Abstract
Background
Over the past decade, the prognosis of patients with metastatic renal cell carcinoma (mRCC) has significantly improved owing to the development of anti-angiogenic targeted drugs such as sunitinib, pazopanib, and sorafenib. However, biomarkers that can identify patients with mRCC who may rapidly develop drug resistance are still lacking. Approximately 25% of patients experience rapid disease progression [progression-free survival (PFS) ≤3 months]. Currently, there is a lack of effective noninvasive biomarkers to identify resistant patients prior to treatment. This study aimed to identify plasma biomarkers associated with therapeutic resistance and develop a predictive model for clinical decision-making.
Methods
Plasma proteomic analysis was conducted in 159 patients with mRCC treated with targeted therapy. Patients were divided into training and validation cohorts. Candidate protein biomarkers were initially screened using liquid chromatography-tandem mass spectrometry (LC-MS/MS) and further validated using enzyme-linked immunosorbent assay (ELISA). A predictive nomogram was subsequently developed using logistic regression analysis and assessed for discrimination and calibration performance. Immunohistochemistry (IHC) was performed to compare protein expression levels in corresponding tumor tissue samples.
Results
Four plasma proteins—chitotriosidase-1 (CHIT1), interleukin-6 receptor (IL-6R), neuronal cell adhesion molecule (NRCAM), and ecto-5’-nucleotidase (NT5E)—were identified as significant predictors of intrinsic resistance. The nomogram incorporating these biomarkers exhibited a high predictive accuracy, with a concordance index (C-index) of 0.956 and 0.869 for the training and validation cohorts, respectively. Notably, while the plasma concentrations of these proteins were significantly elevated in resistant patients, their expression levels in tumor tissues showed no significant differences, underscoring their utility as circulating, noninvasive biomarkers.
Conclusions
We developed and validated a plasma protein-based nomogram to predict mRCC resistance to targeted therapy. The four identified biomarkers allow noninvasive identification of high-risk patients and offer a practical tool for early clinical stratification. This model may assist clinicians in avoiding ineffective treatment and optimizing therapeutic strategies for patients with mRCC.
Keywords: Metastatic renal cell carcinoma (mRCC), drug resistance, plasma proteomic analysis, prediction
Highlight box.
Key findings
• The study identified four plasma proteins (chitotriosidase-1, interleukin-6 receptor, neuronal cell adhesion molecule, and ecto-5’-nucleotidase NT5E) as significant predictors of intrinsic resistance to targeted therapy in metastatic renal cell carcinoma (mRCC).
• A predictive nomogram incorporating these biomarkers demonstrated high accuracy, with a concordance index of 0.956 and 0.869 in the training and validation cohorts, respectively.
What is known and what is new?
• While tyrosine kinase inhibitors (TKIs) play a crucial role in mRCC treatment, approximately 25% of patients experience rapid disease progression due to intrinsic resistance. Effective noninvasive biomarkers to predict this resistance are currently lacking.
• This manuscript adds a novel, validated plasma protein-based nomogram that accurately and noninvasively predicts intrinsic TKI resistance (progression-free survival ≤3 months) prior to treatment initiation.
What is the implication, and what should change now?
• This predictive model enables early clinical stratification of patients with mRCC.
• For patients identified as high-risk for TKI resistance, clinicians should change practice by increasing follow-up monitoring frequency or prioritizing alternative therapeutic regimens, such as immune-targeted combination therapies, to avoid ineffective treatments and optimize outcomes.
Introduction
Renal cell carcinoma (RCC) is a common malignant tumor originating from the kidney tubular epithelial cells. It accounts for 80% of adult kidney cancer cases, and 5% and 3% of all adult malignancies in men and women, respectively (1). Adult kidney cancer leads to approximately 179,000 deaths worldwide annually (2). Most RCC cases are found incidentally on imaging, usually with abdominal ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI) (3,4). Although the majority of patients are diagnosed with RCC at an early stage, approximately one-third of cases are diagnosed with metastatic renal cell carcinoma (mRCC), which has a 5-year relative survival rate of 12% (5,6). For patients with mRCC, targeted tyrosine kinase inhibitors (TKIs), which provide a reliable survival benefit, have been the first-line systemic therapy over the last decade (7). However, the long-term therapeutic efficacy of TKIs is limited due to the development of drug resistance. Clinically, drug resistance can be broadly categorized as acquired or primary resistance (8). In routine clinical practice, the first radiological tumor assessment to evaluate TKI response is typically performed 8–12 weeks (approximately 2–3 months) after treatment initiation. Consequently, patients who exhibit progressive disease at this initial evaluation—corresponding to a progression-free survival (PFS) of ≤3 months—are clinically defined as having intrinsic resistance (9). Identifying patients who may prematurely develop drug resistance and switching them to an alternative treatment in a timely manner are thought to be essential for prolonging survival.
With the development of translational medicine, several biomarkers related to the diagnosis, treatment efficacy, and prognosis of RCC have been discovered (7). Current predictive prognostic biomarkers that have been reported include histological biomarkers, serological biomarkers, and genetic biomarkers (10-13). Blood-derived biomarkers are thought to have great development potential because they can be assessed noninvasively and are less costly. It has been reported that baseline plasma VEGF and soluble VEGFR levels during treatment are related to TKI treatment response in patients with mRCC. In addition, plasma osteopontin, interleukin-6, and TIMP-1 baseline levels were independently associated with overall survival (OS) in another study (14,15). However, reliable biomarkers for predicting PFS in patients with mRCC receiving first-line TKI treatment are still lacking.
In this study, we used a proteomic approach to analyze differentially expressed plasma proteins between patients who were sensitive to targeted drugs and those who were resistant to targeted drugs and then selected and tested the feasibility and reliability of these differentially expressed proteins as biomarkers for predicting the PFS of mRCC patients. We present this article in accordance with the TRIPOD reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-1-0041/rc).
Methods
Patient data
This study enrolled patients with mRCC who had not previously received any systemic treatment, regardless of whether surgery was performed on the primary lesion. Other eligibility criteria included age 18–70 years; an Eastern Cooperative Oncology Group (ECOG) performance status ≤1; and adequate renal and hepatic function. Patients were excluded if they were pregnant or lactating, had poorly controlled hypertension or diabetes mellitus, severe hepatic or renal insufficiency, severe cardiovascular or cerebrovascular events, severe autoimmune diseases, or uncontrollable mental illness.
Patients for this multicenter study were sourced from five institutions in Shanghai, China: Changhai Hospital, Changzheng Hospital, Zhongshan Hospital, Renji Hospital, and Shanghai General Hospital. Patients were recommended to receive sunitinib, pazopanib, or sorafenib based on clinical guidelines and the attending surgeon’s clinical experience. The endpoint was PFS, which was defined as the time interval between the date of medication initiation and the date of disease progression or death. Basic demographic and disease characteristic data were collected during the first visit. Follow-up data were collected during regular monthly follow-up, and patients were divided into a targeted drug-sensitive (TDS) group and a targeted drug-resistant (TDR) group, as defined by a PFS of ≤3 months.
Data collection and processing
The transcriptome and clinical data of patients with kidney renal clear cell carcinoma (KIRC) were downloaded from The Cancer Genome Atlas (TCGA) database (https://cancergenome.nih.gov). Proteomic data were obtained from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database. TCGA data were collected for 35 patients with renal clear cell carcinoma who were sensitive to targeted TKIs and nine patients who were resistant to TKIs. Corresponding clinical information, including survival outcomes, was also obtained. Targeted drug resistance was defined as PFS of less than 100 days after targeted drug therapy for RCC, including the time the patient experienced complete or partial remission and the time the patient experienced disease stabilization.
Blood collection and plasma preparation
Peripheral blood samples from all enrolled patients were collected using EDTA-containing vacuum tubes prior to the initiation of targeted therapy. The whole blood was immediately centrifuged at 4,000 rpm for 10 minutes at 4 ℃ to separate the plasma. The resulting plasma supernatants were carefully aliquoted to avoid multiple freeze-thaw cycles and stored at −80 ℃ until subsequent analyses.
Mass spectrometry (MS)-based proteomic analysis
Plasma samples randomly selected from the TDR group (n=14) and TDS group (n=13) were subjected to label-free quantitative liquid chromatography-tandem mass spectrometry (LC-MS/MS). Standardized sample preparation, peptide digestion, and LC-MS/MS sequencing procedures were performed by PTM Biolab Co., Ltd. (Hangzhou, China).
For data processing and bioinformatics analysis, raw MS data were searched against the UniProt Human reference database. To ensure high-quality identification, the false discovery rate (FDR) was strictly controlled at <1% at both peptide and protein levels. Relative quantitative protein intensities were log2-transformed and normalized to eliminate systematic bias. To identify differentially expressed proteins (DEPs) between the drug-resistant and drug-sensitive groups, statistical comparisons were performed using Student’s t-test. Proteins meeting the predefined screening criteria of a fold change >1.5 or <0.67 and a P value <0.05 were defined as significantly differentially expressed.
Enzyme-linked immunosorbent assay (ELISA)
The expression levels of the 11 candidate plasma proteins identified by MS were quantified across the entire cohort of 159 plasma samples using commercially available ELISA kits purchased from Enzyme-Linked Biotechnology Co., Ltd. (Shanghai, China). All assays were performed in strict accordance with the manufacturer’s instructions.
Briefly, all reagents, standard proteins, and plasma samples were equilibrated to room temperature (18–25 ℃) prior to use. Plasma samples were appropriately diluted with the assay diluent. Aliquots of 100 µL of standard solutions (prepared via serial dilution) and diluted plasma samples were added in duplicate to 96-well microplates pre-coated with target-specific antibodies. The plates were incubated for 2.5 hours at room temperature with gentle shaking. After aspirating the liquid, the wells were washed four times with 300 µL of 1× wash buffer using an autowasher to remove unbound substances. Subsequently, 100 µL of biotinylated detection antibody was added to each well and incubated for 1 h at room temperature. Following another four washes, 100 µL of horseradish peroxidase (HRP)-conjugated streptavidin solution was added and incubated for 45 minutes. The wells were washed again, and 100 µL of 3,3',5,5'-tetramethylbenzidine (TMB) substrate reagent was added, followed by a 30-minute incubation in the dark with gentle shaking to allow for color development. The enzymatic reaction was terminated by adding 50 µL of stop solution (0.2 M sulfuric acid) to each well, which induced a color change from blue to yellow. The optical density (OD) of each well was immediately measured at 450 nm using a microplate reader. The absolute concentrations of the candidate proteins were calculated based on the standard curves generated for each assay.
Immunohistochemistry (IHC) staining and image analysis
IHC staining was performed to assess the spatial expression and localization of the four candidate proteins—ecto-5'-nucleotidase (NT5E), chitotriosidase-1 (CHIT1), interleukin-6 receptor (IL-6R), and neuronal cell adhesion molecule (NRCAM) in kidney cancer tissues from 60 patients. Formalin-fixed, paraffin-embedded (FFPE) tissue blocks were sectioned into 4 µm-thick slices, which were deparaffinized in xylene and rehydrated via a graded ethanol series. Heat-induced antigen retrieval was performed in a microwave oven with standard citrate buffer (pH 6.0), followed by endogenous peroxidase quenching with 3% hydrogen peroxide for 10 min.
To eliminate nonspecific binding, the sections were blocked with 5% normal goat serum in Tris-buffered saline containing Tween 20 (TBST) for 1 h at room temperature. They were then incubated overnight at 4 ℃ with primary antibodies: anti-NT5E (ab257311, 1:100 dilution, Abcam), anti-CHIT1 (21432-1-AP, 1:200 dilution, Proteintech), anti-IL-6R (ab271042, 1:2,000 dilution, abcam), and anti-NRCAM (ab191418, 1:500 dilution, abcam). After three washes with TBST on the following day, the sections were incubated with a HRP-conjugated secondary antibody for 1 h at room temperature. Immune complexes were visualized using a 3,3'-diaminobenzidine (DAB) substrate kit, and cell nuclei were counterstained with hematoxylin. The slides were dehydrated, cleared in xylene, and mounted.
For quantitative analysis, the stained sections were imaged using an optical microscope. IHC staining intensity was quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).
Statistical analysis
Categorical variables are presented as frequency and percentage, and continuous variables are presented as median and interquartile range (IQR). Categorical variables were compared between the two groups using the chi-squared test. For the comparison of continuous variables, if the data conformed to a normal distribution and there was homogeneity of variance, Student’s t-test was used; otherwise, the Wilcoxon signed-rank test was used. The median of the plasma proteins was used as the cutoff for the high- and low-expression groups. The prediction model was developed based on data from the training set. Univariate analysis was performed to select parameters that showed significant differences between the TDR and TDS groups as candidate proteins. Multivariate analysis using binary logistic regression was performed to screen candidate variables and establish a prediction model. A nomogram was constructed accordingly. The data from the training and validation sets were used for the internal and external validation of the model, respectively. The concordance index (C-index) and 95% confidence interval (CI) were used to assess discrimination. A calibration curve based on 1000 bootstrap resamples was used to evaluate the agreement between the actual and predicted probabilities of resistance. The optimal cutoff value for the nomogram-derived risk score was determined using receiver operating characteristic (ROC) curve analysis by calculating the maximum Youden index (sensitivity + specificity − 1).
Ethics statement
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Changzheng Hospital (No. 2018SL031). All patients signed an informed consent form before the start of systemic treatment and study participation. All participating hospitals were informed of and agreed to the study. This trial has been registered with the Chinese Clinical Trial Registry (No. ChiCTR1800016836. Registered June 28, 2018, https://www.chictr.org.cn/bin/project/edit?pid=24862).
Results
Study cohort and baseline clinicopathological characteristics
Figure 1 illustrates the screening and enrollment processes used in this study. Based on predefined inclusion and exclusion criteria, 159 treatment-naive patients with mRCC from five participating centers were enrolled; treatment-naive status was defined as having received no prior systemic therapy. Patients were stratified into a TDR cohort (PFS ≤3 months, n=58) and a TDS cohort (PFS >3 months, n=101) based on their response to targeted therapy. Forty patients from Chinghai Hospital were assigned to the external validation set. Detailed baseline demographic and clinicopathological characteristics of both groups are summarized in Table 1. Baseline analyses demonstrated no significant differences in demographic and clinicopathological characteristics, such as age, body mass index (BMI), prior surgery, and pathological grade, between the two cohorts.
Figure 1.
Flow of this study. Inclusion and exclusion criteria are shown. DEPs, differentially expressed proteins; ELISA, enzyme-linked immunosorbent assay; IHC, immunohistochemistry; PCA, principal component analysis; PFS, progression-free survival; TCGA, The Cancer Genome Atlas.
Table 1. Baseline demographic and clinicopathological characteristics of training and validation cohort.
| Characteristics | Training cohort (n=119) | Validation cohort (n=40) | P value |
|---|---|---|---|
| Gender | 0.52 | ||
| Male | 95 (59.7) | 30 (18.9) | |
| Female | 24 (15.1) | 10 (6.3) | |
| Age (years) | 60 [53, 66] | 59.5 [52.8, 67.2] | 0.75 |
| Hypertension | 0.55 | ||
| No | 68 (42.8) | 25 (15.7) | |
| Yes | 51 (32.1) | 15 (9.4) | |
| Diabetes | 0.72 | ||
| No | 92 (57.9) | 32 (20.1) | |
| Yes | 27 (17.0) | 8 (5.0) | |
| BMI (kg/m2) | 0.67 | ||
| ≤25 | 79 (49.7) | 25 (15.7) | |
| >25 | 40 (25.2) | 15 (9.4) | |
| Pathology | >0.99 | ||
| ccRCC | 114 (71.7) | 38 (23.9) | |
| Non-ccRCC | 5 (3.1) | 2 (1.3) | |
| IMDC score | 2 [1, 3] | 1.5 [1, 3] | 0.46 |
| Treatment response | 0.88 | ||
| No | 43 (27.0) | 15 (9.4) | |
| Yes | 76 (47.8) | 25 (15.7) | |
| Fuhrman grade | 0.55 | ||
| Grade 3 | 59 (37.1) | 16 (10.1) | |
| Grade 4 | 22 (13.8) | 8 (5.0) | |
| Grade 2 | 38 (23.9) | 16 (10.1) | |
| Targeted therapy | 0.02 | ||
| Sunitinib | 99 (62.3) | 40 (25.2) | |
| Sorafenib | 4 (2.5) | 0 (0.0) | |
| Pazopanib | 16 (10.1) | 0 (0.0) | |
| Prior nephrectomy | >0.99 | ||
| Yes | 112 (70.4) | 37 (23.3) | |
| No | 7 (4.4) | 3 (1.9) | |
| ECOG | 0.82 | ||
| Score 1 | 41 (25.8) | 13 (8.2) | |
| Score 0 | 78 (49.1) | 27 (17.0) |
Data are presented as n (%) or median [IQR]. BMI, body mass index; ccRCC, clear cell renal cell carcinoma; ECOG, Eastern Cooperative Oncology Group; IMDC, International Metastatic RCC Database Consortium; IQR, interquartile range.
Identification and screening of differentially expressed plasma proteins associated with targeted therapy resistance
To identify potential plasma protein biomarkers indicative of targeted therapy resistance in patients with RCC, we randomly selected plasma samples from 14 patients in the TDR group (labeled as ‘R’ in the figures) and 13 patients in the TDS group (labeled as ’s’) for high-throughput proteomic sequencing. Initially, principal component analysis (PCA) was employed to evaluate the overall expression profiles of the samples. The results (Figure 2A) demonstrated acceptable intragroup reproducibility for both cohorts despite the presence of a few outliers. Concurrently, the relative standard deviation (RSD) boxplot (Figure 2B) indicated that the degree of variation was largely consistent between the two groups, confirming the stability and reliability of the proteomic sequencing data.
Figure 2.
The results of the mass spectrometry-based proteomic analysis. (A) Principal component analysis of samples for proteomic analysis. (B) The relative standard deviation of samples for proteomic analysis. (C) A total of 172 differentially expressed proteins were identified in the proteomic analysis, including 26 upregulated proteins and 146 downregulated proteins in TKI-resistant patients. The 11 candidate proteins were specifically tagged. (D) Cluster analysis of the differentially expressed proteins in the proteomic analysis. The color scale ranging from −4 to 4 represents the normalized relative protein expression levels. ccRCC, clear cell renal cell carcinoma; PFS, progression-free survival; R, resistance; RSD, relative standard deviation; S, sensitive; TKI, tyrosine kinase inhibitor.
Subsequently, we analyzed DEPs between the two groups. Using a screening threshold of P<0.05 and a fold change >1.5 or <0.67, 172 proteins were identified as being significantly differentially expressed between the TDR and TDS groups. The overall distribution of these 172 DEPs is illustrated in a volcano plot and a hierarchical clustering heatmap. The heatmap revealed that the expression patterns of these DEPs effectively distinguished between resistant and sensitive samples. To further narrow down the core potential biomarkers, we ranked all DEPs by their fold changes and applied a more stringent criterion of a fold change >3 or <0.2. This led to the identification of 11 promising candidate protein biomarkers (Figure 2C,2D; Table 2). Among these, RASA1, CHIT1, ATG9A, IL-6R, KRT13, PRCP, NT5E, and NRCAM were significantly upregulated (red dots), whereas HADH, PDIA5, and MAVS were significantly downregulated (blue dots) in the TDR group. A review of the existing literature revealed that these 11 core DEPs (e.g., IL-6R, ATG9A, and HADH) play critical roles in pathways closely associated with targeted therapy resistance, such as RCC progression, tumor microenvironment remodeling, and metabolic reprogramming. These established biological findings were highly consistent with our quantitative data, further validating the reliability and clinical translational potential of our proteomic screening results.
Table 2. Information about the 11 candidate proteins.
| Protein name | Gene ID | Fold change (TDR/TDS) | P value (U test) |
|---|---|---|---|
| ATG9A | 79065 | 3.5007 | 0.03 |
| CHIT1 | 1118 | 5.6588 | 0.02 |
| HADH | 3033 | 0.1060 | 0.001 |
| IL-6R | 3570 | 3.2435 | 0.03 |
| KRT13 | 3860 | 2.6015 | 0.04 |
| MAVS | 57506 | 0.1431 | 0.02 |
| NRCAM | 4897 | 1.5373 | 0.01 |
| NT5E | 4907 | 3.2622 | 0.02 |
| PDIA5 | 10954 | 0.1208 | 0.007 |
| PRCP | 5547 | 1.6009 | 0.005 |
| RASA1 | 5921 | 7.6116 | 0.01 |
TDR, targeted drug-resistant; TDS, targeted drug-sensitive.
Prognostic value of individual candidate proteins and model integration
To further investigate the prognostic value of each candidate protein for PFS, we quantified the levels of these proteins in plasma samples from the training cohort using ELISA. The results indicated that, within this cohort, only five plasma proteins (CHIT1, IL-6R, NRCAM, NT5E, and RASA1) exhibited significantly different expression levels between the TDR and TDS groups (P<0.0001), suggesting their independent prognostic significance (Figure 3). The expression of the remaining proteins (ATG9A and HADH) showed no significant differences between the two groups.
Figure 3.
Scatter plots of differential protein levels. The levels of eleven proteins differentially expressed between the drug-resistant group and drug-sensitive group were detected by enzyme-linked immunosorbent assay and are shown by scatter plots with mean lines and standard deviation lines. Drug-resistant group, n=43. Drug-sensitive group, n=76. ns, not significance; ****, P<0.0001. TDR, targeted drug-resistant; TDS, targeted drug-sensitive.
To construct a prognostic model, we performed multivariate analysis to evaluate the correlations between candidate proteins, potentially relevant clinical characteristics, and PFS. The median plasma protein concentration was used as the cutoff value to dichotomize patients into high- and low-expression groups, followed by binary logistic regression analysis. The results revealed that clinical features previously hypothesized to be associated with shorter PFS (including primary tumor resection, preoperative metastasis, tumor pathological type, and type of targeted therapy) were not significantly correlated with actual PFS; however, the International Metastatic RCC Database Consortium (IMDC) score and Fuhrman Grade differed between the TDS and TDR groups (Table 3). In the multivariate model, RASA1 was excluded because its P-value in the binary logistic regression analysis exceeded 0.05. The final model demonstrated that high plasma concentrations (> median) of four proteins were independent predictors of a PFS ≤3 months (Table 4): CHIT1 (>1,973.83 ng/L), IL-6R (>17.0054 ng/mL), NRCAM (>2.4306 ng/mL), and NT5E (>4.4277 ng/mL). Based on logistic regression results, we developed a nomogram model for risk prediction (Figure 4A).
Table 3. Possibly related clinical features of modeling patient set.
| Characteristics | Targeted-drug-resistant (n=43) | Targeted-drug-sensitive (n=76) | P value |
|---|---|---|---|
| Pathology | 0.11 | ||
| ccRCC | 39 (32.8) | 75 (63.0) | |
| Non-ccRCC | 4 (3.4) | 1 (0.8) | |
| IMDC score | 3 [2, 4] | 1 [1, 2] | <0.001 |
| Fuhrman grade | <0.001 | ||
| Grade 3 | 21 (17.6) | 38 (31.9) | |
| Grade 4 | 22 (18.5) | 0 (0.0) | |
| Grade 2 | 0 (0.0) | 38 (31.9) | |
| Targeted therapy | 0.89 | ||
| Sunitinib | 36 (30.3) | 63 (52.9) | |
| Pazopanib | 6 (5.0) | 10 (8.4) | |
| Sorafenib | 1 (0.8) | 3 (2.5) | |
| Prior nephrectomy | 0.11 | ||
| Yes | 38 (31.9) | 74 (62.2) | |
| No | 5 (4.2) | 2 (1.7) |
Data are presented as n (%) or median [IQR]. ccRCC, clear cell renal cell carcinoma; IMDC, International Metastatic RCC Database Consortium; IQR, interquartile range.
Table 4. Proteins independently associated with drug resistance.
| Protein name | Targeted-drug-resistant | Targeted-drug-sensitive | P value |
|---|---|---|---|
| CHIT1 | <0.001 | ||
| Low | 7 (16.28) | 53 (69.74) | |
| High | 36 (83.72) | 23 (30.26) | |
| Total | 43 (100.00) | 76 (100.00) | |
| IL-6R | <0.001 | ||
| Low | 13 (30.23) | 68 (89.47) | |
| High | 30 (69.77) | 8 (10.53) | |
| Total | 43 (100.00) | 76 (100.00) | |
| NRCAM | <0.001 | ||
| Low | 9 (20.93) | 60 (78.95) | |
| High | 34 (79.07) | 16 (21.05) | |
| Total | 43 (100.00) | 76 (100.00) | |
| NT5E | <0.001 | ||
| Low | 11 (25.58) | 58 (76.32) | |
| High | 32 (74.42) | 18 (23.68) | |
| Total | 43 (100.00) | 76 (100.00) |
Data are presented as n (%).
Figure 4.
The nomogram, calibration curves and receiver operating characteristic curves of our prediction model. (A) The nomogram of our prediction model. The plasma level of each of the 4 biomarkers was given a score on the point scale axis. A total score could be calculated by adding each single score and, by projecting the total score to the lower total point scale, we were able to estimate the probability of drug-resistance. (B) The calibration curves for the nomogram. The x-axis represents the nomogram-predicted probability and y-axis represents the actual probability of drug-resistance. Perfect prediction would correspond to the 45° dashed line. The dashed curve represents the entire modeling set (n=119) and solid curve is bias-corrected by bootstrapping (B=1,000 repetitions), indicating observed nomogram performance. (C) The ROC curves for the predictive model. The ROC curve was performed according to the external validate data of the predictive model. The x-axis represents the specificity of the predictive model and y-axis represents the sensitivity of the predictive model. The AUC was 0.869. AUC, area under the curve; ROC, receiver operating characteristic.
To assess the predictive performance of the model, we conducted an internal validation using a bootstrap resampling procedure (N=1,000) with the same selection criteria as the original model. The calibration curve demonstrated excellent concordance between the predicted and actual observed probabilities (Figure 4B). For internal validation of the model, the C-index was 0.956 (95% CI: 0.918–0.994). To further validate the accuracy of the model, the levels of the four core plasma proteins in the external validation set were measured using the same methodology. The results indicated that our novel model achieved a C-index of 0.869 (95% CI: 0.754–0.985) in the external validation set. The final ROC curve was plotted (Figure 4C) to further confirm the reliability of the model. Based on ROC curve analysis, the optimal cutoff value for the risk score was determined to be 171. Accordingly, patients were stratified into a high-risk group (total score >171) and a low-risk group (total score ≤171) for clinical decision-making.
Plasma biomarker elevations are independent of tumor tissue expression
To determine whether the elevated levels of plasma biomarkers were attributable to increased intratumoral expression, we performed immunohistochemical (IHC) staining of RCC tissue sections. No significant differences were observed in the protein expression of CHIT1, IL-6R, NRCAM, or NT5E between the targeted therapy-resistant (TDR) and targeted therapy-sensitive (TDS) groups (Figure 5A,5B). Furthermore, we investigated the transcriptional profiles using public datasets from the CPTAC and TCGA. Although the CPTAC data indicated significant correlations between the protein and mRNA levels of CHIT1, NRCAM, and NT5E (Figure S1), an analysis of the TCGA-KIRC cohort revealed no significant differences in the mRNA expression of these four genes between the resistant and sensitive patients (Figure 5C). Collectively, these findings suggest that elevated plasma protein levels may be mediated by active secretion or shedding into the circulation, rather than by differential accumulation within tumor tissues.
Figure 5.
Immunohistochemical analysis of 4 proteins. (A) Immunohistochemical sections of four mRCC TDR and TDS samples incubated with CHIT1, IL-6R, NRCAM and NT5E primary antibodies were selected. Scale bar: 50 μm. (B) Expression of CHIT1, IL-6R, NRCAM, and NT5E in the TDR groups and TDS group. P values were greater than 0.05 and not statistically significant. (C) Differential expression of CHIT1, IL-6R, NRCAM, and NT5E genes between TDR and TDS groups in the TCGA-KIRC cohort. mRCC, metastatic renal cell carcinoma; ns, not significance; TCGA-KIRC, The Cancer Genome Atlas-Kidney Renal Clear Cell Carcinoma; TDR, targeted drug-resistant; TDS, targeted drug-sensitive; TPM, transcripts per million.
Discussion
With the continuous emergence of novel therapeutic regimens for mRCC, there is an urgent clinical need to prospectively identify populations that will benefit from TKI therapy and to elucidate the underlying mechanisms of treatment resistance (16-19). Since the era of cytokine therapy, clinical prognostic models for mRCC survival have continuously evolved, most notably the Memorial Sloan Kettering Cancer Center (MSKCC) risk model (7,20,21). Following the advent of targeted therapies, Heng et al. validated the widely used IMDC risk score (22,23). However, existing models primarily reveal the correlation between risk factors and OS and fail to provide sufficient evidence for the early identification of patients who rapidly develop TKI resistance. The present study aimed to identify plasma biomarkers capable of distinguishing patients with poor responses to targeted therapy, thereby providing a reference for clinical decision-making.
In this study, we collaborated with five hospitals with expertise in the diagnosis and treatment of urological diseases, and enrolled 159 patients with mRCC. The inclusion criteria strictly limited patients to an ECOG performance status of ≤1. This design was intended to exclude early mortality and non-tumor-specific progression caused by extremely poor baseline health or severe comorbidities. By strictly controlling for this variable, we minimized the interference of confounding factors, ensuring that the observed shortened PFS was primarily attributable to the intrinsic resistance of the tumor to targeted drugs rather than patient intolerance to treatment or rapid deterioration of the general condition, thereby enhancing the reliability of our findings.
We initially performed a proteomic analysis of 27 plasma samples to identify differentially expressed proteins between the targeted therapy-resistant and therapy-sensitive groups. Subsequently, 11 candidate proteins were selected and validated for their correlation with PFS using ELISA in plasma samples from 119 patients. The results revealed that CHIT1, IL-6R, NRCAM, NT5E, and RASA1 were independently associated with PFS. Further multivariate analysis demonstrated that high plasma concentrations of four proteins (above the median, specific thresholds: CHIT1 >1,973.83 ng/L, IL-6R >17.0054 ng/mL, NRCAM >2.4306 ng/mL, NT5E >4.4277 ng/mL) were independent predictors of a PFS ≤3 months, serving as potential biomarkers for the early identification of TKI-resistant patients.
To clarify the expression characteristics of these proteins in tumor tissues, we conducted immunohistochemical (IHC) analysis; however, no significant differences in the expression of these four proteins were observed in renal cancer tissues. We analyzed the public TCGA-KIRC database and found no significant differences in the mRNA expression levels of CHIT1, IL-6R, NRCAM, and NT5E between the resistant and sensitive groups. Regarding this discrepancy between plasma and tissue protein expression, we propose the following potential mechanisms. First, these four proteins may be secreted proteins that primarily exert their functions by being released into blood circulation, thus not exhibiting significant accumulation in local tumor tissues.
Furthermore, this inconsistency in expression may be closely related to the origin of the tested samples and evolutionary characteristics of the tumor. The tissue samples used for IHC in this study were mostly derived from surgical resection of primary lesions. However, the majority of enrolled mRCC patients had already undergone primary lesion resection before receiving targeted therapy, indicating that the residual or progressing tumors in their bodies were primarily metastatic lesions. Therefore, the differences in protein levels detected in blood may predominantly reflect secretion by metastatic lesions. Given the significant spatial and temporal heterogeneity in biological characteristics and clonal evolution between primary and metastatic lesions, protein expression in the primary lesion may not comprehensively reflect plasma protein changes induced by metastases. This phenomenon highlights the unique advantages and feasibility of utilizing peripheral plasma proteins to dynamically and comprehensively assess systemic tumor burden and predict the efficacy of targeted drugs in mRCC.
Previous studies have reported associations of IL-6R and NT5E with TKI resistance in RCC. Following TKI treatment, the level of secreted IL-6 in renal cancer cells increases, which binds to IL-6R on the cancer cell surface and activates the IL-6/JAK/STAT3 pathway, leading to elevated VEGF levels and increased cell survival (24). Additionally, in various malignancies, IL-6 can promote the epithelial-mesenchymal transition (EMT) of cancer cells via TGF-β, a process that plays a critical role in tumor metastasis (15,25,26). Meanwhile, in the context of sunitinib treatment, NT5E inhibitors have been demonstrated in both in vitro and vivo experiments to suppress the growth of sunitinib-resistant cells, the EMT process, and the AKT/GSK-3β pathway (27). Currently, there are limited reports on the correlation between CHIT1, NRCAM, and TKI resistance in RCC. Nevertheless, studies have shown that CHIT1 can regulate the TGF-β pathway in various benign pulmonary diseases (28-30). In human colon cancer cells and malignant melanoma, NRCAM promotes cell growth, enhances cell motility, induces cell transformation, and forms rapidly growing tumors in nude mice, suggesting its potential involvement in tumor resistance and progression (31).
Although this study has made progress, it has several limitations. First, patients in the study cohort received either sunitinib or pazopanib, and our analysis showed no significant difference in resistance prediction between the two drugs. Possible reasons include: (I) the relatively small sample size of patients receiving pazopanib may have led to insufficient statistical power to detect potential differences; and (II) both sunitinib and pazopanib are multitarget TKIs with highly consistent core mechanisms of action, primarily exerting anti-tumor effects by blocking angiogenesis pathways. Therefore, there may be an extensive overlap in the molecular mechanisms by which these drugs induce resistance, making the biomarkers screened in this study universally predictive of both drugs. Second, during the study period, first-line treatment for mRCC underwent major advances. Two phase III clinical trials confirmed that the efficacy of immune checkpoint inhibitors combined with targeted therapy was superior to that of sunitinib monotherapy, establishing immune-targeted combination therapy as the new preferred first-line regimen, which may, to some extent, reduce the current clinical utility of our findings (32,33). Other limitations include the following: (I) the enrolled patients were exclusively of Chinese descent, lacking racial diversity, and the generalizability of the results requires further validation; (II) the findings need to be externally validated in larger cohorts to enhance their reliability; and (III) although we utilized bootstrap resampling for internal validation and a temporal split cohort for external validation to ensure the reliability of the nomogram, we did not perform additional internal validation strategies such as k-fold cross-validation. Although our current validation methods demonstrated the robustness of the model, future large-scale multicenter studies incorporating cross-validation are warranted to further verify and optimize our predictive model.
Further prospective studies are necessary to validate the prognostic utility of this model in the evolving mRCC treatment landscape. As the standard of care has transitioned to immune-targeted combination therapies, it is crucial to investigate whether the expression dynamics of these four plasma proteins correlate with treatment efficacy in patients receiving TKI plus immunotherapy. Future research should aim to evaluate the predictive value of these biomarkers in this new clinical context, thereby determining their potential to facilitate optimal patient selection and personalized therapeutic strategies in modern clinical practice.
Conclusions
Our findings support the use of plasma CHIT1, IL-6R, NRCAM, and NT5E as potential biomarkers to identify patients with mRCC who may derive minimal benefit from first-line TKI monotherapy. For such patients, we recommend increasing the frequency of follow-up monitoring or considering alternative therapeutic regimens as early as possible to improve prognosis.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Changzheng Hospital (No. 2018SL031). All patients signed an informed consent form before the start of systemic treatment and study participation. All participating hospitals were informed of and agreed to the study.
Footnotes
Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-1-0041/rc
Funding: This study was supported by the Clinical Medicine Research Special Project (No. 2024LYA05).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-1-0041/coif). All authors report support from the Clinical Medicine Research Special Project. The authors have no other conflicts of interest to declare.
Data Sharing Statement
Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-1-0041/dss
References
- 1.Siegel RL, Kratzer TB, Giaquinto AN, et al. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10-45. 10.3322/caac.21871 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rumgay H, Nethan ST, Shah R, et al. Global burden of oral cancer in 2022 attributable to smokeless tobacco and areca nut consumption: a population attributable fraction analysis. Lancet Oncol 2024;25:1413-23. 10.1016/S1470-2045(24)00458-3 [DOI] [PubMed] [Google Scholar]
- 3.Bex A, Ghanem YA, Albiges L, et al. European Association of Urology Guidelines on Renal Cell Carcinoma: The 2025 Update. Eur Urol 2025;87:683-96. 10.1016/j.eururo.2025.02.020 [DOI] [PubMed] [Google Scholar]
- 4.Sadaghiani MS, Baskaran S, Gorin MA, et al. Utility of PSMA PET/CT in Staging and Restaging of Renal Cell Carcinoma: A Systematic Review and Metaanalysis. J Nucl Med 2024;65:1007-12. 10.2967/jnumed.124.267417 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ali A, Adams DL, Kasabwala DM, et al. Cancer associated macrophage-like cells in metastatic renal cell carcinoma predicts for poor prognosis and tracks treatment response in real time. Sci Rep 2023;13:10544. 10.1038/s41598-023-37671-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Baston C, Parosanu AI, Stanciu IM, et al. Metastatic Kidney Cancer: Does the Location of the Metastases Matter? Moving towards Personalized Therapy for Metastatic Renal Cell Carcinoma. Biomedicines 2024;12:1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Heng DY, Xie W, Regan MM, et al. Prognostic factors for overall survival in patients with metastatic renal cell carcinoma treated with vascular endothelial growth factor-targeted agents: results from a large, multicenter study. J Clin Oncol 2009;27:5794-9. 10.1200/JCO.2008.21.4809 [DOI] [PubMed] [Google Scholar]
- 8.Astore S, Baciarello G, Cerbone L, et al. Primary and acquired resistance to first-line therapy for clear cell renal cell carcinoma. Cancer Drug Resist 2023;6:517-46. 10.20517/cdr.2023.33 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Riaz IB, He H, Ryu AJ, et al. A Living, Interactive Systematic Review and Network Meta-analysis of First-line Treatment of Metastatic Renal Cell Carcinoma. Eur Urol 2021;80:712-23. 10.1016/j.eururo.2021.03.016 [DOI] [PubMed] [Google Scholar]
- 10.Delahunt B, Cheville JC, Martignoni G, et al. The International Society of Urological Pathology (ISUP) grading system for renal cell carcinoma and other prognostic parameters. Am J Surg Pathol 2013;37:1490-504. 10.1097/PAS.0b013e318299f0fb [DOI] [PubMed] [Google Scholar]
- 11.Zurita AJ, Gagnon RC, Liu Y, et al. Integrating cytokines and angiogenic factors and tumour bulk with selected clinical criteria improves determination of prognosis in advanced renal cell carcinoma. Br J Cancer 2017;117:478-84. 10.1038/bjc.2017.206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chehrazi-Raffle A, Meza L, Alcantara M, et al. Circulating cytokines associated with clinical response to systemic therapy in metastatic renal cell carcinoma. J Immunother Cancer 2021;9:e002009. 10.1136/jitc-2020-002009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gudbrandsdottir G, Aarstad HH, Bostad L, et al. Serum levels of the IL-6 family of cytokines predict prognosis in renal cell carcinoma (RCC). Cancer Immunol Immunother 2021;70:19-30. 10.1007/s00262-020-02655-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Guðbrandsdottir G, Hjelle KM, Frugård J, et al. Preoperative high levels of serum vascular endothelial growth factor are a prognostic marker for poor outcome after surgical treatment of renal cell carcinoma. Scand J Urol 2015;49:388-94. 10.3109/21681805.2015.1021833 [DOI] [PubMed] [Google Scholar]
- 15.Wang Y, Zhang Y. Prognostic role of interleukin-6 in renal cell carcinoma: a meta-analysis. Clin Transl Oncol 2020;22:835-43. 10.1007/s12094-019-02192-x [DOI] [PubMed] [Google Scholar]
- 16.Zarrabi K, Fang C, Wu S. New treatment options for metastatic renal cell carcinoma with prior anti-angiogenesis therapy. J Hematol Oncol 2017;10:38. 10.1186/s13045-016-0374-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Atkins MB, Tannir NM. Current and emerging therapies for first-line treatment of metastatic clear cell renal cell carcinoma. Cancer Treat Rev 2018;70:127-37. 10.1016/j.ctrv.2018.07.009 [DOI] [PubMed] [Google Scholar]
- 18.Frank I, Blute ML, Cheville JC, et al. An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score. J Urol 2002;168:2395-400. 10.1016/S0022-5347(05)64153-5 [DOI] [PubMed] [Google Scholar]
- 19.Négrier S, Escudier B, Gomez F, et al. Prognostic factors of survival and rapid progression in 782 patients with metastatic renal carcinomas treated by cytokines: a report from the Groupe Français d’Immunothérapie. Ann Oncol 2002;13:1460-8. 10.1093/annonc/mdf257 [DOI] [PubMed] [Google Scholar]
- 20.Guo J, Jin J, Oya M, et al. Safety of pazopanib and sunitinib in treatment-naive patients with metastatic renal cell carcinoma: Asian versus non-Asian subgroup analysis of the COMPARZ trial. J Hematol Oncol 2018;11:69. 10.1186/s13045-018-0617-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Davis ID, Xie W, Pezaro C, et al. Efficacy of Second-line Targeted Therapy for Renal Cell Carcinoma According to Change from Baseline in International Metastatic Renal Cell Carcinoma Database Consortium Prognostic Category. Eur Urol 2017;71:970-8. 10.1016/j.eururo.2016.09.047 [DOI] [PubMed] [Google Scholar]
- 22.Heng DY, Xie W, Regan MM, et al. External validation and comparison with other models of the International Metastatic Renal-Cell Carcinoma Database Consortium prognostic model: a population-based study. Lancet Oncol 2013;14:141-8. 10.1016/S1470-2045(12)70559-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Manola J, Royston P, Elson P, et al. Prognostic model for survival in patients with metastatic renal cell carcinoma: results from the international kidney cancer working group. Clin Cancer Res 2011;17:5443-50. 10.1158/1078-0432.CCR-11-0553 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ishibashi K, Koguchi T, Matsuoka K, et al. Interleukin-6 induces drug resistance in renal cell carcinoma. Fukushima J Med Sci 2018;64:103-10. 10.5387/fms.2018-15 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ishibashi K, Haber T, Breuksch I, et al. Overriding TKI resistance of renal cell carcinoma by combination therapy with IL-6 receptor blockade. Oncotarget 2017;8:55230-45. 10.18632/oncotarget.19420 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Weng YS, Tseng HY, Chen YA, et al. MCT-1/miR-34a/IL-6/IL-6R signaling axis promotes EMT progression, cancer stemness and M2 macrophage polarization in triple-negative breast cancer. Mol Cancer 2019;18:42. 10.1186/s12943-019-0988-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Peng D, Hu Z, Wei X, et al. NT5E inhibition suppresses the growth of sunitinib-resistant cells and EMT course and AKT/GSK-3β signaling pathway in renal cell cancer. IUBMB Life 2019;71:113-24. 10.1002/iub.1942 [DOI] [PubMed] [Google Scholar]
- 28.Gao S, Hu J, Wu X, et al. PMA treated THP-1-derived-IL-6 promotes EMT of SW48 through STAT3/ERK-dependent activation of Wnt/β-catenin signaling pathway. Biomed Pharmacother 2018;108:618-24. 10.1016/j.biopha.2018.09.067 [DOI] [PubMed] [Google Scholar]
- 29.Cha JH, Park NR, Cho SW, et al. Chitinase 1: a novel therapeutic target in metabolic dysfunction-associated steatohepatitis. Front Immunol 2024;15:1444100. 10.3389/fimmu.2024.1444100 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hong JY, Kim M, Sol IS, et al. Chitotriosidase inhibits allergic asthmatic airways via regulation of TGF-β expression and Foxp3+ Treg cells. Allergy 2018;73:1686-99. 10.1111/all.13426 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zhou L, He L, Liu CH, et al. Liver cancer stem cell dissemination and metastasis: uncovering the role of NRCAM in hepatocellular carcinoma. J Exp Clin Cancer Res 2023;42:311. 10.1186/s13046-023-02893-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Rini BI, Plimack ER, Stus V, et al. Pembrolizumab plus Axitinib versus Sunitinib for Advanced Renal-Cell Carcinoma. N Engl J Med 2019;380:1116-27. 10.1056/NEJMoa1816714 [DOI] [PubMed] [Google Scholar]
- 33.Motzer RJ, Penkov K, Haanen J, et al. Avelumab plus Axitinib versus Sunitinib for Advanced Renal-Cell Carcinoma. N Engl J Med 2019;380:1103-15. 10.1056/NEJMoa1816047 [DOI] [PMC free article] [PubMed] [Google Scholar]