Identifying new cancer biomarkers is crucial for cancer diagnosis. The modification sites of N6-methyladenosine (m6A) significantly influence its regulatory effects on RNA expression and cellular function, suggesting that site-specific m6A modifications may serve as more valuable cancer biomarkers than global m6A or m6A signatures of individual RNA transcripts. However, this potential has largely remained unexplored. In this study, we developed a comprehensive strategy for assessing the clinical value of site-specific m6A modifications via combining the high specificity of an m6A-sensitive DNAzyme with the ultrasensitivity of surface-enhanced Raman scattering (SERS) detection and imaging. Using this strategy, we first identified three m6A sites that could serve as biomarkers for pancreatic cancer diagnosis and breast cancer subtyping through single-cell and tissue extract assays. Further, we spatially mapped m6A modifications on clinical formalin-fixed, paraffin-embedded (FFPE) tissue sections. The SERS imaging of the three m6A sites on tissue sections enabled not only the differentiation of pancreatic ductal adenocarcinoma (PDAC) and breast cancer from normal tissues but also the differentiation between triple-negative breast cancer (TNBC) and luminal A breast cancer (LABC). Our strategy exhibits multiple significant merits, including high sensitivity and specificity, preservation of cellular context, versatility across sample types, spatial mapping capability, and minimal sample requirement. It provides a new research paradigm for exploring the clinical implications of epitranscriptomic modifications, holding great promise for advancing cancer diagnosis and precision treatment.