Metabolic memory in diabetic complications: evidence for persistent cellular states and limits of therapeutic reversal.

Metabolic memory is defined here as an exposure-history-dependent cellular phenotype that persists beyond immediate carry-over after verified withdrawal of the initiating exposure, principally high glucose and is accompanied by a specified functional impairment or an altered response to the same defined secondary stimulus. This operational definition separates cellular memory from clinical legacy effects, continuing local stimulation and residual structural injury. We critically examine evidence from longitudinal clinical studies, human biospecimens, animal models and controlled cell experiments, distinguishing molecular persistence, measured signaling activity, causal maintenance and treatment-free durability. DCCT/EDIC and UKPDS establish lasting benefits of earlier glycemic management but do not independently identify cellular memory as the mechanism. Experimental evidence supports selected endothelial redox circuits, epigenetic regulation and hematopoietic progenitor training; human causal evidence is substantially more limited. Persistent chromatin marks do not necessarily imply continuously active signaling, and latent priming may become apparent only after rechallenge. In kidney, cardiovascular, retinal and wound disease, candidate cellular memory coexists with hypoxic or inflammatory niches and structural loss. Pancreatic β-cell plasticity is considered a comparative endocrine model rather than an established complication-memory circuit. The proposed interactions among epigenetic, mitochondrial, metabolic and inflammatory mechanisms remain a framework supported by selected local feedback loops, not a universally demonstrated network. Therapeutic evidence is organized into prevention, modification or suppression of established abnormalities, and sustained molecular and functional recovery after treatment cessation after intervention withdrawal. The latter remains unestablished in patients for the mechanisms reviewed. Early glycemic and multifactorial management remains essential, while mechanistic translation requires explicit recovery, perturbation and post-treatment follow-up designs.