Discussion
We report here the compared performance of two sIgE platforms and their clinical cut-offs for 10 common food allergens. Clinical cut-offs for sIgE have been previously proposed multiple times, in particular for food allergens, as indicators of the probability of presenting with allergic symptoms, rather than thresholds accurately predicting the occurrence of symptoms (14, 23, 24). Thus, the quantitative nature of sIgE measurements is essential for allergy diagnosis, and physicians must be aware of the characteristics of the employed methods. Indeed, several routine methods of sIgE quantitation co-exist, because they belong to successive generations and to different times of availability for clinical use. First generation tests were radioimmunoassays which used an anti-IgE reagent labelled with a radio-isotope, usually125I: the RAST (RadioAllergoSorbent Test, Pharmacia Diagnostics AB, Uppsala, Sweden ) was commercialized in 1974 (25). The main second-generation test is based on the ImmunoCAP™ technology (originally from Pharmacia AB, now ThermoFisher Scientific), where allergens are covalently attached to a nitrocellulose sponge (4). Third generation sIgE tests are represented by the IMMULITE 2000 system (Siemens Healthcare SAS, Saint-Denis, France), which uses biotinylated soluble allergens bound to a large diameter (25mm) avidin-coated unique bead, and a chemiluminescent signal is used for detection (26). Recently (2020), fourth-generation technologies for sIgE determination were made available (NOVEOS, Hycor, Garden Grove, CA, USA and IDS-iSYS, Bolton, UK), differing from third generation tests through the use of biotin-labelled allergens bound to avidin micro-beads and chemiluminescent detection (16, 27).
We found that NOVEOS sIgE results were significantly lower than those obtained with ImmunoCAP™, by a mean value of 15%. This discrepancy is not due to a defect in the linearity of NOVEOS technology (28) which is similar to that of ImmunoCAP™ (27). The differences we observed between the two methods could be due to an underestimation by NOVEOS, to an overestimation by ImmunoCAP™, or both. Significant discrepancies had been reported previously between ImmunoCAP™ and Immulite™, a third-generation technology developed by SIEMENS (29). It is notable that our data show very few differences for low values (0.1 to 1 kUA/L). These levels of sIgE are important for early detection of sensitization against food allergens in children (30, 31). By contrast, if the two methods produce significantly different results for high sIgE concentrations, these discrepancies are of lesser clinical significance, especially if they are greater than the clinically relevant cut-offs.
For clinical performance, NOVEOS is better able to exclude allergy in sensitized individuals having low sIgE values (0.35-3.5 kUA/L) and we also found some false positive results in non-allergic patients tested with ImmunoCAP™ for peanut and hazelnut extracts. Despite these discrepancies, we report that NOVEOS and ImmunoCAP™ have mostly similar performance for discriminating between food allergic and food tolerant individuals. It is impossible to determine clinically relevant universal thresholds of sIgE concentrations due to important variations from one population to another one. For example, the cut-off for rAra h 2 sIgE concentration in peanut allergy varies from 0.10 to 42.2 kUA/L between studies (15). However, the establishment of “local” clinical cut-offs is of utmost importance for the management of a given population of patients including the design of OFC protocols. In support of this assertion, a 2002 study conducted in our center found a clinical threshold for ImmunoCAP™ peanut extract sIgE (cut-off of 15kUA/L with 95% specificity and 44% sensitivity) which was similar to the values we report here (cut-off of 14kUA/L, 86% specificity and 51% sensitivity) (24). Thus, cut-off values can be established for a given population on the condition of using similar protocols and seem to be stable for extended periods (20 years in this example). Our study further supports that sIgE measurements by both ImmunoCAP™ and NOVEOS, are highly informative on the risk of allergy in the patients we studied based on OR values >10 and RR>2.
We investigated two potential causes of discrepancies in clinical performance between NOVEOS and ImmunoCAP™ with hazelnut and peanut extracts, namely Cor a 1 spiking of ImmunoCAP™ hazelnut extract and the presence of CCD. While Cor a 1 spiking of ImmunoCAP™ hazelnut extract did not contribute to clinical performance discrepancies, CCD did. Thus, our study supports the view that glycosylated epitopes are more accessible to sIgE with ImmunoCAP™ than with NOVEOS. This could be due to the avidin-coated beads and the biotinylation of NOVEOS allergens. Another possibility is that anti-CCD sIgE react both with CCD determinants on allergen molecules and also with the nitrocellulose sponge matrix on ImmunoCAP™ (32, 33). Unlike the animal-derived galactose-α−1,3-galactose epitope, plant CCD (e.g. MUXF3), are currently considered devoid of clinical relevance in allergy (34). This could explain our findings of better sIgE - confirmed plant food allergy correlation with NOVEOS than with ImmunoCAP™. For recombinant MA which are non-glycosylated in both systems we found similar clinical performance as expected.
There are several limitations to our study. Firstly, this is a retrospective, monocentric study. Secondly, the population was mainly comprised of children (85%). In addition, and depending on the allergen, the percentage of patients under a strict avoidance diet varied from 12% (peanut) to 53% (seafood) and an OFC was not systematically performed for food allergy diagnosis, except for peanut. On the other hand, our study encompasses a large number of comparisons including both extracts and molecular food allergens. Moreover, the heterogeneity of the patients and of their therapeutic protocols mirrors our regular clinical practice.
In conclusion, we demonstrate here that, for 10 common food allergen extracts and molecules, assayed in a large pediatric cohort, sIgE determination performed with NOVEOS or with ImmunoCAP™ are highly correlated with and predictive of the actual diagnosis of food allergy or tolerance. Despite a ten-fold lower test sample volume requirement (4 µL) compared to ImmunoCAP™ (40 µL), NOVEOS has an overall better capacity to identify patients at risk of allergy versus asymptomatic sensitization (p=0.03 when AUC are compared). Further confirmatory studies are warranted including more allergens (i.e. other food allergens, respiratory, venom, drugs) and both adult and pediatric patients from other geographical areas.