2159 Impact of Oncological Biomarkers and Clinical Factors on Survival Outcomes in Renal Cell Carcinoma With Spinal Metastases: A Retrospective Analysis

View at publisher Summarise with VIRAT
Authors: Lansaol Yang, Abdul Karim Ghaith, Taha Khalilullah, Abdel-Hameed W. Al-Mistarehi, Anthony J. Davidson, Shaan Bhandarkar, Linda Tang, John Theodore, Nicholas Theodore, Daniel Lubelski
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

INTRODUCTION: Renal cell carcinoma (RCC) has a high propensity for spinal metastasis, often presenting with substantial tumor burden, rich vascularity, and resistance to conventional systemic therapies. Despite advances in surgical and oncologic management, survival outcomes remain highly variable. The prognostic relevance of tumor biomarker expression in RCC spinal metastases remains insufficiently characterized. METHODS: We conducted a retrospective analysis of RCC patients who underwent surgical resection of spinal metastases at Johns Hopkins Hospital between 2013 and 2022. Clinical, surgical, and treatment-related data were extracted from medical records. Immunohistochemical expression of the ten most commonly observed biomarkers—along with Ki-67 and tumor mutational burden (TMB)—was assessed from resected spinal tumor specimens. The primary outcome was overall mortality. Predictive modeling was performed using logistic regression, decision tree analysis, and a tuned random forest algorithm. Model performance was assessed using area under the receiver operating characteristic curve (AUC), and unsupervised K-means clustering was used to define biomarker-driven subgroups. Conditional depth and feature interaction analysis were conducted to evaluate synergistic biomarker relationships. RESULTS: The most frequently expressed biomarkers were TTF-1 (58.3%), CK7 (38.9%), and Napsin-A (33.3%). Less commonly expressed markers such as CAM5.2 and KRAS were associated with the highest observed mortality rates (100% and >80%, respectively), whereas EGFR expression correlated with the lowest mortality rate (∼20%). Random forest analysis identified CK7, AE1/AE3, and KRAS as the top predictors of mortality, based on minimal depth. CK7 and AE1/AE3 also demonstrated statistical significance in multivariable logistic regression (p < 0.01) and ranked highest in multi-metric feature importance analysis. Conditional depth interaction analysis revealed that AE1/AE3–CK7 was the most frequent and predictive biomarker pair, with additional influential combinations including AE1/AE3–Napsin-A and CK7–TTF-1. Unsupervised clustering further highlighted that the TTF-1-positive and AE1/AE3-positive biomarker cluster was associated with the poorest survival (77.8% mortality rate). The final tuned random forest model demonstrated robust predictive performance (AUC = 80%). CONCLUSIONS: This study demonstrates that specific biomarker expression patterns—particularly involving CK7, AE1/AE3, and KRAS—are significantly associated with mortality in RCC spinal metastases. Machine learning analysis underscores the predictive strength of biomarker combinations over single-marker models, offering a framework for molecular risk stratification. These findings support the integration of immunohistochemical and mutational profiling into preoperative prognostication and treatment planning for patients with metastatic RCC involving the spine.

Fields of Study & Subjects

Renal cell carcinoma treatment Management of metastatic bone disease Metastasis and carcinoma case studies

Publication details

Journal / source
Neurosurgery
Volume
72
Issue
Supplement_1
Pages
289–289
Publication date
26 March 2026
Language
EN