Gene profiling and clinicopathological features for prognostic modeling of recurrence in non-metastatic clear-cell renal cell carcinoma

Introduction

Renal cell carcinoma (RCC), comprising about 3% of all adult malignancies worldwide (1), is predominantly clear-cell RCC (ccRCC), which constitutes 70–80% of cases and is marked by extensive histopathological and genetic variability (2). Despite the majority of ccRCC patients presenting with localized disease at diagnosis and undergoing potentially curative surgery or ablation, 20–30% experience postoperative relapse, which critically impacts prognosis (3). The lack of a universally accepted postoperative adjuvant therapy regimen, backed by robust evidence, poses a challenge in managing ccRCC. While recent studies suggest that postoperative immunotherapy with pembrolizumab can enhance disease-free survival (DFS) for patients at intermediate to high risk of recurrence (4), there is a pressing need for more precise methods to predict recurrence.

While studies have identified risk factors such as tumor-node-metastasis (TNM) stage, nuclear grade and patient performance status as predictors of postoperative recurrence and reduced DFS in RCC (5,6), current prognostic systems that include these clinical and pathological parameters fall short in terms of accuracy, limiting their utility in post-surgical outcome prediction (7). Moreover, these traditional markers may not fully capture the complexity of individual tumor biology (8). In light of these limitations, next-generation sequencing (NGS) has emerged as a powerful tool in cancer research, offering insights into tumorigenesis and progression (9). Extensive research, including The Cancer Genome Atlas (TCGA) project (10), has utilized NGS to delineate the molecular profiles of RCC subtypes, particularly ccRCC, revealing genes such as PBRM1, SEDT2, BAP1 (11), and others as significant markers linked to clinical and pathological outcomes post-treatment. Moreover, circulating tumor DNA (ctDNA) has emerged as a promising non-invasive biomarker for the early detection of recurrence and real-time monitoring of tumor dynamics. Therefore, merging molecular data with clinicopathological features may improve high-risk tumor identification, enhance recurrence prediction, and refine management strategies for non-metastatic ccRCC (NMCCRCC) patients.

In this study, we performed NGS on tumor tissue DNA from a cohort of 73 RCC patients to explore the correlations between clinicopathological and genetic features. We then developed a novel prognostic model aimed at predicting postoperative recurrence in patients with NMCCRCC. Furthermore, we assessed the utility of ctDNA as a promising non-invasive biomarker for detecting mutations and prognosticating RCC outcomes. We present this article in accordance with the TRIPOD reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-2025-177/rc).

Methods

Patients and samples

Tumor tissue and peripheral blood (PB) samples were procured from 73 RCC patients (54 with ccRCC and 19 with non-ccRCC) for NGS at Daping Hospital, Chongqing, China, from November 2014 to June 2023. Additionally, PB samples from 13 of these patients were analyzed for ctDNA. In NMCCRCC patients, DFS was calculated from the surgery date to the earliest occurrence of local or regional recurrence, metastasis, contralateral kidney cancer, or death. A total of 18 patients experienced DFS events, with a median DFS of 23.5 months [interquartile range (IQR), 8.25–34 months]. The clinical attributes of the 73 patients were detailed in Table 1. Ethical clearance was granted by the Daping Hospital Ethics Committee (No. 2024_56), and informed consent was obtained from all study participants. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

DNA extraction, target capture, and sequencing data analysis

Sequencing was conducted using a methodology previously described in our earlier publication (12). DNA was isolated from formalin-fixed and paraffin-embedded (FFPE) tumor samples or fresh tissue, and blood samples were processed to separate PB lymphocytes (PBLs) and plasma via centrifugation. DNA from PBLs was extracted to serve as normal controls for somatic mutation detection, while ctDNA was extracted from plasma using the QIAamp Circulating Nucleic Acid Kit. DNA fragmentation, quantification, and size distribution were assessed following standard protocols and stringent quality control measures. The sequencing library was prepared with the KAPA DNA Library Preparation Kit, and target regions were enriched using the SeqCap EZ library system (Roche NimbleGen, Madison, USA). For tissue and PBL DNA, sequencing was performed using either whole-exome capture (n=6) or custom panels targeting commonly mutated genes in solid tumors, including panels of 1,021 genes (n=38), 89 genes (n=26), 77 genes (n=2), and 642 genes (n=1). In contrast, all ctDNA libraries were analyzed using the 1,021-gene panel. Sequencing was carried out on the MGIseq-2000 (Gene+ Technology, Shenzhen, China). Quality control standards included a minimum of 10% tumor cell content in tissue samples, as determined by hematoxylin and eosin (H&E) staining, with an average sequencing depth of ≥500×. For blood samples, the total DNA yield was required to be ≥15 ng, with an average sequencing depth of ≥4,000×.

Raw reads were processed to remove terminal adaptor sequences and low-quality reads using NCfilter (version 2.0.0; in-house software). Subsequent to cleaning, reads were aligned to the human genome (hg19) with the Burrows-Wheeler Aligner (BWA) (version 0.7.15-r1140) (13). Duplicate reads from cancer samples, a result of PCR amplification, were identified using realSeq software, while duplicates in control samples were marked with the MarkDuplicates module in Picard tools (http://broadinstitute.github.io/picard/). Somatic single nucleotide variants (SNVs) and small insertions and deletions (indels) were identified by comparing tumor-normal pairs with MuTect (1.1.4) (14), considering only variants supported by ≥5 high-quality sequencing reads (mapQthres >30, baseQthres >30). Somatic copy number alterations (CNAs) were analyzed with GATK software (version 4.0) (15), using matched PB cell samples as controls. Copy number gains were annotated as median log₂ ratio >1.25, and losses as log₂ ratio <0.75.

A variant in cfDNA was classified as a ctDNA mutation if it met the following criteria: (I) tumor-specific driver gene variants present in matched tumor tissue with at least 2 high-quality reads; (II) non-recurrent variants in matched tumor tissue with at least 4 high-quality reads; (III) hot-spot variants unique to blood samples with at least 4 high-quality reads; (IV) non-hotspot variants unique to blood samples with at least 8 high-quality reads.

Validation cohorts

We obtained the “Kidney Renal Clear Cell Carcinoma (TCGA, Firehose Legacy)” dataset from the cBioPortal website (https://www.cbioportal.org/datasets), hereinafter referred to as the TCGA-KIRC cohort, which includes mutation data for 424 patients (10). From this cohort, 344 NMCCRCC patients with complete clinical data were selected as an external validation set for our recurrence prediction model. In addition, we collected a validation dataset for our prognostic model from the study published by Vasudev et al. in the Clinical Cancer Research (CCR) journal. Patients with NMCCRCC, who had complete TNM stage information and mutational profiles were enrolled. The validation dataset was referred to as the CCR cohort (N=745) (16). Detailed patient information is provided in Table S1.

Mutational concordance between ctDNA and tumor tissue DNA

At the individual level, the concordance between ctDNA and paired tissue DNA was quantified as the ratio of identical mutations to the total number of mutations identified in both samples (17). This metric reflects the genetic overlap between ctDNA and tumor tissue DNA within individual patients and the ability of ctDNA to capturing tumor genetic characteristics. The overall concordance rate in study cohort was calculated as the proportion of patients exhibiting at least one shared mutation between ctDNA and tissue DNA (17). This method assesses how often ctDNA mutations mirror those in tumor tissue DNA across the patient population, reflecting the accuracy and reliability of ctDNA’s application in non-invasive diagnostics.

Statistical analyses

To standardize the analysis and minimize potential biases, we focused on a core set of 76 genes (Table S2) that intersected across the four different gene panels used for tissue DNA sequencing. Categorical variables were described as percentages with frequencies, and continuous non-parametric data were presented as medians with IQRs. The Fisher’s exact test was used to compare categorical variables.

Cox proportional hazards (PHs) models were employed to determine the association between gene alterations and DFS in NMCCRCC patients. Multivariable survival models were constructed to examine the independent effects of these associations. The stepwise approach (forward-backward) in Cox PHs regression was then used to refine the selection of potential risk factors for the predictive model. The predictive accuracy of the model was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC). All statistical analyses were conducted using R software (version 4.2.3, R Foundation for Statistical Computing, Vienna, Austria), with a significance level set at P<0.05.

Results

Demographic and clinical characteristics of RCC patients

A total of 73 RCC patients were enrolled, with a median age of 53 years (IQR, 47–61 years). Forty-eight patients had localized diseases, while 25 had distant metastases. Surgical intervention was performed on 64 patients, including 24 partial nephrectomies and 40 radical nephrectomies; nine patients with multiple metastases received tyrosine kinase inhibitor (TKI) therapy only. Sequencing data from two separate lesions in the same patient’s kidney were combined, regardless of clonal correlation analysis suggesting independent origins. Patient characteristics are summarized in Table 1.

Gene mutations in RCC patients

Sequencing analysis of the 73 RCC patients revealed mutations in 54 genes. The mutational landscape of these patients is depicted in Figure 1. VHL was the most commonly mutated gene, with a frequency of 54.79% (40/73). In ccRCC patients, VHL mutations were prevalent (72.22%, 39/54), followed by PBRM1 (25.93%, 14/54), BAP1 (20.37%, 11/54), aligning with patterns reported in the TCGA database (10). However, notable variations in mutation frequencies for genes such as FH, BRCA1, and TP53 were observed, suggesting unique genetic characteristics in our cohort (Figure 2A).

Figure 1 Mutation landscape across RCC patient groups. This figure illustrates the mutation profiles of 73 RCC patients categorized into NMCCRCC (N=34), mccRCC (N=20), and non-ccRCC (N=19). Each row corresponds to a unique gene, with colored squares denoting distinct mutation types. Mutation frequencies are listed beside each gene, and the legend explains the color coding. Clinical data including T, N, M, and Fuhrman grade are provided at the top. ccRCC, clear cell renal cell carcinoma; M, metastasis; mccRCC, metastatic clear cell renal cell carcinoma; N, node; NA, not available; NMCCRCC, non-metastatic clear cell renal cell carcinoma; RCC, renal cell carcinoma; T, tumor.

Figure 2 Comparisons between genetic alterations and clinical features in ccRCC. Subfigures (A) to (F) compare gene mutation frequencies across different contexts: (A) between our ccRCC cohort and TCGA-KIRC; (B) between ccRCC and non-ccRCC; and stratified by (C) tumor stage, (D) lymph node status, (E) distant metastasis, and (F) Fuhrman grade in ccRCC patients. Statistical significance is marked with asterisks: *, P<0.1; **, P<0.05; ***, P<0.01. ccRCC, clear cell renal cell carcinoma; mccRCC, metastatic clear cell renal cell carcinoma; N, node; NMCCRCC, non-metastatic clear cell renal cell carcinoma; T, tumor; TCGA, The Cancer Genome Atlas.

In the non-ccRCC cohort, the most frequently mutated genes were TP53 (26.32%). A comparison of mutation frequencies between the ccRCC and non-ccRCC cohorts revealed significant differences (Figure 2B). Genes like JAK1 and MDM2 were mutated in the non-ccRCC cohort (10.53% each) but not in the ccRCC cohort. Moreover, BAP1 and PBRM1 mutations were less common in the non-ccRCC cohort compared to the ccRCC cohort. VHL mutations were significantly more frequent in the ccRCC cohort (72.22%) than in the non-ccRCC cohort (5.26%, P<0.001). These discrepancies indicate distinct mutation profiles between ccRCC and non-ccRCC, potentially indicating different genetic mechanisms and implications for treatment strategies.

The associations between genetic alterations and clinical characteristics in ccRCC patients

Analyzing the correlation between gene mutations and clinical characteristics is essential for elucidating their influence on disease progression and patient outcomes. Patients were categorized into groups based on T stage (T1–2, n=33; T3–4, n=21), lymph node metastasis (N0, n=43; N1, n=11), and distant metastasis (M0, n=34; M1, n=20). TP53 mutations were more frequent in T3–4 (23.81%) vs. T1–2 (3.03%, P=0.03) and in N1 (36.36%) vs. N0 (4.65%, P=0.01), suggesting a role in ccRCC progression. This underscores their potential as prognostic markers and therapeutic targets (Figure 2C,2D). FH mutations were also notable, being absent in the non-metastasis group but present in the metastasis group (15.00%, P=0.046, Figure 2E).

To further examine the relationship between gene mutations and Fuhrman grades, Fuhrman grades 1–2 were defined as low-grade (n=33) and grades 3–4 as high-grade (n=21). Commonly mutated genes, such as VHL, PBRM1, and SETD2, showed no significant differences in mutation frequencies between these two groups (Figure 2F). However, BAP1 mutations were significantly more frequent in the high-grade group (38.10%) compared to the low-grade group (9.09%, P=0.02), suggesting a potential role for BAP1 in the progression and severity of ccRCC. Additionally, there was no association between the frequencies of genetic alterations and tumor necrosis features.

Risk factors for postoperative recurrence in NMCCRCC

To identify clinical and molecular risk factors for postoperative recurrence in NMCCRCC and enhance prognostic assessments, we followed up on 34 patients’ post-surgery. The median follow-up duration was 35 months. Univariate Cox regression revealed several factors associated with recurrence. Advanced T stage and genetic alterations in SETD2, ALK, BAP1, and BRCA2 were significantly associated with an increased risk of recurrence. Additionally, clinical factors such as age, necrosis, and lymph node involvement, along with other genetic alterations including TP53, APC, and ATM mutations, were near-significant risk factors, with P values less than 0.1 (Table 2).

To identify independent risk factors, a multivariate Cox regression analysis was conducted, including factors with P values less than 0.1 from the univariate analysis. This analysis revealed that advanced T stage [hazard ratio (HR) =11.2, 95% confidence interval (CI): 1.24–101.05, P=0.03], BAP1 (HR =4.98, 95% CI: 1.31–18.99, P=0.02), and SETD2 (HR =5.04, 95% CI: 1.15–22.12, P=0.03) mutations remained significant independent predictors of recurrence (Table 2). These findings underscore the importance of these factors in predicting recurrence in NMCCRCC patients and their relevance in prognostic assessments and patient management strategies.

Construction and validation of a predictive model for DFS in NMCCRCC

Although we identified independent risk factors for postoperative recurrence in NMCCRCC, a practical approach was needed to translate these findings into clinical practice. A multivariate prognostic model was developed to address this. Predictors with P values <0.1 from the univariate analysis were selected, excluding ALK due to its low mutation frequency (1/34 patients). An optimal multivariate model was constructed, which included three key predictors: T stage, BAP1 mutation, and SETD2 mutation. The model’s concordance index (C-index) was 0.833, indicating high predictive accuracy. Statistical significance and robustness were confirmed through likelihood ratio tests (χ²=24.34, df=3, P<0.001), Wald tests (χ²=18.92, df=3, P<0.001), and Score (log-rank) tests (χ²=27.93, df=3, P<0.001).

To facilitate clinical application, a nomogram based on the multivariate Cox model was developed to estimate 1-, 3-, and 5-year DFS probabilities for NMCCRCC patients (Figure S1). Risk scores were calculated for each patient, and ROC curves for 1 to 5 years were plotted, yielding AUC values of 0.900, 0.887, 0.843, 0.829, and 0.821, respectively, indicating good predictive accuracy (Figure 3A). An optimal cutoff of 5.00 was identified for risk scores using the “surv_cutpoint” function in the R package “survminer”, dividing patients into high-risk (scores >5.00, n=12) and low-risk (scores ≤5.00, n=22) groups. Kaplan-Meier survival analysis and Log-rank tests revealed that high-risk patients had significantly shorter median DFS compared to low-risk patients (12 vs. 75 months, P<0.001, Figure 3B).

Figure 3 Predictive model performance for DFS in NMCCRCC patients. ROC curves for the predictive model are presented for (A) our NMCCRCC cohort, (C) the CCR cohort, and (E) the TCGA-KIRC cohort, with AUC values indicating 1- to 5-year DFS predictions. Kaplan-Meier curves (B), (D), and (F) compare DFS between high- and low-risk groups in the respective cohorts. AUC, area under the curve; CCR, Clinical Cancer Research; DFS, disease-free survival; NMCCRCC, non-metastatic clear cell renal cell carcinoma; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.

External validation was conducted using two independent datasets. In the CCR cohort of 745 NMCCRCC patients, the model demonstrated AUC values of 0.751, 0.717, 0.699, 0.700, and 0.688 for 1 to 5 years, respectively (Figure 3C). Kaplan-Meier survival analysis and Log-rank tests indicated that high-risk patients had significantly shorter DFS compared to low-risk patients (110 months vs. not reached, P<0.001, Figure 3D). In the TCGA-KIRC cohort, ROC curves showed AUC values of 0.756, 0.768, 0.736, 0.721, 0.738 for 1 to 5 years, respectively (Figure 3E). Kaplan-Meier survival analysis and Log-rank tests confirmed that high-risk patients had significantly shorter median DFS compared to low-risk patients (28.7 months vs. not reached, P<0.001, Figure 3F).

Mutational concordance and clinical relevance of ctDNA in RCC

In our study, NGS was performed on ctDNA and tumor tissue DNA from 13 RCC patients. All 13 tumor tissue samples harbored mutations across 45 genes, totaling 64 mutations. ctDNA mutations were detected in 8 patients (61.54%), encompassing 29 mutations in 25 genes (Figure S2). A comparison of ctDNA and tumor tissue DNA mutation frequencies revealed no significant differences across all genes (Figure S3).

Each of the 8 ctDNA-positive blood samples shared at least one identical mutation with its paired tissue sample, yielding an overall concordance rate of 61.54% (8/13). The average and median concordance rates across the 13 patients were 30.40% and 15.38%, respectively (Tables S3). In patient NMCCRCC-33, a scatter plot of ctDNA and tissue DNA mutation variant allele fractions (VAFs) revealed a positive correlation (R²=0.819, R=0.905, P<0.001), suggesting consistency in mutation detection between ctDNA and tissue DNA (Figure S4A).

Higher ctDNA detection rates were observed in patients with advanced T3–4 stage (6/6 vs. 2/7, P=0.02) and distant metastases (5/5 vs. 3/8, P=0.08, Figure S4B). However, the detection rates were comparable between patients with positive and negative lymph node metastases (100% vs. 44.44%, P=0.11). Median concordance between ctDNA and tissue DNA was similar between advanced (T3–4) and early (T1–2) stages (42% vs. 0%, P=0.21, Figure S4C) and between patients with and without lymph node metastasis (53% vs. 0%, P=0.18, Figure S4D). Distant metastasis patients exhibited a higher median concordance (55.60%) compared to those without metastasis (0%, P=0.042, Figure S4E). ctDNA detection did not correlate with DFS in the 8 NMCCRCC patients (Figure S4F). Moreover, univariate Cox regression analysis indicated that concordance between ctDNA and tumor tissue DNA was not a significant factor affecting DFS in NMCCRCC (HR =0.96; P=0.99; 95% CI: 0.0123–19.314).

Discussion

In this study, we analyzed gene mutations and their clinical associations in 73 RCC patients, with a focus on postoperative recurrence in 34 NMCCRCC patients. We identified T stage, BAP1, and SETD2 mutations as independent risk factors for postoperative recurrence in NMCCRCC patients. These factors were integrated into a prognostic model that demonstrated high predictive accuracy (C-index =0.833) and were validated internally and across two external cohorts. Furthermore, the high concordance between ctDNA and tumor tissue DNA mutations, particularly in patients with distant metastases, highlights the potential of ctDNA as a non-invasive prognostic biomarker.

Despite the promising surgical outcomes for NMCCRCC patients, with a 5-year survival rate of 70–90% (18), the postoperative recurrence rate remains high at 30% (3). The KEYNOTE-564 trial, although demonstrating a 10% improvement in delaying recurrence with pembrolizumab, highlights the inadequacy of current risk stratification methods (19). The current clinicopathological methods used in trials may not fully capture the complexity of tumor biology, leading to suboptimal risk prediction (8). The development of prognostic models for recurrence has been ongoing (5), with Kattan et al.’s initial nomogram in 2001 (C-index 0.74) as a starting point (20). However, these models have been variably validated and their performance has not met expectations, as evidenced by the ASSURE trial, where various risk stratification systems demonstrated low C-indices (21). In response to these challenges, our study has constructed and validated a multivariate prognostic model for NMCCRCC patients, integrating both genetic and clinical factors.

In this study, we identified T stage, BAP1 mutations, and SETD2 mutations as independent risk factors for postoperative recurrence in patients with NMCCRCC. Our findings align with previous reports by Hakimi et al. (22), who highlighted the prognostic significance of BAP1 and SETD2 mutations in ccRCC. Expanding on this foundation, we developed a clinically practical, mutation-based prognostic model focused on DFS. Compared to the 16-gene expression-based recurrence score by Rini et al. (23), our streamlined model combining BAP1/SETD2 mutations with clinical T stage demonstrated superior prognostic accuracy, achieving 1- to 5-year AUC values of 0.900–0.821 and a C-index of 0.833. Our study also builds on the SSPN scoring system proposed by Ohsugi et al. (24), which incorporated PBRM1 expression, pathological stage, and tumor necrosis. Although PBRM1 was not retained in our final model, BAP1 and SETD2 mutations emerged as more powerful and consistent predictors of recurrence. The robustness and generalizability of our model were further validated through external testing in the CCR and TCGA-KIRC cohorts, where it maintained favorable performance with AUCs ranging from 0.688 to 0.751 and 0.721 to 0.768, respectively. Moreover, our results extend the work of Vlachostergios et al. (25), who emphasized the prognostic importance of non-VHL mutations—particularly PBRM1, BAP1, and SETD2—in primary ccRCC, and their potential role in selecting candidates for adjuvant therapy. Similarly, Vasudev et al. demonstrated that tumors with VHL mutations alone were associated with improved prognosis (16), whereas co-mutations with genes like SETD2 conferred a higher risk of recurrence. By integrating these genetic markers, our model captures and refines these insights, underscoring the pivotal role of BAP1 and SETD2 mutations in recurrence risk stratification. Collectively, these findings not only validate and build upon previous studies but also provide a clinically applicable tool for predicting postoperative recurrence and guiding individualized management in NMCCRCC.

SETD2, a histone methyltransferase, is integral to the maintenance of genomic stability. Its loss has been implicated in various cancers (26), including its role in the epigenetic reprogramming of RCC, as highlighted by Xie et al. (27), which promotes metastasis and creates dependencies on histone chaperone complexes. Additionally, Xue et al. (28) demonstrated that SETD2-mediated trimethylation of histone H3 at lysine 36 (H3K36me3) regulates the transcription of ferrochelatase (FECH), influencing ferroptosis-related pathways and tumor cell apoptosis. Additionally, BAP1, a tumor suppressor gene, functions as a chromatin regulator and modulates cell proliferation through its histone deubiquitinase activity (29). Its prognostic value has been recognized in ccRCC (30). Langbein et al. (31) showing that BAP1 supports hypoxia-inducible factor (HIF)-dependent interferon beta induction to suppress tumor growth. Furthermore, Friedhoff et al. (32) correlated BAP1 mutations with a reduction in tumor-infiltrating lymphocytes and increased programmed death-ligand 1 (PD-L1) expression, facilitating immune evasion. Kapur et al. (33) have also linked mTORC1 activation with BAP1 loss, driving tumor grade and aggressiveness. The outcomes of these foundational studies collectively indicate that the proteins SETD2 and BAP1 play pivotal roles in the normal maintenance of kidney function, and their mutations leading to loss of function are intimately linked to malignant tumor behavior. Therefore, the integration of these two mutations in our model is well-founded and justified, providing a robust prognostic tool for NMCCRCC patients.

Our study revealed that the concordance between plasma ctDNA and paired tissue biopsies can reflect the potential of ctDNA to overcome genetic heterogeneity in RCC. Previous studies have reported ctDNA detection rates in RCC patients ranging from 17% to 54% (34), with a mutational concordance of only 8.6% noted by Hahn et al. (35). These studies characterized RCC as a ctDNA-low malignancy. However, our study observed a ctDNA detection rate and overall mutational concordance of 61.54% (8/13), which may be attributed to advancements in sequencing technology enhancing ctDNA detection sensitivity. Consistent with earlier research (36), we found that ctDNA detection rates and mutational concordance were significantly higher in patients with advanced T3–4 stages and distant metastases. Even in NMCCRCC patients, ctDNA demonstrated the ability to capture the genetic heterogeneity and burden of cancer, as evidenced by the positive correlation in VAFs between ctDNA and tissue DNA in patient NMCCRCC-33. Several studies have consistently shown that ctDNA-positive RCC patients across various stages are significantly correlated with increased mortality risk, diminished DFS (37), cancer-specific survival (37), or overall survival (36). These findings suggest that ctDNA holds promise as a reliable, non-invasive method for diagnosing and monitoring disease progression, although the need for larger studies to confirm its prognostic value remains.

It is crucial to recognize the limitations of our study. Although our model demonstrated high predictive accuracy in the validation cohorts, further prospective studies are required to confirm its utility in diverse clinical settings. Additionally, the incorporation of novel biomarkers and advanced imaging techniques could potentially enhance the model’s predictive performance in future iterations. Nevertheless, despite being a preliminary exploration based on a small sample size, our study underscores the significant role of molecular characteristics in predicting the postoperative recurrence of NMCCRCC.

Conclusions

Our study introduces a novel and robust prognostic model for NMCCRCC that integrates genetic and clinicopathological factors to offer a highly accurate prediction of DFS. This model has significant potential for enhancing patient stratification and guiding clinical decision-making, thereby contributing to personalized treatment strategies in the management of NMCCRCC.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-177/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-177/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-2025-177/prf

Funding: This work was supported by the National Natural Science Foundation of China (Nos. 82172807 and 82172721); and University Research Project of Army Medical University (No. 2019CXLCB006).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2025-177/coif). The authors have no conflicts of interest to declare.

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. Ethical clearance was granted by the Daping Hospital Ethics Committee (No. 2024_56). Informed consent was obtained from all individual participants enrolled in the study. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Ferlay J, Colombet M, Soerjomataram I, et al. Cancer incidence and mortality patterns in Europe: Estimates for 40 countries and 25 major cancers in 2018. Eur J Cancer 2018;103:356-87. [Crossref] [PubMed]
2. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med 2012;366:883-92. [Crossref] [PubMed]
3. Dabestani S, Beisland C, Stewart GD, et al. Long-term Outcomes of Follow-up for Initially Localised Clear Cell Renal Cell Carcinoma: RECUR Database Analysis. Eur Urol Focus 2019;5:857-66. [Crossref] [PubMed]
4. Powles T, Tomczak P, Park SH, et al. Pembrolizumab versus placebo as post-nephrectomy adjuvant therapy for clear cell renal cell carcinoma (KEYNOTE-564): 30-month follow-up analysis of a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial. Lancet Oncol 2022;23:1133-44. [Crossref] [PubMed]
5. Zisman A, Pantuck AJ, Dorey F, et al. Improved prognostication of renal cell carcinoma using an integrated staging system. J Clin Oncol 2001;19:1649-57. [Crossref] [PubMed]
6. Sorbellini M, Kattan MW, Snyder ME, et al. A postoperative prognostic nomogram predicting recurrence for patients with conventional clear cell renal cell carcinoma. J Urol 2005;173:48-51. [Crossref] [PubMed]
7. Usher-Smith JA, Li L, Roberts L, et al. Risk models for recurrence and survival after kidney cancer: a systematic review. BJU Int 2022;130:562-79. [Crossref] [PubMed]
8. Delahunt B, Sika-Paotonu D, Bethwaite PB, et al. Grading of clear cell renal cell carcinoma should be based on nucleolar prominence. Am J Surg Pathol 2011;35:1134-9. [Crossref] [PubMed]
9. Ayanambakkam A, Basu A. Future Directions for Circulating Tumor DNA Studies in Renal Cell Carcinoma. Eur Urol Focus 2025;S2405-4569(25)00082-3.
10. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 2013;499:43-9. [Crossref] [PubMed]
11. Joseph RW, Kapur P, Serie DJ, et al. Loss of BAP1 protein expression is an independent marker of poor prognosis in patients with low-risk clear cell renal cell carcinoma. Cancer 2014;120:1059-67. [Crossref] [PubMed]
12. Wang Z, Yan X, Tang P, et al. Genetic profiling of hormone-sensitive and castration-resistant prostate cancers and identification of genetic mutations prone to castration-resistant prostate cancer. Prostate Cancer Prostatic Dis 2023;26:180-7. [Crossref] [PubMed]
13. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009;25:1754-60. [Crossref] [PubMed]
14. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213-9. [Crossref] [PubMed]
15. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010;20:1297-303. [Crossref] [PubMed]
16. Vasudev NS, Scelo G, Glennon KI, et al. Application of Genomic Sequencing to Refine Patient Stratification for Adjuvant Therapy in Renal Cell Carcinoma. Clin Cancer Res 2023;29:1220-31. [Crossref] [PubMed]
17. Sabari JK, Offin M, Stephens D, et al. A Prospective Study of Circulating Tumor DNA to Guide Matched Targeted Therapy in Lung Cancers. J Natl Cancer Inst 2019;111:575-83. [Crossref] [PubMed]
18. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin 2020;70:7-30. [Crossref] [PubMed]
19. Choueiri TK, Tomczak P, Park SH, et al. Adjuvant Pembrolizumab after Nephrectomy in Renal-Cell Carcinoma. N Engl J Med 2021;385:683-94. [Crossref] [PubMed]
20. Kattan MW, Reuter V, Motzer RJ, et al. A postoperative prognostic nomogram for renal cell carcinoma. J Urol 2001;166:63-7. [Crossref] [PubMed]
21. Correa AF, Jegede OA, Haas NB, et al. Predicting Disease Recurrence, Early Progression, and Overall Survival Following Surgical Resection for High-risk Localized and Locally Advanced Renal Cell Carcinoma. Eur Urol 2021;80:20-31. [Crossref] [PubMed]
22. Hakimi AA, Ostrovnaya I, Reva B, et al. Adverse outcomes in clear cell renal cell carcinoma with mutations of 3p21 epigenetic regulators BAP1 and SETD2: a report by MSKCC and the KIRC TCGA research network. Clin Cancer Res 2013;19:3259-67. [Crossref] [PubMed]
23. Rini B, Goddard A, Knezevic D, et al. A 16-gene assay to predict recurrence after surgery in localised renal cell carcinoma: development and validation studies. Lancet Oncol 2015;16:676-85. [Crossref] [PubMed]
24. Ohsugi H, Yoshida T, Ohe C, et al. The SSPN Score, a Novel Scoring System Incorporating PBRM1 Expression, Predicts Postoperative Recurrence for Patients with Non-metastatic Clear Cell Renal Cell Carcinoma. Ann Surg Oncol 2021;28:2359-66. [Crossref] [PubMed]
25. Vlachostergios PJ, Papathanassiou M, Anagnostou M, et al. Mutational profile of primary clear cell renal cell carcinoma predicts recurrence and potential candidacy for adjuvant immune checkpoint inhibition. F1000Res 2023;12:918. [Crossref] [PubMed]
26. Lu M, Zhao B, Liu M, et al. Pan-cancer analysis of SETD2 mutation and its association with the efficacy of immunotherapy. NPJ Precis Oncol 2021;5:51. [Crossref] [PubMed]
27. Xie Y, Sahin M, Sinha S, et al. SETD2 loss perturbs the kidney cancer epigenetic landscape to promote metastasis and engenders actionable dependencies on histone chaperone complexes. Nat Cancer 2022;3:188-202. [Crossref] [PubMed]
28. Xue W, Jian W, Meng Y, et al. Knockdown of SETD2 promotes erastin-induced ferroptosis in ccRCC. Cell Death Dis 2023;14:539. [Crossref] [PubMed]
29. Daou S, Hammond-Martel I, Mashtalir N, et al. The BAP1/ASXL2 Histone H2A Deubiquitinase Complex Regulates Cell Proliferation and Is Disrupted in Cancer. J Biol Chem 2015;290:28643-63. [Crossref] [PubMed]
30. Galtung KF, Lauritzen PM, Baco E, et al. Predictive Performance of Prospectively Applied ISUP and Fuhrman Grade in Nonmetastatic Renal Cell Carcinoma. Anticancer Res 2022;42:2967-75. [Crossref] [PubMed]
31. Langbein LE, El Hajjar R, He S, et al. BAP1 maintains HIF-dependent interferon beta induction to suppress tumor growth in clear cell renal cell carcinoma. Cancer Lett 2022;547:215885. [Crossref] [PubMed]
32. Friedhoff J, Schneider F, Jurcic C, et al. BAP1 and PTEN mutations shape the immunological landscape of clear cell renal cell carcinoma and reveal the intertumoral heterogeneity of T cell suppression: a proof-of-concept study. Cancer Immunol Immunother 2023;72:1603-18. [Crossref] [PubMed]
33. Kapur P, Rajaram S, Brugarolas J. The expanding role of BAP1 in clear cell renal cell carcinoma. Hum Pathol 2023;133:22-31. [Crossref] [PubMed]
34. Lasseter K, Nassar AH, Hamieh L, et al. Plasma cell-free DNA variant analysis compared with methylated DNA analysis in renal cell carcinoma. Genet Med 2020;22:1366-73. [Crossref] [PubMed]
35. Hahn AW, Gill DM, Maughan B, et al. Correlation of genomic alterations assessed by next-generation sequencing (NGS) of tumor tissue DNA and circulating tumor DNA (ctDNA) in metastatic renal cell carcinoma (mRCC): potential clinical implications. Oncotarget 2017;8:33614-20. [Crossref] [PubMed]
36. Lin YL, Wang YP, Li HZ, et al. Aberrant Promoter Methylation of PCDH17 (Protocadherin 17) in Serum and its Clinical Significance in Renal Cell Carcinoma. Med Sci Monit 2017;23:3318-23. [Crossref] [PubMed]
37. Yamamoto Y, Uemura M, Fujita M, et al. Clinical significance of the mutational landscape and fragmentation of circulating tumor DNA in renal cell carcinoma. Cancer Sci 2019;110:617-28. [Crossref] [PubMed]