ADAM19 Promotes EMT‐Related Changes and Serves as a Prognostic Marker in Circulating Tumor Cells of Renal Cell Carcinoma

ABSTRACT

Detection of circulating tumor cells (CTCs), particularly characterization of their epithelial‐mesenchymal transition (EMT) phenotypes, may improve prognostic accuracy for malignancies. Identification of novel biomarkers for CTC classification may improve the characterization of tumor progression and provide potential therapeutic insights for renal cell carcinoma (RCC). Herein, we identified that a disintegrin and metalloproteinase domain‐containing protein 19 (ADAM19) was associated with EMT phenotypes of CTCs. In vitro functional experiments, Western blotting, and RNA sequencing demonstrated that ADAM19 promoted RCC malignant progression and EMT‐related changes and was associated with NOTCH and NF‐κB signaling pathways. Our findings further suggested potential bidirectional association between ADAM19 and these signaling pathways. CTCs isolated from xenograft mouse models exhibited enhanced proliferative, migratory, and invasive abilities, accompanied by increased ADAM19 expression compared with parental tumor cells. Clinical CTC analysis further revealed that ADAM19‐Low‐CTCs were associated with unfavorable clinical outcomes in RCC patients. Notably, higher time‐updated ADAM19‐Low‐CTC counts were identified as an independent risk factor for progression‐free survival (HR = 1.243, 95% CI: 1.006–1.536).

ADAM19 is associated with EMT‐related malignant phenotypes and NOTCH/NF‐κB pathway activity in RCC and may serve as a prognostic biomarker in CTCs.

Abbreviations

ADAM19

A disintegrin and metalloprotease domain‐containing protein 19

AUC

Area under curve

CTCs

Circulating tumor cells

DEGs

Differentially expressed genes

E‐CTC

Epithelial circulating tumor cell

EMT

Epithelial‐mesenchymal transition

GEO

Gene Expression Omnibus

GEPIA

Gene expression profiling interactive analysis

GSEA

Gene Set Enrichment Analysis

H‐CTC

Hybrid circulating tumor cell

M‐CTC

Mesenchymal circulating tumor cell

MSigDB

Molecular Signatures Database

PFS

Progression free survival

RCC

Renal cell carcinoma

ROC

Receiver operator characteristic

TSS

Transcription start site

1. Introduction

Renal cell carcinoma (RCC), originating from the renal tubular epithelium, accounts for approximately 90% of all primary renal tumors. The latest data indicate that RCC accounts for approximately 2.2% of all newly diagnosed cancers worldwide [1]. Due to its often asymptomatic or nonspecific presentation, RCC is often diagnosed at advanced stages when patients present with abdominal pain, hematuria, or an abdominal mass; many of them are diagnosed incidentally while undergoing routine physical examination [2, 3]. Current treatment guidelines generally recommend surgical resection as the first‐line therapy for RCC. However, tumor recurrence and metastasis occur in approximately 30% of postoperative patients. Therefore, early detection of recurrence or metastasis is of great significance for intervening in tumor progression and improving prognosis [4].

Circulating tumor cells (CTCs) represent a promising noninvasive prognostic biomarker [5]. Higher CTC counts before and after treatment are associated with worse prognosis in nasopharyngeal carcinoma, and higher mesenchymal CTC counts are significantly associated with poorer overall survival [6]. In addition, CTC detection is of great significance in making treatment planning. Metastatic breast cancer patients with HER2‐positive CTCs tend to gain more benefit from anti‐HER2 treatment, such as longer median progression‐free survival (PFS) [7]. Furthermore, dynamic changes in CTC counts or phenotypes serve as indicators of treatment efficacy, as seen in prostate cancer patients in whom a decline in CTCs following abiraterone or chemotherapy translates to improved survival [8]. As potential seeds of metastasis, CTCs provide important insights into tumor progression. Previous studies have indicated that CTCs are closely related to epithelial‐mesenchymal transition (EMT), which promotes the migration and invasion of tumor cells [9]. Classic EMT markers such as E‐cadherin, N‐cadherin, Vimentin, and Twist are widely used, yet they exhibit considerable heterogeneity across tumor types and might fail to capture the full spectrum of EMT plasticity in CTCs [10, 11]. Therefore, identifying novel EMT‐related markers is critical for improving the sensitivity of CTC‐based prognosis and for uncovering targetable drivers of metastatic progression.

In this study, we aimed to identify a novel EMT‐related marker in CTCs and to investigate its clinical significance. Through bioinformatics analyses, we identified a disintegrin and metalloproteinase domain‐containing protein 19 (ADAM19) as a candidate EMT‐associated gene. ADAM19 belongs to ADAM family and has been reported to be involved in epithelial–myofibroblast transdifferentiation [12]. Previous studies indicated that ADAM19 shares cleavage functions similar to those of ADAM17 and ADAM15, which play important roles in tumor progression [13]. We further evaluated its expression pattern, functional roles in RCC progression, and clinical relevance as a potential prognostic marker of CTCs in RCC.

2. Material and Methods

2.1. Bioinformatics Analysis

mRNA profiles of primary tumors and CTCs from four malignancies (breast cancer: GSE188893; prostate cancer: GSE106363; non‐small cell lung cancer: GSE50991; colon cancer: GSE82198) were obtained from the GEO database. Differentially expressed genes (DEGs) between primary tumors and CTCs were identified using thresholds of |log₂FC| ≥ 1.5 and adjusted p < 0.05, and overlapping DEGs were further analyzed. Prognostic and correlation analyses were performed in Gene expression profiling interactive analysis (GEPIA3) [14]. EMT gene sets were obtained from the Molecular Signatures Database (MSigDB), and pan‐cancer CTC mRNA/lncRNA expression data were retrieved from ctcRbase [15]. ChIP‐seq data from the ChIP‐Atlas database were used to identify potential transcription factors occupancy near the ADAM19 transcription start site (TSS), and candidate binding motifs were analyzed [16].

2.2. Isolation and Culture of 786‐O‐CTCs

CTCs were isolated using density gradient centrifugation as previously described [17, 18]. 786‐O cells were injected into the tail veins of BALB/c‐nude (5 × 10⁷/100 μL). After 30 days, blood samples were collected from the heart and diluted with an equal volume of PBS. CTCs were isolated using Ficoll density gradient centrifugation (500 g, 25 min, room temperature), washed with PBS, and resuspended in RPMI 1640 complete medium for culture in 24‐well plates (Figure 7a). The figure was partly generated using Biovisart and BioGDP [19]. The animal experimental protocol was approved by the Xi'an Jiaotong University Animal Ethics Committee (No. XJTUAE2025‐1030). All animal procedures were conducted in accordance with the ARRIVE guidelines.

FIGURE 7.

786‐O‐CTCs exhibit enhanced malignant phenotype (a). Isolation and culture of 786‐O‐CTCs (b). 786‐O‐CTCs exhibit increased proliferative capacity compared with parental 786‐O cells (c). 786‐O‐CTCs exhibit enhanced migratory and invasive abilities compared with parental 786‐O cells (d). ADAM19 is highly expressed in 786‐O‐CTCs (e). ADAM19 knockdown inhibits proliferation of 786‐O‐CTCs (f). ADAM19 knockdown inhibits migration and invasion of 786‐O‐CTCs. Data are presented as the mean ± SD and are representative of at least three independent experiments. nsp > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

2.3. Culture and Transfection of Cells

The 786‐O, ACHN, OS‐RC2, and HK‐2 cell lines were obtained from Procell (China). 786‐O and OS‐RC2 cells were cultured in RPMI 1640 complete culture medium (10% fetal bovine serum, 1% penicillin–streptomycin), while ACHN and HK‐2 cells were cultured in DMEM/F12 complete medium (10% fetal bovine serum, 1% penicillin–streptomycin). All cells were cultured at 37°C, 5% CO₂.

ADAM19 overexpression plasmids and siRNAs targeting indicated genes were synthesized by GenePharma (China) and Genecarer (China), respectively. Cells were transfected at 60% (plasmids) or 40% (siRNAs) confluency using the corresponding reagents. Protein extraction was performed 48 h after transfection. NF‐κB and NOTCH signaling pathways were activated using PMA (100 nM, GlpBio, USA) and Jagged‐1 (5 μg/mL, MedChemExpress, USA), respectively.

2.4. Western Blotting

Total proteins were extracted using RIPA lysis buffer and then centrifuged to remove the precipitate. The protein concentration was determined by BCA assay. A total of 30 μg of protein was loaded per lane for electrophoresis. After transfer, 5% skim milk was used for blocking, and the primary antibodies (ADAM19, GAPDH, N‐cadherin, E‐cadherin, Vimentin, Slug, NICD, Hes1, p‐P65, P65, p‐IκBα, IκBα) were incubated overnight at 4°C. The membranes were washed, incubated with secondary antibodies for 1 h, and protein bands were visualized using a ChemiDoc MP imaging system (Bio‐Rad, USA) for imaging.

2.5. Cell Viability Assay

Cell viability was assessed using the CCK‐8 assay kit (Agbio, China). Cells were seeded into 96‐well plates at a density of 3000 cells/well. After inoculation with CCK‐8 reagent diluted in culture medium (1:10) for 2 h, absorbance was measured at 450 nm using a microplate reader.

2.6. Transwell Assay

For migration assays, pretreated cells were resuspended at 2.5 × 10⁵ cells/mL, and 200 μL of cell suspension was seeded into the upper chamber of Transwell inserts. The lower chamber contained 600 μL of medium containing 10% FBS. After 24 h, migrated cells were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet, and quantified under a microscope. For invasion assays, Transwell chambers were precoated with Matrigel (Corning, USA, diluted 1:6) and cells were seeded at 5 × 10⁵ cells/mL for 48 h. The subsequent procedures were performed as described for the migration assay.

2.7. Wound Healing Assay

Cells were seeded into six‐well plates and grown to 90%–100% confluence. A linear scratch was generated in the center of the plate using a 100 μL pipette tip, followed by washing with PBS and incubation in serum‐free medium. Images were captured at 0, 12, and 24 h after scratching, and wound closure was analyzed to evaluate cell migration.

2.8. RNA Sequencing and Bioinformatics Analysis

Total RNA was isolated from cells using TRIzol reagent (Takara Bio, Japan) according to the manufacturer's instructions. RNA quality was assessed using NanoDrop 2000 and Agilent 4200 TapeStation. RNA‐seq libraries were prepared and sequenced on an Illumina platform by SHBIO Technology (Shanghai, China). Differential expression was performed using DESeq2 with thresholds of |log2FC| > 1 and P _adj < 0.05. Gene Ontology (GO) enrichment analysis was performed using the clusterProfiler package in R (version 4.2.2). Gene Set Enrichment Analysis (GSEA) was conducted with default parameters.

2.9. Exogenous co‐Immunoprecipitation (Co‐IP)

A total of 293 T cells were separately transfected with tagged overexpression plasmids for ADAM19, NOTCH1, TNF‐α, or corresponding negative‐control plasmids. After 48 h, cells were lysed using NP‐40 lysis buffer, and the lysates were incubated with tag‐specific antibody and protein A/G magnetic beads. Immunoprecipitated proteins were detected by Western Blotting.

2.10. Clinical and CTC Detection Data

Clinical data of RCC patients who underwent postoperative CTC detection at the Department of Urology, the Second Affiliated Hospital of Xi'an Jiaotong University (Dec 2024–Sep 2025) were retrospectively collected. Patients with metastatic disease at diagnosis or previous neoadjuvant/adjuvant therapy were excluded. Clinical characteristics, including gender, age, tumor stage, pathological type, and tumor diameter, were obtained. Peripheral blood samples (5 mL/patient) were collected for CTC enrichment using the CanPatrol CTCs enrichment technology (China SurExam Biotechnology Company) via RNA‐based fluorescence in situ hybridization. CTCs were classified as epithelial CTCs (E‐CTCs), mesenchymal CTCs (M‐CTC), or hybrid CTCs (H‐CTC) based on the expression of epithelial markers (EpCAM and CK8/18/19) and mesenchymal markers (vimentin and twist). ADAM19 expression was detected by fluorescence labeling and graded according to fluorescent foci counts as follows: none (0), low (1–2), medium (3–9), high (≥ 10). This study was approved by the Ethics Committee of the Second Affiliated Hospital of Xi'an Jiaotong University (ID: 2021033) and registered in the Chinese Clinical Trial Registry (ChiCTR2000035394). All patients who underwent CTC detection signed relevant informed consent forms.

2.11. Statistical Analysis

Statistical analyses were performed using GraphPad Prism 9 and R 4.2.2. Normally distributed continuous variables were analyzed using Student's t‐test or one‐way ANOVA, while non‐parametric tests were applied otherwise. Correlations were assessed using Pearson or Spearman analysis, and categorical variables were compared using chi‐square or Fisher's exact tests. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the predictive ability of baseline postoperative ADAM19‐Low‐CTC counts for disease progression, and the optimal cutoff was determined by the maximum Youden's index (Sensitivity + Specificity−1). Area under curve (AUC) was calculated to assess discrimination performance. Kaplan–Meier curves and log‐rank tests were used for survival analysis. R packages including “survival”, “corrplot”, and “pROC” were used for statistical analysis and visualization. p < 0.05 was considered statistically significant.

3. Results

3.1. ADAM19 Is Associated With EMT and Differentially Expressed in CTCs and Primary Tumor Cells

Differential expression analysis was performed on mRNA profiles of primary tumors and CTCs across breast cancer, non‐small cell lung cancer, prostate cancer, and colon cancer. The intersection of 4 groups of DEGs included 78 DEGs (Figure 1a). Univariate Cox regression analysis of these 78 DEGs in RCC indicated 28 of them were significantly associated with the prognosis of RCC. Subsequently, correlation analysis was performed between the expression of the 28 DEGs and EMT‐related genes, and we found that ADAM19 was highly correlated with EMT signatures, with a maximum correlation coefficient of 0.86 (p < 0.05) (Figure 1b). Survival analysis indicated that high ADAM19 expression in RCC patients was significantly associated with shorter overall survival and disease‐free survival (Figure 1c). These findings suggested a correlation between ADAM19 and malignant phenotypes and EMT in RCC. Furthermore, we searched ADAM19 in ctcRbase and found that, according to existing data, ADAM19 had higher levels in CTCs than in primary tumors in breast cancer, hepatocellular carcinoma, and melanoma (Figure 1d). Results above suggested that ADAM19 may also be expressed at higher levels in CTCs than in primary tumors and may be associated with the malignant phenotypes of CTCs.

FIGURE 1.

ADAM19 is differentially expressed in CTCs and primary tumor cells and associated with EMT‐related signatures (a). Venn diagram showing overlapping DEGs between CTCs and primary tumor cells (b). Correlation between ADAM19 and EMT‐related genes (c). Association between ADAM19 expression and survival outcomes in RCC patients (d). ADAM19 expression in CTCs and primary tumor cells across various cancer types.

3.2. ADAM19 Promotes Proliferation, Migration, Invasion and EMT in RCC Cells

First, we compared protein expression of ADAM19 between RCC cell lines (786‐O, ACHN, OS‐RC2) and an immortalized proximal tubule epithelial cell (HK‐2). ADAM19 was highly expressed in RCC cells, especially in 786‐O cells (Figure 2a). Next, ADAM19 was overexpressed in ACHN and OS‐RC2 and knocked down in 786‐O cells (Figure 2b). In 786‐O cells, ADAM19 knockdown downregulated protein expression levels of the mesenchymal markers N‐cadherin, Vimentin, and Slug, while upregulated the epithelial marker E‐cadherin. Conversely, ADAM19 overexpression in ACHN and OS‐RC2 cells produced the opposite effects (Figure 2c). The CCK‐8 assay revealed that ADAM19 overexpression significantly promoted the proliferation of RCC cells (Figure 2d). Wound healing and Transwell migration/invasion assays indicated that ADAM19 expression was positively correlated with the migration and invasion abilities of RCC cells (Figure 2e–g).

FIGURE 2.

ADAM19 promotes EMT‐related changes and malignant phenotypes in RCC cells (a). Protein expression of ADAM19 in RCC cells and HK‐2 cells (b). ADAM19 knockdown and overexpression validation (c). ADAM19 affects EMT‐related marker expression (d). ADAM19 promotes proliferation (e, f). ADAM19 promotes migration (g). ADAM19 promotes invasion. Data are presented as the mean ± SD and are representative of at least three independent experiments. nsp > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

3.3. Association Between ADAM19 and NOTCH/ NF‐κB Signaling Pathway

To investigate how ADAM19 may be involved in EMT, we performed RNA sequencing and bioinformatic analyses. GSEA further supported the association between ADAM19 and EMT‐related gene signatures (Figure 3a). Combined KEGG enrichment and GSEA analyses revealed that ADAM19 was significantly associated with the NOTCH and NF‐κB signaling pathways (Figure 3b–d). We further examined this observation in 786‐O and OS‐RC2 cell lines. ADAM19 knockdown in 786‐O cells markedly reduced the levels of pathway activation‐associated proteins, including NICD, Hes1, p‐P65/P65, and p‐IκBα/IκBα ratios, whereas ADAM19 overexpression in OS‐RC2 cells showed opposite results (Figure 3e). Functional assays showed that the activation of the NOTCH or NF‐κB pathway significantly promoted the migration and invasion of RCC cells (Figure 3f,g). Exogenous Co‐IP experiments suggested potential physical associations between ADAM19 and NOTCH1 or membrane‐bound TNF‐α under overexpression conditions (Figure S1).

FIGURE 3.

Association between ADAM19 and NOTCH/NF‐κB pathways (a). Association between ADAM19 and EMT‐related signatures (b). Top 30 hub genes from DEGs (c). Results of KEGG analysis (d). Results of GSEA (e). ADAM19 affects NOTCH/NF‐κB pathway‐related proteins (f). Activating NOTCH signaling pathway promotes RCC cell migration and invasion (g). Activating NF‐κB signaling pathway promotes RCC cell migration and invasion. Data are presented as the mean ± SD and are representative of at least three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001.

Subsequent rescue experiments indicated that ADAM19 overexpression promoted EMT (Figure 4a,c), migration, and invasion (Figure 4b,d) in RCC cells, whereas knockdown of NOTCH1 or P65 to inhibit NOTCH or NF‐κB signaling pathway partially reversed the effects induced by ADAM19 overexpression. In addition, activation of the NOTCH and NF‐κB signaling pathways also upregulated the ADAM19 expression in RCC cells (Figure 5a). Knockdown of ADAM19 partly reduced the EMT‐related changes (Figure 5b,d), migration, and invasion (Figure 5c,e) induced by activation of these pathways. Correlation analysis indicated that ADAM19 mRNA expression was positively correlated with transcription factors RELA, RBPJ, related to these pathways (Figure 6a). ADAM19 expression also tended to be lower in groups with low expression of these transcription factors (Figure 6b). Public ChIP‐seq datasets revealed RELA and RBPJ occupancy signals near the ADAM19 TSS (Figure 6c). Motif analysis further identified potential binding motifs within these regions (Table S1, Figure 6d). Other transcription factors (NFKB1, NFKB2, RELB) in NF‐κB pathway were also found positively associated with ADAM19 expression; peaks of NFKB1 and NFKB2 were found in the ±5 kb region near the TSS site of ADAM19 (Figure S2). Together, these findings suggest a potential bidirectional association between ADAM19 expression and the NOTCH/NF‐κB signaling activities, which may contribute to EMT, migration, and invasion in RCC cells.

FIGURE 4.

NOTCH/NF‐κB inhibition reduces overexpression of ADAM19‐induced malignant phenotypes in RCC cells (a, b). NOTCH1 knockdown inhibits ADAM19‐induced malignant phenotypes (c, d). P65 knockdown inhibits ADAM19‐induced malignant phenotypes. Data are presented as the mean ± SD and are representative of at least three independent experiments. nsp > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

FIGURE 5.

NOTCH and NF‐κB pathway activation is associated with increased ADAM19 expression (a). Activating of NOTCH and NF‐κB pathways increases ADAM19 expression (b, c). ADAM19 knockdown inhibits NOTCH pathway‐induced EMT‐related changes and malignant phenotypes (d, e). ADAM19 knockdown inhibits NF‐κB pathway‐induced EMT‐related changes and malignant phenotypes. Data are presented as the mean ± SD and are representative of at least three independent experiments. nsp > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 6.

Bioinformatic identification of potential transcriptional associations between NOTCH/NF‐κB‐related transcription factors and ADAM19 expression (a). Correlation analysis of TCGA data between ADAM19 and RELA, RBPJ. (b) ADAM19 expression in groups stratified by the expression levels of RELA, RBPJ. (c) Analysis of RELA and RBPJ occupancy signals near the ADAM19 locus using public ChIP‐seq databases. (d) Motif Analysis of RELA and RBPJ. Data are presented as mean ± SD, Figure 6a,b, n = 1025. ****p < 0.0001.

3.4. 786‐O‐CTCs Exhibit Enhanced Malignant Phenotypes

786‐O‐CTCs isolated from an animal model (Figure 7a) were cultured, and their proliferative, migratory, and invasive abilities were compared with those of 786‐O cells. The CCK‐8 assay revealed that 786‐O‐CTCs exhibited enhanced proliferative capacity (Figure 7b). Transwell migration and invasion assays showed that 786‐O‐CTCs exhibited enhanced migratory and invasive abilities (Figure 7c). Meanwhile, we examined the protein expression level and found that ADAM19 was expressed at higher levels in 786‐O‐CTCs than in 786‐O cells (Figure 7d). ADAM19 knockdown significantly suppressed the proliferation, migration, and invasion of 786‐O‐CTCs (Figure 7e,f).

3.5. Correlation Between ADAM19 Expression and EMT Phenotypes of CTCs in RCC

We collected 181 CTC samples from the Department of Urology, the Second Affiliated Hospital of Xi'an Jiaotong University, with 86 surgically treated RCC patients possessing full clinical data included. ADAM19 mRNA expression levels in CTCs were categorized as none, low, medium, or high. E‐CTCs and H‐CTCs were detected in nearly all 181 samples, whereas M‐CTCs were rarely observed (Figure 8a). ADAM19 expression was predominantly none/low in E‐CTCs and medium/low in H‐CTCs (Figure 8b). As no sample was dominated by high ADAM19 expression, patients were grouped into None, Low, and Medium cohorts. The proportion of E‐CTCs was significantly increased in the ADAM19‐None group compared with ADAM19‐Low or ADAM19‐Medium groups, whereas the proportion of H‐CTCs was significantly higher in ADAM19‐Low or ADAM19‐Medium groups (Figure 8c). Correlation analyses revealed E‐CTC proportion was positively correlated with none/low ADAM19 and inversely correlated with medium/high ADAM19, whereas the H‐CTC proportion showed positive relations with low/medium/high ADAM19 (Figure 8d).

FIGURE 8.

ADAM19 expression is associated with EMT phenotype of CTCs in RCC patients (a). CTCs classification (b). The proportion of different levels of ADAM19 in CTCs is significantly different in E‐CTC and H‐CTC (c). Changes in E‐CTC and H‐CTC proportions according to ADAM19 expression (d). Correlation between ADAM19 expression and proportion of different phenotype of CTCs. Data are presented as the mean ± SD, Figure 8c: N = 181. nsp > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

3.6. Prognostic Value of CTC‐ADAM19 in Postoperative RCC Patients

To avoid temporal bias from variable timing of initial postoperative CTC detection, we restricted the analyses to patients with baseline CTC detection within 6 months postoperatively (Table 1). The chi‐square test or Fisher's exact test indicated that no clinical parameters differed significantly among the groups. However, tumor size tended to be smaller in the ADAM19‐None and ADAM19‐Low group than in the ADAM19‐Medium groups (Figure 9a), and the percentage of medium ADAM19 expression was higher in Stages II–III than in Stage I (Figure 9b). Baseline univariate Cox regression identified ADAM19‐Low‐CTC, ADAM19‐High‐CTC and M‐CTC counts as prognostic risk factors, though none remained independently predictive in baseline multivariate Cox regression (Table 2).

TABLE 1.

Clinical data of renal cell carcinoma patients.

Variables	Total	CTC‐ADAM19 expression level			p
		None	Low	Medium
All cases	54	16	26	12
Age					0.238
> 58	29	8	12	9
≤ 58	25	8	14	3
Gender					0.169
Male	42	15	19	8
Female	12	1	7	4
Stage					0.336
Stage I	47	14	24	9
Stages II–III	7	2	2	3
Tumor size					0.901
> 4.4 cm	19	6	9	4
≤ 4.4 cm	35	10	17	8
Pathology
ccRCC	46	12	24	10	0.302
No ccRCC	8	4	2	2
Targeted therapy
Yes	15	3	9	3	0.522
No	39	13	17	9
Progression					0.623
Yes	3	1	2	0
No	51	15	24	12

FIGURE 9.

Prognostic association of CTC‐associated ADAM19 expression in RCC (a, b). Association between tumor size/stage and ADAM19 expression levels in CTCs (c). ROC analysis of CTC subsets for prediction of progression d. Kaplan–Meier survival analysis according to ADAM19‐Low‐CTC counts. Data are presented as the mean ± SD, Figure 9a: N = 54. nsp > 0.05.

TABLE 2.

Univariate and multivariate Cox regression analysis.

Variables	Univariate Cox regression			Multivariate Cox regression
	Exp (B)	p	95% Exp (B) CI	Exp (B)	p	95% Exp (B) CI
Age				—	—	—
≤ 58	(reference)	0.455	0.226–27.570	—	—	—
> 58	2.498			—	—	—
Gender				—	—	—
Male	(reference)	0.999		—	—	—
Female	3.647e‐9		0‐Inf	—	—	—
Stage				—	—	—
Stage I	(reference)	0.344		—	—	—
Stages II–III	3.187		0.289–35.160	—	—	—
Tumor size				—	—	—
≤ 4.4 cm	(reference)	0.999		—	—	—
> 4.4 cm	1.464e+4		0‐Inf	—	—	—
Pathology				—	—	—
No ccRCC	(reference)	0.999		—	—	—
ccRCC	1.192e‐8		0‐Inf	—	—	—
Targeted therapy				—	—	—
No	(reference)	0.829		—	—	—
Yes	1.302		0.118–14.360	—	—	—
E‐CTC count	1.052	0.570	0.883–1.253	—	—	—
H‐CTC count	1.072	0.402	0.912–1.260	—	—	—
M‐CTC count	15.469	0.027	1.369–174.800	0.808	0.949	0.001–562.877
ADAM19‐None‐CTC count	0.941	0.646	0.728–1.218	—	—	—
ADAM19‐Low‐CTC count	1.408	0.013	1.074–1.847	1.330	0.143	0.908–1.948
ADAM19‐Meidum‐CTC count	1.019	0.892	0.780–1.331	—	—	—
ADAM19‐High‐CTC count	2.957	0.026	1.142–7.658	1.427	0.740	0.174–11.705
ADAM19‐Positive‐CTC count	1.179	0.056	0.997–1.396	—	—	—

Note: Bold values indicate statistically significant of p < 0.05.

Subsequently, the ROC curve demonstrated that the ADAM19‐Low‐CTC count could distinguish patients with or without progression during follow‐up, with an AUC value of 0.833 (Figure 9c). Based on the maximum Youden's Index, a cutoff value of 12 was selected; Log‐rank analysis confirmed that baseline ADAM19‐Low‐CTC  count ≥ 12 was associated with inferior PFS (Figure 9d).

Given the dynamic nature of disease progression, changes in CTC counts were analyzed in 52 of the 86 patients who underwent repeated detection. ADAM19‐Low‐CTC counts declined significantly during follow‐up, whereas ADAM19‐Medium‐CTC counts increased (Figure S3a). Among the 16 patients receiving postoperative Vorasidenib targeted therapy, changes in ADAM19‐Low/Medium‐CTC counts were consistent with those of the total cohort, whereas E‐CTC and ADAM19‐None‐CTC exhibited opposite dynamic changes compared with patients without targeted therapy (Figure S3b,c). Time‐dependent Cox regression showed that the time‐updated ADAM19‐Low‐CTC count, disease stage, and targeted therapy were prognostic in univariate analysis; only higher time‐updated ADAM19‐Low‐CTC count served as an independent adverse prognostic factor (HR = 1.243, 95% CI: 1.006–1.536, Table 3).

TABLE 3.

Time‐dependent Cox regression analysis.

Variables	Univariate Cox regression			Multivariate Cox regression
	Exp (B)	p	95% Exp (B) CI	Exp (B)	p	95% Exp (B) CI
Age				—	—	—
≤ 58	(reference)	0.520	0.110–3.053	—	—	—
> 58	0.580			—	—	—
Gender				—	—	—
Male	(reference)	0.998		—	—	—
Female	2.592e‐9		0‐Inf	—	—	—
Stage				—	—	—
Stage I	(reference)	0.003		(reference)	0.106	0.755–19.178
Stages II–III	8.783		2.061–37.430	3.804	—	—
Tumor size				—	—	—
≤ 4.4 cm	(reference)	0.999		—	—	—
> 4.4 cm	2.562e+9		0‐Inf	—	—	—
Pathology				—	—	—
No ccRCC	(reference)	0.998		—	—	—
ccRCC	8.341e+7		0‐Inf	—	—	—
Targeted therapy				—	—	—
No	(reference)	0.034		(reference)	0.264	0.456–17.520
Yes	5.172		1.132–23.640	2.827	—	—
E‐CTC count	1.093	0.169	0.963–1.240	—	—	—
H‐CTC count	1.068	0.336	0.934–1.220	—	—	—
M‐CTC count	1.337	0.332	0.744–2.402	—	—	—
ADAM19‐None‐CTC count	1.031	0.736	0.865–1.228	—	—	—
ADAM19‐Low‐CTC count	1.293	0.008	1.069–1.564	1.243	0.044	1.006–1.536
ADAM19‐Meidum‐CTC count	1.014	0.860	0.866–1.188	—	—	—
ADAM19‐High‐CTC count	1.218	0.350	0.806–1.843	—	—	—
ADAM19‐Positive‐CTC count	1.119	0.078	0.988–1.267	—	—	—

Note: Bold values indicate statistically significant of p < 0.05.

4. Discussion

CTC detection provides valuable information for monitoring tumor progression and guiding clinical decisions. However, its clinical application is limited by the rarity and heterogeneity of CTCs. EMT plays an important role in tumor progression and contributes to CTC generation, survival, and metastatic colonization. Previous studies have shown that EMT enables CTCs to resist anoikis, shear stress, and immune cell‐mediated killing, resulting in exhibiting more mesenchymal characteristics compared with primary tumor cells [20, 21, 22, 23]. However, EMT is a dynamic and reversible process, and CTCs can exist in epithelial, mesenchymal, and hybrid phenotypes. These subpopulations may coexist within the same patient and change during tumor progression or therapeutic intervention. In many cancers, higher CTC counts are associated with poor prognosis [24]. Meanwhile, M‐CTCs and H‐CTCs are associated with poor prognosis, whereas E‐CTCs are generally associated with a more favorable prognosis [25, 26]. Traditional CTC isolation methods rely on epithelial markers such as EpCAM and cytokeratins, which underestimate mesenchymal CTC populations. Therefore, identification of novel EMT‐related markers may improve CTC classification and facilitate clinical application.

In this study, we found that ADAM19 was differentially expressed between CTCs and primary tumor cells, with higher expression levels observed in CTCs. ADAM19 expression was associated with the EMT phenotypes of CTCs in RCC. Functional experiments further suggested that ADAM19 may contribute to EMT‐related changes, as well as the proliferation, migration, and invasion of RCC cells. We also identified an association between ADAM19 and the NOTCH and NF‐κB signaling pathways. Our results suggest that ADAM19 may contribute to EMT‐related changes, migration, and invasion of RCC cells in association with these pathways and also indicate a potential reciprocal relationship between ADAM19 and NOTCH/NF‐κB pathway activity. Given the proteolytic functions of ADAM family members, we speculate that ADAM19 may influence downstream signaling through proteolytic processing of pathway‐associated molecules. Our preliminary investigation showed potential interactions between ADAM19 and NOTCH1 or membrane‐bound TNF‐α under overexpression conditions. However, these data do not determine whether the interactions are direct or whether they contribute to pathway activation. Further studies, such as endogenous Co‐IP and proximity ligation assays, are needed to clarify these possibilities. Publicly available data also suggested a potential regulatory association between transcription factors of NOTCH/NF‐κB pathways and ADAM19, especially RELA and RBPJ. However, these analyses still need further experiments such as dual‐luciferase reporter assays and ChIP‐qPCR to determine whether these transcription factors directly regulate ADAM19 transcription.

In addition, animal‐derived 786‐O‐CTCs exhibited significantly enhanced proliferative, migratory, and invasive capacities compared with parental cells, which is consistent with observations reported in other tumor types. Additionally, in 786‐O‐CTCs ADAM19 was upregulated, and ADAM19 knockdown significantly inhibited proliferation, migration, and invasion of 786‐O‐CTCs. Clinically, CTC data from patients identified a correlation between ADAM19 expression and EMT phenotypes of CTCs. Correlation analysis showed that ADAM19‐None‐CTC counts were significantly associated with E‐CTC counts, ADAM19‐Low‐CTC counts were significantly associated with both H‐CTC and E‐CTC counts, and ADAM19‐Medium/High‐CTC counts were significantly associated with H‐CTC counts. Although the count of M‐CTCs was insufficient for reliable statistical analysis in our cohort, the observed distribution patterns suggest that different levels of ADAM19 expression may be associated with distinct EMT states of CTCs. These observations require further validation in larger cohorts. Collectively, cellular functional assays indicated that ADAM19 may contribute to EMT‐related changes, and significant correlations were observed between ADAM19 expression and EMT phenotypes of CTCs. However, existing in vitro results in RCC cells and single‐center CTC data are insufficient to establish ADAM19 as a reliable indicator reflecting EMT status. Further large‐scale multicenter clinical validation and in‐depth mechanistic investigations are required to determine its biomarker potential and understand the underlying regulatory mechanisms.

In survival analysis, baseline postoperative ADAM19‐Low‐CTC count was associated with PFS, with a cutoff value of 12 derived from our cohort. Baseline Cox regression showed that M‐CTC, ADAM19‐Low‐CTC, and ADAM19‐High‐CTC counts at the first postoperative CTC detection were associated with PFS, although none remained independent prognostic factors. During longitudinal follow‐up, ADAM19‐Low‐CTC counts decreased, whereas ADAM19‐Medium‐CTC counts increased among patients with repeated CTC assessments. In addition, E‐CTC and ADAM19‐None‐CTC showed different dynamic trends between patients with and without targeted therapy. To evaluate the prognostic impact of longitudinal CTC changes, we further performed time‐dependent Cox regression analysis. This analysis showed that higher time‐updated postoperative ADAM19‐Low‐CTC count was an independent risk factor for PFS, whereas time‐updated ADAM19‐Medium‐CTC, ADAM19‐None‐CTC, and E‐CTC counts were not significantly associated with PFS. The prognostic pattern of ADAM19‐Low‐CTC appears inconsistent with functional assays showing that high ADAM19 expression promotes malignant phenotypes in cultured 786‐O and 786‐O‐CTC cells. This discrepancy may be explained by the limited sample size, short follow‐up duration, and low detection frequency of ADAM19‐Medium/High‐CTCs, which may reduce the statistical power to identify their prognostic effects. Alternatively, ADAM19‐Low‐CTCs may represent an intermediate EMT state with both epithelial and mesenchymal characteristics, potentially facilitating survival in circulation and metastatic progression. However, this hypothesis requires further validation through functional characterization of isolated CTC subsets. Although our findings suggest that ADAM19 expression in CTCs may have prognostic potential, further validation is required before clinical application. The ADAM19‐Low‐CTC cutoff was derived from a single‐center cohort and requires confirmation in independent cohorts with standardized follow‐up protocols. Larger studies with longer follow‐up are also needed to clarify the prognostic significance of rare CTC subsets, including ADAM19‐High‐CTCs and CTC clusters. Future investigations should further evaluate ADAM19 expression across different CTC subsets, including CTC clusters, which may provide additional prognostic information.

In summary, our findings suggest that ADAM19 contributes to RCC migration, invasion, and EMT‐related changes and is associated with NOTCH/NF‐κB pathway activity. ADAM19 expression in CTCs correlates with EMT phenotypes and unfavorable prognosis, suggesting its potential value as an EMT‐associated CTC parameter (Figure 10). Further multicenter validation and mechanistic studies are required to determine its clinical utility and biological functions.

FIGURE 10.

ADAM19 is associated with EMT‐related malignant phenotypes and NOTCH/NF‐κB pathway activity in RCC and may serve as a prognostic biomarker in CTCs.

Author Contributions

Minxin He: writing – original draft, data curation, formal analysis, visualization. Mingrui Li: writing – original draft, formal analysis, data curation. Yifei Ma: formal analysis, data curation. Zhenzhuo Chen: formal analysis, data curation. Lingyu Guo: data curation, methodology. Xiangyu Wang: formal analysis, data curation. Wenjing Wang: formal analysis. Tie Chong: writing – review and editing, methodology, validation, supervision, conceptualization. Zhenlong Wang: writing – review and editing, validation, supervision, methodology, funding acquisition, conceptualization.

Funding

This study was supported by the Wu Jieping Medical Foundation Scientific Research Special Project (320.6750.2025‐16‐19) and Key Project of IIT Clinical Research Fund, The Second Affiliated Hospital of Xi'an Jiaotong University (z014).

Ethics Statement

This study was conducted in accordance with the principles of the Declaration of Helsinki. All patients signed an informed consent form to voluntarily participate in the study before the start of the experiment. All procedures involving human participants and animals were conducted in compliance with institutional ethics regulations and approved by the Research Ethics Committee of the Second Affiliated Hospital of Xi'an Jiaotong University (ID: 2021033) and Xi'an Jiaotong University Animal Ethics Committee (No. XJTUAE2025‐1030). This study was registered in the Chinese Clinical Trial Registry (ChiCTR2000035394).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Exogenous Co‐IP analysis of the potential interactions between ADAM19 and NOTCH1 or TNF‐α.

Figure S2: Association between ADAM19 expression and transcription factors involved in the NOTCH and NF‐κB pathways. (a) Correlation analysis of TCGA data between ADAM19 and RELB, NFKB1, NFKB2. (b) ADAM19 expression in groups stratified by the expression levels of NOTCH/NF‐κB pathway‐related transcription factors. (c) Analysis of the binding of NFKB1 and NFKB2 near the ADAM19 locus using public ChIP‐seq databases. Data are presented as mean ± SD, (a) n = 1025. ****p < 0.0001.

Figure S3: Dynamic changes in CTC subsets between initial and latest postoperative assessments. (a) Dynamic changes in CTC subsets in total cohort. (b) Dynamic changes in CTC subsets in patients without targered therapy cohort. (c) Dynamic changes in CTC subsets in patients receiving targeted therapy. Data are presented as the mean ± SD, Figure S3a: n = 52, Figure S3b: n = 36, Figure S3c: n = 16. nsp > 0.05; *p < 0.05.

Table S1: Top3 predicted binding sites of each transcription factors for ADAM19.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.