Abstract
Mannings et al. 1 present data from seven patients where serum alphafetoprotein (AFP) results from a second trimester Down's screening programme were discordant between the autoDELFIA® hAFP assay (PerkinElmer Life and Analytical Sciences, Turku, Finland) and the Roche E170 Modular Analytics E170 AFP assay (Roche GmbH, Mannheim, Germany). After incubation at 4°C for one week, the results obtained by the autoDELFIA method were higher, which these authors take as evidence for complement interference in this assay. We have also run the autoDELFIA assay for 20 years as part of a second trimester screening programme; we have seen similar cases and arrive at the same conclusion. 2 We have found that dilution studies easily identify this interference which all but disappears at four-fold dilution. We have also shown that the complex is labile using freeze–thaw cycles and by heating the sample (56° for 30 min). While this is consistent with complement interference, more definitive evidence for the involvement of complement comes from the use of calcium chelators to prevent complement activation. In our experience, the addition of 5 mol/L EDTA to the assay buffer negates this interference (the DELFIA assay needs to be re-configured to prevent EDTA from chelating the europium-labelled detection antibody). By using antihuman IgG or anticomplement secondary detection antibodies, we have been able to demonstrate that both human IgG and complement bind to the DELFIA AFP plate in the presence of these serum samples. As complement activation is required to cause the interference, we suggest that steric hindrance by a complement complex is a more plausible mechanism for this observation than a simple blocking antibody. We have not seen this interference in other DELFIA assays using the same serum samples. If this effect is unique to the AFP assay, it may be due to an endogenous anti-idiotype interfering antibody or due to a unique feature or the AFP plate such as antibody density which enables the binding of complement directly to the capture antibody. As suggested by Mannings et al., 1 this type of interference is fortunately rather rare. Based on a strategy of checking all low AFP results, we estimate an incidence of five per 100,000 samples analysed for AFP. We would recommend analysis following serial dilution for all samples giving a result of less than 0.2 times the gestation age corrected median (0.2 multiples of median [MoM]). We routinely use doubling dilutions to 16-fold to ensure that excess capture antibody-binding sites are available to the assay. The result, when corrected for dilution, should stabilize as more open capture antibody sites become available relative to complement blocked sites at higher dilution. It is conceivable that this type of interference may affect patients with serum AFP MoMs higher than 0.2, so potentially affecting the calculated Down's risk in more patients. A more extensive method comparison study would be required to confirm this. Although rare, the potential consequences of this type of error emphasizes the need for continuing vigilance when interpreting results from immunoassay methods and for continuing improvement in immunoassay design.
DECLARATIONS