| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Laboratorio de Endocrinología, Hospital de Pediatría Garrahan, Buenos Aires, Argentina 1245
aaddress correspondence to this author at: Laboratorio de Endocrinología, Hospital de Pediatría Garrahan, Buenos Aires, Argentina, Combate de los Pozos 1881 (1245); fax 54-11-4308-5325, e-mail echaler{at}yahoo.com
The diagnosis of childhood growth hormone (GH) deficiency is controversial. The usefulness of provocative tests of GH secretion has been questioned for several reasons, one of which involves the large discrepancies in GH measurements among methods and laboratories. Some reports have proposed 10 µg/L as the acceptable GH cutoff value (1), but other values have also been used.
GH immunoassays show poor interassay agreement. These assay discrepancies may occur for several reasons: (a) the use of different GH calibrators (2); (b) the heterogeneity of the GH molecule in human serum (3); (c) the interference of endogenous GH binding protein (4); (d) the various GH epitope specificities of anti-GH antibodies; and (e) the addition of serum to GH calibrators. Because all of these reasons are closely interrelated, their interactions may also accentuate interlaboratory differences in results. The use of methods able to quantify the 22-kDa form of GH exclusively has been proposed as a means of approaching uniformity in results (5). The 22-kDa form is the major circulating fraction and carries the dominant bioactivity.
We analyzed serum GH cutoff values, using different immunoassays. We defined the commercial assay SER 66/217 as the Reference Method, and 10 µg/L as the reference cutoff value for this test, based on several years of clinical experience. In the present study, we analyzed 80 samples to compare the Reference Method with nine other commercially available assays. In addition, results of all assays were compared with those of the Delfia Wallac assay (DELFIA 80/505), which utilizes a monoclonal antibody specific for 22-kDa GH.
Blood samples (n = 80) were from 42 different individuals (26 males and 16 females; age range, 3.2–16 years). All samples were selected from arginine and clonidine pharmacologic tests (6). Patients had either idiopathic short stature or GH deficiency, and basal samples were also selected to include both spontaneous and provoked GH secretion. Seventy percent of the collected samples had values close to 10 µg/L as measured by the SER 66/217 assay. Samples were fractionated once, and aliquots were kept frozen until assayed, up to 3 years.
The types of assay, antibodies, reference standards (2), assay specificities for the GH 22-kDa molecule as claimed by the manufacturers, and CVs are shown in Table 1
. Serum GH was measured in every sample by 10 different commercial assays. In the case of the Serono immunoradiometric assays (Bio Chem Immuno Systems, Rome, Italy), the assay (SER 66/217) was used with WHO International Reference Preparation (IRP) for human GH for immunoassay 66/217, as provided by the manufacturer. However, it was also modified to generate two additional assays. Both modified assays used the original reagents, except for the reference standard: one assay used the WHO IRP for human GH for bioassay 80/505 (SER 80/505), and the other used the WHO IRP international standard for somatotropin 88/624 (SER 88/624). The other seven commercial assays were as follows (abbreviation is followed by type of IRP): four other IRMAs, IMM 80/505 (Immunotech, Marseilles, France), DSL 80/505 (Diagnostic System Laboratories, Webster, TX), NI 80/505 (Nichols Institute Diagnostics, San Juan Capistrano, CA), and DSL 88/624, a time-resolved fluoroimmunoassay; DELFIA 80/505 (Wallac Oy, Turku, Finland); a chemiluminescent enzyme immunometric assay, DPC Q 80/505 (Diagnostic Products Corporation, Los Angeles, CA); and a RIA, DPC RIA 66/217. Assay CVs were calculated from duplicates.
|
To compare cutoff values, linear regression analyses between the 80 samples measured with the SER 66/217 (initial Reference Method) and with each of nine different commercially available assays were performed. The y value for 10 µg/L in the SER 66/217 assay was calculated from each equation line.
The influence of specific determination of 22-kDa GH in the assays was studied by choosing the 22-kDa GH-specific DELFIA 80/505 as our 22-kDa GH Reference Method. Ratio plots were used for between-assay comparisons (7). Ratios were calculated by dividing individual GH results determined by the DELFIA 80/505 assay by individual GH results determined by each of the other assays. A 95% prediction interval was calculated as proposed by Andersen et al. (7). Acceptability criteria were based on inherent analytical imprecision (8) and on analytical quality specifications for imprecision. The allowable CV was calculated from CVwithin-subject, as proposed by Cotlove et al. (8): CVallowable = 1.96 x [(CVwithin-subject x 0.5)2 + (CVwithin-subject x 0.5)2]1/2. Bland–Altman difference plots (9) were also used: the difference between two methods (y axis) was plotted as a function of their mean GH concentration (x axis). To study possible effects of GH dose, linear regression was calculated for every difference plot.
The results of linear regression analyses between SER 66/217 (Reference Method) and the other nine assays are shown in Table 1
. y values for a fixed x value of 10 µg/L are shown. Comparisons between the DELFIA 80/505 (22-kDa GH Reference Method) and the SER 88/624, IMM 80/505, and DSL 80/505 assays using ratio plots (individual GH results by DELFIA 80/505 divided by results of other assays) are shown in Fig. 1
, which also shows both 95% prediction intervals and allowable limits. The mean ratio was 1.07 ± 0.16, 0.98 ± 0.20, and 1.31 ± 0.31 for the SER 88/624, IMM 80/505, and DSL 80/505 assays, respectively, whereas the within-subject biologic variation was 44%, 50%, and 49%, respectively. Additional figures showing the ratio plots for the remaining six assays are supplied in the data supplement for this report, available at Clinical Chemistry Online (http://www.clinchem.org/content/vol47/issue9).
|
Bland–Altman difference plots between the DELFIA 80/505 and three other assays were also constructed (data not shown). The mean bias for SER 88/624 (-0.41 µg/L), IMM 80/505 (0.22 µg/L), and DSL 80/505 (-0.64 µg/L) was similar and relatively small, and no concentration effect was observed.
In summary, our sample universe included patients with and without GH deficiency, both before and after two different provocative tests of GH secretion, in an effort to include samples with an ample variety of GH circulating forms. Differences among cutoff values of the 10 assays were striking. However, differences were smaller among the four assays specific for 22-kDa GH. Therefore, the information provided by our study could be useful in clinical practice to assess provocative tests when different assays are used. However, it is advisable that cutoff equivalents be checked in individual laboratories before being accepted.
The use of an assay system that exclusively quantifies 22-kDa GH has been recommended as part of an ideal GH assay (8). We therefore chose the DELFIA 80/505 as our Reference Method for 22-kDa GH and applied both the ratio plot, as recommended by Andersen et al. (7), and Bland–Altman difference plots (9) to compare differences among assays. With the ratio plots, only three other assays showed values with ratios close to 1 and variations within acceptable limits: SER 88/624, IMM 80/505, and DSL 80/505. Mean ratios were close to 1, and most of the 80 points were within the 95% prediction interval. This interval was narrower than the biologic variation of the sample. The last two assays share the same reference standard with the DELFIA 80/505, and the three are specific for 22-kDa GH. The explanation for the similarity with SER 88/624 is not apparent. When we compared these methods, we also found that there was no dose effect in the Bland–Altman difference plots, making them more reliable, at least in the concentration ranges studied. We conclude that the DELFIA 80/505, IMM 80/505, DSL 80/505, and SER 88/624 could be used interchangeably, but with a cutoff between 4.2 and 5 µg/L, to assess provocative tests of GH secretion in children.
Acknowledgments
This work was supported by grants from Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Agencia Nacional de Promoción Científica (FONCYT), and Ministerio de Salud Pública (Beca Carrillo-Oñativia) of Argentina.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  A. Bernasconi, R. Marino, A. Ribas, J. Rossi, M. Ciaccio, M. Oleastro, A. Ornani, R. Paz, M. A. Rivarola, M. Zelazko, et al. Characterization of Immunodeficiency in a Patient With Growth Hormone Insensitivity Secondary to a Novel STAT5b Gene Mutation Pediatrics, November 1, 2006; 118(5): e1584 - e1592. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Bakker, W. Oostdijk, J. M. Wit, and J. E. Sanders Final height after transplantation in childhood Blood, October 1, 2005; 106(7): 2592 - 2593. [Full Text] [PDF]
|
|
|
|
 |
|
 M Rix, P Laurberg, A S Hoejberg, and B Brock-Jacobsen Pegvisomant therapy in pituitary gigantism: successful treatment in a 12-year-old girl Eur. J. Endocrinol., August 1, 2005; 153(2): 195 - 201. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Mochel, A. Slama, G. Touati, I. Desguerre, I. Giurgea, D. Rabier, M. Brivet, P. Rustin, J.-M. Saudubray, and P. DeLonlay Respiratory Chain Defects May Present Only with Hypoglycemia J. Clin. Endocrinol. Metab., June 1, 2005; 90(6): 3780 - 3785. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Marino, E. Chaler, M. Warman, M. Ciaccio, E. Berensztein, M. A. Rivarola, and A. Belgorosky The Serum Growth Hormone (GH) Response to Provocative Tests Is Dependent on Type of Assay in Autosomal Dominant Isolated GH Deficiency because of an ARG183HIS (R183H) GH-I Gene Mutation Clin. Chem., June 1, 2003; 49(6): 1002 - 1005. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
