Background When evaluating peripubertal short stature patients, the interpretation of insulin-like growth factor 1 (IGF-1) levels based on chronological age (CA) can be inaccurate due to the influence of sex steroids and, presently, there is no evidence to support the assessment of IGF-1 values according to bone age (BA) and pubertal status (PS). Our objective was to assess the discriminatory performance of IGF-1 levels based on CA, BA and PS in the diagnosis of growth hormone (GH) deficiency. Methods We evaluated IGF-1 levels from 154 peripubertal short stature patients classified as GH deficient (GHD, n=23) or non-GHD (n=131). IGF-1 was assayed by a chemiluminescent immunometric assay and transformed into standard deviation scores (SDS) according to CA (IGF-1-SDS-CA), BA (IGF-1-SDS-BA) and PS (IGF-1-SDS-PS). Results The performances of IGF-1-SDS-CA, IGF-1-SDS-BA and IGF-1-SDS-PS in the receiver operator characteristics (ROC) curves were similar. There were greater accuracy and specificity of IGF-1-SDS-PS (98.4% and 93.3%, respectively) and IGF-1-SDS-BA (92.7% and 90.1%, respectively) when compared to IGF-1-SDS-CA (65.6% and 69.5%, respectively). The post-test probability of the IGF-1-SDS was also improved when compared to PS and BA - 44.8% (IGF-1-SDS-PS), 16.8% (IGF-1-SDS-BA) and 5.1% (IGF-1-SDS-CA), with similar negative predictive values. Conclusions The evaluation of IGF-1 levels based on CA has a higher sensitivity than those based on BA or PS, which justify its use as a screening tool. Additionally, IGF-1 assessed by PS has the best positive predictive power for GHD diagnosis in peripubertal age and could reduce the necessity of a second GH stimulation test.
Keywords: GH deficiency; IGF-1; diagnosis; evaluation; positive predictive value; short stature; test performance.