Evaluation of an ELISA Method for the Measurement of Chromogranin a and Comparison with an Immunoradiometric Method

Chromogranin A (CgA) is an acidic protein of 49 kDa produced by different neuroendocrine cells that plays the role of a prohormone, generating several bioactive peptides with autocrine and endocrine functions. Elevated circulating levels of CgA were found in different neuroendocrine tumors (NET), and the determination of the protein concentration is considered to be a support in the diagnosis and treatment of patients with both functional and nonfunctional forms of NET (1).

The immunoradiometric assay CGA-RIA CT (Cisbio Bioassays, Parc Technologique et Scientifique M. Boiteux, Bagnols sur Cèze, France) is a widely used method for the measurement of CgA and exhibits good analytical and clinical characteristics, as also shown by the results of a pilot EQA study (2). The same company manufacturing the abovementioned assay recently developed an ELISA method using the same monoclonal antibodies, raised against the amino acid sequences 145-197 and 198-245 (3), as to facilitate their use in laboratories not equipped for handling radioactivity.

The aim of this study was to evaluate the analytical performances of the new ELISA method and to compare the results of routine samples measured with the IRMA CGA-RIA CT, of the same company. Both the assays were performed according the manufacturer's instructions.

The precision of the ELISA assay was evaluated by testing 2 pools of EDTA plasma (P1 and P2) in quadruplicates, for 5 different runs, according to the CLSI EP-15 standard (4). Linearity was evaluated according to dilution tests (6 dilutions from 1:1.5 to 1:30) of 5 samples with concentrations ranging from 929 μg/L to 133 μg/L. Agreement between methods was determined on 94 routine samples drawn in EDTA plasma and analyzed by the Bland-Altman difference plot and the Passing-Bablok test. The reference values were evaluated by testing 120 normal subjects (64 males and 56 females, aged between 26 years and 64 years), according the CLSI C28-A2 standard (5).

As for the intra-assay and inter-assay coefficients of variation (CV), P1 showed a mean concentration of 72.5 μg/L with an intra-assay CV of 4.7% and a total CV of 6.8%. P2 showed a mean concentration of 389 μg/L with an intra-assay CV of 4.4% and a total CV of 7.9%. The results confirmed the data described in the product insert (intra-assay CV <6%, and total CV <10%).

The linearity test showed acceptable results. The recovery percentages ranged between 90% and 115% for 3 samples in all dilution ratios, with the highest concentrations ranging from 763 μg/L to 173 μg/L and the lowest concentrations from 82.7 μg/L to 18.3 μg/L (a value close to the claimed functional sensitivity of 11 μg/L). However, the dilution of 1 sample at a concentration of 133 μg/L showed inadequate recoveries under 25 μg/L. On the contrary, the sample with a concentration of 929 μg/L showed a good linearity, but the results were constantly overestimated, thus suggesting a possible underestimation of the concentrations greater than 900 μg/L, a value that is close to the highest concentration of the calibration curve (1′100 μg/L).

The comparison with the IRMA assay was performed both using the overall data and excluding the 4 samples with concentrations above the highest calibrator of 1′100 μg/L, to emphasize the agreement within the measuring range. The comparison showed the following Passing-Bablok regression (confidence intervals are indicated in parenthesis): ELISA = −0.53 (-5.5/3.7) + 0.789 (0.757/0.820) IRMA, with a significant underestimation of the values obtained by ELISA. The Bland-Altman plot confirmed this finding, showing a mean bias of −25.3%, with ±2 ds range from −51.6% to 1.0%. The comparison within the measuring range (n=90) showed the following Passing-Bablok regression (confidence intervals are indicated in parenthesis): ELISA = −0.64 (-5.45/3.48) + 0.789 (0.759/0.820) IRMA (Fig. 1). The Bland-Altman plot showed a mean bias of −25.2%, with ±2 ds range from −51.4% to 1.0% (Fig. 2). The results of the analysis of the whole data set and of the cases within the reportable range are superimposable.

Fig. 1

Passing-Bablok regression between IRMA and ELISA methods of the 90 cases within the reportable range.

Fig. 2

Bland-Altman plot of the difference between IRMA and ELISA CgA methods of the 90 cases within the reportable range.

In the product insert the manufacturer reports a reference range for serum samples, while there are no data for samples processed with plasma EDTA. Although less common, plasma EDTA is a matrix that is used as an alternative to serum by IRMA CIS users (6) and that shows correlated but quantitatively higher results for the same patients (7). In fact, the presence of the calcium chelator EDTA may decrease the in vitro processing of CgA by protease inhibition or, alternatively, can reduce a possible aggregation of CgA molecules due to the presence of free calcium ions (8).

The evaluation of plasma EDTA samples from healthy subjects carried out in the present study indicated a reference limit (95^th percentile of the distribution) of 109 μg/L, without differences between genders.

In conclusion, the CGA-ELISA assay has been shown to be sufficiently correlated with the original CGA-IRMA, although it appears to result, at least for plasma EDTA samples, in a relative underestimation. The CGA-ELISA assay has the advantage of not using radioactivity and seems to be robust and reliable for the introduction in the routine setting. It is worth noting that, in the reagents released after this study, the calibration range was restricted by the manufacturing company to 870 μg/L, and this should have improved the linearity.