Allelic phenotype values: a model for genotype-based phenotype prediction in phenylketonuria

Purpose: The nature of phenylalanine hydroxylase (PAH) variants determines residual enzyme activity, which modifies the clinical phenotype in phenylketonuria (PKU). We exploited the statistical power of a large genotype database to determine the relationship between genotype and phenotype in PKU. Methods: A total of 9336 PKU patients with 2589 different genotypes, carrying 588 variants, were investigated using an allelic phenotype value (APV) algorithm. Results: We identified 251 0-variants encoding inactive PAH, and assigned APVs (0 = classic PKU; 5 = mild PKU; 10 = mild hyperphenylalaninaemia) to 88 variants in PAH-functional hemizygous patients. The genotypic phenotype values (GPVs) were set equal to the higher-APV allele, which was assumed to be dominant over the lower-APV allele and to determine the metabolic phenotype. GPVs for 8872 patients resulted in cut-off ranges of 0.0-2.7 for classic PKU, 2.8-6.6 for mild PKU and 6.7-10.0 for mild hyperphenylalaninaemia. Genotype-based phenotype prediction was 99.2% for classic PKU, 46.2% for mild PKU and 89.5% for mild hyperphenylalaninaemia. The relationships between known pretreatment blood phenylalanine levels and GPVs (n = 4217), as well as tetrahydrobiopterin responsiveness and GPVs (n = 3488), were significant (both P < 0.001). Conclusions: APV and GPV are powerful tools to investigate genotype-phenotype associations, and can be used for genetic counselling of PKU families.