Edited by: Antonella Muraro
Clinical value of component-resolved diagnostics in peanut-allergic patients
Article first published online: 14 DEC 2012
© 2012 John Wiley & Sons A/S. Published by Blackwell Publishing Ltd
Volume 68, Issue 2, pages 190–194, February 2013
How to Cite
Clinical value of component-resolved diagnostics in peanut-allergic patients. Allergy 2013; 68: 190–194., .
- Issue published online: 16 JAN 2013
- Article first published online: 14 DEC 2012
- Manuscript Accepted: 14 OCT 2012
- Ara h 2;
- diagnostic values;
As replacement for the oral food challenge, decision-points for sensitization test have been established, but suboptimal sensitivity and/or specificity, as well as regional differences, have reduced the clinical usability. IgE toward specific peanut protein components has been reported to be of value, but data on correlation with clinical data are sparse. Our aim was to correlate IgE values with the outcome of peanut challenges.
Data from 175 positive and 30 negative peanut challenges in patients aged 1–26 years were retrospectively correlated with the levels of specific IgE to peanut and peanut components (Ara h 1–3, h 8, and h 9).
The best correlation between IgE and clinical thresholds was found for Ara h 2 (ρs = −0.30, P < 0.01). A cutoff of Ara h 2 > 1.63 kU/l yielded a specificity = 1.00, with a corresponding sensitivity of 0.70. Symptom severity elicited during challenge correlated significantly with the levels of Ara h 2 (ρs = 0.60, P < 0.0001), but large individual variation was found.
The level of IgE toward Ara h 2 can improve diagnostic accuracy by introducing a more clear-cut decision-point with an optimal specificity maintaining a high sensitivity. In our study, this would have reduced the necessary number of challenges to be performed from 205 to 92. Extrapolation between centers is difficult and decision-points need to be addressed in relation to settings and population. Further component-resolved diagnostic cannot replace oral challenge neither in determining thresholds nor in the assessment of severity of symptoms elicited during challenge.
Peanut is one of the most common foods eliciting allergic reactions both in children and in adults [1-4]. Double-blind, placebo-controlled food challenge (DBPCFC) is the gold standard today in the clinical diagnostic workup , but is a time-consuming procedure and not always safe for the patient. Thus, a faster and simpler test to replace oral challenge would be advantageous.
Outcome of challenges has been correlated with sensitization parameters such as skin prick test (SPT) or level of specific IgE (s-IgE) in order to establish decision-points, which enables discrimination between positive or negative peanut challenges; however, these vary between centers and depend on the study population, settings, and procedures, making extrapolation and comparison between centers difficult. For SPT, a mean wheal diameter of 7–8 mm has been suggested [6-8], and for s-IgE, decision-points at 10–15 kU/l have been proposed [7, 9, 10]. For both SPT and s-IgE, a higher cutoff and thereby increased specificity will reduce the sensitivity to as low as 30%. Attempts to increase sensitivity by including case history and thereby applying different cutoff points to different patient groups have also been proposed .
Peanut consists of a variety of different peanut proteins, 11 of which have been identified so far . Traditionally, s-IgE toward whole peanut protein (f13) measures the amount of circulating IgE toward this mixture of different proteins. The introduction of natural purified or recombinant peanut proteins and the measurement of circulating IgE directed toward these specific protein components could introduce the next leap into diagnosing peanut allergy. Only a few attempts to explore the clinical potential of component-resolved diagnostic (CRD) in peanut allergy have been made. It has been shown that Ara h 2 is the most important component in relation to peanut allergy , and recently, Codreanu et al.  have shown that a cutoff of 0.23 ku/l would yield a sensitivity of 0.93 and a specificity of 0.97. A specificity of 1.00 for Ara h 2 was obtained by raising the cutoff to 0.29 kU/l. The correlation between CRD and the outcomes of peanut challenges, such as clinical thresholds and severity of symptoms, has, to our knowledge, not been established. In order to benefit from the full clinical potential of peanut components, this needs to be performed. The purpose of this study was to correlate absolute values of IgE to peanut components with challenge-verified clinical data, that is, outcome of challenges, thresholds, and symptoms.
Data were evaluated from patients routinely investigated for peanut allergy, that is, referred with a positive case history ranging from mild to very severe reactions, during the period from March 2003 to March 2009 at the Allergy Centre, Odense University Hospital, Denmark. Final inclusion criterion was a positive or negative oral peanut challenge combined with available sensitization data, that is, SPT, s-IgE, and CRD, obtained within a year from the challenge date. In case of repetitive challenges in a patient, the initial challenge was used. Routinely, when blood was drawn for clinical purposes, an extra blood sample was stored, with written permission from patients or parents. All patients were evaluated according to routine standards in the department.
All patients were challenged according to European Academy of Allergy and Clinical Immunology (EAACI) guidelines  by trained staff. Children ≤3 years of age were exposed to foods by standardized open food challenge (OFC), whereas the DBPCFC was planned in all participants older than 3 years of age. Challenges were performed with whole roasted unsalted peanuts and were in case of a DBPCFC masked in a chocolate bar matrix. Challenge regimen and administrated doses have previously been described . Symptom severity was classified into 5 groups according to Sampson et al. , ranging from localized urticaria, nausea, abdominal pain, pruritus, OAS (Group 1), generalized urticaria, angioedema and emesis (Group 2), repetitive vomiting and rhinorrhea (Group 3), diarrhea and asthma (Group 4) to anaphylaxis (Group 5).
SPT was performed on the volar surface of the forearm with a 1-mm lancet according to EAACI guidelines [16, 17] applying skin ‘prick-prick’ technique with commercially obtained unsalted, roasted peanut. A positive SPT was defined as a mean wheal diameter ≥3 mm. Serum IgE was analyzed by ImmunoCAP (Thermo Fisher Scientific, Uppsala, Sweden). This included total IgE, specific IgE toward peanut (f13) and peanut component Ara h 1, Ara h 2, Ara h 3, Ara h 8, and Ara h 9. Analyses were performed in accordance with the manufacturer's instructions.
Distribution of thresholds and estimation of eliciting dose in 10% (ED10) were modulated according to previous publication [4, 18]. Differences in frequency of sensitized challenged-positive and challenged-negative patients were tested with chi-square and comparison of mean levels of IgE between the two groups was made using a Wilcoxon–Mann–Whitney test. Spearman's rank correlation coefficient (ρs) was used to correlate components and thresholds. Best cutoff was found using receiver operating characteristic (ROC) curves. Symptom score was correlated with age and threshold using ordinal logistic regression. All calculations were made in Stata version 10 (Stata corp, Collage Station, TX, USA) on a windows 7 platform.
A total of 205 patients were included in this study: 165 had OFCs and 40 had DBPCFCs. Mean age of patients was 5.6 years [range 1–26 y], with a male predominance (sex ratio = 1.7). Thirty challenges were negative and 175 were positive; 158 of these resulted in objective signs. Thresholds for 133 of the 175 positive challenges have previously been described . The estimated eliciting dose in 10% of peanut-allergic patients (ED10) in this study was 108 (77.1–150.7) mg peanut, which is within the previously reported CI (95.9–186.6 mg peanut).
IgE correlation with thresholds
Patients with a positive challenge had significantly more frequently IgE levels >0.35 kU/l toward Ara h 2 and Ara h 3 than patients with a negative challenge, whereas no statistical differences were detected for specific IgE to whole peanut (f13), Ara h 1, Ara h 8, and Ara h 9 (Fig. 1). Mean IgE level for the positive challenges was significantly higher for f13, Ara h 1, Ara h 2, and Ara h 3, compared to those with a negative challenge (Fig. 1). All 3 major components, Ara h 1, Ara h 2, and Ara h 3, correlated well with f13 (ρs = 0.79–0.88; P < 0.01) (Table 1); best correlation was found with Ara h 2. Levels of Ara h 8 and Ara h 9 showed no correlation with f13. The best correlation between thresholds leading to objective signs and IgE was found for Ara h 2 (ρs = −0.30, P < 0.01) and f13 (ρs = −0.29, P < 0.01) (Table 1); however, this correlation was mainly driven by low-sensitive patients, that is, patients with high thresholds. Thresholds above 1g peanut correlated significantly with low IgE levels (ρs = −0.58 to −0.41, P = 0.01), whereas no correlation with IgE levels was found in sensitive patients with low thresholds.
|Correlation with f13 (n = 205)||Correlation with objective threshold (n = 158)||ROC curve analysis||If specificity = 1.00||Correlation with symptoms (n = 158)|
|ρ (Rs)||P||ρ (Rs)||P||Best cutoff||AUC||sens||Spec||Cutoff >||sens||ρ (Rs)||P|
|Ara h 1||0.79||<0.01||−0.28||<0.01||2.23||0.64||0.41||0.97||8.32||0.30||0.36||<0.01|
|Ara h 2||0.88||<0.01||−0.35||<0.01||1.28||0.90||0.76||0.97||1.63||0.72||0.60||<0.01|
|Ara h 3||0.82||<0.01||−0.24||<0.01||1.13||0.65||0.27||0.97||11.00||0.08||0.35||<0.01|
|Ara h 8||0.20||<0.01||0.08||0.32||0.80||0.43||0.27||0.67||54.2||0.01||−0.02||0.74|
|Ara h 9||0.34||<0.01||−0.04||0.60||0.01||0.44||0.64||0.30||14.8||0.01||0.04||0.54|
The optimal cutoff points for the outcome of challenges, that is, predicting positive vs negative challenges, were determined by ROC analysis (Fig. 2) and were estimated to be 2.6 kU/l for f13 and 1.28 kU/l for Ara h 2 (marked by solid lines in Fig. 3). Sensitivity using these cutoffs was 0.76 for both f13 and Ara h 2, whereas specificity was higher for Ara h 2 than for IgE to whole peanut (f13) (0.97 vs 0.80). The cutoff yielding a specificity = 1.00 was for f13 > 16.7 kU/l and for Ara h 2 > 1.63 (marked by a dashed line in Fig. 3). Sensitivity corresponding to a specificity of 1.00 was 0.43 for f13 and 0.70 for Ara h 2. Neither Ara h 1, Ara h 3, Ara h 8, nor Ara h 9 demonstrated better performance characteristics than Ara h 2 (Fig. 2). Combinations in which there was addition or extraction of IgE values from any of the three major components, for example, Ara h 1 + Ara h 2 + Ara h 3, did not improve accuracy compared to Ara h 2.
Severity of symptom in 133 of the 175 patients with a positive challenge has previously been described , but has not been correlated with IgE levels to any components. Symptom severity was classified into 5 groups  and correlated significantly with increasing age (ρ = 0.18, P = 0.03), but not with thresholds. For the group as whole, an increasing symptom score was correlated with increasing levels of IgE to whole peanut (f13) as well as with all 3 major components (Fig. 4); best correlation with symptom score was found for Ara h 2 (ρs = 0.60, P < 0.0001 (Table 1).
This study is, so far, the largest study publishing clinical data on the relationship between levels of IgE components and outcomes of peanut challenges. We showed that IgE to Ara h 2 is a superior predictor for the determination of the outcome of challenge than the other single components including s-IgE to whole peanut (f13). ROC curve analysis determined the optimal decision-point separating positive from negative challenges to be 1.28 kU/l, whereas a cutoff at 1.63 kU/l would yield a specificity of 1.00, but still retain a high sensitivity (0.70). Correlation between threshold and severity was poor.
Replacement for oral challenge
Different cutoff points for SPT and s-IgE have been presented in a number of studies [6-10], but so far either a low sensitivity or low specificity has reduced the reliability as replacement for oral challenges and thereby the clinical impact. Introduction of purified and recombinant peanut allergens and the usage of component-resolved diagnostics (CRD) have been reported as more useful compared to s-IgE toward whole peanut protein (f13) . In line with previous findings [13, 14], we found that Ara h 2 was the best predictor of outcomes of a challenge, that is, positive versus negative. One reason for the previously published low specificity of SPT and IgE is the cross-reactivity found between many members of the legume family, for example driven by Ara h 8, which is the pathogenesis-related protein family 10 (PR-10) homologue to Bet v1 in birch and is known to cause cross-sensitization. We found, in line with previous findings , that patients with a negative peanut challenge were more frequently sensitized to Ara h 8 (Fig. 1), to birch pollen, and to the major component in birch pollen (Bet v1) than peanut-allergic patients (data not shown). This higher frequency was not reflected in the level (kU/l) of birch sensitization, where no difference was detected. No difference was found in the frequency nor in the level of grass sensitization between patients challenged positive or negative with peanut.
IgE toward peanut protein components has been correlated with outcome of peanut challenges in two studies. Codreanu et al.  showed in 237 patients that a cutoff of Ara h 2 > 0.23 would be optimal to separate peanut allergic from those who were tolerant. This cutoff yielded a sensitivity/specificity of 0.96/0.93. To achieve a sensitivity = 1.00 in their study, a cutoff for Ara h 2 > 0.29 kU/l should be applied. In another study by Nicolaou , CRD in 81 patients were correlated with peanut allergy; oral challenges were, however, not performed systematically, and only 7 of 29 assumed positive were challenge-verified. They reported that a level of Ara h 2 > 0.35 kU/l would yield a sensitivity of 1.00 with a corresponding specificity of 0.96, whereas a cutoff > 0.55 kU/l yielded a sensitivity = 0.93 and specificity = 1.00.
The decision-points presented in these two papers are considerably lower than that found in our study. It has been shown that the pattern of cross-reactivity and thereby the pattern of peanut sensitization differ geographically [21, 22]. Therefore, it is not unexpected that these differences are also reflected in component-resolved diagnosis – again stressing the importance of taking regional differences into account when trying to establish decision-points from published data from other regions of the world. Finally, case history and s-IgE > 15 kU/l were used to establish a diagnosis of peanut allergy in the English study . In our study, we found 1 patient with s-IgE (f13) >15 kU/l and in total 3 patients with f13 > 11 kU/l, all with negative peanut challenges. All three patients had a case history of peanut reactions and would mistakenly have had a diagnosis of peanut allergy, if they were diagnosed on serology and case history solely. Only oral provocation could identify that these patients had outgrown their clinical peanut allergy. A false-positive diagnoses based on high level of s-IgE to f13 could therefore lead to selection bias, which due to the close relationship between levels f13 and Ara h 2 (ρs = 0.88) eventually could lower the decision-points for Ara h 2.
By using the cutoff of Ara h 2 > 1.63 kU/l to identify patients with definite clinical peanut allergy, we would be able to avoid challenges in 55% of the cases in our center. For the remaining 45% (92 patients with Ara h2 < 1.63 kU/l), we were unable to predict outcomes of challenges, even if IgE levels of other components such as Ara h 1 or Ara h 3 were included.
Threshold and symptom score
We found no relationship between threshold and level of IgE to components. Seen as a group, thresholds correlated significantly with IgE values for f13 and the 3 major allergens Ara h 1, Ara h 2, and Ara h 3 (Table 1). This was driven by patients with high threshold (>1000 mg), who had low IgE values. No significant correlation was found in patients with lower thresholds (<1000 mg) and their IgE values. Symptoms were classified into 5 groups . For the group with a positive challenge (n = 175), higher levels of IgE meant more severe symptoms (Fig. 2A), but large individual variation was found. Levels of IgE to Ara h 2 as well as other components can thus not replace an oral challenge neither in determining threshold nor in the assessment of severity of symptoms appearing at threshold during challenge.
In the study by Codreanu et al. , 4% of the positive challenges have undetectable levels of Ara h 2 (0.00), indicating the existence of different phenotypes; we also found 3 patients (2%) with an undetectable levels of measurable Ara h2, but with positive challenges. The clinical impact of this finding remains unknown.
Introduction of measurement of IgE levels to purified peanut components, especially Ara h 2, and the usage of CRDs can be an advantage in the clinic. The level of IgE toward Ara h 2 can improve diagnosing by introducing a more clear-cut decision-point with a specificity of 1.00 and still maintaining a high sensitivity (> 0.70). In our study, a cutoff for Ara h 2 > 1.63 kU/l would have reduced the number of challenges performed from 205 to 92. However, decision-points for CRD are age and geographically related to a study population, making them difficult to compare between centers. Ideally, they should be established at each center for their relevant patient groups. Further decision-points even using CRDs cannot replace oral challenge neither in determining thresholds nor in the assessment of the severity of symptoms and signs after the ingestion of a dose similar to threshold.
We would like to thank Thermo Fisher Scientific, Sweden, for supporting this study by performing ImmunoCAP analysis for all patients, and Jodie Urcioli for help and revision.
Conflict of interest
Both authors have received lecture honoraria from Thermo Fisher Diagnostics. E. Eller has additionally received travel grants and C. Bindslev-Jensen has received financial support for other projects.
- 17Position paper: allergen standardization and skin tests. Allergy 1993;48(Suppl 14):48–82., , , , , et al.