| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Articles |
Departments of Bioanalytical Technology (V.Q.), Biostatistics (P.C.), and Analytical Chemistry (C.Q., V.L., E.C.D.), Genentech, Inc., South San Francisco, California 94080
Address all correspondence and requests for reprints to: Dr. V. Quarmby, Bioanalytical Technology, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080.
Abstract
There is a significant systematic difference between the normal range obtained from ethylenediamine tetraacetate plasma samples using the Genentech total insulin-like growth factor I (IGF-I) RIA and normal ranges for other total IGF-I RIAs. To determine whether the quality of the assay standard was the cause of this systematic difference, we analyzed commercially available preparations of recombinant human IGF-I (rhIGF-I) typical of those used as IGF-I immunoassay standards along with our own well characterized rhIGF-I assay standard. For the commercial standards, high performance liquid chromatography-derived purities were low, and some vendor-assigned protein concentrations were inconsistent with values from quantitative amino acid analysis. The Genentech rhIGF-I assay standard was highly pure and quantitatively correct. However, the poor quality of some commercial rhIGF-I preparations was not the primary reason for the systematic discrepancy between the Genentech total IGF-I RIA normal range and most other normal ranges.
Most assays for total IGF-I are calibrated against the WHO International Reference Reagent (IRR) for IGF-I Immunoassays (87/518). The Genentech total IGF-I RIA is not calibrated against WHO IRR 87/518. The protein content assigned to WHO IRR 87/518 was a consensus value from a multicenter collaborative study. Physico-chemical analyses showed that WHO IRR 87/518 is Met-1-IGF-I of low purity (44%), and that the assigned protein content is higher than the value determined by quantitative amino acid analysis. Thus, assays that are calibrated against WHO IRR 87/518 will report total IGF-I concentrations in excess of actual values. We believe that calibration against WHO IRR 87/518 is the cause of the systematic discrepancy between the Genentech IGF-I assay normal range and most other normal ranges, and that much of the plasma IGF-I concentration data in the literature are of questionable accuracy.
INSULIN-LIKE growth factor I (IGF-I), also known as somatomedin C, is a basic polypeptide of 70 amino acids with a molecular mass of 7649 Da (1, 2). This peptide mediates the somatotropic actions of GH (3, 4) and appears to play a pivotal role in the regulation of glycemic control (5). IGF-I is one component of the highly complex IGF system, which also includes IGF-II, a family of high affinity binding proteins, binding protein proteases, and IGF receptors (3, 6).
Measurements of circulating total IGF-I levels are an important part of many clinical studies of growth and development. Several clinical reference laboratories and commercial kit vendors offer assays for determination of human total IGF-I levels in serum or plasma samples (7, 8). In addition, many research laboratories have their own assays for this key analyte (9, 10, 11). The technical challenges in accurately and precisely measuring circulating IGF-I levels by immunoassay have been well documented (12). In addition, a consensus statement has described a series of IGF-I immunoassay validation recommendations (13). There is currently no College of American Pathologists-mandated proficiency-testing program for clinical reference laboratories measuring this peptide by immunoassay in the U.S. Moreover, there is no National Committee for Clinical Laboratory Standards-recommended definitive method or reference method for IGF-I measurement against which IGF-I immunoassays can be calibrated (14).
Circulating levels of IGF-I are GH dependent and are also impacted by age, gender, and nutritional status. Endogenous total IGF-I concentrations in normal children and adults (7, 8, 10, 15, 16) and in different disease states (11, 17, 18, 19, 20) have been extensively documented. These data have formed the foundation upon which our current understanding of the pleiotropic nature of IGF-I is based.
Our ability to accurately and precisely measure IGF-I concentrations in biological matrices becomes increasingly important as recombinant human (rh) IGF-I enters therapeutic use. rhIGF-I is being evaluated as a therapeutic for several indications, including growth promotion in patients with GH receptor deficiency (21), preservation of lean body mass in critically ill patients (22), and improvement of glycemic control in patients with diabetes mellitus (23).
At Genentech, we have been evaluating the safety and efficacy of rhIGF-I as a therapeutic for type I and type II diabetes mellitus. A total IGF-I RIA was developed to monitor levels of IGF-I in ethylenediamine tetraacetate plasma samples from patients receiving rhIGF-I therapy (24). Baseline levels of endogenous IGF-I measured by this assay in 12 healthy adult men (24) are lower than would be expected from normative data in the literature (10, 15).
In this study, we measured circulating levels of IGF-I in a population of 100 normal healthy adults using the Genentech assay. We show that the normal range for this assay is significantly lower than normal range data from other assays. We have investigated the involvement of IGF-I immunoassay standardization in this discrepancy, and we have generated quantitative evidence for the basis of this discrepancy.
Subjects and Methods
Experimental subjects and sampling
Ethylenediamine tetraacetate plasma samples were obtained from 100 healthy adults. Donors were paid volunteers from California and Tennessee, and represented an ethnically diverse population. All donors had given their informed consent. This study was performed in compliance with the Declaration of Helsinki Recommendations. Blood samples were obtained in the morning from fasting donors. Samples were centrifuged, and plasma was stored at -70 C for up to 4 months before analysis.
Preparation of rhIGF-I standard solutions
rhIGF-I was biosynthesized in Escherichia coli at Genentech (South San Francisco, CA). Genentech rhIGF-I reference material is a well characterized aqueous formulation. Lyophilized rhIGF-I was purchased from Bachem Bioscience (catalog no. H-5555, lots C00152 and C00155), Bachem California (Torrance, CA; rhIGF-I catalog no. DGR012A, lot QN326; Met-1-rhIGF-I catalog no. DGR 010A, lot QM323), and Gro-Pep (Adelaide, Australia; receptor grade, catalog no. CU100, lot GJECMS01).
The WHO International Reference Reagent (IRR) for IGF-I Immunoassays (WHO IRR 87/518) was provided by Dr. Adrian Bristow of the National Institute for Biological Standards and Controls (NIBSC; South Mimms, UK) in ampules and as bulk material. Each WHO IRR 87/518 ampule contained 3.1 µg (nominal) lyophilized rhIGF-I (25). Bulk WHO IRR 87/518 was provided in solution. All lyophilized rhIGF-I samples were reconstituted according to the recommendations of the supplier, using nominal protein mass. Results for the commercial rhIGF-I samples are reported in code.
IGF-I immunoassays
Plasma total IGF-I concentrations were measured in duplicate in the Genentech RIA after acid-ethanol extraction. The performance parameters of this assay have been previously reported (24). This assay uses a high purity Genentech rhIGF-I assay standard, an 125I-labeled Genentech rhIGF-I tracer, and a rabbit polyclonal antibody that was raised against Genentech rhIGF-I. This assay is not calibrated against WHO IRR 87/518.
Concentrations of rhIGF-I in standard solutions were determined in duplicate directly (i.e. without acid ethanol extraction) using Nichols Institute Diagnostics IGF-I by extraction RIA (8), Stanford total IGF-I RIA (9), Endocrine Sciences (Calabasas, CA) total IGF-I RIA (26), and Genentech IGF-I RIA (described above). All assays except the Genentech RIA are calibrated against WHO IRR 87/518. Results from these assays are reported in code.
High performance liquid chromatography (HPLC)
IGF-I standard solutions were analyzed by neutral pH, reverse phase HPLC (RP-HPLC) on a HP1090 II system (Hewlett Packard, Palo Alto, CA) with a YMC (Wilmington, NC) ODS-AP column. The mobile phase was: A, 68 mmol/L sodium phosphate, pH 6.8; and B, 70% acetonitrile-30% isopropyl alcohol. Isocratic elution was at 73% A-27% B, and samples were detected by absorbance at 214 nm.
Quantitative amino acid analysis (QAAA)
Amino acid compositions of commercial rhIGF-I standards and the WHO reference materials were determined by QAAA (27). Analyses were performed in triplicate. After acid hydrolysis, samples were dried under vacuum, reconstituted in a sodium citrate buffer, and then run on a Beckman (Palo Alto, CA) 6300 amino acid analyzer. Amino acids were quantitated by peak area comparison to an external standard mixture containing 2 nmol of each component.
N-terminal sequence analysis
Bulk WHO IRR 87/518 reference material was subjected to N-terminal sequence analysis. The sample was loaded onto the reverse phase support of a Hewlett-Packard G1000A sequencer (Palo Alto, CA) and subjected to 10 cycles of Edman degradation (28).
Mass spectrometry
For analysis by mass spectrometry (29), reconstituted samples were loaded onto a Vydac (Separations Group, Hesperia, CA) C18 column and desalted with 0.1% trifluoroacetic acid in water for 7 min. The proteins were then eluted with 0.1% trifluoroacetic acid in 60% acetonitrile for 8 min at 0.2 mL/min into a Finnigan MAT TSQ 7000 (Finnigan, San Jose, CA) electrospray triple quadrupole mass spectrometer. The instrument was operated in the positive mode, with an electrospray voltage of 4500 V and a capillary temperature of 225 C. Scans were from 1000–2250 atomic mass units in 2.5 s. BioToolBox software (PE/Sciex, Foster City, CA) was used for data analysis and mass assignment.
UV-visible spectroscopy
The concentration of a Genentech rhIGF-I standard solution was determined using UV-visible spectroscopy. An extinction coefficient of 0.65, determined by QAAA, was used to convert the measured absorbance at 276 nm into a protein concentration in milligrams per mL.
Statistical analysis
All analyses were performed in S-Plus (version 3.4, StatSci, Seattle, WA). The regression equation and 95% prediction intervals describing the relationship of IGF-I concentration (measured in the Genentech RIA) to age were estimated using log-transformed IGF-I data. IGF-I levels obtained from the Genentech assay were compared to published normal ranges after conversion to standard deviation (SD) scores using age-specific or age- and sex-specific means and SDs. Data transformations were used in situations when the normal ranges were determined using transformed data. For example, the square root transformation was applied to Genentech IGF-I levels before calculating the SD score using the mean and SD derived from the square root transformed data of Juul et al. (10). The IGF-I SD score for a subject’s sample was calculated as (IGF-I - y)/SD, where IGF-I is transformed to conform to the metric of y and SD, the mean and SD of the comparison standard for the age (or age and sex) of the subject.
Results
The age distribution of plasma IGF-I levels in healthy adults of
both sexes, measured using the Genentech IGF-I RIA, is shown in Fig. 1A
. Plasma IGF-I concentrations gradually
decreased with increasing age in adulthood. The graph shows individual
data points along with a calculated mean value and 95% prediction
intervals at each age. Fig. 1B compares data from the Genentech RIA
with one very extensive set of published IGF-I normal range data
measured using a different RIA (10). The graph clearly shows that
normal values from the Genentech assay are significantly lower than the
published normal range values. If the Genentech normal values are
compared with these published data using age-independent SD
scores, on the average the Genentech data are 2.8 SD below
the expected values based on published normal range data
(P < 0.003). In absolute terms, the mean expected
plasma IGF-I concentration for a 30-yr-old adult is about 100 ng/mL
from the Genentech RIA normal range and approximately 250 ng/mL in the
other RIA shown here.
|
IGF-I quantitation can be impacted by many parameters, which vary
between immunoassays, including antibody specificity and affinity,
standard quantity and quality, and IGF-I extraction procedure and
efficiency. Antibody and extraction characteristics were compared
across several IGF-I immunoassays and were eliminated as the potential
cause of this major discrepancy (data not shown). Immunoassay standard
quality could have a major impact on IGF-I quantitation. To determine
whether the systematic difference between our results and other
reported assay normal ranges could be reconciled by differences in
IGF-I assay standards, we obtained and analyzed five commercially
available preparations of rhIGF-I. These were selected as being
representative of materials typically used in commercially available
IGF-I assays in the U.S. These samples were analyzed by neutral pH
RP-HPLC and compared with well characterized rhIGF-I reference
material, which is used as the standard for the Genentech IGF-I assay
(Fig. 2A and Table 1). The reference material has a main
peak retention time of about 25 min and is of high purity, with 97% of
the material in the main peak. Two of the commercial IGF-I samples (A
and B) had much longer main peak retention times than the reference
material (
39 min each); the other three (C–E) had main peak
retention times similar to that of the reference material. All five
samples were less pure than the reference material, with 74%, 69%,
78%, 63%, and 81% of material in the main peak, respectively.
|
|
); both of
these samples had masses of 7780 Da (consistent with
rhMet-1-IGF-I), rather than 7649 (consistent with
rhIGF-I). Interestingly, sample B was labeled as
rhMet-1-IGF-I, but sample A was labeled as rhIGF-I. QAAA
also showed that two of the commercial rhIGF-I samples contained less
IGF-I than the vendor labels claimed (Table 1); samples D and E
contained approximately 50% of the nominal amount of rhIGF-I. The commercial rhIGF-I samples analyzed here were selected as being representative of local standards used for IGF-I immunoassays. These samples are of low purity; in addition, some, but not all, were incorrectly mass assigned. An rhIGF-I standard that has been assigned an incorrect, approximately 2-fold higher concentration would cause a competitive assay to overreport IGF-I concentrations by a factor of about 2. However, an rhIGF-I standard such as C, which has been correctly value assigned and is 78% pure, would not have this effect. Thus, our data on local standards do not explain the systematic, approximately 2-fold difference between results from the Genentech IGF-I RIA and those from other IGF-I RIAs that we have observed.
As there is no National Committee for Clinical Laboratory Standards reference method for IGF-I immunoassay calibration, immunoassays for total IGF-I are typically calibrated against WHO IRR 87/518 for IGF-I Immunoassays (25). In contrast, the Genentech assay is not calibrated against this material. We analyzed WHO IRR 87/518 to determine whether this calibration was the source of our systematic assay discrepancy. WHO reference materials are generally supplied to clinical reference laboratories as ampules containing very small amounts of lyophilized formulated material, which would be insufficient for physico-chemical analyses. For this reason, we analyzed WHO IRR 87/518 bulk material, which was provided in solution and had been stored at -70 C.
WHO IRR 87/518 bulk material was compared with Genentech rhIGF-I
reference material by neutral pH, RP-HPLC (Fig. 2B and Table 1).
Surprisingly, WHO IRR 87/518 had a retention time of about 39 min,
rather than the expected 25 min. In addition, WHO IRR 87/518 was of low
purity (44%) by HPLC. These samples were also analyzed by mass
spectrometry. Genentech rhIGF-I reference material contained one main
peak with the expected mass of 7649 Da (Fig. 3A). The main peak in WHO IRR 87/518 had
a mass of 7780 Da, consistent with the presence of IGF-I plus an extra
methionine (Fig. 3B). Other peaks were seen in WHO IRR 87/518,
suggesting that other IGF-I variants were present; this was also
suggested by RP-HPLC (Fig. 2B). QAAA and N-terminal sequence analysis
(data not shown) confirmed that WHO IRR 87/518 contained an additional
methionine that was at the N-terminal end of the molecule; it is
71-amino acid Met-1-IGF-I, rather than 70-amino acid
IGF-I. QAAA also revealed that WHO IRR 87/518 bulk material had a
protein concentration of 0.66 mg/mL, rather than the expected protein
concentration of 1.00 mg/mL.
|
). The two curves were
parallel in this assay, suggesting that they contain the same material.
However, the Genentech standard was approximately 2-fold more potent
than the calibrated kit standard. This means that unknown samples will
be assigned values approximately 2-fold higher than actual values when
they are obtained from standard curves that have been calibrated to WHO
IRR 87/518.
|
shows that
the standard stock solution has measured protein concentrations that
are close to nominal. This material recovers at 108% of nominal in the
Genentech RIA, but overrecovers at 200–234% (i.e.
2-fold) in the three other assays. This experiment shows that
standard curves made from highly pure, correctly value-assigned
Genentech reference material consistently overrecover when run as
unknowns in assays calibrated against WHO IRR 87/518.
|
Preliminary normal range data for the Genentech total IGF-I RIA show that circulating IGF-I levels decrease with increasing age in adulthood. This same pattern of age-related change in circulating IGF-I levels has been reported from other IGF-I immunoassays (10, 11, 15). The circulating IGF-I levels reported here for healthy young adults are consistent with baseline values in a pharmacokinetic study of rhIGF-I administration to healthy adults reported previously using this assay (24). However, the circulating IGF-I levels measured by the Genentech RIA are significantly lower than those reported with other IGF-I immunoassays (7, 8, 10, 11, 15).
We have found that commercial rhIGF-I preparations typical of those used as local standards for IGF-I immunoassays are of low purity, and that in some cases their actual identity and/or mass are inconsistent with label claims. However, these findings alone cannot explain the systematic discrepancy between the Genentech total IGF-I RIA and other assays that we have documented here.
The primary reason for this systematic difference is that most immunoassays for total IGF-I are calibrated against the WHO IRR for IGF-I Immunoassays (WHO IRR 87/518). In contrast, the Genentech assay is not calibrated against this material. Physico-chemical analyses have shown that WHO IRR 87/518 is actually Met-1-rhIGF-I (71 amino acids) and not rhIGF-I (70 amino acids). Moreover, WHO IRR 87/518 is of low purity (44%), and the assigned protein content is higher than the value determined by QAAA. The protein content assigned to this IRR was the mean value obtained from an international multicenter collaborative study (25). In this study, WHO IRR 87/518 was measured as an unknown in IGF-I RIAs run by 9 laboratories in 4 countries. Interestingly, the estimates of IGF-I measured in WHO IRR 87/518 by the laboratories in the collaborative study showed considerable heterogeneity. The discrepancy between the assigned or nominal and actual protein contents may have arisen during the value assignment of WHO IRR 87/518 if the IGF-I immunoassays used in the collaborative study used impure or incorrectly value-assigned local standards. The protein content assigned to WHO IRR 87/518 by the collaborative study was not verified by physico-chemical analyses. Thus, IGF-I immunoassays that are calibrated against the WHO IRR 87/518 will report total IGF-I concentrations that are in excess of actual values by approximately 2-fold.
Much of the plasma IGF-I concentration data in the literature are based on assays calibrated against WHO IRR 87/518. Although these data are inaccurate, it is important to note that within-assay results are still valid and will correctly reflect relative changes in IGF-I levels. However, it may not be appropriate to compare or pool IGF-I values based on different assays. Data from any IGF-I assay should always be carefully interpreted within the context of assay-specific reference interval (i.e. normal range) data.
Genentech and NIBSC/WHO are working together to produce a new international standard for IGF-I immunoassays, containing a high purity rhIGF-I preparation whose assigned protein content will be directly confirmed by physico-chemical analyses.
Acknowledgments
The authors thank the following individuals: Adrian Bristow of NIBSC/WHO for providing WHO reagents and information; Bonnie Baker, Janice Lee, Kristen Nakamura, Azin Shahzamani, and Genentech’s Assay Services for assistance with IGF-I immunoassays; Reed Harris, May Kwong, Felicity Shen, and Long Truong for analytical chemistry support; Wayne Anstine for graphics; and Ray Hintz, Susan Kramer, and Seth Porter for many helpful discussions.
Received September 25, 1997.
Revised December 23, 1997.
Accepted January 7, 1998.
References
This article has been cited by other articles:
 |
|
 |
|
  M. P. Brugts, M. B. Ranke, L. J. Hofland, K. van der Wansem, K. Weber, J. Frystyk, S. W. J. Lamberts, and J. A. M. J. L. Janssen Normal Values of Circulating Insulin-Like Growth Factor-I Bioactivity in the Healthy Population: Comparison with Five Widely Used IGF-I Immunoassays J. Clin. Endocrinol. Metab., July 1, 2008; 93(7): 2539 - 2545. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. P Hauffa, N. Lehmann, M. Bettendorf, O. Mehls, H.-G. Dorr, N. Stahnke, H. Steinkamp, E. Said, M. B Ranke, and participating members of the German KIGS Board/Med Central laboratory reassessment of IGF-I, IGF-binding protein-3, and GH serum concentrations measured at local treatment centers in growth-impaired children: implications for the agreement between outpatient screening and the results of somatotropic axis functional testing Eur. J. Endocrinol., November 1, 2007; 157(5): 597 - 603. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. E. Wieringa and P. J. Trainer Harmonizing Growth Hormone Measurements: Learning Lessons for the Future J. Clin. Endocrinol. Metab., August 1, 2007; 92(8): 2874 - 2875. [Full Text] [PDF]
|
|
|
|
 |
|
 N. Mauras, V. Quarmby, and D. C. Bloedow Pharmacokinetics of insulin-like growth factor I in hypopituitarism: correlation with binding proteins Am J Physiol Endocrinol Metab, October 1, 1999; 277(4): E579 - E584. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
