Sensitive detection of low levels of serum glycans is essential for diagnosing various urological cancers with ambiguous clinical symptoms, but this remains challenging due to the lack of affordable, user-friendly methods with adequate accuracy. Here, microemulsion-guided assembly of hierarchical porous cerium-based metal–organic frameworks and iron ion absorption are exploited to develop a novel graphitized carbon matrix (HPC-Ce/Fe) with high specific surface area and hierarchical porosity through controllable high-temperature treatment, which effectively promotes the diffusion and adsorption of N-glycans, resulting in an outstanding improvement in the detection limit for N-glycans. Leveraging the high enrichment sensitivity of HPC-Ce/Fe, high-throughput mass spectrometry is used to rapidly acquire high-quality glycan profiles from over a hundred serum samples of urological cancers, including prostate, bladder, and renal cancer. Machine learning is employed to screen and evaluate differential-specific glycans, thereby developing a comprehensive diagnostic system for urological cancers, capable of distinguishing cancer patients from healthy donors (area under the curve values (AUCs) of 0.987–1.000) as well as differentiating among cancer types (AUCs of 0.960–0.993).