Hepatocellular carcinoma (HCC) treatment options remain limited despite advances in targeted therapies for molecularly-defined cancers. To address tumor heterogeneity, we reconstructed HCC clonal evolution through single-cell RNA sequencing trajectory analysis, identifying 902 signature genes across seven cellular states. Weighted gene co-expression network analysis of public HCC datasets revealed tumor-grade-associated modules and established a 14-gene prognostic model linked to clonal evolution. Central to this model is the LDHA-NDRG1 axis - two hypoxia-responsive regulators showing coordinated spatiotemporal expression patterns during cancer progression. Dual-expressing cell lineages correlated with poor prognosis and senescence resistance through LDHA-mediated lactylation of histone H2B at K58 on NDRG1, an epigenetic mechanism connecting metabolic reprogramming to senescence evasion. Therapeutically, dual inhibition of this axis extended survival in metastatic HCC murine models. Our findings reveal that lactate-driven epigenetic modification via the LDHA-NDRG1 axis creates a molecularly distinct subpopulation enabling senescence resistance, providing mechanistic insights into HCC heterogeneity. This work proposes a precision medicine strategy targeting lactylation-mediated epigenetic regulation, with implications for developing combination therapies and patient stratification based on clonal evolution patterns.