Novel promoter polymorphism in insulin-like growth factor-binding protein-3: Correlation with serum levels and interaction with known regulators

Insulin-like growth factor (IGF)-binding protein-3 (IGFBP-3) is a major det erminant of circulating levels of the IGFs and is clinically useful for the evaluation of GH deficiency and for predicting the response to GH treatmen t. Recent studies provide evidence that the circulating level of IGFBP-3 is inversely related to the risk of several common cancers, and that antiprol iferative agents such as antiestrogens and retinoids act in part by up-regu lating IGFBP-3 gene (IGFBP-3) expression. Although approximately 50% of the substantial interindividual variability in circulating IGFBP-3 levels is k nown to have a genetic basis, the specific loci involved are unknown. Direct sequencing of genomic DNA specimens from a multiethnic population id entified several single nucleotide polymorphisms in the promoter region of IGFBP3. For the most common single nucleotide polymorphism (nucleotide -202 ) found to be in Hardy-Weinberg equilibrium, genotype was highly correlated to circulating level of IGFBP-3 in 478 men from the Physicians' Health Stu dy. In vitro, we documented significantly higher promoter activity of the A allele at the -202 locus compared with the C allele, consistent with the r elationship observed between genotype and circulating IGFBP-3 (AA > AC > CC ). A positive correlation was observed between circulating retinol levels and circulating IGFBP-3 levels; subset analysis by genotype showed that this re lationship was only present among individuals carrying an A allele at -202 (AA > AC > CC). Tall individuals or individuals with a body mass index of 2 7 or greater had levels of circulating IGFBP-3 that were significantly high er when they possessed at least one A allele (AA > AC > CC). The IGFBP3 promoter region deserves investigation as a locus where polymorp hic variation occurs frequently and may influence GH responsiveness, somati c growth, and possibly cancer risk.