Immunometric IGF-1 Measurements Frequently Misclassify Clinical Samples Compared to a Mass Spectrometry IGF-1 Assay

Background: Accurate measurement of insulin like growth factor 1 (IGF-1) is critical for diagnosis and management of pituitary growth hormone axis abnormalities. As anomalous results, poor reproducibility , low precision and assay bias are often reported with currently  available IGF-1 radioimmunoassays(RIA), mass spectrometry, was employed as an alternative  assay technology.

 Objective: IGF-1 was measured by both RIA and mass spectrometry (LC-MS/MS ) in serum samples from patients with pituitary disorders .

 Methods: Serum samples  prospectively obtained from 101 patients with both naïve and treated secretory and non-secretory pituitary tumors, cysts, hypophysitis,and craniopharyngioma, were evaluated by IGF-1 mass spectrometry (MS) and by Siemens Immulite 2000 IGF-1 assay (reagent lot #512; solid phase, enzyme-labeled chemiluminscent immunometric assay).  After sample accrual, samples were submitted for each assay as a single batch.  

Results: Although results of the two assays were highly correlated (R²=0.97), a significant positive bias was observed in the RIA  compared to the MS assay, reflecting a systematic proportional difference observed between  Siemens and LCMS assays, with the Siemens typically reporting higher results than the LCMS. This is reflected in the slope of the relationship observed in Deming regression y=1.57x – 54.0 and linear regression analyses y=1.55x – 48.3, . where y is the  Siemens assay  value . For classification into normal or abnormal values from each assay, results  were interpreted according to their respective reference intervals. IGF-1 MS assay results were interpreted using published age-adjusted ranges (1) and for the Siemens assay, according to age-adjusted ranges provided in the directional insert. Twenty six of 99 (27.2%) samples for which reference ranges were available for both assays, were classified differently (in range, low, or elevated) by the Siemens and the IGF-1 LCMS assay.  All 26 disagreements occurred with LCMS in the reference range: 14 were with elevated RIA  and 12 with low RIA.  LCMS was in the reference range significantly more frequently than RIA, 72.7% versus 46.5% of the cases, McNemar P < 0.0001.

 Conclusion: While results of  IGF-1 testing with the RIA  and LC-MS/MS assays  were highly correlated, a strong positive bias with higher results was reported using RIA compared to LC-MS/MS.  Twenty seven percent of IGF-1 samples measured by the two assays  were differently classified indicating a large degree of inter-assay variability. Inter-assay variability for IGF-1 results is well established and caution should be used when interpreting results of differing methodologies.The  accuracy of the IGF-1 LCMS methodology, large reference ranges, and concordance of  published LCMS reference ranges with a large published reference population (2 ) reinforces the clinical utility  of the mass spectrometry based IGF-1 assay