Hollow Dodecahedral Nanocages for Reliable Multi-Level Differentiation of Urological Cancers through Non-Invasive Metabolic Fingerprinting.

Urological cancers, including bladder cancer (BCa), prostate cancer (PCa), and renal cancer (RCa), represent the most common malignancies of the urinary system and account for a significant proportion of global cancer-related morbidity and mortality. Despite advances in imaging and pathology, achieving accurate and tiered differentiation among these entities remains clinically challenging due to overlapping symptoms, heterogeneous progression patterns, and limited noninvasive diagnostic tools. Herein, we developed hollow dodecahedral trimetallic oxide nanocages (TONCs) via a metal-organic framework (MOF)-derived synthesis strategy. Leveraging synergistic compositional and structural features, the TONCs exhibited enhanced photoresponsive behavior, thermal conductivity, and stability. As a high-performance matrix for LDI-MS, TONCs enabled the direct analysis of urine samples within seconds while minimizing background interference and improving ionization efficiency. Combined with Gradient Boosting, the approach achieved robust multiclass classification with area under the curve (AUC) values of 0.940 (training set) and 0.944 (testing set). Furthermore, nine m/z features were selected to construct a simplified yet effective urological cancer differentiation platform (UC-D9), which retained a high classification performance. Pathway analysis elucidated the underlying metabolic mechanisms associated with these malignancies. Together, this study provides a clinically translatable strategy for high-throughput metabolomic profiling, offering a promising tool for the noninvasive, accurate, and multilevel differentiation of urological cancers.