Comparison of analytical performance of ImmunoCAP™ and NOVEOS
Prior to analysing the clinical performance of the two methods, we compared ImmunoCAP and NOVEOS sIgE results at the analytical level. A total of 570 comparisons were made encompassing 6 different allergen extracts and 4 associated MA: egg white/ovomucoid (nGal d 1), cow’s milk/casein (nBos d 8), peanut/rAra h 2, hazelnut/rCor a 14, fish and shrimp (Figure 1). The correlation between the two methods was evaluated using the Spearman’s test which showed very high rho coefficients for both allergen extracts (r=0.92) and MA (r=0.96) (Figure 1A, B). Using a Bland-Altman approach to test agreement between the two methods (Figure 1C, D), we observed a significant divergence (p<0.0001) between absolute differences only for values between 10 and 100 kUA/L, but not between residuals (difference/average). When considering values between 0.1 and 100 kUA/L, NOVEOS sIgE results were lower than ImmunoCAP™ results, by a mean of -15%, from -13% (extracts, p<0.0001) to -17% (MA, p=0.0006).
Then, we found a very good level of agreement (κ index 0.84; agreement 0.92) between the two methods (Figure S1). Highest or lowest levels of concordance were found for egg white extract (κ 1.0, agreement 1.00), and for shrimp (κ 0.74) and hazelnut extracts (agreement 0.81), respectively. Next and to gain further insight in the analysis of discordances, sIgE values were subdivided according to the sIgE reactivity classes from class 0 (≤0.35 kUA/L) to classes 5-6 (>50 kUA/) (Figure 2A). The percentage of agreement within classes was 72% (n = 410/570 sIgE results), while 23% (133/570) results differed by one class. Only 4.7% (27 results) differed by 3 to 4 IgE reactivity classes and corresponded to 25 patients (two patients were discordant for both extract and MA).