Abstract
Background
Clear cell renal cell carcinoma (ccRCC) exhibits heterogeneous postoperative outcomes. This study aimed to evaluate systemic immunoinflammatory biomarkers and pathological features and develop an integrated prognostic model for overall survival (OS) in patients with ccRCC.
Methods
This multicenter retrospective study initially screened 1,662 patients who underwent surgery between January 2010 and December 2024. After excluding 27 patients with substantial missing data, 1,635 were included in baseline analyses, of whom 1,429 with complete data were included in prognostic modeling. Candidate predictors identified by univariable Cox regression were selected using elastic net regression and subsequently evaluated by multivariable Cox regression. Model performance was assessed using the C-index, time-dependent area under the curve (AUC), calibration curves, and decision curve analysis.
Results
Sixteen variables with p < 0.20 in univariable analyses were considered candidate predictors, and 12 were retained by elastic net regression. ISUP grade, tumor necrosis, platelet count, recurrence/metastasis status, and systemic inflammation response index (SIRI) were independently associated with OS. The nomogram showed good discrimination, with C-indices of 0.849 and 0.869 in the training and internal testing cohorts, respectively. The 1-, 3-, and 5-year AUCs ranged from 0.829 to 0.897. Calibration was satisfactory, and risk stratification identified groups with significantly different survival outcomes (p < 0.001).
Conclusions
An internally evaluated nomogram integrating systemic inflammatory indices and pathological features demonstrated good prognostic performance for OS in patients with ccRCC.
Keywords: Multicenter retrospective study, pathology, postoperative prognosis, systemic immunoinflammatory biomarkers, systemic inflammatory indices
Introduction
Clear cell renal cell carcinoma (ccRCC) is the most common pathological subtype of renal carcinoma, accounting for approximately 70–80% of all renal cell carcinoma cases [1,2]. Its global incidence is on the rise, with significant regional, gender, and ethnic differences—the incidence rate in males is about 1.5–2 times that in females [3]. Although early-stage renal cell carcinoma can achieve a favorable prognosis through surgical resection, approximately 20% of patients present with metastasis at initial diagnosis [4], and about 25% of localized patients experience recurrence or metastasis after surgery [5,6]. For patients with advanced or metastatic ccRCC, the prognosis remains poor, with a significantly reduced 5-year survival rate. Therefore, in-depth exploration of key factors closely related to the occurrence, development, and prognosis of ccRCC is of great significance for patient risk stratification, formulation of individualized treatment strategies, and improvement of overall survival (OS).
In recent years, immune and inflammatory responses in the tumor microenvironment have become a research hotspot in ccRCC [7]. Notably, systemic inflammatory and immune indices derived from routine peripheral blood examinations have demonstrated important prognostic value due to their non-invasiveness, low cost, and high reproducibility [8]. By integrating information from different immune cell subsets, these composite indices provide a more comprehensive reflection of the host’s pro- versus anti-tumor immune balance than any single biomarker alone [9]. The neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) are among the most extensively investigated systemic inflammatory indices [8]. NLR serves as a marker of nonspecific inflammation and immunosuppressive status, while PLR reflects the interplay between prothrombotic tendency and immune response [7]. To provide a more integrated assessment of host inflammation and immunity, composite indices such as the systemic immune-inflammation index (SII), calculated as platelet count × neutrophil count/lymphocyte count, and the systemic inflammatory response index (SIRI), derived from neutrophil count × monocyte count/lymphocyte count, have been developed [10,11]. By incorporating multiple cellular components of the peripheral blood—including neutrophils, platelets, lymphocytes, and monocytes—these indices offer a more comprehensive representation of the systemic immune-inflammatory milieu than single-parameter measures. The lymphocyte–monocyte ratio (LMR) may be associated with the balance between lymphocyte-mediated anti-tumor immunity and monocyte-mediated pro-tumor activity [12,13]. In ccRCC, studies have confirmed that preoperative high SIRI is an independent risk factor for OS in metastatic patients undergoing cytoreductive nephrectomy, and its predictive value may be superior to NLR and PLR [7]. Furthermore, these inflammatory indices are correlated with malignant tumor characteristics—for example, high NLR, PLR, and SIRI values are significantly associated with poorer ISUP grade and T stage [7]. Therefore, constructing prognostic models using these easily accessible clinical inflammatory and immune factors has become a powerful tool for ccRCC risk stratification and auxiliary individualized treatment decision-making [14].
In clinical practice, pathological staging is the cornerstone for evaluating ccRCC prognosis. In clinical practice, anatomical TNM stage and histological nuclear grade are among the major clinicopathological determinants of prognosis in ccRCC. The TNM staging system characterizes the anatomical extent of disease, whereas nuclear grading reflects tumor differentiation and biological aggressiveness [15]. Fuhrman grade reflects the degree of malignancy by assessing the nuclear morphology of tumor cells—the higher the grade (G3/G4), the stronger the tumor invasiveness and the poorer the prognosis [16]. Additionally, the International Society of Urological Pathology (ISUP) refined and unified the Fuhrman grade to develop the ISUP grading system, which further improves the consistency of interpretation and the accuracy of prognostic prediction [17]. High grade (ISUP grade 3–4) has become a recognized independent risk factor [18,19].
Beyond established macroscopic staging and nuclear grading, a detailed histopathological assessment provides indispensable insights into the tumor’s biological aggressiveness in ccRCC [17,19]. Substantial evidence confirms that specific microscopic features within ccRCC specimens are robustly associated with more aggressive disease phenotypes. These features encompass tumor necrosis (particularly coagulative necrosis), microscopic vascular/lymphatic invasion, sarcomatoid differentiation, and the presence of undifferentiated giant cells, alongside secondary changes such as intratumoral hemorrhage and cystic degeneration [20]. Among these, tumor necrosis and vascular invasion constitute cornerstone elements in several validated prognostic algorithms, including the University of California, Los Angeles Integrated Staging System (UISS) and the Stage, Size, Grade, and Necrosis (SSIGN) score, serving as independent predictors of high recurrence and metastatic risk [21,22]. Sarcomatoid differentiation represents a particularly ominous finding, characterized by rapid disease progression and profoundly poor survival outcomes. This transformation is underpinned by distinct genomic alterations, including frequent mutations in TP53and BAP1, and specific chromosomal imbalances such as loss of chromosome 9p, which are strongly correlated with high-grade tumors and the presence of necrosis [23]. The integration of these histopathological features with emerging biomarkers, such as systemic immune-inflammation indices, and macroscopic staging is paramount. Contemporary research demonstrates that multimodal prognostic models, which synthesize data from histopathology, radiology, and clinical features, significantly outperform models relying on a single data type [24,25]. Therefore, a systematic evaluation of key pathological features and their incorporation into composite prognostic frameworks will enable more precise identification of high-risk patients, thereby guiding decisions regarding aggressive adjuvant therapy and intensifying follow-up surveillance strategies.
In summary, combining the epidemiological characteristics of ccRCC, comprehensively utilizing clinical inflammatory and immune indices (such as SII, SIRI, NLR, PLR, LMR), precise pathological staging (TNM and ISUP grading), and detailed microscopic pathological features to construct a multi-dimensional comprehensive prognostic evaluation system is an important direction in current clinical and basic research on ccRCC. This study aims to further explore the correlation between the aforementioned factors and the prognosis of ccRCC patients, in order to provide more robust evidence for optimizing patient management and improving clinical outcomes.
Materials and methods
Study population
This retrospective cohort study included patients who underwent surgical resection for renal tumors at The First People’s Hospital of Yunnan Province and Xiangya Hospital of Central South University. The data between January 2010 and December 2024 were retrospectively analysed on January 2025. A total of 1,662 patients were enrolled in the study.
Inclusion criteria were as follows: (1) Histopathological confirmation of ccRCC following surgery; (2) Availability of complete clinical records and follow-up data.
Exclusion criteria included: (1) History of other malignancies; (2) Receipt of neoadjuvant therapy prior to surgery; (3) Patients who underwent emergency surgery.
A total of 1,662 patients with pathologically confirmed ccRCC were initially screened. Twenty-seven patients were excluded because of substantial missing baseline clinical, pathological, or follow-up information. Consequently, 1,635 patients were included in the baseline descriptive analyses.
For prognostic modeling, a complete-case analysis strategy was adopted. Patients with missing values in variables required for Cox regression and nomogram development were excluded from model construction. As a result, 1,429 patients remained available for survival modeling.
These 1,429 patients were randomly assigned to a training cohort (n = 1,000) and an internal testing cohort (n = 429) using a random number table method.
Data collection
A systematic and comprehensive chart review was conducted to retrospectively abstract data from the institutional electronic medical record (EMR) systems. The extracted dataset encompassed patient demographics, including age, sex, height, and weight. Preoperative laboratory parameters were collected, covering hematological indices—such as white blood cell count, red blood cell count, hemoglobin level, platelet count, neutrophil count, lymphocyte count, and monocyte count—from which inflammatory indices including the NLR, PLR, SII, and SIRI were derived. The biochemical profile included total protein, albumin, globulin, total bilirubin, direct bilirubin, blood urea nitrogen (BUN), creatinine, uric acid, and fasting blood glucose levels. Serum electrolytes (potassium, sodium, chloride, calcium, phosphorus, magnesium) and coagulation parameters—prothrombin time (PT), international normalized ratio (INR), and thrombin time (TT)—were also recorded.
Surgical and pathological data involved variables such as laterality (left or right kidney), type of nephrectomy (radical or partial), and surgical approach (laparoscopic or open). Pathological features included tumor grading based on the World Health Organization/International Society of Urological Pathology (WHO/ISUP) classification and Fuhrman grade, along with assessments for tumor necrosis, giant cells, sarcomatoid differentiation, cystic change, hemorrhage, and vascular invasion.
Patients were followed through scheduled outpatient visits, with primary outcomes being OS, recurrence-free survival (RFS), tumor recurrence, and metastasis. The follow-up endpoint was defined as either patient death or the last documented contact. The median follow-up duration was 3.4 years (interquartile range: 2.1–6.2 years).
To ensure consistency in pathological assessment across participating institutions, all available histopathological slides were independently re-reviewed by four experienced genitourinary pathologists, including two senior pathologists from The First People’s Hospital of Yunnan Province and two senior pathologists from Xiangya Hospital, Central South University.
The pathologists were blinded to the original pathological reports and patient outcomes during reassessment. Fuhrman grade, WHO/ISUP grade, tumor necrosis, sarcomatoid differentiation, giant cells, cystic change, hemorrhage, and lymphovascular invasion were re-evaluated according to contemporary diagnostic criteria.
In cases of disagreement, a consensus review was performed. Cases for which consensus could not be achieved were excluded from further analysis.
This centralized pathological review enabled specimens obtained before widespread implementation of the WHO/ISUP grading system (including cases diagnosed during 2010–2011 to be reclassified according to the revised WHO/ISUP criteria, thereby improving pathological consistency throughout the study period.
All patients had undergone routine postoperative pathological evaluation, including immunohistochemical staining when clinically indicated. Because of the retrospective nature of the study and limited availability of archived specimens, additional immunohistochemical or molecular testing was not repeated during the re-review process.
Statistical analysis
All statistical analyses were performed using R software (version 4.3.2) and SPSS Statistics (version 27.0). Key R packages utilized included ‘survival’ for survival analysis, ‘rms’ for regression modeling, ‘pROC’ for receiver operating characteristic curve analysis, ‘rmda’ for decision curve analysis, and ‘ggplot2’, ‘dplyr’, and ‘tidyverse’ for data visualization and manipulation. Descriptive statistics were computed as follows: continuous variables following a normal distribution were presented as mean ± standard deviation, while non-normally distributed variables were expressed as median with interquartile range; categorical variables were summarized as frequencies and percentages. For prognostic factor identification, univariable Cox proportional hazards regression analyses were initially performed for all candidate variables to evaluate their associations with OS. Variables with a p value <0.20 in univariable analyses were considered potential candidate predictors for subsequent model development. These variables were further selected using elastic net regression (α = 0.8). Variables retained by elastic net regression were subsequently entered into multivariable Cox proportional hazards regression analysis to identify independent prognostic factors. Statistical significance in multivariable analyses was defined as p < 0.05.
For survival analyses, continuous laboratory and biomarker variables were dichotomized according to standard clinical reference ranges or cohort-derived cut-off values, such as the maximum Youden index for immunoinflammatory biomarkers (e.g. NLR, PLR, LMR, SII, SIRI). Kaplan–Meier survival curves were generated for these dichotomized variables, and differences between groups were assessed using the log-rank test.
Because no universally accepted cut-off values have been established for NLR, PLR, LMR, SII, and SIRI in patients with ccRCC, optimal thresholds were determined within the present study cohort using receiver operating characteristic (ROC) curve analysis. The cut-off value for each biomarker was selected according to the maximum Youden index.
The resulting optimal cut-off values were as follows: NLR = 3.05, PLR = 138.86, LMR = 8.17, SII = 99.60, and SIRI = 1.22. Based on the multivariable analysis results, a nomogram was developed to estimate 1-, 3-, and 5-year OS probabilities
Model validation and performance assessment
The predictive performance of the nomogram was rigorously assessed in terms of discrimination, calibration, and clinical utility. Discrimination, which reflects the model’s ability to distinguish between outcome groups, was quantified using Harrell’s concordance index (C-index) and time-dependent receiver operating characteristic (ROC) curves, with the area under the curve (AUC) calculated at 1, 3, and 5 years. To obtain robust internal validation, bootstrapping with 1,000 resamples was applied, and performance metrics were reported for both the training cohort and the internal testing cohort generated through random dataset partitioning.
The entire dataset consisted of patients from two participating institutions. Patients were randomly assigned to the training cohort (70%) and internal testing cohort (30%) using a random number table method. Institutional origin was not used as the criterion for cohort assignment. Therefore, the testing cohort represents an internal validation cohort rather than a truly independent external validation cohort. These curves were visually and analytically compared to the ideal reference line (45° line). Furthermore, Platt scaling was employed to improve the calibration of predicted survival probabilities, thereby enhancing the reliability of the probability estimates.
To evaluate the potential clinical value of the nomogram, decision curve analysis (DCA) was performed. This analysis estimates the net clinical benefit across a range of threshold probabilities, weighing the benefits of true positives against the harms of false positives, and thus facilitating assessment of the model’s utility in clinical decision-making.
All statistical tests were two-sided, and a p value less than 0.05 was considered statistically significant.
Results
Baseline characteristics of the study population
A total of 1,635 patients were included in the baseline analyses with baseline characteristics summarized as follows.
The median age of the patients was 55 years (interquartile range [IQR]: 47–64 years), including 1,126 (68.9%) were male and 508 (31.1%) were female. Sex information was unavailable for one patient. Regarding anthropometric measurements: the median height was 167 cm (IQR: 162–171 cm) for males and 155 cm (IQR: 152–159 cm) for females; the median weight was 68 kg (IQR: 62–75 kg) for males and 56 kg (IQR: 51–62 kg) for females.
Regarding tumor and surgical characteristics, tumor laterality was relatively balanced, with 49.4% of tumors located in the left kidney (n = 804). The median follow-up duration was 3.4 years (IQR: 2.1–6.2 years). During the follow-up period, the recurrence rate was 2.3% (n = 37), metastasis rate was 6.9% (n = 113), and mortality rate was 9.5% (n = 155) (Table 1).
Table 1.
Baseline clinical features of the study population.
| Demographics | Overall | Alive | Dead | p |
|---|---|---|---|---|
| Age, years | 55 (47,64) | 54 (47,63) | 60 (52,66) | <0.001 |
| Sex, male (%) | 68.91% (1126/1634) | 68.42% (1012/1479) | 73.55% (114/155) | 0.190 |
| Height, cm | 0.270 | |||
| Male | 167 (162,171) | 168 (163,171) | 165 (161,170) | |
| Female | 155 (152,159) | 155 (152,160) | 153 (152,157) | |
| Weight, kg | <0.001 | |||
| Male | 68 (62,75) | 69 (62,76) | 63 (57,70) | |
| Female | 56 (51,62) | 57 (52,62) | 53 (46,61) | |
| Side, left (%) | 49.39% (804/1628) | 49.69% (732/1473) | 46.45% (72/155) | 0.442 |
| Surgical procedure, radical (%) | 58.91% (959/1628) | 55.74% (821/1473) | 89.03% (138/155) | <0.001 |
| Surgical approach, laparoscopic (%) | 91.6% (1492) | 93.28% (1375/1474) | 75.48% (117/155) | <0.001 |
| Follow-up duration, years | 3.4 (2.1,6.2) | |||
| Recurrence (%) | 2.3% (37/1635) | 0.9% (14/1480) | 14.8% (23/155) | <0.001 |
| Metastasis (%) | 6.9% (113/1635) | 3.3% (49/1480) | 41.3% (64/155) | <0.001 |
| Death (%) | 9.5% (155/1635) |
Regarding pathological characteristics, Fuhrman grade 2 tumors were most frequently observed (60.9%, n = 974), followed by grade 1 (25.7%, n = 411), grade 3 (10.9%, n = 174), and grade 4 (2.5%, n = 40). A similar distribution was noted for the ISUP (International Society of Urological Pathology) grading system, with grade 2 being predominant (54.0%, n = 855), followed by grades 1 (23.5%, n = 372), 3 (18.3%, n = 290), and 4 (4.2%, n = 66). Among the evaluated pathological alterations, hemorrhage was commonly identified (36.3%, n = 594), while cystic changes were observed in 20.2% of cases (n = 331). Necrosis was present in 16.2% of cases (n = 266). Less common features included vascular invasion (2.4%, n = 40), sarcomatoid differentiation (1.5%, n = 25), and giant cells (1.2%, n = 20) (Table 2).
Table 2.
Tumor pathological grading and pathological components.
| Pathological grading | Overall | Alive | Dead | p |
|---|---|---|---|---|
| Fuhrman grade (%) | 0.036 | |||
| 1 | 25.7% (411/1599) | 27.4% (395/1444) | 10.3% (16/155) | |
| 2 | 60.9% (974/1599) | 61.2% (884/1444) | 58.1% (90/155) | |
| 3 | 10.9% (174/1599) | 9.6% (138/1444) | 23.2% (36/155) | |
| 4 | 2.5% (40/1599) | 1.9% (27/1444) | 8.4% (13/155) | |
| WHO/ISUP grade (%) | <0.001 | |||
| 1 | 23.5% (372/1583) | 25.5% (365/1429) | 4.5% (7/154) | |
| 2 | 54.0% (855/1583) | 55.0% (786/1429) | 44.8% (69/154) | |
| 3 | 18.3% (290/1583) | 16.2% (231/1429) | 38.3% (59/154) | |
| 4 | 4.2% (66/1583) | 3.3% (47/1429) | 12.3% (19/154) | |
| Pathological components (%) | ||||
| Necrosis | 16.2% (266/1635) | 13.4% (199/1480) | 43.2% (67/155) | <0.001 |
| Giant cells | 1.2% (20/1635) | 0.7% (11/1480) | 5.8% (9/155) | <0.001 |
| Sarcomatoid differentiation | 1.5% (25/1635) | 1.1% (17/1480) | 5.2% (8/155) | <0.001 |
| Cystic change | 20.2% (331/1635) | 20.9% (309/1480) | 14.2% (22/155) | 0.049 |
| Hemorrhage | 36.3% (594/1635) | 36.4% (538/1480) | 36.1% (56/155) | 0.956 |
| Lymphovascular invasion | 2.4% (40/1635) | 2.2% (33/1480) | 4.5% (7/155) | 0.080 |
Hematological analysis revealed a median white blood cell (WBC) count of 6.3 × 109/L (IQR: 5.2–7.9 × 109/L), with elevated and reduced levels observed in 13.4% (n = 216) and 1.8% (n = 29) of patients, respectively. The median red blood cell (RBC) count was 4.41 × 1012/L (IQR: 4.01–4.79 × 1012/L), showing elevation in 3.5% (n = 57) and reduction in 24.3% (n = 391) of cases. Hemoglobin (Hb) levels averaged 134 g/L (IQR: 121–145 g/L), with a notable reduction observed in 31.5% (n = 506) of patients (Table 3).
Table 3.
Baseline hematological parameters of the study population.
| Hematological parameters | Overall | Alive | Dead | p |
|---|---|---|---|---|
| White blood cell count (×10⁹/L) | 6.3 (5.2,7.9) | 6.2 (5.1,7.7) | 6.8 (5.5,8.4) | 0.011 |
| Elevated (%) | 13.4% (216) | |||
| Reduced (%) | 1.8% (29) | |||
| Red blood cell count (×10¹²/L) | 4.41 (4.01,4.79) | 4.44 (4.04,4.81) | 4.10 (3.68,4.57) | <0.001 |
| Elevated (%) | 3.5% (57) | |||
| Reduced (%) | 24.3% (391) | |||
| Hemoglobin (g/L) | 134 (121,145) | 135 (123,146) | 125 (107,139) | <0.001 |
| Elevated (%) | 0.5% (8) | |||
| Reduced (%) | 31.5% (506) | |||
| Platelet count (×10⁹/L) | 197 (160,245) | 196 (160,242) | 216 (164,266) | 0.007 |
| Elevated (%) | 2.1% (33) | |||
| Reduced (%) | 10.1% (162) | |||
| Hematocrit (L/L) | 40.1 (35.7,43.3) | 40.4 (36.3,43.4) | 36.7 (28.9,41.6) | <0.001 |
| Elevated (%) | 1.7% (27) | |||
| Reduced (%) | 4.7% (75) | |||
| Neutrophil count (×10⁹/L) | 3.9(3.1,5.3) | 3.8(3.0,5.2) | 4.5(3.4,6.0) | <0.001 |
| Elevated (%) | 12.1% (194) | |||
| Reduced (%) | 3.2% (52) | |||
| Neutrophil percentage (%) | 64.3% (56.5%,72.0%) | 63.5% (56.1%,71.4%) | 67.7% (59.5%,74.3%) | <0.001 |
| Lymphocyte count (×10⁹/L) | 1.5(1.1,1.9) | 1.5(1.2,1.9) | 1.5(1.1,1.8) | 0.118 |
| Elevated (%) | 7.7% (124) | |||
| Reduced (%) | 0 | |||
| Lymphocyte percentage (%) | 25.0% (17.8%,31.8%) | 25.7% (18.6%,32.2%) | 22.1% (15.9%,28.4%) | <0.001 |
| Eosinophil count (×10⁹/L) | 0.1(0.1,0.2) | 0.1(0.1,0.2) | 0.1(0.1,0.2) | 0.777 |
| Elevated (%) | 2.3% (37) | |||
| Reduced (%) | 19.7% (315) | |||
| Eosinophil percentage (%) | 1.9% (1.1%,3.2%) | 1.9% (1.2%,3.2%) | 1.6% (1.0%,2.9%) | 0.046 |
| Basophil count (×10⁹/L) | 0.0(0.0,0.2) | 0.0(0.0,0.3) | 0.0(0.0,0.0) | 0.064 |
| Elevated (%) | 0.9% (14) | |||
| Reduced (%) | 0 | |||
| Basophil percentage (%) | 0% (0%,0.3%) | 0% (0%,0.5%) | 0% (0%,0%) | 0.023 |
| Monocyte count (×10⁹/L) | 0.5 (0.4,0.6) | 0.5 (0.4,0.6) | 0.5 (0.4,0.6) | 0.021 |
| Elevated (%) | 6.6% (106) | |||
| Reduced (%) | 2.4% (38) | |||
| Monocyte percentage (%) | 7.5% (6.2%,8.9%) | 7.5% (6.2%,8.9%) | 7.6% (6.0%,9.3%) | 0.692 |
| Plateletcrit (%) | 0.20 (0.16,0.23) | 0.20 (0.16,0.23) | 0.20 (0.15,0.22) | 0.896 |
Inflammatory markers demonstrated a median NLR of 2.55 (IQR: 1.79–4.06) and a median PLR of 133.08 (IQR: 97.78–182.32) (Table 4).
Table 4.
Systemic immunoinflammatory biomarkers of the study population.
| Systemic immunoinflammatory biomarkers | Overall | Alive | Dead | p |
|---|---|---|---|---|
| Neutrophil-to-lymphocyte ratio (NLR) | 2.55 (1.79,4.06) | 2.47 (1.73,3.81) | 3.06 (2.09,4.63) | <0.001 |
| Platelet-to-lymphocyte ratio (PLR) | 133.08 (97.78,182.32) | 131.54 (97.22,176.57) | 152.86 (106.06,220.45) | 0.002 |
| Lymphocyte-to-monocyte ratio (LMR) | 3.33 (2.33,4.50) | 3.40 (2.40,4.50) | 2.83 (2.00,4.00) | <0.001 |
| Systemic immune-inflammation index (SII) | 76.46 (46.97,115.41) | 77.70 (48.56,116.95) | 73.60 (44.41,104.00) | 0.125 |
| Systemic inflammation response index (SIRI) | 1.15 (0.75,2.16) | 1.09 (0.73,1.97) | 1.57 (0.98,2.54) | <0.001 |
Biochemical evaluation indicated a median total protein level of 68.3 g/L (IQR: 64.1–73.3 g/L), which was reduced in 24.1% (n = 382) of subjects. The median albumin level was 41.3 g/L (IQR: 38.0–44.4 g/L), with a reduction rate of 29.5% (n = 468) (Table 5). Electrolyte and metabolic parameters were generally within normal reference ranges. However, hyperuricemia was observed in 37.9% (n = 602) of patients, and elevated blood glucose levels were present in 20.8% (n = 331). (Table 6) Coagulation function, as assessed by prothrombin time (PT), showed a median value of 12.6 s (IQR: 12.0–13.2 s), with a low overall rate of abnormality (Table 6).
Table 5.
Baseline blood biochemistry parameters of the study population.
| Blood biochemistry | Overall | Alive | Dead | p |
|---|---|---|---|---|
| Total protein (g/L) | 68.3 (64.1,73.3) | 68.4 (64.3,73.3) | 67.8 (63.0,73.8) | 0.285 |
| Elevated (%) | 2.4% (38) | |||
| Reduced (%) | 24.1% (382) | |||
| Albumin (g/L) | 41.3 (38.0,44.4) | 41.5 (38.5,44.5) | 38.1 (34.3,42.6) | <0.001 |
| Elevated (%) | 0.5% (8) | |||
| Reduced (%) | 29.5% (468) | |||
| Albumin/total protein ratio | 60.7% (57.4%,63.5%) | 60.9% (57.8%,63.6%) | 57.4% (52.4%,61.3%) | <0.001 |
| Globulin (g/L) | 26.6 (23.9,29.8) | 26.5 (23.9,29.6) | 28.2 (24.5,32.6) | 0.001 |
| Elevated (%) | 13.0% (216) | |||
| Reduced (%) | 7.7% (128) | |||
| Globulin/total protein ratio | 39.3% (36.5%,42.6%) | 39.1% (36.4%,42.2%) | 42.7% (38.7%,47.7%) | <0.001 |
| Total bilirubin (µmol/L) | 10.0 (7.4,13.6) | 10.2 (7.5,13.6) | 8.5 (6.4,12.5) | 0.002 |
| Elevated (%) | 12.2% (195) | |||
| Reduced (%) | 0.1% (2) | |||
| Direct bilirubin (µmol/L) | 4.2 (3.1,5.8) | 4.3 (3.2,5.9) | 3.4 (2.6,4.7) | <0.001 |
| Elevated (%) | 13.4% (213) | |||
| Reduced (%) | 0 | |||
| Direct bilirubin/total bilirubin ratio | 43.1% (37.8%,48.8%) | 43.3% (38.2%,48.8%) | 40.1% (34.1%,48.4%) | 0.009 |
| Indirect bilirubin (µmol/L) | 5.5 (3.8,7.9) | 5.6 (3.9,7.9) | 4.8 (3.1,7.2) | 0.003 |
| Indirect bilirubin/total bilirubin Ratio | 56.9% (51.2%,62.2%) | 56.7% (51.2%,61.8%) | 59.1% (51.6%,65.9%) | 0.009 |
| Urea (mmol/L) | 5.1 (4.2,6.1) | 5.1 (4.2,6.1) | 5.6 (4.4.6.8) | 0.002 |
| Elevated (%) | 12.3% (195) | |||
| Reduced (%) | 3.7% (59) | |||
| Creatinine (µmol/L) | 86 (74,101) | 85 (73,99) | 88 (76,110) | 0.030 |
| Elevated (%) | 6.0% (95) | |||
| Reduced (%) | 2.2% (35) | |||
| Uric acid (µmol/L) | 339 (286,402) | 339 (286,402) | 342 (284,394) | 0.937 |
| Elevated (%) | 37.9% (602) | |||
| Reduced (%) | 1.3% (21) | |||
| Glucose (mmol/L) | 5.3 (4.8,6.0) | 5.3 (4.8,6.0) | 5.3 (4.7,6.0) | 0.982 |
| Elevated (%) | 20.8% (331) | |||
| Reduced (%) | 0.4% (7) |
Table 6.
Baseline serum electrolyte levels and coagulation profile of the study population.
| Electrolytes and coagulation profile | Overall | Alive | Dead | P |
|---|---|---|---|---|
| Potassium (mmol/L) | 3.96 (3.74,4.17) | 3.95 (3.73,4.16) | 4.05 (3.82,4.36) | <0.001 |
| Elevated (%) | 9.6% (153) | |||
| Reduced (%) | 0.4% (7) | |||
| Sodium (mmol/L) | 141.7 (139.9,143.2) | 141.8 (140.0,143.3) | 141.0 (139.3,142.6) | <0.001 |
| Elevated (%) | 0.3% (5) | |||
| Reduced (%) | 3.6% (57) | |||
| Chloride (mmol/L) | 103.7 (101.8,105.5) | 103.8 (102.0,105.6) | 102.4 (99.8,104.8) | <0.001 |
| Elevated (%) | 2.4% (38) | |||
| Reduced (%) | 5.4% (86) | |||
| Calcium (mmol/L) | 2.28 (2.20,2.37) | 2.29 (2.20,2.37) | 2.25 (2.16,2.36) | 0.006 |
| Elevated (%) | 1.7% (27) | |||
| Reduced (%) | 2.0% (32) | |||
| Phosphate (mmol/L) | 1.14 (1.00,1.27) | 1.14 (1.00,1.27) | 1.13 (1.01,1.24) | 0.844 |
| Elevated (%) | 1.1% (17) | |||
| Reduced (%) | 2.9% (46) | |||
| Magnesium (mmol/L) | 0.87 (0.82,0.92) | 0.87 (0.82,0.92) | 0.89 (0.84,0.94) | 0.024 |
| Elevated (%) | 0.8% (13) | |||
| Reduced (%) | 1.3% (21) | |||
| Prothrombin time (s) | 12.6 (12.0,13.2) | 12.6 (12.0,13.2) | 12.7 (12.1,13.3) | 0.238 |
| Elevated (%) | 0.9% (14) | |||
| Reduced (%) | 0.2% (3) | |||
| International normalized ratio (INR) | 0.98 (0.94,1.04) | 0.98 (0.94,1.04) | 0.98 (0.92,1.04) | 0.440 |
| Thrombin time (s) | 17.1 (16.2,18.3) | 17.1 (16.1,18.3) | 17.5 (16.6,19.0) | <0.001 |
| Elevated (%) | 2.7% (42) | |||
| Reduced (%) | 0.8% (13) |
Kaplan–Meier survival analysis results
Kaplan–Meier analysis showed that both surgical procedure and surgical approach were significantly associated with OS (both p < 0.001). Patients undergoing radical nephrectomy showed poorer survival than those undergoing partial nephrectomy, while patients treated laparoscopically showed better survival than those undergoing open surgery.
Analysis of pathological characteristics indicated that both Fuhrman grade (p < 0.001) and ISUP grade (p < 0.001) were significant prognostic factors. Higher grades were associated with lower survival rates. Additionally, the presence of pathological features including necrosis (p < 0.001), giant cells (p < 0.001), sarcomatoid differentiation (p < 0.001), and vascular tumor thrombus (p < 0.001) was significantly correlated with poor prognosis.
Evaluation of laboratory parameters showed that systemic inflammation markers, including the NLR (p < 0.001), PLR (p < 0.001), and SIRI (p < 0.001), were significantly associated with OS. Abnormalities in blood cell counts, including red blood cell count (p < 0.001), hemoglobin level (p < 0.001), and platelet count (p < 0.001), were also indicative of an unfavorable prognosis.
Among biochemical indicators, albumin levels (p < 0.001) and creatinine levels (p = 0.004) were significantly associated with OS, whereas direct bilirubin levels were not significantly associated with OS (p = 0.834).
Recurrence/metastasis status was significantly correlated with patient survival (p < 0.001), with survival curves declining markedly in patients who developed recurrence or distant metastasis.
In contrast, several variables were not significantly associated with OS in this cohort, including sex (p = 0.103), cystic change (p = 0.451), and pathological hemorrhage (p = 0.790). Detailed results are summarized in Table 7 and Figures 1 and 2.
Table 7.
Kaplan–Meier overall survival analysis according to clinical, pathological, and laboratory characteristics.
| Variable | Compared group | Patients, n | Deaths, n | Median OS, years | p value |
|---|---|---|---|---|---|
| Sex | Female | 508 | 41 | NR | 0.103 |
| Male | 1126 | 114 | NR | ||
| Tumor side | Right | 824 | 83 | NR | 0.632 |
| Left | 804 | 72 | NR | ||
| Surgical procedure | Partial nephrectomy | 669 | 17 | NR | <0.001 |
| Radical nephrectomy | 959 | 138 | NR | ||
| Surgical approach | Open | 137 | 38 | 9.66 | <0.001 |
| Laparoscopic | 1492 | 117 | NR | ||
| Recurrence/metastasis | No | 1495 | 74 | NR | <0.001 |
| Yes | 140 | 81 | 5.16 | ||
| Fuhrman grade | Grade 1 | 411 | 16 | NR | <0.001 |
| Grade 2 | 974 | 90 | NR | ||
| Grade 3 | 174 | 36 | NR | ||
| Grade 4 | 40 | 13 | 3.93 | ||
| WHO/ISUP grade | Grade 1 | 372 | 7 | NR | <0.001 |
| Grade 2 | 855 | 69 | NR | ||
| Grade 3 | 290 | 59 | NR | ||
| Grade 4 | 66 | 19 | 3.93 | ||
| Necrosis | Absent | 1369 | 88 | NR | <0.001 |
| Present | 266 | 67 | 9.66 | ||
| Giant cells | Absent | 1615 | 146 | NR | <0.001 |
| Present | 20 | 9 | NR | ||
| Sarcomatoid differentiation | Absent | 1610 | 147 | NR | <0.001 |
| Present | 25 | 8 | 3.23 | ||
| Cystic change | Absent | 1304 | 133 | NR | 0.451 |
| Present | 331 | 22 | NR | ||
| Hemorrhage | Absent | 1041 | 99 | NR | 0.790 |
| Present | 594 | 56 | NR | ||
| Lymphovascular invasion | Absent | 1595 | 148 | NR | <0.001 |
| Present | 40 | 7 | NR | ||
| White blood cell count | Normal | 1354 | 122 | NR | 0.823 |
| Outside reference range | 225 | 23 | NR | ||
| Red blood cell count | Normal | 1141 | 78 | NR | <0.001 |
| Outside reference range | 438 | 67 | NR | ||
| Hemoglobin | Normal | 1079 | 65 | NR | <0.001 |
| Outside reference range | 500 | 80 | NR | ||
| Platelet count | Normal | 1385 | 112 | NR | <0.001 |
| Outside reference range | 194 | 33 | NR | ||
| Neutrophil count | Normal | 1354 | 127 | NR | 0.135 |
| Outside reference range | 225 | 18 | NR | ||
| Lymphocyte count | Normal | 1469 | 127 | NR | 0.034 |
| Outside reference range | 110 | 18 | NR | ||
| Eosinophil count | Normal | 1244 | 114 | NR | 0.473 |
| Outside reference range | 332 | 29 | NR | ||
| Basophil count | Normal | 1563 | 141 | NR | 0.565 |
| Outside reference range | 13 | 2 | NR | ||
| Monocyte count | Normal | 1441 | 122 | NR | 0.042 |
| Outside reference range | 135 | 21 | NR | ||
| Plateletcrit | Normal | 670 | 68 | NR | 0.504 |
| Outside reference range | 889 | 72 | NR | ||
| Total protein | Normal | 1159 | 95 | NR | 0.202 |
| Outside reference range | 406 | 48 | NR | ||
| Albumin | Normal | 1104 | 67 | NR | <0.001 |
| Outside reference range | 462 | 76 | NR | ||
| Globulin | Normal | 1300 | 113 | NR | 0.022 |
| Outside reference range | 334 | 42 | NR | ||
| Total bilirubin | Normal | 1372 | 126 | NR | 0.847 |
| Outside reference range | 193 | 17 | NR | ||
| Direct bilirubin | Normal | 1357 | 129 | NR | 0.834 |
| Outside reference range | 208 | 14 | NR | ||
| Creatinine | Normal | 1438 | 119 | NR | 0.004 |
| Outside reference range | 130 | 24 | NR | ||
| Uric acid | Normal | 952 | 83 | NR | 0.102 |
| Outside reference range | 615 | 60 | NR | ||
| Glucose | Normal | 1097 | 99 | NR | 0.975 |
| Outside reference range | 333 | 29 | NR | ||
| Potassium | Normal | 1413 | 127 | NR | 0.839 |
| Outside reference range | 158 | 15 | NR | ||
| Sodium | Normal | 1511 | 131 | NR | 0.054 |
| Outside reference range | 59 | 11 | NR | ||
| Chloride | Normal | 1447 | 121 | NR | 0.009 |
| Outside reference range | 123 | 21 | NR | ||
| Calcium | Normal | 1516 | 131 | NR | 0.017 |
| Outside reference range | 54 | 11 | NR | ||
| Phosphorus | Normal | 1426 | 126 | NR | 0.534 |
| Outside reference range | 62 | 7 | NR | ||
| Magnesium | Normal | 1532 | 137 | NR | 0.250 |
| Outside reference range | 28 | 5 | NR | ||
| Prothrombin time | Normal | 1520 | 135 | NR | 0.045 |
| Outside reference range | 17 | 4 | NR | ||
| Thrombin time | Normal | 1476 | 130 | NR | 0.067 |
| Outside reference range | 55 | 9 | NR | ||
| NLR | <3.0513 | 985 | 71 | NR | <0.001 |
| ≥3.0513 | 593 | 74 | NR | ||
| PLR | <138.8562 | 862 | 59 | NR | <0.001 |
| ≥138.8562 | 716 | 86 | NR | ||
| LMR | <8.1667 | 1540 | 138 | NR | 0.632 |
| ≥8.1667 | 26 | 4 | NR | ||
| SII | <99.5991 | 1029 | 105 | NR | 0.049 |
| ≥99.5991 | 550 | 40 | NR | ||
| SIRI | <1.2165 | 852 | 51 | NR | <0.001 |
| ≥1.2165 | 726 | 94 | NR |
Figure 1.
Kaplan–Meier survival curves of pathological and laboratory variables in ccRCC. One-minus survival function curves depicting OS disparities among ccRCC patients, stratified by diverse clinicopathological and inflammation-associated characteristics. (A) Stratified by surgical procedure (with vs. without); (B) stratified by surgical approach (with vs. without); (C) stratified by Fuhrman grade; (D) stratified by WHO/ISUP grade; (E) stratified by necrosis (with vs. without); (F) stratified by giant cells (with vs. without); (G) stratified by sarcomatoid differentiation (with vs. without); (H) stratified by cystic change (with vs. without); (I) stratified by hemorrhage (with vs. without); (J) stratified by lymphovascular invasion (with vs. without); (K–O) stratified by additional laboratory characteristics relevant to ccRCC.
Figure 2.
Kaplan–Meier survival curves for stratifying patients with ccRCC based on clinical features, recurrence/metastasis status, and systemic inflammatory immune indicators and cross-validation graphs of nomograms.
Nomogram analysis results
After initial screening using univariable Cox regression, 16 variables with p < 0.20 were considered candidate predictors. Elastic net regression (α = 0.8) subsequently retained 12 variables for multivariable Cox regression analysis and nomogram development. ISUP grade was treated as a single categorical predictor, with individual grade levels represented by indicator terms in the Cox model. Multivariable Cox regression analysis revealed that ISUP grade (Grade 3: HR = 4.159, p = 0.002; Grade 4: HR = 6.011, p < 0.001), absence of necrosis (HR = 0.603, p = 0.011), normal platelet count (HR = 0.572, p = 0.011), no recurrence or metastasis (HR = 0.143, p < 0.001), and low SIRI (<1.22) (HR = 0.625, p = 0.013) were independently associated with OS in patients with ccRCC.
To ensure the model’s generalization ability, the total dataset (n = 1,429) was randomly divided into a training set (70%, n = 1,000) and a testing set (30%, n = 429) using a random splitting method. After reconstructing the predictive model on the training set, internal validation was performed using the internal testing cohort.
The 12 variables retained by elastic net regression were ISUP grade, necrosis status, sarcomatoid differentiation, lymphovascular invasion, red blood cell count, platelet count, monocyte count, albumin level, globulin level, recurrence/metastasis status, PLR, and SIRI.
Multivariable Cox regression analysis in the training cohort showed that, relative to the reference category of ISUP grade, Grade 3 (HR = 3.221, p = 0.022) and Grade 4 (HR = 4.432, p = 0.012) were associated with significantly increased mortality risk. Absence of recurrence or metastasis was associated with a lower mortality risk (HR = 0.157, p < 0.001). The overall model goodness-of-fit test was highly significant (likelihood ratio test χ2=177.8, p < 0.001) (Figure 3).
Figure 3.
Development and risk stratification of a prognostic nomogram for OS in ccRCC. A: A nomogram for predicting 1-, 3-, and 5-year OS probabilities in patients with ccRCC. B: Kaplan–Meier survival analysis for risk groups stratified by the nomogram-predicted scores in the training cohort. Patients were categorized into low-, intermediate-, and high-risk groups. The survival curves show a significant separation among the three groups (log-rank test, p < 0.0001).
The model showed good discriminative performance in both cohorts. The C-index was 0.849 (SE = 0.022) in the training cohort and 0.869 in the internal testing cohort.
Time-dependent ROC analysis further validated the model’s predictive performance. In the training set, the AUC values at 1, 3, and 5 years were 0.829, 0.862, and 0.871, respectively; the corresponding AUC values in the testing set were 0.893, 0.897, and 0.886. These results demonstrated stable predictive accuracy of the model at different time points (Figure 4).
Figure 4.
Receiver operating characteristic (ROC) curves for 1-, 3-, and 5-year survival prediction in training and test cohorts. A: 1-year survival prediction ROC curve (training cohort), B: 3-year survival prediction ROC curve (training cohort), C: 5-year survival prediction ROC curve (training cohort), D: 1-year survival prediction ROC curve (test cohort), E: 3-year survival prediction ROC curve (test cohort), F: 5-year survival prediction ROC curve (test cohort).
The proportional hazards assumption test showed that all variables satisfied the proportional hazards assumption (p > 0.05) except for albumin (p = 0.007), with a global test p value of 0.440, indicating that the model overall conformed to the Cox proportional hazards assumption.
Calibration curves of the model showed good consistency between predicted probabilities and actual observed probabilities. The calibration curves of both the training and testing sets were close to the diagonal line, indicating that the model could provide accurate predictions for patients at different risk levels.
To further evaluate model calibration, calibration curves were generated for the training cohort and internal testing cohort at 1, 3, and 5 years (Figure 5). Calibration curves assess the model’s calibration by comparing predicted survival probabilities with actually observed survival proportions.
Figure 5.
Comprehensive calibration analysis of the prognostic nomogram across multiple time points and datasets. Calibration plots for 1-year (A), 3-year (C), and 5-year (E) survival predictions in the training set. Corresponding calibration plots for 1-year (B), 3-year (D), and 5-year (F) survival predictions in the internal testing cohort. The proximity of the observed calibration curves to the ideal diagonal indicates good agreement between predicted and observed survival probabilities in the internal testing cohort. Simple calibration plot (G) showing the relationship between predicted risk and the actual event rate, providing an overview of the model’s predictive accuracy. Direct comparison of model calibration before (H) and after (I) applying the Platt Scaling post-processing method. The improved alignment of data points with the diagonal line in panel I illustrates the effectiveness of the calibration adjustment in refining the model’s probability estimates. Collectively, these analyses confirm the nomogram’s strong and consistent calibration across different time horizons and in both training cohort and internal testing cohort.
Calibration results of the training set showed that the model exhibited good calibration ability at all time points. The mean absolute error (MAE) of the 1-year survival calibration curve was 0.014 with a 0.9 quantile of 0.028; the MAE of the 3-year survival calibration curve was 0.044 with a 0.9 quantile of 0.12; and the MAE of the 5-year survival calibration curve was 0.071 with a 0.9 quantile of 0.203. The calibration curves at all time points were close to the ideal diagonal line, indicating excellent calibration accuracy of the model on the training set.
Evaluation in the internal testing cohort further supported the stability of the model’s predictive performance. The MAE of the 1-year survival calibration in the testing set was 0.015 with a 0.9 quantile of 0.029; the MAE of the 3-year survival calibration was 0.055 with a 0.9 quantile of 0.118; and the MAE of the 5-year survival calibration was 0.084 with a 0.9 quantile of 0.204. The calibration curves of the testing set were highly consistent with those of the training set, indicating good generalization ability of the model and no overfitting.
Based on the distribution characteristics of risk scores in the training set, patients were divided into three risk groups (low-risk, intermediate-risk, and high-risk) using the 25th and 75th percentiles as cut-off points. The risk score ranged from −5.915 to 0.644, with a median of −4.696. The 25th percentile (Q1) was −5.225, and the 75th percentile (Q3) was −3.984.
Risk stratification results showed 260 patients (26.0%) in the low-risk group, 490 patients (49.0%) in the intermediate-risk group, and 250 patients (25.0%) in the high-risk group. Kaplan–Meier survival analysis revealed significant differences in survival curves among the three groups (p < 0.001), confirming the effectiveness of risk stratification. The 5-year survival rate of patients in the low-risk group was significantly higher than that in the intermediate-risk and high-risk groups, indicating a high consistency between risk stratification and patients’ actual prognosis.
The clinical utility of the model was further validated by manually calculated simple calibration plots. Correlation analysis between predicted risks and actual event rates showed that the model maintained good calibration accuracy across the entire risk spectrum. The risk score distribution was approximately normally distributed, supporting its application value in clinical practice.
Discussion
Based on a large-scale, multi-center retrospective cohort, this study systematically evaluated the impact of clinicopathological characteristics and systemic inflammatory/immune indices on the prognosis of patients with ccRCC, and constructed and validated a comprehensive prognostic prediction model. ISUP grade emerged as an important prognostic factor in the present study. Compared with the reference grade, patients with ISUP Grade 3 and Grade 4 had approximately 3.2-fold and 4.4-fold higher mortality risks, respectively. Meanwhile, systemic inflammatory indices represented by the NLR, PLR, SII and SIRI were significantly associated with patient prognosis. Among these, SIRI integrates information on neutrophils, monocytes, and lymphocytes, more comprehensively reflecting the balance between pro-tumor and anti-tumor immune responses, and patients with SIRI values below the cohort-derived threshold (1.22) showed better survival outcomes. Regarding pathological features, tumor necrosis, sarcomatoid differentiation, giant cells, and vascular invasion were closely correlated with poor prognosis, whereas cystic change and intratumoral hemorrhage did not show significant prognostic significance in this cohort [20].
Following univariable Cox screening and elastic net selection, 12 predictor variables were retained for multivariable model development. The resulting prognostic model demonstrated good discriminative performance, with a C-index of 0.849 in the training cohort and 0.869 in the internal testing cohort, together with satisfactory calibration. Variable selection was performed via elastic net regression (α = 0.8), and the final model integrated multidimensional information—including ISUP grade, tumor necrosis, platelet count, recurrence/metastasis status, and SIRI—into an easy-to-use nomogram. The time-dependent area under the receiver operating characteristic curve (AUC) for 1-, 3-, and 5-year outcomes remained at a high level (training set: 0.829–0.871; testing set: 0.886–0.897). Patients were stratified into low-, intermediate-, and high-risk groups based on risk scores, achieving effective risk stratification and providing a reliable tool for clinical decision-making.
Compared with existing studies, the findings of this research are consistent with previous evidence. As a revised version of the Fuhrman grade, the ISUP grading system is widely recognized for its superior inter-observer consistency and prognostic accuracy; high grades (ISUP 3–4) are often accompanied by genomic instability and enhanced tumor invasiveness. Although the Fuhrman grade showed prognostic relevance in univariate analysis, it was not included in the final model due to potential collinearity with the ISUP grade. Additionally, the adverse prognostic association of tumor necrosis observed in this study is consistent with the core findings of established prognostic models such as the Stage, Size, Grade, and Necrosis (SSIGN) score. Conversely, the absence of tumor necrosis was associated with a lower risk of death (HR = 0.603). The prognostic value of systemic inflammatory indices (e.g. SIRI and PLR) further supports the hypothesis that ‘tumor-associated systemic inflammatory responses promote cancer progression,’ with mechanisms potentially involving immunosuppression and angiogenesis induced by the release of pro-inflammatory cytokines [7].
The strengths of this study include its large multicenter cohort, with 1,635 patients available for baseline analyses and 1,429 patients included in prognostic modeling, comprehensive coverage of clinical and pathological variables, and the use of regularized regression methods for variable selection and internal model evaluation.
Several limitations of this study should be acknowledged.
First, because of its retrospective design, the study is inherently susceptible to selection bias, information bias, and residual confounding. Although comprehensive clinicopathological variables were collected, some potentially important prognostic factors may not have been available and therefore could not be adjusted for in the analyses.
Second, despite the multicenter design and centralized pathological re-review, differences in patient populations, clinical practice patterns, and data collection procedures between institutions may have introduced a degree of heterogeneity.
Third, the study period spanned more than a decade (2010–2024), during which substantial advances occurred in surgical techniques, perioperative management, targeted therapy, and immunotherapy. These treatment-related changes may have influenced patient outcomes and were not fully captured in the present model.
Fourth, although all pathological specimens were re-reviewed by experienced pathologists using standardized criteria, some degree of interobserver variability cannot be completely excluded.
Fifth, molecular and genomic biomarkers, such as VHL, PBRM1, BAP1 mutations, and PD-L1 expression, were not routinely available and therefore were not incorporated into the model.
Sixth, although the model demonstrated good performance in the internal testing cohort, the testing cohort was generated through random partitioning of the same pooled dataset and does not constitute a true external validation cohort. Therefore, the generalizability of the nomogram requires further confirmation in geographically and institutionally independent populations.
Finally, the median follow-up duration was 3.4 years, which may limit the assessment of long-term oncological outcomes.
Clinically, this model may provide a practical tool for postoperative risk stratification in patients with ccRCC. It may help identify individuals at higher risk of adverse outcomes and support individualized follow-up strategies and future treatment-related decision-making. Future research directions include conducting external validation in independent multi-center cohorts, integrating molecular biomarkers to construct a multi-omics prognostic system, exploring the biological mechanisms of prognostic factors, designing prospective clinical trials to verify the model’s utility, and standardizing the evaluation criteria for pathological features.
In conclusion, this study systematically analyzed the value of clinical inflammatory indices, pathological grading, and microscopic features in ccRCC prognosis, and constructed a high-performance prognostic model. It confirms that the ISUP grade, systemic inflammatory status, and key pathological features are core factors influencing prognosis, laying the foundation for clinical risk stratification and individualized treatment.
Additional information
This study was conducted in strict accordance with the Declaration of Helsinki. It has been registered with the Chinese Clinical Trial Registry(China, Approval No. ChiCTR2600117319) and approved by the Clinical Research Ethics Committee of the First People’s Hospital of Yunnan Province (Approval No. KHLL2025-KY340). Written informed consent was obtained from all participants. The written consent forms used were standardized templates of the First People’s Hospital of Yunnan Province. Requests for access to these forms may be directed to the corresponding author.
Acknowledgments
Conceptualization, D.C. and Z.Y.; software, D.C. and J.Y.; resources, D.C.; data curation, D.C.; writing—original draft preparation, D.C.; writing—review and editing, D.C. and X.L.; visualization, D.C.; supervision, D.C. and J.Z.; project administration, J.F.; and funding acquisition, D.C. and J.F. All authors have read and agreed to the published version of the manuscript.
Funding Statement
This study was supported by Science and Technology plan project of the First People’s Hospital of Yunnan Province. Project number: 2024-KHRCBZ-B14. Xingdian Talent Support plan for medical and health personnel. Project number: XDYC-YLWS-2023-0071.
Disclosure statement
The authors declare no conflicts of interest.
Data availability declaration
The raw data supporting the conclusions of this article will be made available by the authors(D.C.), without undue reservation. For specific data, please contact the author or obtain it from the Chinese Clinical Trial Registry.
References
- 1.Bezwada D, Perelli L, Lesner NP, et al. Mitochondrial complex I promotes kidney cancer metastasis. Nature. 2024;633(8031):923–931. doi: 10.1038/s41586-024-07812-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Li W, Shi J, Lv Q, et al. Identification of the LCOR–PLCL1 pathway that restrains lipid accumulation and tumor progression in clear cell renal cell carcinoma. Int J Biol Sci. 2025;21(5):2296–2312. doi: 10.7150/ijbs.107981. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Brannon AR, Haake SM, Hacker KE, et al. Meta-analysis of clear cell renal cell carcinoma gene expression defines a variant subgroup and identifies gender influences on tumor biology. Eur Urol. 2012;61(2):258–268. doi: 10.1016/j.eururo.2011.10.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bharthuar A. Metastatic renal cell carcinoma: current scenario and future trends. South Asian J Cancer. 2012;1(1):30–35. doi: 10.4103/2278-330X.96505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Capitanio U, Montorsi F.. Renal cancer. Lancet. 2016;387(10021):894–906. doi: 10.1016/S0140-6736(15)00046-X. [DOI] [PubMed] [Google Scholar]
- 6.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(5):857–866. doi: 10.1016/j.euf.2018.02.010. [DOI] [PubMed] [Google Scholar]
- 7.Chen D, Zhang Z, Dong L, et al. Prognostic significance of systemic immunoinflammatory biomarkers in patients with clear cell renal cell carcinoma: a retrospective multicenter analysis. Front Immunol. 2025;16:1575497. doi: 10.3389/fimmu.2025.1575497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lee A, Lee HJ, Huang HH, et al. Prognostic significance of inflammation-associated blood cell markers in nonmetastatic clear cell renal cell carcinoma. Clin Genitourin Cancer. 2020;18(4):304–313. doi: 10.1016/j.clgc.2019.11.013. [DOI] [PubMed] [Google Scholar]
- 9.Mantovani A, Allavena P, Sica A, et al. Cancer-related inflammation. Nature. 2008;454(7203):436–444. doi: 10.1038/nature07205. [DOI] [PubMed] [Google Scholar]
- 10.Hu B, Yang X-R, Xu Y, et al. Systemic immune-inflammation index predicts prognosis of patients after curative resection for hepatocellular carcinoma. Clin Cancer Res. 2014;20(23):6212–6222. doi: 10.1158/1078-0432.CCR-14-0442. [DOI] [PubMed] [Google Scholar]
- 11.Qi Q, Zhuang L, Shen Y, et al. A novel systemic inflammation response index (SIRI) for predicting the survival of patients with pancreatic cancer after chemotherapy. Cancer. 2016;122(14):2158–2167. doi: 10.1002/cncr.30057. [DOI] [PubMed] [Google Scholar]
- 12.Mei P, Feng W, Zhan Y, et al. Prognostic value of lymphocyte-to-monocyte ratio in gastric cancer patients treated with immune checkpoint inhibitors: a systematic review and meta-analysis. Front Immunol. 2023;14:1321584. doi: 10.3389/fimmu.2023.1321584. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hutterer GC, Stoeckigt C, Stojakovic T, et al. Low preoperative lymphocyte-monocyte ratio (LMR) represents a potentially poor prognostic factor in nonmetastatic clear cell renal cell carcinoma. Urol Oncol. 2014;32(7):1041–1048. doi: 10.1016/j.urolonc.2014.04.001. [DOI] [PubMed] [Google Scholar]
- 14.Xu W, Lu J, Liu W-R, et al. Heterogeneity in tertiary lymphoid structures predicts distinct prognosis and immune microenvironment characterizations of clear cell renal cell carcinoma. J Immunother Cancer. 2023;11(12):e006667. doi: 10.1136/jitc-2023-006667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Nakajima R, Nozaki S, Kondo T, et al. Evaluation of renal cell carcinoma histological subtype and Fuhrman grade using (18)F-fluorodeoxyglucose-positron emission tomography/computed tomography. Eur Radiol. 2017;27(11):4866–4873. doi: 10.1007/s00330-017-4875-z. [DOI] [PubMed] [Google Scholar]
- 16.Ficarra V, Novara G, Martignoni G.. The use of simplified versions of the Fuhrman nuclear grading system in clinical practice requires the agreement of a multidisciplinary panel of experts. Eur Urol. 2009;56(5):782–784; discussion 784–5. doi: 10.1016/j.eururo.2009.07.024. [DOI] [PubMed] [Google Scholar]
- 17.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(10):1490–1504. doi: 10.1097/PAS.0b013e318299f0fb. [DOI] [PubMed] [Google Scholar]
- 18.Li S, Zhou Z, Gao M, et al. Incremental value of automatically segmented perirenal adipose tissue for pathological grading of clear cell renal cell carcinoma: a multicenter cohort study. Int J Surg. 2024;110(7):4221–4230. doi: 10.1097/JS9.0000000000001358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Moch H, Cubilla AL, Humphrey PA, et al. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs – Part A: renal, penile, and testicular tumours. Eur Urol. 2016;70(1):93–105. doi: 10.1016/j.eururo.2016.02.029. [DOI] [PubMed] [Google Scholar]
- 20.Guo Z, Zhou H, Lv D, et al. Tumor necrosis-informed prognostic nomogram for clear cell renal cell carcinoma model development and clinical validation. BMC Urol. 2025;25(1):252. doi: 10.1186/s12894-025-01940-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zisman A, Pantuck AJ, Dorey F, et al. Improved prognostication of renal cell carcinoma using an integrated staging system. J Clin Oncol. 2001;19(6):1649–1657. doi: 10.1200/JCO.2001.19.6.1649. [DOI] [PubMed] [Google Scholar]
- 22.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(6):2395–2400. doi: 10.1016/S0022-5347(05)64153-5. [DOI] [PubMed] [Google Scholar]
- 23.El Zarif T, Semaan K, Eid M, et al. Epigenomic signatures of sarcomatoid differentiation to guide the treatment of renal cell carcinoma. Cell Rep. 2024;43(6):114350. doi: 10.1016/j.celrep.2024.114350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Calderaro J, Žigutytė L, Truhn D, et al. Artificial intelligence in liver cancer – new tools for research and patient management. Nat Rev Gastroenterol Hepatol. 2024;21(8):585–599. doi: 10.1038/s41575-024-00919-y. [DOI] [PubMed] [Google Scholar]
- 25.Steyaert S, Pizurica M, Nagaraj D, et al. Multimodal data fusion for cancer biomarker discovery with deep learning. Nat Mach Intell. 2023;5(4):351–362. doi: 10.1038/s42256-023-00633-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors(D.C.), without undue reservation. For specific data, please contact the author or obtain it from the Chinese Clinical Trial Registry.