Development and validation of a SEER-based prognostic nomogram for patients with early-onset upper tract urothelial carcinoma

Technical Summary: Development and Validation of a SEER-based Prognostic Nomogram for Early-Onset Upper Tract Urothelial Carcinoma

Problem Statement

Upper tract urothelial carcinoma (UTUC) accounts for 5–10% of all urothelial carcinomas. While prognostic factors such as lymphovascular invasion, tumor size, and TNM stage are well-established for the general UTUC population, there is a significant gap in understanding the specific prognostic determinants for early-onset patients (<70 years). Current literature focuses heavily on elderly cohorts (≥70 years), and traditional models (e.g., standard TNM staging) fail to account for the distinct biological behaviors and treatment patterns characteristic of younger patients. Furthermore, early-onset UTUC often presents with non-specific symptoms, leading to potential misdiagnosis and advanced-stage detection. There is an urgent need for a prognostic tool tailored specifically to the early-onset demographic to facilitate personalized risk assessment and treatment planning.

Methodology

This study employed a retrospective design utilizing two primary data sources:

1. SEER Database (Training/Internal Validation): Data from 2004 to 2015 covering 30% of the US population.
   • Cohort: 2,694 early-onset patients (<70 years) and 4,144 elderly patients (≥70 years) were identified.
   • Exclusion Criteria: Non-primary tumors, missing staging/grading/tumor size data, non-radical nephroureterectomy, and unclear cause of death.
   • Split: Early-onset patients were randomly divided into a training cohort (n=1,886) and an internal validation cohort (n=808) at a 7:3 ratio.
2. External Validation Cohort: Retrospective data from 255 early-onset UTUC patients treated at the Second Hospital of Lanzhou University (January 2011–June 2023).

Statistical Analysis:

• Comparative Analysis: Clinicopathological characteristics between early-onset and elderly groups were compared.
• Survival Analysis: Overall Survival (OS) and Cancer-Specific Survival (CSS) were analyzed using Kaplan-Meier curves and Cox proportional hazards regression.
• Model Construction: Univariate and multivariate Cox regressions identified independent risk factors. A nomogram was constructed based on these factors to predict 1-, 3-, 5-, and 8-year OS and CSS.
• Validation: The model's performance was evaluated using the Concordance Index (C-index), Receiver Operating Characteristic (ROC) curves (AUC), Decision Curve Analysis (DCA), and calibration plots.
• Risk Stratification: Patients were categorized into low-risk and high-risk groups based on nomogram scores, with survival differences analyzed via Kaplan-Meier curves.

Key Contributions

• First Dedicated Nomogram for Early-Onset UTUC: This study presents the first prognostic model specifically designed for patients under 70 years old, addressing a critical gap in precision medicine for this demographic.
• Identification of Distinct Prognostic Factors: The study identified independent risk factors specific to early-onset patients, including age, tumor size, T/N/M stages, histologic subtype, and the impact of chemoradiotherapy.
• Comprehensive Validation: The model underwent rigorous internal validation (training/validation split) and external validation using an independent clinical cohort, demonstrating robustness across different populations.
• Risk Stratification System: A clear stratification system was developed, allowing clinicians to categorize patients into distinct risk groups with statistically significant differences in survival outcomes.

Results

• Cohort Characteristics: Early-onset patients exhibited a higher proportion of males (63.4% vs. 52.9%), unmarried individuals, and Black patients compared to the elderly cohort. They also received chemotherapy more frequently (28.2% vs. 14.5%).
• Survival Outcomes: Early-onset patients demonstrated significantly better OS and CSS compared to elderly patients (HR for OS: 2.08; HR for CSS: 1.61).
• Independent Predictors:
  • For OS: Age, tumor size, T stage, N stage, M stage, histologic subtype, and chemoradiotherapy.
  • For CSS: Age, tumor grade, tumor size, T stage, N stage, M stage, histologic subtype, and radiotherapy.
• Model Performance:
  • Training Cohort (OS): AUC values for 1, 3, 5, and 8 years were 0.807, 0.774, 0.760, and 0.736, respectively.
  • Internal Validation (OS): AUC values were 0.811, 0.793, 0.763, and 0.752.
  • External Validation (OS): AUC values were 0.793, 0.767, 0.754, and 0.730.
  • CSS Performance: The nomogram consistently outperformed individual factors (e.g., T stage alone) in predicting CSS across all cohorts, with AUCs ranging from 0.780 to 0.863.
• Risk Stratification: Kaplan-Meier analysis confirmed that low-risk groups had significantly higher survival rates than high-risk groups across all cohorts (e.g., OS HR = 2.82 in training, P < 0.001).

Significance and Claims

The authors claim that this study successfully developed a validated nomogram to predict survival outcomes for early-onset UTUC patients, a group previously underrepresented in prognostic modeling. The tool aims to:

1. Quantify Risk: Precisely estimate 1-, 3-, 5-, and 8-year OS and CSS probabilities for individual patients.
2. Guide Treatment: Assist clinicians in formulating individualized treatment plans, particularly regarding the use of adjuvant therapies.
3. Address Clinical Gaps: Provide a specialized assessment tool that integrates variables often overlooked by conventional models, thereby improving prognostic accuracy for younger patients.

The study acknowledges limitations, including its retrospective nature, the lack of granular treatment details in the SEER database (e.g., specific drug regimens), and potential selection bias due to the single-institution external validation. However, the authors assert that the model offers a valuable reference for prognostic assessment and supports the integration of emerging therapies (such as immunotherapy) into the management of early-onset UTUC.