation-cancer transition”.⁴¹ The chemokine signalling pathway, on the other hand, delivers signals to cells through a series of molecular events, regulating cell migration, immune response, and inflammatory response initiation and maintenance.⁴² Additionally, CXCL16 activates the NF-κB pathway via trimeric G proteins, PI 3-kinase, Akt, and IκB kinases (PI3K, Akt, IKK, and IB phosphorylation) to induce TNF-α expression.⁴³ Based on these studies, we deduced that CXCL16 treatment might significantly activate the pro-inflammatory pathway (NF-κB pathway), induce MI macrophage polarisation, and release numerous inflammatory factors (eg, TNF-α, IL-6, and so on) that could cause gastric mucosa injury, thus promoting CAG onset and progression. In other words, CXCL16 interacts with immune cells through multiple mechanisms, regulates immune responses, and modulates CAG immunopathology. Overall, pro-inflammatory chemokines play an additional role in CAG as inflammatory promoters, and besides inflammatory factor upregulation, CAG occurrence and the effect of CXCL16 in macrophage polarisation regulation might also be dependent on NF-κB activation and the positive feedback regulation of the CXCL16/CXCR6 axis.

This study highlights the critical role of CXCL16 in CAG via macrophage polarisation regulation, presenting it as a promising target for preventing gastric mucosal inflammation aggravation. Therefore, CXCL16-neutralising antibodies or CXCR6 receptor antagonists could be leveraged to suppress inflammatory progression. However, antibodies that could specifically target CXCL16 remain clinically unavailable. Furthermore, although the recombinant Tumour Necrosis Factor Receptor:Fc (rhTNFR:Fc) fusion protein can attenuate inflammation in patients with ankylosing spondylitis via CXCL16/CXCR6 pathway inhibition,³¹ studies on its potential efficacy in downregulating CXCL16 expression during CAG progression are scarce, underscoring the need for additional research in the future.

Despite its valuable insights, this study has several notable limitations. First, due to the restricted disease types and research focus, the datasets included were limited in both scope and sample size. To further elucidate the diagnosis and immune responses associated with CAG, future research should prioritize selecting additional datasets from diverse databases and incorporating larger sample sizes. Second, macrophages in vivo typically exhibit distinct tissue-specific phenotypes. Our experimental validation was conducted using in vitro cultured macrophages without subsequent in vivo validation through mouse models. The growth processes and environmental conditions of ex vivo macrophages may differ significantly from those observed in vivo. Third, mIHC staining of CXCL16 and macrophage-related markers may demonstrate heterogeneity; additionally, depending on storage duration, loss of antigenicity in stored paraffin sections could impact results. Fourth, while our current findings provide preliminary insights into the relationship between CXCL16 and macrophage polarization, the observed dose-response pattern indicates a degree of specificity. However, establishing a direct causal relationship will require future studies employing functional knockout experiments (eg, CXCL16 neutralization or CXCR6 siRNA/CRISPR knockout) for more compelling validation. Furthermore, there is a lack of fundamental experimental verification regarding the transcriptional regulatory mechanisms underlying the expression patterns of CXCL16 and macrophage markers. Additionally, it remains unclear which molecular mechanisms or signaling pathways are involved by which CXCL16 modulates macrophage subpopulations in CAG. Consequently, further research will be necessary to deepen our understanding of potential biomechanisms. Many other chemokines could also lead to CAG onset and should be explored in future research. Moreover, the expression and potential functions of CXCL16 in other immune cells remain unclear.

Conclusion

In summary, this study employed bioinformatics techniques to elucidate the immunobiological functions and associated pathways related to CAG, identifying macrophage-related markers and their potential influencing factors. The findings revealed that the M1 macrophage marker CD86 and the chemokine CXCL16 are significantly upregulated in CAG, demonstrating a positive correlation between these two entities. Furthermore, experimental validation not only confirmed but also illuminated for the first time the potential role of CXCL16 in inducing M1 macrophage polarization while simultaneously suppressing M2 macrophage polarization. As a regulator of macrophage inflammatory responses, CXCL16 influences both the onset and progression of CAG. This discovery provides fundamental insights into how CXCL16 modulates immune responses in disease contexts, suggesting its potential as a novel target for regulating M1 macrophage polarization in CAG. It enhances our understanding of CAG pathogenesis and offers new perspectives for identifying biomarkers and therapeutic interventions. Moving forward, we can explore innovative diagnostic approaches and therapeutic strategies for CAG by targeting the regulation of macrophage polarization. Concurrently, implementing loss-of-function studies and animal models will further clarify the precise roles and molecular mechanisms of chemokines and macrophages in CAG, thereby advancing our comprehensive understanding of the disease’s pathogenesis.

Data Sharing Statement

The datasets analyzed in this study are available in the GEO under accession number [GSE153224](https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE153224) and [GSE27411] (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE27411), as well as in the ImmPort database (https://www.immport.org/shared/genelists). All data are publicly accessible.

Human Ethics and Consent to Participate Declarations

The study protocol was reviewed and approved for consent by the Ethics Committee of Gansu Provincial Hospital (Approval number: 2025-410) and was conducted in accordance with the principles of the Declaration of Helsinki.

Acknowledgments

The authors would like to thank all the reviewers who participated in the review and MJEditor (www.mjeditor.com) for its linguistic assistance during the preparation of this manuscript.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data and analysis, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This study was supported by the Gansu Provincial Health Industry Research Project (GSWSQN2025-04), the Research Project of the Eighth Affiliated Hospital of Southern Medical University (SRSP2025013) and the National Natural Science Foundation of China (81560093).

Disclosure

All authors declare that there are no competing conflicts of interest in this study.

References

1. Kothari S, Afshar Y, Friedman LS, et al. AGA clinical practice update on pregnancy-related gastrointestinal and liver disease. Exp Rev Gastroenterol. 2024;167(5):1033–1045.PMID: 39140906. doi:10.1053/j.gastro.2024.06.014

2. Song H, Ekheden IG, Zheng Z, et al. Incidence of gastric cancer among patients with gastric precancerous lesions: observational cohort study in a low risk Western population. BMJ. PMID: 26215280; PMCID: PMC4516137. doi:10.1136/bmj.h4134

3. Murray PJ, Wynn TA. Protective and pathogenic functions of macrophage subsets. Nat Rev Immunol. 2011;11(11):723–737.PMID:21997792; PMCID: PMC3422549. doi:10.1038/nri3073

4. Mohammadi A, Blesso CN, Barreto GE, et al. Macrophage plasticity, polarization and function in response to curcumin, a diet-derived polyphenol, as an immunomodulatory agent. J Nutr Biochem. 2019;66:1–16. PMID:30660832. doi:10.1016/j.jnutbio.2018.12.005

5. He L, Jhong JH, Chen Q, et al. Global characterization of macrophagepolarization mechanisms and identification of M2-type polarization inhibitors. Cell Rep. 2021;37(5):109955.PMID:34731634; PMCID: PMC8783961. doi:10.1016/j.celrep.2021.109955

6. Yang T, Wang R, Liu H, et al. Berberine regulates macrophage polarization through IL-4-STAT6 signaling pathway in Helicobacter pylori-induced chronic atrophic gastritis. Life Sci. 2021;266:118903. Epub 2020 Dec 16. PMID: 33340526. doi:10.1016/j.lfs.2020.118903

7. Chen X, Wan Z, Yang L, et al. Exosomes derived from reparative M2-like macrophages prevent bone loss in murine periodontitis models via IL-10 mRNA. J Nanobiotechnology. 2022;20(1):110.PMID: 35248085; PMCID: PMC8898524. doi:10.1186/s12951-022-01314-y

8. Matloubian M, David A, Engel S, et al. A transmembrane CXC chemokine is a ligand for HIV-coreceptor Bonzo. Nat Immunol. 2000;1(4):298–304.PMID: 11017100. doi:10.1038/79738

9. Kolaczkowska E, Kubes P. Neutrophil recruitment and function in health and inflammation. Nat Rev Immunol. 2013;13(3):159–175.PMID:23435331. doi:10.1038/nri3399

10. El-Zaatari M, Bishu S, Zhang M, et al. Aim2-mediated/IFN-β-independent regulation of gastric metaplastic lesions via CD8+ T cells. JCI Insight. 2020;5(5):e94035.PMID: 32053518; PMCID: PMC7141403. doi:10.1172/jci.insight.94035

11. Han J, Fu R, Chen C, et al. CXCL16 promotes gastric cancer tumorigenesis via ADAM10-Dependent CXCL16/CXCR6 axis and activates akt and MAPK signaling pathways. Int J Biol Sci. 2021;17(11):2841–2852.PMID: 34345211; PMCID:PMC8326113. doi:10.7150/ijbs.57826

12. Echizen K, Horiuchi K, Aoki Y, et al. NF-κB-induced NOX1 activation promotes gastric tumorigenesis through the expansion of SOX2-positive epithelial cells. Oncogene. 2019;38(22):4250–4263.PMID: 30700829; PMCID: PMC6756228. doi:10.1038/s41388-019-0702-0

13. Sabatti C, Service S, Freimer N. False discovery rate in linkage and association genome screens for complex disorders. Genetics. 2003;164(2):829–833.PMID: 12807801; PMCID:PMC1462572. doi:10.1093/genetics/164.2.829

14. Gooden MJ, Wiersma VR, Boerma A, et al. Elevated serum CXCL16 is an independent predictor of poor survival in ovarian cancer and may reflect pro-metastatic ADAM protease activity. Br J Cancer. 2014;110(6):1535–1544.PMID: 24518602; PMCID: PMC3960624. doi:10.1038/bjc.2014.55

15. Tabata S, Kadowaki N, Kitawaki T, et al. Distribution and kinetics of SR-PSOX/CXCL16 and CXCR6 expression on human dendritic cell subsets and CD4+ T cells. J Leukoc Biol. 2005;77(5):777–786.PMID: 15703197. doi:10.1189/jlb.1204733

16. Borst O, Abed M, Alesutan I, et al. Dynamic adhesion of eryptotic erythrocytes to endothelial cells via CXCL16/SR-PSOX. Am J Physiol Cell Physiol. 2012;302(4):C644–51.PMID: 22173866. doi:10.1152/ajpcell.00340.2011

17. O’Connor RA, Li X, Blumerman S, et al. Adjuvant immunotherapy of experimental autoimmune encephalomyelitis: immature myeloid cells expressing CXCL10 and CXCL16 attract CXCR3+CXCR6+ and myelin-specific T cells to the draining lymph nodes rather than the central nervous system. J Immunol. 2012;188(5):2093–2101.PMID: 22287719; PMCID:PMC3987917. doi:10.4049/jimmunol.1101118

18. Kong YF, Sha WL, Wu XB, et al. CXCL10/CXCR3 Signaling in the DRG exacerbates neuropathic pain in mice. Neurosci Bull. 2021;37(3):339–352.PMID: 33196963; PMCID:PMC7955022. doi:10.1007/s12264-020-00608-1

19. Ali RA, Saber LM, Al-Harbi AM. A novel association between IL1-Ra (receptor antagonist) gene polymorphism and T1DM in Al-Madina Al-Mounawra. Eur Rev Med Pharmacol Sci. 2015;19(19):3701–3708. PMID: 26502861.

20. Korbecki J, Kojder K, Simińska D, et al. CC chemokines in a tumor: a review of pro-cancer and anti-cancer properties of the ligands of receptors CCR1, CCR2, CCR3, and CCR4. Int J Mol Sci. 2020;21(21):8412.PMID:33182504; PMCID: PMC7665155. doi:10.3390/ijms21218412

21. Lachner C, Steinle NI, Regenold WT. The neuropsychiatry of vitamin B12 deficiency in elderly patients. J Neuropsychiatry Clin Neurosci. 2012;24(1):5–15.PMID: 22450609. doi:10.1176/appi.neuropsych.11020052

22. Ford ES, Smith SJ, Stroup DF, et al. Homocyst(e)ine and cardiovascular disease: a systematic review of the evidence with special emphasis on case-control studies and nested case-control studies. Int J Epidemiol. 2002;31(1):59–70.PMID: 11914295. doi:10.1093/ije/31.1.59

23. Yu B, Wu Y, Li Z. Retraction notice to KLF4/Ch25h axis activated by metformin suppresses EndoMT in human umbilical vein endothelial cells [Biochem. Biophys Res Commun. 2019;522(4):838–844].PMID: 34774192. doi:10.1016/j.bbrc.2021.10.011

24. Hardbower DM, Asim M, Murray-Stewart T, et al. Arginase 2 deletion leads to enhanced M1 macrophage activation and upregulated polyamine metabolism in response to Helicobacter pylori infection. Amino Acids. 2016;48(10):2375–2388.PMID: 27074721; PMCID: PMC5042810. doi:10.1007/s00726-016-2231-2

25. Zhou LN, Bi CS, Gao LN, et al. Macrophage polarization in human gingival tissue in response to periodontal disease. Oral Dis. 2019;25(1):265–273.PMID: 30285304. doi:10.1111/odi.12983

26. Uttamani JR, Kulkarni V, Valverde A, et al. Dynamic changes in macrophage polarization during the resolution phase of periodontal disease. Immun Inflamm Dis. 2024;12(10):e70044.PMID: 39436204; PMCID: PMC11494820. doi:10.1002/iid3.70044

27. Wynn TA, Vannella KM. Macrophages in tissue repair, regeneration, and fibrosis. Immunity. 2016;44(3):450–462.PMID: 26982353; PMCID: PMC4794754. doi:10.1016/j.immuni.2016.02.015

28. Wehr A, Baeck C, Heymann F, et al. Chemokine receptor CXCR6-dependent hepatic NK T Cell accumulation promotes inflammation and liver fibrosis. J Immunol. 2013;190(10):5226–5236.PMID: 23596313. doi:10.4049/jimmunol.1202909

29. Ma Z, Yu R, Zhu Q, et al. CXCL16/CXCR6 axis promotes bleomycin-induced fibrotic process in MRC-5 cells via the PI3K/AKT/FOXO3a pathway. Int Immunopharmacol. 2020;81:106035. PMID: 31753588. doi:10.1016/j.intimp.2019.106035

30. Liang H, Liao M, Zhao W, et al. CXCL16/ROCK1 signaling pathway exacerbates acute kidney injury induced by ischemia-reperfusion. Biomed Pharmacother. 2018;98:347–356. PMID: 29275176. doi:10.1016/j.biopha.2017.12.063

31. Zhang P, Zhou S, Chen Z, et al. TNF receptor: fc fusion protein downregulates rankl/opg ratio by inhibitingcxcl16/cxcr6 in active ankylosing spondylitis. Curr Pharm Biotechnol 2021. 2021;22(2):305–316.PMID: 32116188. doi:10.2174/1389201021666200302104418

32. Ma KL, Wu Y, Zhang Y, et al. Activation of the CXCL16/CXCR6 pathway promotes lipid deposition in fatty livers of apolipoprotein E knockout mice and HepG2 cells. Am J Transl Res. 2018;10(6):1802–1816. PMID:30018721; PMCID: PMC6038063.

33. Xu H, Xu W, Chu Y, et al. Involvement of up-regulated CXC chemokine ligand 16/scavenger receptor that binds phosphatidylserine and oxidized lipoprotein in endotoxin-induced lethal liver injury via regulation of T-cell recruitment and adhesion. Infect Immun. 2005;73(7):4007–4016.PMID: 15972488; PMCID: PMC1168580. doi:10.1128/IAI.73.7.4007-4016.2005

34. Engin AB, Engin A, Gonul II. The effect of adipocyte-macrophage crosstalk in obesity-related breast cancer. J Mol Endocrinol. 2019;62(3):R201–R222.PMID: 30620711. doi:10.1530/JME-18-0252

35. Schramme A, Abdel-Bakky MS, Kämpfer-Kolb N, et al. The role of CXCL16 and its processing metalloproteinases ADAM10 and ADAM17 in the proliferation and migration of human mesangial cells. Biochem Biophys Res Commun. 2008;370(2):311–316.PMID: 18373975. doi:10.1016/j.bbrc.2008.03.088

36. Wang B, Zhou W, Zhang H, et al. Exploring the effect of Weifuchun capsule on the toll-like receptor pathway mediated HES6 and immune regulation against chronic atrophic gastritis. J Ethnopharmacol. 2023;303:115930. PMID: 36403744. doi:10.1016/j.jep.2022.115930

37. Tong Y, Zhao X, Wang R, et al. Therapeutic effect of berberine on chronic atrophic gastritis based on plasma and urine metabolisms. Eur J Pharmacol. 2021;908:174335. PMID: 34265298. doi:10.1016/j.ejphar.2021.174335

38. Tong Y, Liu L, Wang R, et al. Berberine attenuates chronic atrophic gastritis induced by MNNG and its potential mechanism. Front Pharmacol. 2021;12:644638. PMID: 33841162; PMCID: PMC8026873. doi:10.3389/fphar.2021.644638

39. Shimaoka T, Kume N, Minami M, et al. Molecular cloning of a novel scavenger receptor for oxidized low density lipoprotein, SR-PSOX, on macrophages. J Biol Chem. 2000;275(52):40663–40666.PMID: 11060282. doi:10.1074/jbc.C000761200

40. Hofnagel O, Engel T, Severs NJ, et al. SR-PSOX at sites predisposed to atherosclerotic lesion formation mediates monocyte-endothelial cell adhesion. Atherosclerosis. 2011;217(2):371–378.PMID: 21612780. doi:10.1016/j.atherosclerosis.2011.04.021

41. Lawrence T. The nuclear factor NF-kappaB pathway in inflammation. Cold Spring Harb Perspect Biol. 2009;1(6):a001651.PMID: 20457564; PMCID: PMC2882124. doi:10.1101/cshperspect.a001651

42. Xu M, Wang Y, Xia R, et al. Role of the CCL2-CCR2 signalling axis in cancer: mechanisms and therapeutic targeting. Cell Prolif. 2021;54(10):e13115.PMID: 34464477;PMCID: PMC8488570. doi:10.1111/cpr.13115

43. Chandrasekar B, Bysani S, Mummidi S. CXCL16 signals via Gi,phosphatidylinositol 3-kinase, Akt, I kappa B kinase, and nuclear factor-kappa B and induces cell-cell adhesion and aortic smooth muscle cell proliferation. JBiol Chem. 2004;279(5):3188–3196.PMID: 14625285. doi:10.1074/jbc.M311660200