The Diagnosis of Growth Hormone Deficiency in Adults^a

We were most interested to read the recommendation for using the insulin tolerance test (ITT) for diagnosing growth hormone deficiency (GHD) (1), and we would like to comment on the choice of this test and of the recommended GH cut-off limit of 3 μg/L. The evaluation or introduction of a GH-stimulation test requires the definition of a reference interval for healthy adults, taking into consideration the possible influences of sex, age, obesity, and substitution therapy. The current choice of method for diagnosing GHD in adults is clearly arbitrary. In earlier studies of the clinical effect of GH therapy on patients suffering from GHD, very low stimulated GH-levels were used as inclusion criteria (2–6). In addition, it was, at that time, impossible to base the cut-off limits for GH on scientific data, as no reference interval had been published for stimulated GH-responses to GH-stimulation tests. It has been suggested recently that the criterion for the diagnosis of GHD in adults should be a peak GH-response to the ITT of less than 3 μg/L (1). This limit was based on the further reduction of reported cut-off limit of 5 μg/L in normal subjects (7) to “allow for the influences of age and adiposity” (1). Hoffmann et al. (7) chose age, sex, and body mass index (BMI)-matched controls, as these parameters have been considered to be important to the peak GH-responses to the ITT. However, the matching of control subjects does not imply that these results can be automatically extrapolated to healthy adults, as the controls included obese and elderly individuals. It should be noted that the mean BMI for this group of individuals was 25.1 kg/m² (min-max: 15.6–38.9), and that obesity in otherwise healthy adults has been associated with markedly impaired GH-responses to the ITT (8, 9). Furthermore, although the mean age of this group was 47.3 yr (min-max: 17–78), the peak GH-responses of the most elderly subjects to the ITT were not significantly lower than those of the rest of the group (7). The use of the lowest peak GH-response obtained from a control group that includes aged and obese subjects risks reducing the sensitivity of the ITT in normal-weight patients with suspected GHD.

The marked discrepancies among GH-assays did not become apparent until after monoclonal assays came into widespread use. This is not surprising, as different monoclonal antibodies would be expected to recognize different GH forms in a disparate manner (10), and problems caused by incorrect calibration and the use of inappropriate mass units have been recorded (11). Nevertheless, the relationship between the two assays that have been studied (i.e. the polyclonal Pharmacia assay and the monoclonal Dissociation-Enhanced Lanthanide Fluoro-Immuno-Assay (DELFIA) assay) is reasonably constant, and it is, in fact, possible to use a conversion factor to compare the results of the two assays. The conversion factor for converting RIA (Pharmacia AB, Uppsala, Sweden) to DELFIA (Wallac, Turku, Finland) is: DELFIA = RIA × 0.63. The conversion factor proved to be constant throughout the concentration range, and the distribution around the mean accorded with the analytical imprecision of the two methods (Andersen et al. to be published).

Although more than 15 different GH-stimulation tests are available, opinion is divided concerning which test is to be used for specific patient groups. Moreover, many of the tests are unreliable.

The effect of the combination of pyridostigmine (PD) and growth hormone-releasing hormone (GHRH) on peak GH-levels has been studied by teams including Ghigo et al. (12), Arvat et al. (13), and Andersen et al. (14). The effect produced is based on direct GHRH stimulation of the pituitary gland and PD-inhibition of somastostatin (15–19). 120 mg PD is administered orally 60 min before the injection of GHRH (1 μg/kg, time 0) as an iv bolus. Serum samples for the measurement of GH-levels are then taken at 0, 20, 30, 45, 60, and 90 min. The reference interval for the peak GH responses to the PD-GHRH test was established in 40 healthy adults (14), all of whom were 22–58 yr of age and within 10% of ideal body weight. The GH responses were Log-Gaussian distributed and were unaffected by factors such as age, sex, and the use of oral contraceptives. The 95% reference interval was 42–422 mU/L. Please note that the RIA (Pharmacia AB, Uppsala, Sweden) was used in this study. The 2.5 percentile (−1.96 sd) of 42 mU/L was used as the cut-off limit for the biological GH responses to the PD-GHRH test for adults aged 22–58 yr, and the 90% Cl for this percentile ranged from 31–56 mU/L. Using the established cut-off limit of 42 mU/L for the PD-GHRH test (14), and the conventional cut-off limit of 20 mU/L for the ITT (20), a high level of agreement between the two tests, 44/47 (94%), was found (14). This is a particularly convincing result, especially considering the characteristics of the patients examined. In contrast to the results of the study carried out by Andersen et al. (14), most studies using GH stimulation tests included only patients who were considered GH-deficient (7, 12). The ITT may not be the best tool for the diagnosis of GHD in adults. Hoeck et al. (21) have questioned the reproducibility of the ITT in normal adults. A reference interval for the ITT will be difficult to establish, and it may prove impossible to evaluate the possible influences of sex, age, body composition, and cortisol levels on ITT-stimulated GH levels, as patients find the ITT unpleasant. Furthermore, the ITT is potentially hazardous (22). In contrast to this, no severe side effects and only minor, transitory complaints were recorded in over 400 PD-GHRH tests carried out in Milan and Odense. Of those tested, approximately 10% suffered gastrointestinal discomfort, with 1/10 of these experiencing diarrhea and nausea, and the remainder experiencing heartburn and meteorism. Approximately 10% of those tested displayed tics, and the injection of GHRH caused flushing lasting 30 seconds in approximately 50% of the subjects. Obesity in otherwise healthy adults has been associated with markedly impaired GH-responses to the ITT (8, 9), to the PD-GHRH test (23–26), and to the arginine-GHRH test (27). However, to establish reference intervals for this group of patients it is essential to study the impact of over-weight on peak GH-responses.

Elderly subjects also pose new diagnostic problems. Ghigo et al. found the PD-GHRH test to be a very reliable tool for the biological measurement of GHD in 20–40-yr-olds (12), but normal adults aged 40–65 yr were not included in this study. Some subjects from the 65+ age group who were considered normal still displayed low GH-responses to the PD-GHRH test (12). The combination of arginine and GHRH produced high GH-responses in the 65+ age group, and even in 80-yr-olds. Because of its effectiveness in even very elderly subjects, Ghigo et al. preferred the combination of arginine and GHRH for elderly patients with suspected GHD (12). It must be remembered that basal GH and insulin-like growth factor (IGF)-I levels are reduced in normal, elderly individuals (28, 29), and that GH-levels in some normal individuals will almost certainly be low at a relatively young age. Early somatopause thus makes the diagnosis of GHD in patients over the age of 65 a discussion of the general attitude toward “the diagnosis GHD” in healthy old people. This is an area which merits thorough investigation.

Among the many tests available for the diagnosis of GHD in adults, a particularly strong case can be made for choosing the PD-GHRH test, as it is easy to perform, safe, potent, reproducible, and reliable (14). Furthermore, the cut-off limit of 42 mU/L (90% Cl 31–56 mU/L) has already been established for this test (14). The arginine-GHRH test might also be considered as a diagnostic test (12).

GHD-evaluation should be a part of the total evaluation procedure for patients with suspected hypothalamic and/or pituitary diseases.

In conclusion: 1) Important information can be gathered from the clinical characteristics of the patients. Insufficiency of ACTH will imply diminished peak GH-responses to the ITT in approximately 91% (30) to 100% (14) of patients. However, it should be noted that Toogood et al. (30) reported that 28% of patients suffered from only GHD and not from insufficiency of any other pituitary hormones. Therefore, GHD does not necessarily imply deficiency of other pituitary hormones; 2) A subnormal total IGF-I value essentially confirms the diagnosis, whereas IGF-l levels within the normal range do not preclude GHD (7, 14); 3) We recommend using the PD-GHRH test to diagnose GHD and would like to stress that in normal subjects aged 22–58 yr, the PD-GHRH-stimulated GH-responses were not significantly affected by factors such as age, sex, or the use of oral contraceptives (14).

However, the influence of body composition still remains to be evaluated, and it is still necessary to clarify the state of somatotrophs in healthy elderly people.