CD44 is associated with papillary thyroid carcinoma metastasis via potential modulation of the immunosuppressive tumor microenvironment

Abstract

To investigate the role of CD44 in papillary thyroid carcinoma (PTC) lymph node metastasis and its connotation with Treg-mediated immunosuppression through TGF-β and IL-10 signaling. Eighty-two benign thyroid nodule patients were compared with 122 PTC patients. Differentially expressed proteins (DEPs) were identified in both tissue and serum. CD44 expression was validated by ELISA, IHC and flow cytometry. The proportion of Treg and cytokines were analyzed in parallel. The tumor microenvironment architecture was examined using 3D tissue clearing and whole-mount imaging. The functional role of CD44 was further explored using TPC-1 human PTC cell line through gain-of-function and loss-of-function assays in vitro. Proteomic analysis identified 34 tissue and 17 serum DEPs involved in metastasis. CD44 levels in serum were significantly higher in PTC patients than in controls (345.45 ± 44.88 pg/mL vs. 73.33 ± 25.64 pg/mL, P < 0.001). Furthermore, CD44 expression in the serum of CLNM patients was further elevated (782.01 ± 168.38 pg/mL vs. 248.77 ± 55.33 pg/mL, P < 0.001). Multivariate analysis identified CD44 as an independent risk factor for PTC (OR = 5.271, 95% CI: 1.741 ~ 15.959) and CLNM (OR = 3.995, 95% CI: 1.298 ~ 12.298). CD44 levels showed positive correlations with increased TGF-β, IL-10 levels, and Treg frequency. CD44 levels were connected with amplified TGF-β, IL-10, and Treg frequency. IHC and 3D imaging also demonstrated colocalization of CD44⁺ tumor cells and Tregs in H&E sections and whole-mount. Importantly, CD44 knockdown suppressed, while CD44 overexpression enhanced, TPC-1 cell proliferation in vitro, suggesting a direct role of CD44 in promoting tumor cell growth. Our data suggest that CD44 may contribute to PTC metastasis by modulating the immunosuppressive microenvironment through TGF-β/IL-10 mediated Treg accumulation and by directly driving tumor cell proliferation. These findings highlight a potential pathogenic role of CD44 in PTC progression.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10238-026-02101-x.

Introduction

Papillary thyroid carcinoma (PTC) is the highest mutual subtype of thyroid cancer and represents more than 80% of thyroid cancers worldwide [1]. There were an estimated 586,000 thyroid carcinoma patients worldwide in 2020 [2]. In the absence of cervical lymph node metastasis (CLNM), distant metastasis, or extrathyroidal extension [3], PTC typically follows an indolent course [4]. However, with lymph node involvement and distant metastasis, PTC has a poor prognosis. When considering whether or not to undergo operation for thyroid cancer, the presence of CLNM is recommended by the American Thyroid Association as a treatment for PTC [5]. Both ultrasound and CT had value in detecting thyroid cancer metastasis, but with different performance; ultrasound had higher sensitivity (85–90%) [6] but lower specific (70–75%) for different lymph node metastases especially for cervical lymph nodes [7]. CT had better specific (80–85%) than ultrasound did, although with moderately lower sensitivity (75–80%) for assessment of distantly metastatic disease [5]. Therefore, biomarkers are crucial for identifying small metastases undetectable by conventional imaging modalities such as ultrasound and CT. It can help us to identify PTC patients with greater risks of metastasis and make decision to select surgery or not. So, we firstly performed the proteomic analysis on metastatic thyroid cancer tissues and nonmetastatic thyroid cancer tissues and adjoining tissues as well as serum to identify differentially expressed proteins (DEPs). And then we chose CD44 as the potential biomarker to test. CD44 is overexpressed in many other cancer types including PTC, as well as breast cancer, colorectal carcinoma, gastric cancer, pancreatic ductal adenocarcinoma and glioblastoma etc [8, 9]. Chu KH et al. reported that CD44 expression was associated with Treg function [10]. CD44 high Tregs enlarge after injury and are more oppressive than CD44low Tregs [11]. Tregs, previously identified as suppressor T cells, are a subset of CD4 + T cells that control the activity of the immune system and direct a transcription issue identified as forkhead box P3 (Foxp3), which is a specific marker for Tregs. Tregs and Th17 cells allocate mutual precursor cells (naive CD4 + T cells) and both of them need a mutual tumor growth factor-β (TGF-β) signal to mediate original discrepancy [12]. Tregs depend on the IL-10 signaling pathway to maintain expression of FoxP3, survival and suppressive activity [13, 14].

We summarized key investigational drugs targeting CD44 and its variants (CD44v5, CD44v6), including antibody-drug conjugates (ADCs), monoclonal antibodies, and CAR-T therapies. Listed parameters include drug name, type, target, mechanism of action, indication, highest development phase, timeline, developing institution, and relevant references. These agents demonstrate the clinical potential of CD44 inhibition across multiple cancer types, supporting its role as a therapeutic target in oncology [15–17] (Supplementary Table 1).

Although CD44 has been implicated in PTC progression and metastasis, the precise mechanisms by which it modulates the tumor microenvironment, particularly through immune regulation, remain unclear. Given that CD44high Tregs exhibit enhanced immunosuppressive activity, we hypothesize that CD44 promotes immune escape in PTC via the TGF-β/IL-10/Foxp3 signaling axis, thereby facilitating lymph node metastasis. Furthermore, we postulate that CD44 may also exert a direct mitogenic effect on PTC cells. This study aims to explore potential associations between CD44 expression, TGF-β/IL-10 signaling, Treg accumulation, and metastatic progression in PTC using a multi-modal correlative approach, and these findings may provide preliminary insights for the future development of targeted therapies to inhibit metastatic progression in PTC patients.

Materials and methods

Clinical study design

A total of 204 patients were included, 82 with benign thyroid nodule and 122 with histologically proven PTC (50 with PTC-CLNM, 73 with PTC-NCLNM) who presented at the First Affiliated Hospital of Xinjiang Medical University from December 2021 to January 2024 (Fig. 1). Patients with autoimmune diseases or history of malignancies were excluded as well as patients with incomplete clinical data. Recorded clinical data included age, sex, body mass index (BMI), smoking, nodules (single/multiple), aspect ratio, thyroid capsule invasion, boundary (fuzzy/clear), blood flow, echogenicity (hypoechoic/mixed and strong echo), characteristic of calcification.

Fig. 1.

Flowchart of this study. A total of 204 patients (82 with nodular goiter (NG) and 122 with papillary thyroid carcinoma (PTC), including 49 with PTC-cervical lymph node metastasis (CLNM) and 73 with PTC-noncervical lymph node metastasis (NCLNM) were analyzed. Proteomic screening was performed on 24 tissue and serum samples (NG = 3, PTC-CLNM = 13, PTC-NCLNM = 8), followed by validation using enzyme-linked immunosorbent assay (ELISA), flow cytometry, and immunohistochemistry (IHC). Bioinformatics analysis, hematoxylin and eosin (H&E) staining, and 3D tissue clearing were conducted to examine CD44 expression and its correlation with IL-10, TGF-β, and Tregs. Statistical analysis included both univariate and multivariate approaches

Fresh tumor tissues, adjacent normal tissues and serum instances were composed during surgery closely snap-frozen in liquid nitrogen and stored at − 80 °C. Clinicopathological parameters (tumor size, extrathyroidal extension and lymph node status) were recorded. Sample size was calculated based on G*Power (α = 0.05, power = 0.8, effect size = 0.5).

Bioinformatics analysis

Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and protein‒protein interaction (PPI) network studies were executed using the PTM-BIO Shiny tool and STRING (v11.0, confidence score ≥ 0.7). Functional enrichment was as-sessed via Fisher’s exact test (P < 0.05).

Enzyme-linked immunosorbent assay (ELISA)

Blank wells (sample diluents), standard wells (gradient-diluted standards) and test sample wells were set on the microplate. 100 µL of reagent were added into each well individually (avoid bubble; finished in 5–10 min; final volume of 1 mL per well). The plate was sealed and incubated at 37 °C for 2 h. Subsequently, the sealing film was removed, the liquid was decanted, and the wells were tapped dry on absorbent paper. Reservoir was washed 4× with 350–400 µL of 1× buffer, and samples were blotted to be dry. 100 µL of detection antibody were added into each well, samples were sealed and incubated at 37 °C for 1 h. Reservoir was washed 4× with 350–400 µL of 1× buffer, and samples were blotted to be dry. 100 µL of HRP-streptavidin were added into each well, sample was sealed, and mixture was incubated at 37 °C for 40 min. Reservoir was washed, and 100 µL of TMB substrate were added into each well. Mixture was incubated at 37 °C (in the dark for 15–20 min). Finally, 100 µL of stop solution were added into each well (color changed to yellow). OD was measured at 450 nm within 5 min since the moment of coloring, and data was analyzed at 450 nm.

Hematoxylin and Eosin (H&E) staining

Preparation of paraffin sections of benign thyroid nodule, PTC cancer, thyroid cancer adjacent tissue, noncervical lymph node metastasis (NCLNM), and cervical lymph node metastasis samples were made, and heated in the 60 °C oven for 1 h. Samples were dewaxed in xylene and dehydrated in ethanol for 10 min in xylene, 5 min in 100% ethanol, 5 min in 90% ethanol, 5 min in 80% ethanol, 5 min in 70% ethanol, and 5 min in distilled water. Samples were stained for 5 min in hematoxylin, and then washed for 1 min in distilled water. Instances were stained for several seconds in 1% HCl, stained in eosin for 10 min, and next dehydrated in gradient ethanol, cleared in xylene, and dried and sealed in neutral resin. Morphological and structural changes in cells were observed, and images were acquired under light microscope.

3D tissue clearing

Paraformaldehyde fixed tissues were washed with PBS for 24 h (change every 2 h) and subsequently cleared in 50% Solution A (37 °C, 70–80 rpm, 6 h), followed by 100% Solution A (28 days). After a few washes with PBS (24 h with buffer changes every 2 h), tissues were incubated with primary antibodies (CD44, Abcam Cat# AB316123, RRID: AB_3665134; CD25, Abcam Cat# ab238272, RRID: AB_3677706; CD127, Santa Cruz Biotechnology Cat# sc-514445, RRID: AB_3662581) in PBS/0.5% Triton-X100/0.02% NaN₃ (RT, 4 weeks). After PBS washing, tissues were incubated in fluorescent secondary antibody (Abcam Cat# ab175652, RRID: AB_2687498; Abcam Cat# ab150116, RRID: AB_2650601) solutions (light-protected, 4 weeks), followed by final PBS wash and incubation in Solution B (7–10 days) until fully transparent. Tissues were then immersed in imaging solution (refraction index: 1.523) for light-sheet microscopy [18].

Immunohistochemistry (IHC)

Immunohistochemical staining was expended to examine the protein illustration and distribution of CD44 (Proteintech Cat# 15675-1-AP, RRID: AB_2076198), TGF-β (Proteintech Cat# 21898-1-AP, RRID: AB_2811115) and IL-10 (Proteintech Cat# 60269-1-Ig, RRID: AB_2881389) in PTCs with or without CLNM and benign thyroid nodule tissues. After being incubated with 3% H2O2 for 10 min, CD44 polyclonal antibody (Proteintech Cat# 15675-1-AP, RRID: AB_2076198), TGF-β polyclonal antibody (Proteintech Cat# 21898-1-AP, RRID: AB_2811115) and IL-10 monoclonal antibody (Proteintech Cat# 60269-1-Ig, RRID: AB_2881389) was added, and the samples were incubated for 12 h at 4 °C. Then, the samples were incubated with secondary antibodies for 1 h at 37 °C, and stained with diaminobenzidine. Negative control (PBS instead of primary antibody) was performed in parallel with the samples.

Flow cytometry

Peripheral leukocytes were separated by Ficoll-Paque gradient centrifugation. The PBMCs from PTC patients (with/without CLNM) and benign thyroid nodule patients were resuspended in PBS at 2 × 106 cells/tube, centrifuged (400 g, 10 min, 4 °C) and evaluated by flow cytometry (BD LSRFortessa). Tregs were stained with CD4 (BioLegend Cat# 300518, RRID: AB_314086), CD25 (BioLegend Cat# 302606, RRID: AB_314276) and CD127 (BioLegend Cat# 351342, RRID: AB_2564137) antibodies (30 min, 4 °C, dark) for 30 min and then stained with intracellular Foxp3 staining (FITC anti-human Foxp3) after permeabilization. Finally, the cells were washed with PBS (400 × g, 10 min) resuspended in 350 µL PBS and acquired (5,000 lymphocytes/tube). The data were analyzed with Kaluza software: CD4 + T cells were gated (FSC-A/SSC-A, CD4 APC-A750-A) and the number of Tregs (CD25 + CD127−) was calculated in this population.

Cell viability assay

To discover the functional outcome of CD44 in the process of PTC progression, we performed in vitro cell viability assays based on the TPC-1 human papillary thyroid carcinoma cell line. The cells were transfected into the following groups: TPC-1: Wild type cells. TPC-1EV: Empty vector transfected control. TPC-1 - CD44 - KD: CD44 knockdown group (transfected with shRNA targeting CD44). TPC-1 - CD44 - OE: CD44 overexpression group (transfected with CD44 expression plasmid). Cell viability was assayed in line with the manufacturer’s directions for the cell counting kit-8 (CCK-8) assay (Dojindo, Japan). Absorbance was measured at 450 nm using a microplate reader (BioTek, USA). Each experiment was staged in triplicate and recapped three times autonomously.

Statistical analysis

The data are displayed as the means ± SDs. Continuous serum biomarkers (CD44 and thyroglobulin) were dichotomized into ‘high’ and ‘low’ expression groups for logistic regression analyses. The median value of the entire PTC cohort was used as the cut-off point. For comparison between groups, Student’s t test (normal distribution) or the Mann‒Whitney U test (nonnormal distribution) was used. Fisher’s precise test was expended for categorical variables. Spearman’s rank test was expended to analyze the correlations (P < 0.5 indicated significance). SPSS v25.0 (IBM) was used for the analyses.

Results

Identification of DEPs in PTC patients

We used the proteomics study to classify the DEPs. In the cancer tissues, we found 34 differentially expressed proteins, and 19 proteins were upregulated and 15 proteins were downregulated (p < 0.05, Fig. 2A (a)). In the serum samples, we found 17 differentially expressed proteins, and 13 proteins were upregulated and 4 proteins were downregulated (p < 0.05). CD44 was upregulated in the serum samples. (b) According to the heatmap from the unsupervised hierarchical clustering analysis, the proteomics profiles of the PTC-CLNM cancer tissue samples were significantly different from those of the PTC-NCLNM tissue samples (Fig. 2A (c)) and from those of the serum samples (Fig. 2A (d)).

Fig. 2.

Proteomic profiling identifies metastasis-associated differentially expressed proteins (DEPs) and reveals their functional enrichment in Papillary Thyroid Carcinoma (PTC) (A) Identification and unsupervised clustering of DEPs in tumor tissues and serum. (a), DEPs in cancer tissue between patients of PTC with cervical lymph node metastasis (PTC-CLNM) and PTC with noncervical lymph node metastasis (PTC-NCLNM). Red dots indicate upregulated proteins and blue dots indicate downregulated proteins. (b), DEPs in serum between patients of PTC-CLNM and PTC-NCLNM. (c), The representative heatmap of DEPs in cancer tissue between PTC with/without CLNM. Red and blue indicate higher and lower proteins expression, respectively. (d), The representative heatmap of DEPs in serum samples between PTC with/without CLNM. (e), upregulated biological process (BP), cellular component (CC) and molecular function (MF) ontologies in tissue between PTC with/without CLNM. blue dots BP, yellow dots CC and green dots MF. (f), downregulated BP, CC and MF ontologies in tissue between PTC with/without CLNM. (g), upregulated BP, CC and MF ontologies in serum between PTC with/without CLNM. (h), downregulated BP, CC and MF ontologies in serum between PTC with/without CLNM. (B) Functional annotation and pathway analysis highlight key biological processes dysregulated in metastatic PTC. Proteins related to biological process in cancer tissue (a) and in serum sample (f). Proteins related to cellular processes in cancer tissue (b) and in serum sample (g). Proteins related to molecular function in cancer tissue (c) and in serum sample (h). Protein domain in richment-based clustering analysis in cancer tissue (d) and in serum sample (i). Proteins related to KEGG pathway in cancer tissue (e) and in serum sample (j)

Enrichment analysis of functions of DEPs

Next, to explore the functions of the DEPs between PTC-CLNM and PTC-NCLNM, we used the GO enrichment analysis. The enriched GO terms with a P value < 0.05 for upregulated proteins included biological process (BP), cellular component (CC), and molecular function (MF) ontologies (Fig. 2A (e)); we also found the GO (BP, CC, and MF) terms enriched in downregulated proteins between PTC-CLNM and PTC-NCLNM tissues (Fig. 2A (f)) and between cancer tissues and serum samples (Fig. 2A (g)- (h)).

Analysis of GO, protein domain, KEGG in PTC-CLNM and PTC-NCLNM. In cancer tissue, the DEPs were enriched in the following terms: biological procedures: antigen managing/presentation (MHC class I) and TCA/citrate metabolism (Fig. 2B (a)); cellular components: cell surface, coated vesicles and endocytic membranes (Fig. 2B (b)); molecular functions: amide/peptide binding and oxidoreductase activity (Fig. 2B (c)). Protein domain analysis showed enrichment in the immunoglobulin C1/V-set domains (Fig. 2B (d)). Enriched KEGG paths contained graft-versus-host disease and antigen managing (Fig. 2B (e)). In the serum samples, the DEPs were enriched in the following terms: biological processes: leukocyte chemotaxis, the inflammatory response and protein processing (Fig. 2B (f)); cellular components: main axon (Fig. 2B(g)); molecular function: peptide hormone binding (Fig. 2B(h)). Protein domain analysis showed sushi domain enrichment (Fig. 2B (i)). Enriched KEGG pathways included complement/coagulation cascades and Epstein–Barr virus (EBV) infection (Fig. 2B (j)).

Discriminative features and independent predictors of PTC

Supplementary Table 2 demographic, sonographic and serological profiles of patients in NG group (n = 82) and PTC group (n = 122). Univariate analysis PTC patients were more probable to be female (52.4% vs. 68.9%, p = 0.018) and younger than 45 years of age (20.7% vs. 49.2%, p < 0.001). Classically, a taller-than-wide aspect ratio (9.9% vs. 58.8%, p < 0.001), thyroid capsule invasion (3.7% vs. 38.8%, p < 0.001), ill-defined boundaries (2.4% vs. 27%, p < 0.001), hypoechogenicity (32.1% vs. 81.5%, p < 0.001) and presence of calcifications (12.3% vs. 47.9%, p < 0.001) are more frequently observed in malignant thyroid nodules. In our series, these sonographic features were markedly more frequent in PTC group. In addition, serological analysis showed that CD44 (73.33 vs. 345.45 pg/mL, p < 0.001), CD44 ≥ 214 pg/mL (29.3% vs. 63.9%, p < 0.001), IL-10 (3.94 vs. 18.31 pg/mL, p < 0.001) and TGF-β (1685.81 vs. 4709.87 pg/mL, p < 0.001) serum levels were significantly increased in PTC patients.

Then, based on variables with noteworthy dissimilarities in univariate study, we further constructed a multivariate logistic regression model (Supplementary Table 3, Fig. 3A (a)). Results from the multivariate logistic regression study displayed that female (OR = 3.864, 95% CI: 1.266 ~ 11.794, p = 0.018), age < 45 years (OR = 3.56, 95% CI: 1.162 ~ 10.913, p = 0.026), aspect ratio (OR = 10.272, 95% CI: 3.055 ~ 34.533, p < 0.001), Thyroid capsule invasion (OR = 8.347, 95% CI: 1.134 ~ 61.450, p = 0.037), Hypoechoic (OR = 5.298, 95% CI: 1.926 ~ 14.58, p = 0.001), calcifications (OR = 11.660, 95% CI: 2.662 ~ 51.061, p = 0.001) and high serum levels of CD44 ≥ 214 pg/mL (OR = 5.271, 95% CI: 1.741 ~ 15.959, p = 0.003) and IL-10 (OR = 1.080, 95% CI: 1.009 ~ 1.156, p = 0.026) were independent predictors for PTC.

Fig. 3.

CD44 is an independent risk factor for papillary thyroid carcinoma (PTC) metastasis, associated with elevated immunosuppressive cytokines, and organizes the metastatic niche. (A) (a), Logistic regression analysis of risk factors for PTC. (b), Logistic regression analysis of risk factors for cervical lymph node metastasis (CLNM). (B) Serum levels of CD44, TGF-β, and IL-10 are significantly elevated in PTC patients and further increased in those with CLNM. (a), The serum expression levels of CD44 between nodular goiter (NG) and PTC groups. (b), The serum expression levels of IL-10 between NG and PTC groups. (c), The serum expression levels of TGF-β between NG and PTC groups. (d), The serum expression levels of CD44 between PTC-CLNM and PTC with noncervical lymph node metastasis (PTC-NCLNM). (e), The serum expression levels of IL-10 between PTC-CLNM and PTC-NCLNM. (f), The serum expression levels of TGF-β between PTC-CLNM and PTC-NCLNM. (C) Representative hematoxylin and eosin (H&E) staining of thyroid and lymph node tissues (scale bars: 100 μm). (a) Benign thyroid nodule showing regular follicular architecture with uniform epithelial lining and colloid-filled lumens. (b) Papillary thyroid carcinoma displaying characteristic features: papillary fronds with fibrovascular cores, nuclear enlargement, overlapping, and longitudinal nuclear grooves. (c) Non-metastatic lymph node with preserved architecture including germinal centers and sinuses. (d) Metastatic lymph node from PTC patient, showing replacement of normal lymphoid tissue by tumor nests and intralymphatic tumor emboli. D. 3D tissue clearing images of the PTC microenvironment, obtained by confocal microscopy, reveal key spatial interactions. (a), Merged multichannel image with a composite view of the integrated landscape of the nuclei (DAPI, blue), CD44 + tumor cells (green), and CD25 + regulatory T cells (Tregs, red) followed by (b) CD44 + tumor cells (green) with highlighting of stemness-associated metastatic morphology and (c) DAPI-stained nuclei (blue) as an architectural context of cellular distribution; (d), CD25⁺ Tregs (red) illustrating immunosuppressive niche formation.(*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ^NSP > 0.05)

Risk Factors for cervical lymph node metastasis in PTC

We next investigated factors predictive of CLNM within the PTC cohort. As shown in Table 1, comparative analysis between patients without (NCLNM, n = 73) and with (CLNM, n = 49) metastasis revealed that the CLNM group was significantly younger (41.1% vs. 61.2%, p = 0.029), aspect ratio (46.5% vs. 77.1%, p = 0.001), and exhibited substantially higher preoperative serum levels of thyroglobulin (Tg) (9.125 vs. 17.5 mIU/L, p < 0.001), Tg ≥ 10.55 mIU/L (38.2% vs. 65.8%, p = 0.009), CD44 (248.77 vs. 782.01 pg/mL, p < 0.001), CD44 ≥ 367.25 pg/mL (54.8% vs. 77.6%, p = 0.010), and IL-10 (3076.66 vs. 14352.82 pg/mL, p < 0.001).

Table 1.

General characteristics of PTC-NCLNM and PTC-CLNM patients

Characteristics		PTC-NCLNM (n = 73)	PTC-CLNM (n = 49)	t/Z/χ2	P
Gender	Male	22 (30.1%)	16 (32.7%)	0.087	0.769
	Female	51 (69.9%)	33 (67.3%)
Age	≥ 45	43 (58.9%)	19 (38.8%)	4.753	0.029*
	< 45	30 (41.1%)	30 (61.2%)
BMI (kg/m2)	< 24	31 (42.5%)	22 (44.9%)	0.071	0.790
	≥ 24	42 (57.5%)	27 (55.1%)
Smoking	No	65 (89%)	45 (91.8%)	0.258	0.761
	Yes	8 (11%)	4 (8.2%)
Nodules	Single	20 (28.2%)	11 (22.9%)	0.410	0.522
	Multiple	51 (71.8%)	37 (77.1%)
Aspect ratio	< 1	38 (53.5%)	11 (22.9%)	11.074	0.001*
	≥ 1	33 (46.5%)	37 (77.1%)
Thyroid capsule invasion	No	50 (68.5%)	24 (50.0%)	4.169	0.041*
	Yes	23 (31.5%)	24 (50.0%)
Boundary	Clear	57 (78.1%)	32 (65.3%)	2.435	0.119
	Fuzzy	16 (21.9%)	17 (34.7%)
Blood flow	No blood flow	13 (18.8%)	6 (12.5%)	2.999	0.223
	Slight blood flow	51 (73.9%)	34 (70.8%)
	Rich blood flow	5 (7.2%)	8 (16.7%)
Echogenicity	Mixed and Strong echo	16 (22.5%)	6 (12.5%)	1.914	0.167
	Hypoechoic	55 (77.5%)	42 (87.5%)
Calcification	No	39 (56.5%)	22 (45.8%)	1.296	0.255
	Yes	30 (43.5%)	26 (54.2%)
TSH [M (IQR), mIU/L]		2.26 (2.12)	2.28 (2.13)	-0.511	0.609
Tg [M (IQR), mIU/L]		9.125 (10)	17.5 (33.37)	-2.438	< 0.001*
LMR [x ± s, mIU/L]		4.73 ± 1.43	4.54 ± 1.33	1.292	0.199
CD44 [M (IQR), pg/mL]		248.77 (256.03)	782.01 (797.37)	-4.045	< 0.001*
CD44 [pg/mL]	< 214	33 (45.2%)	11 (22.4%)	6.585	0.010*
	≥ 214	40 (54.8%)	38 (77.6%)
IL-10 [M (IQR), pg/mL]		3076.66 (3197.85)	14352.82 (12239.85)	-6.140	< 0.001*
TGF-β [M (IQR), pg/mL]		18.31 (14.39)	19.15 (17.46)	-1.142	0.254

PTC = Papillary Thyroid Carcinoma; NCLNM = noncervical lymph node metastasis; CLNM = cervical lymph node metastasis; BMI = body mass index; TSH = thyroid stimulating hormone; Tg = thyroglobulin; LMR = lymphocyte-to-monocyte ratio; IL-10 = interleukin 10; TGF-β = tumor growth factor-β; *P-value < 0.05 was regarded as statistically significant

A multivariate logistic reversion model was established to adjust for potential confounders as follows: aspect ratio OR = 4.215, 95% CI: 1.196 ~ 14.853, p = 0.025, serum level of CD44 ≥ 367.25 pg/mL OR = 3.995, 95% CI: 1.298 ~ 12.298, p = 0.016, Tg ≥ 10.55 mIU/L OR = 4.291, 95% CI: 1.351 ~ 13.626, p = 0.013, which were also meaningly in univariate analysis (Table 2; Fig. 3B (b)). Finally, the multivariate logistic regression model identified the serum level of CD44 ≥ 367.25 pg/mL as an independent predictive factor for CLNM, indicating that CD44 expression is not only closely related to tumorigenesis but also displays a vital character in the metastatic process of PTC.

Table 2.

Multivariate logistic regression analysis of risk factors of CLNM in PTC patients

Characteristics	B	SE	Waldχ2	P	OR (95%CI)
Gender (Female)	-0.150	0.518	0.084	0.772	0.860 (0.311 ~ 2.377)
Age (< 45 years)	1.029	0.490	4.408	0.035	2.798 (1.071 ~ 7.312)
BMI (≥ 24 kg/m2)	-0.041	0.509	0.006	0.936	0.960 (0.354 ~ 2.605)
Aspect ratio (≥ 1)	1.693	0.522	10.500	0.001	5.436 (1.952 ~ 15.134)
Thyroid capsule invasion (Yes)	0.501	0.480	1.087	0.297	1.650 (0.644 ~ 4.231)
Echogenicity (Hypoechoic)	0.897	0.705	1.618	0.203	2.451 (0.616 ~ 9.761)
Calcification (Yes)	0.639	0.468	1.870	0.171	1.895 (0.758 ~ 4.739)
CD44 (≥ 214 pg/mL)	1.383	0.509	7.369	0.007	3.987 (1.469 ~ 10.823)
IL-10 (pg/mL)	0.033	0.020	2.634	0.105	1.033 (0.993 ~ 1.075)

CLNM = cervical lymph node metastasis; PTC = Papillary Thyroid Carcinoma; OR = Odds Ratio; BMI = body mass index; IL-10 = interleukin 10; *P-value < 0.05 was regarded as statistically significant

CD44, TGF-β and IL-10 levels are higher in PTC patients than NG patients

There were expression level differences of CD44 (99.96(73.86 ~ 251.11) pg/mL vs. 362.31(151.50 ~ 945.28 pg/mL, p < 0.001), IL-10 (7.31(3.81 ~ 15.89) pg/mL vs. 17.98(10.39 ~ 25.25) pg/mL, p < 0.001), and TGF-β (1807.29(1153.94 ~ 2598.90) pg/mL vs. 3813.92(2503.67 ~ 10428.67) pg/mL, p < 0.001) in the serum between the NG patients and PTC patients. There were expression level differences of CD44 (259.35(118.63 ~ 519.70) pg/mL vs. 676.15(205.67 ~ 998.18) pg/mL, p < 0.001) and TGF-β (2938.03(2255.11 ~ 4605.15) pg/mL vs. 10518.43(4175.45 ~ 16776.33) pg/mL, p < 0.001) in the serum of PTC-CLNM patients PTC-NCLNM patients (Fig. 3B).

H&E staining and 3D tissue clearing displayed the thyroid follicle morphology and lymph node metastatic features in PTC

Histopathological assessment confirmed diagnostic features and metastatic patterns (Fig. 3C). (a) shows a benign thyroid nodule with uniform follicular architecture, regular cuboidal epithelium, and colloid-filled lumens. In contrast, (b) displays classic papillary thyroid carcinoma histology: papillary fronds with fibrovascular cores, and characteristic nuclear features including: [1] nuclear enlargement (2–3 × normal) [2], nuclear overlapping and crowding [3], nuclear membrane irregularity [4], longitudinal nuclear grooves and [5] occasional pseudoinclusions (not shown). (c) illustrates a non-metastatic lymph node with preserved architecture, including germinal centers and cortical sinuses. (d) demonstrates cervical lymph node metastasis from PTC, where metastatic tumor nests replace normal lymphoid tissue. Tumor cells within lymphatics confirm vascular invasion as a mechanism of dissemination.

The 3D tissue clearing images (Fig. 3D series) allow us to reconstruct confocal volumes displaying important spatial relationships within the PTC microenvironment using confocal microscopy: [1] CD44 + tumor cells (green) with invasive protrusions containing irregular clusters and filopodial extensions, suggesting epithelial‒mesenchymal transition (EMT)-driven metastatic potential; [2] CD25 + Foxp3+ Tregs (red) localized around lymphatic vessels to form immunosuppressive niches, and the significant CD44-Treg colocalization suggested that PTC might evade the antitumor immune response via TGF-β/IL-10 signaling; and [3] tumor-lymphatic interactions with architectural complexity undetectable by conventional 2D histology CD44, TGF-β, and IL-10 expression correlate with the disease severity of PTC. The high-resolution multicolor imaging (DAPI: nuclei; CD44: tumor stemness; CD25: Tregs) enabled volumetric analysis of Treg infiltration depth and tumor cell spatial organization, supporting CD44 as both a metastatic biomarker and therapeutic target. This 3D visualization platform overcomes limitations of dense stromal imaging in PTC, providing novel insights into microenvironmental drivers of progression (supplementary video).

CD44, TGF-β, and IL-10 expression correlate with the disease severity of PTC

To investigate whether the illustration of CD44/TGF-β/IL-10 is associated with PTC metastasis, we randomly selected PTC with CLNM(n = 30), PTC without CLNM(n = 30), and benign thyroid nodule patients(n = 30). Immunohistochemistry analysis was performed on PTC tissues and benign thyroid nodule tissues. CD44 expression was negative in benign thyroid nodule tissues and positive in most PTC tissues, and CD44 staining was more intense in PTCs with CLNM. Compared with that in benign thyroid nodule patients, TGF-β and IL-10 staining was more intense (higher signal) in PTC patients, and the staining intensity in the group with CLNM was greater than that in the group without CLNM (Fig. 4A).

Fig. 4.

CD44 promotes an immunosuppressive microenvironment and drives tumor cell proliferation in PTC (A) Immunohistochemical analysis reveals upregulated CD44, IL-10, and TGF-β protein expression in PTC tissues, particularly with lymph node metastasis. (a-c), The negative control of PBS instead of first antibody. (d-f), The immunohistochemical results of CD44, IL-10 and TGF-β expression in the thyroid nodule tissue of patients in the nodular goiter (NG) group. (g-i), The immunohistochemical results of CD44, IL-10 and TGF-β expression in the thyroid nodule tissue of patients in the papillary thyroid carcinoma with noncervical lymph node metastasis (PTC-NCLNM) group. (j-l), The immunohistochemical results of CD44, IL-10 and TGF-β expression in the thyroid nodule tissue of patients in the PTC with cervical lymph node metastasis (PTC-CLNM) group. (B) Flow cytometry demonstrates increased circulating Treg frequency in PTC patients, which escalates with metastatic progression. (a-c), The proportion of Treg cells among CD4⁺ T cells in the blood of patients from the NG, PTC-NCLNM, and PTC-CLNM groups. (d) The quantitative comparison of Treg cell proportions in the blood between NG and PTC patients. (e) The quantitative comparison of Treg cell proportions in the blood between PTC-NCLNM and PTC-CLNM patients. (C) Genetic manipulation of CD44 confirms its direct role in enhancing PTC cell proliferation in vitro. Comparison of cell viability among TPC-1-CD44-KD, TPC-1-CD44-KD empty vector control (TPC-1-EV), and CD44 overexpression (TPC-1-CD44-OE) groups. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001)

Increased circulating Treg cell infiltration correlates with lymph node metastasis in PTC

Flow cytometric analysis demonstrated a marked expansion of CD4⁻CD25⁺CD127⁻Foxp3⁺ Tregs in the peripheral blood of PTC patients. The Treg frequency in PTC-CLNM patients (8.03%) was approximately 1.6-fold greater than that in PTC-NCLNM patients (4.97%, p < 0.01) and 3.14-fold greater than that in benign controls (2.55%, p < 0.001). The proportion of Treg cells was significantly higher in the PTC-CLNM group than in the PTC-NCLNM group, indicating a correlation between peripheral Treg frequency and metastatic status in PTC patients. (Fig. 4B).

CD44 knockdown suppresses PTC cell viability

To investigate whether CD44 affects the proliferation of tumor cells directly, we transfected CD44 into TPC-1 cells and analyzed cell viability. The results in FIGURE 4C clearly demonstrated that CD44 knockdown (TPC-1-CD44-KD) markedly decreased cell viability equaled with the wild-type (TPC-1) and empty vector control (TPC-1-EV) groups (p < 0.0001). Meanwhile, CD44 overexpression (TPC-1-CD44-OE) markedly enhanced cell viability compared with the control groups (p < 0.001). The outcomes indicated that CD44 boosted PTC cell proliferation in vitro. Therefore, CD44 is a driver of tumor development (Fig. 4C).

Discussion

Our analysis identified CD44 as a biomarker associated with PTC-CLNM that has been mediated by the regulatory Treg cells that are modulated by CD44 through TGF-β and IL-10. Proteomic and functional studies displayed that the illustration of high dose CD44 was found in the metastatic tissues and serum of PTC and associated with the infiltration of Tregs and presence of the immunosuppressive cytokines. Such results place CD44 as a new biomarker to predict CLNM and indicate its medicinal interest. We discuss the following four points: [1] the innovative use of CD44 as a CLNM biomarker [2], the mechanistic use of CD44 and Treg-based metastasis [3], the CD44-induced IL-10/TGF-β signal [4], the transformative usage of 3D imaging in the study of PTC microenvironment, and [5] the enhancement of PTC cell proliferation using CD44 in vitro.

Elevated CD44 is associated with CLNM in PTC and May complement current diagnostic approaches

Although CD44 levels were elevated in PTC compared to benign nodules, the primary translational relevance of our findings lies in its association with metastatic progression rather than in distinguishing benign from malignant disease. The concept of the tumor microenvironment has evolved alongside advances in tumor immunology. The complex tumor microenvironment comprises tumor cells, various stromal and immune cells, extracellular matrix, blood vessels, lymphatic vessels, and signaling molecules. The tumor microenvironment phenotype has a close relationship with the occurrence, growth, and metastasis of tumors. CD44 has become an innovative preoperative risk marker in PTC patients. In comparison to the traditional imaging modalities, which are unable to identify micro-metastases, CD44 quantification permits molecules to provide accuracy. Our proteomic data are consistent with other reports that EMT correlates with stemness in aggressive thyroid cancer and overexpression of CD44 [9]. The innovation is related to a significant clinical problem because no validated biomarkers are available to predict early CLNM. The specificity and sensitivity of CD44 in a variety of cohorts should be confirmed in future multicenter studies.

CD44 is a glycosylated transmembrane glycoprotein having a strong association with the tumor growth and metastasis [19]. CD44 is also involved in cancer cell division, proliferation, invasion, and angiogenesis and also metabolism alterations through mediating many signaling pathways such as protein kinases, cytoskeleton alterations, intracellular processes, proteinases, and transcription factors [9]. It has been demonstrated that the overexpression of CD44 in thyroid cancer is more than that of benign thyroid nodules and that this overexpression is linked to the oncogenic conversion of ERK signaling path [9]. Han-Ning Li et al. established that CD44 makes a contribution to PTC carcinogenesis and development [20]. Besides, CD44 has also been noted to be conveyed in prospectively found thyroid carcinoma stem cells. When injected orally into the mouse thyroid it could cause tumors [21].The effects of CD44 and its isoforms in tumor progression indicate that CD44 can be used to treat cancer through molecular therapy [22]. Nevertheless, there is no clear indication of the correlation between the CD44 expression and the PTC metastasis. To counter this, first, proteomic screening was run on CD444 to find labeled protein overexpression. Later clinical correlation experiments and validation studies confirmed that Huge CD44 was highly expressed in serum and tumor tissues of PTCs and was thus connected with the development of PTCs and its metastatic progression. Moreover, we suggest that CD44 can enhance PTC invasiveness by stimulating Treg concentration through the IL-10/ TGF-β signaling pathway.

As an active molecule, CD44 is involved in supporting cancer stem cell groups in tumor microenvironment (TME) [23]. Tregs communicate with the infiltrating cell subsets of the immune system, stromal and tumor cells in the TME, and exhibit immunosuppressive activities [24]. The PTC patients were found to have higher contents of CD44, TGF-β, IL-10 and Treg cells than the benign thyroid nodule patients and the level was superior in patients with PTC/CLNM than the patients with PTC/CLNM.

CD44 expression is associated with Th17/Treg imbalance, which May contribute to tumor progression

The Treg with Th17 cells is pertinent in ensuring immune homeostasis in the human body and is strictly keyed under normal health conditions. By activating TGF-β signaling pathway, CD44 could facilitate the difference of both Tregs and Th17 cells which results in an immunosuppressive microenvironment and excretion of pro-inflammatory cytokine (i.e., IL-17, IL-6 and IL-23), increasing the aggressiveness of tumors [25, 26]. Th17 cells increase the immune reaction, while the Treg cells suppress the immune reaction. Treg-mediated immune suppression occurs in two important pathways, one is by suppressing the effector cells in tumor-draining lymph nodes and the other is through inhibiting the capability of effector T lymphocytes to kill tumor cells. Imbalance between Th17/Treg ratio may facilitate tumor immune escape that present as immunosuppression and disregulation of immune checkpoint receptor. Meanwhile, CD44 leads to the release of inflammatory factors including CXCL8 which contributes to angiogenesis and metastasis even more. In the TME, Tregs are able to express granzyme B that inhibits anti-tumor immunity by causing granzyme B/perforin-mediated lysis of the natural killer (NK) cells and cytotoxic T lymphocytes (CTLs). Moreover, being a hyaluronan receptor, CD44 can directly increase the tumor cell motility and immune evasion through immunosuppression mediated by Tregs, which eventually leads to metastasis of thyroid cancer [12].

Tregs are important in ensuring homeostasis of the immune system and self-tolerance. Tregs also participate in the regulation of autoimmunity, contagion, graft-versus-host disease, inflammation, fetal-maternal tolerance as well as tumor immunity [27]. Tregs have the ability to expand well in site in reaction to tumor-secreted factors (TGF-β/ IL-10) in the TME [28]. Tregs can have a suppressive role through many forms of contact-dependent and contact-independent mechanisms. These pathways comprise expression of suppressive cytokines (TGF-β, IL-10, and IL-35), immune checkpoints, and inhibitory receptors (CTLA-4, PD- 1, LAG-3, TIM-3, ICOS, TIGIT, and IDO), express cytotoxicity and metabolic interference with T effector cell activity (IL-2 consumption) [29, 30]. The Tregs have been intensively researched on the peripheral blood and tissues of various cancers. This build-up of Foxp3 + Tregs in the tumor tissue, ovarian cancer, pancreatic ductal adenocarcinoma [30], lung cancer [31], glioblastoma [32], non-Hodgkin’s lymphoma [33], melanoma and other tumor tissues [34] is linked up with a poor prognosis. Our analysis exposed that the percentage of the Treg cells is correlated with the tumor progression. The proportions of Treg cells in the tissues and peripheral blood of the patients with PTC were high in comparison with those in patients with benign nodules of the thyroid, in addition to being significantly increased. The percentage of Tregs was even more enhanced in the conditions of cervical lymph node metastasis in the case of PTC.

Metastatic immune evasion logical driver: the CD44-Treg axis

The correlation between CD44 and Treg infiltration represents a plausible mechanism for CLNM progression. CD44high Tregs have superior immunosuppressive functions by TGF-β/ IL-10 signaling [11]. Our flow cytometry resolution revealed that PTC-CLNM had high levels of CD45RA-Foxp3 + Tregs, which indicated that CD44 mediates the recruitment of Tregs through the hyaluronan attachment. This process resembles the case of pancreatic cancer where the CD44 + Tregs propagate immune tolerance [35]. Anti-CD44 antibodies targeting this axis have the potential to interfere with Treg accumulation, which would recommend a new treatment approach.

CD44 mediates IL-10/TGF-β cross-linkage to maintain immunosuppression

Synergy between CD44 and IL-10 and those between CD44 and TGF-β creates a loop of immunosuppression. TGF-β leads to Foxp3 + Treg differentiation [36] and IL-10 maintains the suppressive action of the Tregs [14]. ELISA and IHC results indicated the coupregulation of these cytokines in metastatic PTC which was probably facilitated by the effects of CD44 on nuclear factor-kappa B (NF- κB). The result is consistent with results in glioblastoma models, in which the interaction of CD44 and TGF-β helps improve Treg infiltration [37, 38]. The combination of TGF-β and CD44 receptor dual blockage can be used to suppress metastasis.

3D imaging: a Spatial revolution in the PTC microenvironment analysis

We discussed that one of the potential opportunities of optical clearing dense connective tissues is being able to discover new microscopic and cellular features who are neither easily observed nor measured with conventional microscopy or histomorphometry techniques. 3D tissue clearing technology offers in-depth ideas into the dynamics of metastatic niches previously unseen. Conventional 2D histopathology does not provide the spatial Treg interactions with tumors, and our 3D -reconstructions displayed the Treg-clustering around lymphatic vessels. The same methods in melanoma pointed out the spatial arrangement of Tregs as a predictor of the metastasis [39]. Combining 3D imaging and spatial transcriptomics would potentially permit to map gradients of CD44 expression as well as identify hotspots of immune evasion and evolve personalized therapeutic design.

Clinical development of CD44- targeted therapies

Tranlation of CD44 is further supported by the current clinical activities aimed at using this pathway. Various therapeutics against the CD44 antigen like humanised monoclonal antibodies like SPL-108 (discovered by Roche and others) have reached clinical trial. Such agents specifically target the CD44 by binding to inhibit the downstream signal transduction, thus suppressing the growth and survival of CD44-positive cancer hyperplastic cells- a process which is consistent with our results on the CD44-mediated Treg-mediated immunosuppression in PTC. These treatments can be used alongside current interventions (i.e., anti-TGF-β/anti-IL-10), to reduce a metastatic course, which provides a two-pronged intervention to subcut the immune suppressive tumour microenvironment in advanced PTC [17].

CD44 increases the PTC cell proliferation in vitro

Direct in vitro evidence that we gathered in our functional assay denotes that CD44 promotes the proliferative ability of PTC cells. The dramatic loss of cell viability caused by the CD44 knockdown and the accommodation of cell viability by the overexpression of CD44 is a good indication of the role of the oncogenic CD44 gene in thyroid carcinogenesis. These results are also in line with the earlier reports in other malignancies wherein CD44 facilitates the existence and proliferation of tumor cells by interacting with hyaluronan and oncogenesis signaling paths such as phosphatidylinositol 3-kinase/protein kinase B (PI3K/AKT) and mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) [40]. Further, the proliferative advantage of CD44 can promote the mesenchymal niches, which further makes CD44 a crucial agent in PTC aggressiveness and lymph node metastasis.

Limitations

Although this study provides insights into the role of CD44 in PTC metastasis, several limitations should be acknowledged: (1) Study Design and Generalizability: The retrospective, single-center design may introduce selection bias and limit the generalizability of our findings. Future prospective, multicenter studies are needed to validate the clinical utility of CD44 as a biomarker. (2) Mechanistic Evidence: While we observed a correlation between CD44 expression and Treg infiltration, direct causal mechanisms remain to be fully elucidated. Further functional experiments, such as in vivo CD44 manipulation and Treg depletion studies, are required to establish causality. (3) Technical Constraints: The 3D tissue clearing technique, though innovative, requires specialized equipment and expertise, which may hinder its widespread clinical application in the near future. (4) Sample Size Considerations: Although our sample size met statistical power requirements, larger cohorts are needed to confirm the predictive value of CD44 across diverse patient populations. (5) Lack of Animal Models: The study lacks in vivo validation using animal models, which would strengthen the translational relevance of our findings.

Despite our efforts to control for known confounders, residual confounding remains possible. First, BRAF V600E mutation status—a strong predictor of PTC aggressiveness—was not available for adjustment. Second, detailed histopathological features (e.g., tumor budding, lymphovascular invasion) were not consistently documented. Third, we cannot exclude the possibility of confounding by unmeasured lifestyle or environmental factors. However, the magnitude of the association between CD44 and metastasis (OR > 3.5) makes it less likely that unmeasured confounders fully explain our findings.

Conclusions

Our study indicates that CD44, TGF-β, IL-10, and Treg cells are associated with PTC progression and metastasis. Further investigation into their interactions may help identify potential therapeutic targets for advanced PTC.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PTC

Papillary Thyroid Carcinoma

CLNM

Cervical Lymph Node Metastasis

Tregs

Regulatory T cells

TGF-β

Tumor Growth Factor-β

IL-10

Interleukin 10

DEPs

Differentially Expressed Proteins

NG

Nodular Goiter

ADCs

Antibody-Drug Conjugates

PTC-NCLNM

PTC with Noncervical Lymph Node Metastasis

IHC

Immunohistochemistry

H&E

Hematoxylin and Eosin

Foxp3

Forkhead box P3

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

PPI

Protein‒Protein Interaction

BP

Biological Process

CC

Cellular Component

MF

Molecular Function

EBV

Epstein–Barr Virus

TME

Tumor Microenvironment

BMI

Body Mass Index

TSH

Thyroid Stimulating Hormone

Tg

Thyroglobulin

LMR

Lymphocyte-to-Monocyte Ratio

PBMCs

Peripheral Blood Mononuclear Cells

CCK-8

Cell Counting Kit-8

Author contributions

Conceptualization, G-L.D.; methodology, Y.A., M-H.H., L.M., X-X.L. and Y.T.; validation, C.B., B-W.W. and X.Q.; formal study, Y.A., M-H.H. and G-L.D.; investigation, X-Y.Y., L-L.M., X-X.G. and J-Y.K.; resources, X-X.L., Y.T., H-Z.L., B-R.Q., R.W., S.J. and G-L.D.; data curation, Y.A., M-H.H. and L.M.; writing—original draft preparation, Y.A.; writing—review and editing, G-L.D.; visualization, J-L.B., S.J. and G-L.D.; supervision, J-L.B., S.J. and G-L.D.; project administration, J-L.B., S.J. and G-L.D.; funding acquisition, B-R.Q., R.W., J-L.B., S.J. and G-L.D. All authors have read and agreed to the issued variety of the manuscript.

Funding

This research was funded by the Outstanding Youth Science Foundation Project (2021D01E28); the National Natural Science Foundation of China (81960078); the State Key Laboratory of Pathogenesis, Prevention and Treatment of High Incidence Diseases in Central Asia, Xinjiang Medical University (SKL-HIDCA-2024-1, SKL-HIDCA-2024-BZ5, SKL-HIDCA-2024-BZ18); the Xinjiang Youth Science and Technology Top Talents Special Project (2022TSYCCX0103); the Xinjiang Science and Technology Innovation Team (Tianshan Innovation Team, 2022TSYCTD0014); the Xinjiang Bayingolin Mongolian Autonomous Prefecture “Open bidding for selecting the best candidates” project (202427) and Science and Technology Research Plan (202515), the Xinjiang Uygur Autonomous Region Health Science and Technology Program Projects (2025001MXICSYTSTGXM652828708), and the Xinjiang Tianshan Talent Medical and Health High-level Talent Project (TSYC202401B046); Science and Technology Aid Program for Xinjiang (2025E02043).

Data availability

The data offered in the report are accessible on demand from the consistent writer. The study dataset has removed mode, but not all safeguarded health material. The dataset is also considered property of First Affiliated Hospital of Xinjiang Medical University and not owned by the researchers. Nevertheless, upon request, we are ready to share rations of the data for proper evaluation. Requirements can be made through the consistent author (Guo-Li Du; email: genemagic@126.com).

Declarations

Competing interests

The authors declare no competing interests.

Conflict of interest

The writers affirm that the study was managed in the absenteeism of any profitable and economic relations that can be interpreted as a possible encounter of relevance.

Institutional review board statement

The revision was directed in accord with the Declaration of Helsinki, and ratified by the Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University (Approval No. K202303-47).

Informed consent statement

Patient consent was waived as the project symbolized a non-interventional study applying routinely gathered data for secondary research devotions.