There is considerable imprecision in the diagnosis of GH deficiency (GHD) in the short child without evident organic etiology. This is illustrated by the high variability in the prevalence of GHD that has been estimated to be from 20 per million in the UK to 287 per million of total population in the United States (1). There are four general causes for the apparent difficulty in assessing GH secretion in childhood: imprecise use of clinical criteria, insufficient use of hypothalamic-pituitary imaging technology, inadequate use of optimal GH testing paradigms, and inappropriate interpretation of laboratory data. This commentary will focus on the laboratory diagnosis of GHD.

It is commonly recommended that GH stimulation tests should be reserved for children with low serum IGF-I and IGFBP-3 and poor growth. While subnormal serum levels of both serum IGF-I and IGFBP-3 may be highly predictive of a subnormal response to a provocative GH test in prepubertal children (2), two recent large retrospective studies have seriously challenged this proposed diagnostic utility (3, 4). Mauras et al. (5) report that normal children had significantly higher mean concentrations of serum IGF-I, IGFBP-3, and ALS than the short stature group, and suggest that these determinations may have greater diagnostic utility in the evaluation of the short child than GH testing. However, at least one of these children had extreme short stature (−7.4 height SDS) that suggests a diagnosis of classical GH-deficiency, making group comparisons invalid since this subject would be expected to have low GH-dependent peptide concentrations.

An evaluation of GH testing for GH deficiency (GHD) in childhood is confounded by the lack of a worldwide consensus on the definition of GHD. Indeed, a recent international report concluded that it was not possible to accurately make the diagnosis of GHD in short children (6). This is largely related to the endogenous cyclicity of normal GH secretion and the variable responses of any given patient to multiple provocative stimuli. Since no one GH stimulation test has 100% sensitivity (no false-negatives) and 100% specificity (no false-positives), most countries have established an arbitrary cut off for a normal peak serum GH response (usually > 8 to 10 μg/L) to at least two provocative GH stimulation tests (1, 7). However, with these criteria, the percentage of children who are retested after the discontinuation of their GH treatment, and who are subsequently found on retesting to have normal GH secretion, may be greater than 40% (1, 8). However, with the implementation of strict auxological criteria and biochemical criteria that included a limited number of specific provocative GH tests, this incidence is less than 5% in Canada (Dean, H., Personal Communication).

The difficulties in assessing GH secretion in children are related primarily to the choice of testing procedures. A striking example of the difficulties in the use of excessive provocative GH tests is summarized by the experience in France with 6373 GH stimulation tests from 3233 short children who had the diagnosis of GHD (9). Eleven different pharmacological tests were used, and 62 of the possible 66 pairs were used at least once; the most frequent combination was used in only 12.7% of patients. It was concluded that only a limited number of standardized GH stimulation tests should be used.

There have been new approaches to improve the diagnostic accuracy of GH testing. It is recognized that endogenous GH Releasing Hormone (GHRH) is required for the normal GH response to each of the following pharmacological stimuli: L-Dopa, arginine, insulin hypoglycemia, and pyridostigmine. Since normal GH secretion is dependent upon both GHRH stimulation and somatostatin inhibition, we as well as others have recently addressed the utility of controlling somatostatin tone before or during the administration of GH secretagogues (10–12). These studies support the hypothesis that control of endogenous somatostatin tone, by agents such as pyridostigmine, arginine, or pretreatment with somatostatin analogs, leads to an augmented and a more reproducible peak serum GH response to secretagogues in short children with normal GH secretion. Children with classical GHD, and especially those with the Pituitary Stalk Interruption Syndrome (PSIS) on magnetic resonance imaging (MRI), have a markedly diminished GH response to standard provocative tests, to GHRH, and to the recently described GH-releasing peptides (GH-RPs) (8, 13).

The data provided by Mauras et al. (5) highlight several significant advances in laboratory technology over the past 10 y that have played an important role in the diagnostic paradigm for childhood GHD, but which have not been universally recognized. These include the use of recombinant GH standards, the development of MAb to GH, and the introduction of newer assay techniques, such as ELIZA (enzyme-linked immunoassay), IRMA (immunoradiometric assay), and LIFA (ligand-induced functional assay) that may detect different isoforms of GH in the circulation. These changes have caused considerable confusion in what may be called a “normal” GH level. It is very important to recognize that these newer GH assays may provide GH levels that are more than two- to 3-fold lower than the older RIA's (14, 15). Despite this systematic lowering of assay results for serum GH levels, the cut off level used to define GHD has not been comparably reduced, indeed it has shown a systematic drift upwards. A recent worldwide survey of GH testing procedures has shown that most countries have not officially adapted revised diagnostic criteria for normal GH cut-off levels that reflect recent advances (7). Thus, there continues to be inappropriate use of GH laboratory values that ensures that short children with normal GH secretion will be enrolled in GH treatment programs. However, there is no evidence that this population benefits significantly in either adult height gain, or psychosocial adaptation (13).

It is strongly recommended that clinicians assess GH secretion only in the slowly growing short child who may have low levels of serum IGF-I and IGFBP-3. They should use a minimum number of standard GH testing procedures, which include the control of somatostatin tone, and have complete knowledge of the assay methodologies used by their reference laboratory. As the data of Mauras et al. (5) clearly demonstrate in a small number of normal statured children, the use of arbitrary cut-off values above 5 μg/L with current assay methods should be abandoned. Indeed, 46% of the normal statured children had GH levels below 5 μg/L using the ELISA method and 33% were below 5 μg/L using the IRMA method. Normative data for GH peak responses should be determined for each center, taking into consideration the recent advances in GH testing paradigms and in diagnostic laboratory technology.