Phenylketonuria from Classical Phenotypes to Base Editing and Targeted Small Molecules: A Critical Narrative Review of Mechanistic Advance and Translational Uncertainty

Abstract

Phenylalanine hydroxylase deficiency, historically described as phenylketonuria, occupies an unusual position in medicine. It was the first inherited metabolic disorder for which dietary intervention prevented intellectual disability, the first to be detected through population newborn screening, and it is now among the first for which in vivo genome correction has reached advanced preclinical validation. This review examines whether the conceptual architecture inherited from that history, above all the classification of patients into discrete phenotypic categories defined by pre-treatment blood phenylalanine concentration, remains adequate for a therapeutic landscape organised around residual enzyme function, variant-specific pharmacology and allele-specific correction. Literature was identified through structured searching of biomedical and multidisciplinary scholarly sources, supplemented by backward and forward citation tracking and by targeted searches for corrections and post-publication correspondence, with critical appraisal focused on design adequacy, endpoint validity and the distance between mechanistic demonstration and demonstrated clinical benefit. Four findings structure the synthesis. First, phenotypic classification predicts population-level severity but performs inconsistently at the level of the individual patient, and its predictive limits are inherited by every therapy selected on its basis. Second, the residual morbidity of early and continuously treated disease is real but modest in average magnitude, heterogeneous in expression and inconsistently attributable to phenylalanine exposure rather than to the restrictive diet itself. Third, pharmacological options have expanded substantially, and a recent head-to-head trial provides the first randomised comparison between cofactor-based agents, yet the evidence base remains dominated by biochemical surrogates over short horizons. Fourth, adenine base editing has achieved durable normalisation of blood phenylalanine in variant-humanised mouse models and improvement in brain biochemistry and motor phenotype, but no corrective editing approach has published peer-reviewed clinical outcomes, and the variant heterogeneity of the disorder poses a scalability problem that current allele-specific strategies do not resolve. The field's principal unresolved question is no longer whether phenylalanine can be lowered, but which outcomes justify which risks, in whom, and at what age.