This paper presents a protocol for detecting serum-specific immunoglobulin E with a microfluidic cartridge-based chemiluminescence system and the evaluation of its usefulness in allergy diagnoses.
Allergic disease is common in both adults and children. Identification of the causative allergens is significant in disease management and prevention. However, a specific immunoglobulin E (IgE) measurement system with a high price-performance ratio is lacking in mainland China, especially in the primary care hospitals. This paper describes the principle and operation procedures of using a microfluidic cartridge-based chemiluminescence system to detect allergen-specific IgE in serum. The results were compared with those from ImmunoCAP (System 1), the industrial standard, and the reproducibility of the system to detect patients sensitized to common allergens is evaluated. The results showed that in comparison with ImmunoCAP (System 1), the BioIC System (System 2) has good precision and sensitivity in detecting serum-specific IgE against various inhalant and food allergens but with a significantly lower cost. It can serve as a good alternative to System 1 in primary care hospitals in mainland China who have lower financial affordability.
The prevalence of allergies has increased steadily in the past decades and is affecting 20%-30% of the global population1. Identification of the causative allergens has important significance in managing the diseases. In China, since registered in vivo skin prick tests are not available in the country, in vitro determination of serum-specific IgE is the most important and commonly used tool for type I allergy diagnoses2. This is similar to practices in the Western world, but although ImmunoCAP system (System 1), a fluorescence enzyme-linked immunosorbent base system, is perceived as the gold standard for in vitro allergy diagnosis3, its usage in China is very limited due to its high equipment and reagent price. Hence a new alternative allergy diagnosis system with a high price-performance ratio is badly needed.
The BioIC System (System 2) is a microfluidic cartridge-based system based on the chemiluminescence principle for multiplexed assays of serum-specific IgE. With a size of 7 cm x 4 cm, the microfluidic cartridge is composed of three layers of injection-molded plastic. The upper part is 3 mm-thick transparent polycarbonate which carries good stability during thermal assembly processes. Together with the 3 mm-thick bottom layer constructed from a copolymer of acrylonitrile, butadiene, and styrene (ABS), it sandwiches the 0.5 mm-thick middle layer made of silicone rubber. Being black in color, the middle layer offers lower background during chemiluminescence detection. On top of the silica gel, a thin layer of nitrocellulose membrane (NC membrane) is sprayed at the position corresponding to the reaction zone, which allows the spotting of different allergenic proteins. The purpose of this study is to evaluate the clinical performance of the microfluidic system for the multiplexed determination of allergen-specific IgE in serum.
This study and the use of human serum samples were approved by the Ethics Committee of The First Affiliated Hospital of Guangzhou Medical University (GYYY-2016-73). All participants have given their written consent independently or through their parents (in the case of children).
1. Basic information of the study group
NOTE: The Allergy Information Repository of the State Key Laboratory of Respiratory Disease (AIR-SKLRD) is a large serum bank established inside the Guangzhou Institute of Respiratory Hospital (GIRH). Initiated in the last decade, the AIR-SKLRD has already started to collect and store serum samples from patients with allergic diseases, together with their clinical information (Table 1)4,5. The current study was performed with sera from the AIR-SKLRD.
2. Study flow and measurements of interest
NOTE: The microfluidic system needs 100 µL of serum for determining 19 allergens. Venous blood (5 mL) was collected from each patient using a vacuum blood vessel containing separating gel. After centrifuging for 10 min at 1,000 x g, the upper layer was collected for testing. Unused serum was stored at -80 °C. Prior to testing, the serum was kept at room temperature for 30 min and was shaken with a vortex mixer. Repeated freeze-and-thaw cycles were avoided.
3. Semi-automation test procedure of the microfluidic system
NOTE: The System 2 is the integration of automatic microfluidic technology, protein microarray, cold light analysis, parallel IgE analysis, and image processing technology. The testing protocol is divided into four parts: preparation of the equipment, sample loading, incubation, and measurement.
4. Definition of sIgE reactivity
NOTE: For an undiluted serum specimen, the detection range of the System 2 is 0.21–100 IU/mL.
5. Statistical analysis
Positive rates for 19 common allergens
The results on 293 sera are shown in Figure 3. Among all the inhalant allergens, D. farinae had the highest positive rate (80.89%, 273/293), followed by D. pteronyssinus (78.84%, 231/293). Among the food allergens, crab has the highest positive rate (20.48%, 60/293), followed by 13.65% (40/293) for shrimp. The total positive rate for inhalant allergens was higher than that for food allergens.
Repeatability of the microfluidic system
Repeatability results for cat dander, dog dander, and cockroach, based on nine rounds of testing, were 32.98 ± 8.94, 1.61 ± 0.48, and 0.76 ± 0.18, respectively, and the consistency levels were 100% (9/9), 100% (9/9), and 67% (6/9). Distribution of the results is shown with the Levey-Jennings curve in Figure 2. All data are within the range of X ± 2 x SD, which is consistent with the maximum allowable clinical error14.
Comparison of two systems
Qualitative results showed that cat dander had the highest concordance to System 1 (95.33%, 243/150). The lowest concordance was seen in shrimp (40.75%, 88/216). The total concordance among inhalant allergens ranged from 92.00% to 95.33%. For food allergens, the concordance range was 40.74%–72.39%. The highest sensitivity for inhalants was seen in Dermatophagoides farinae (93.94%), with a 100% specificity. Among food allergens, the highest sensitivity was seen in peanut (54.55%), with a specificity of 80.65%. Table 2 also shows that all the evaluation results for inhalant allergens were superior to food allergens. Since the AUC values showed a range from 0.613 to 0.984 and the AUC for the three inhalant allergens was greater than 0.950, it can be concluded that the System 2 has a high accuracy with reference to System 1.
Consistency analysis for the two systems showed that the kappa values for the three inhalants were between 0.727–0.876, with the highest value seen in cat dander as 0.876 (95% CI, 0.786–0.965). They were all better than the kappa values for food allergens which, in general, fell <0.400. The lowest kappa value was 0.112 in shrimp (95% CI, 0.062–0.162) (Table 3). Spearman’s correlation analysis showed that the best correlation was seen in peanut and cat dander, with correlation coefficients as r = 0.942 (95% CI, 0.907–0.965; p < 0.0001) and r = 0.927 (95% CI, 0.900–0.947; p < 0.0001), respectively.
In Figure 3, a scatter plot is constructed with System 2’s results along the x-axis and System 1’s along the y-axis to show the distribution of the sIgE concentration results from the two systems for D. pteronyssinus, D. farina, cat dander, milk, shrimp, and peanut. For a concordance and discordance analysis, the allergens that showed ± 1 class difference were D. pteronyssinus (91.60%, 229 vs. 250), D. farinae (81.25%, 91 vs. 112), cat dander (98.00%, 147 vs. 150), milk (83.58%, 112 vs. 134), shrimp (59.72%, 129 vs. 216), and peanut (76.56%, 49 vs. 64). The combined total concordance rate was 81.75% (757 vs. 926).
Figure 1: The positivity rates of the detection of 19 common allergens by the microfluidic assay. d1 - Dermatophagoides pteronyssinus, d2 – Dermatophagoides farinae, d201 – Blomia tropicalis, e1 – cat dander, e5 – dog dander, g2 – Bermuda grass, g6 – timothy grass, i6 – cockroaches, m3 – Aspergillus fumigatus, m5 – Candida albicans, w1 – ragweed, f1 – egg white, f2 – milk, f4 – wheat, f13 – peanut, f14 – soybean, f20 – almond, f23 – crab, and f24 – shrimp. Please click here to view a larger version of this figure.
Figure 2: Levey-Jennings graphs of the three allergens repeatedly detected by the microfluidic system. (A) Cat hair, (B) dog hair, and (C) cockroach were selected for repeatability evaluation. The black, green, yellow, and red lines represent the mean (X), the mean ± the standard deviation (X ± SD), the mean + the standard deviation times two (X ± 2SD), and the mean + the standard deviation times three (X ± 3SD) of multiple measurements, respectively. Please click here to view a larger version of this figure.
Figure 3: Scatter plots of six allergen sIgE concentrations measured by System 1. System 1 (Y-axis) and the System 2 (X-axis). Each line in the plot represents class cutoffs (class 0: 0.35, class 1: 0.35–0.7, class 2: 0.7–3.5, class 3: 3.5–17.5, class 4: 17.5–50, class 5: 50–100, and class 6: >100 IU/mL). Shaded boxes are concordant areas in the concentration class. Please click here to view a larger version of this figure.
Characteristic | No.(%) |
Gender, n(%) | |
Female | 123(41.98%) |
male | 170(58.02%) |
Age, year, n(%) | |
Median (25%,75%) | 23(8,36) |
≤10 | 97(33.11%) |
11-20 | 37(12.63%) |
21-40 | 101(34.47%) |
>41 | 58(19.80%) |
Diagnosis, n(%) | |
Allergic rhinitis | 92(31.40%) |
Allergic asthma | 117(39.93%) |
Allergic rhinitis with asthma | 36(12.29%) |
Others | 48(16.38%) |
Table 1: Patient demographic characteristics. In total, 293 subjects were found who fulfilled the inclusion criteria, with an median age of 23 (interquartile range from 8 to 36 years old). Among them, 170 (58.02%) were male and 123 (41.98%) were female. Also, 92 (31.40%) of them had allergic rhinitis, 117 (39.93%) had allergic asthma, 36 (12.29%) had comorbidity of rhinitis and asthma, and 48 (16.38%) had other allergic diseases, such as a food allergy or skin allergies.
Sample size | CAP + | CAP – | Total agreement | SE | SP | PPV | NPV | AUC(95%, CI) | |||
BioIC + | BioIC – | BioIC + | BioIC – | ||||||||
d1 | 250 | 196 | 20 | 0 | 34 | 92.00% | 90.74% | 100.00% | 100.00% | 62.96% | 0.975(0.947 to 0.991) |
d2 | 112 | 93 | 6 | 0 | 13 | 94.64% | 93.94% | 100.00% | 100.00% | 68.42% | 0.984(0.941 to 0.999) |
e1 | 150 | 34 | 5 | 2 | 109 | 95.33% | 87.18% | 98.20% | 94.44% | 95.61% | 0.968(0.925 to 0.990) |
f2 | 134 | 16 | 27 | 10 | 81 | 72.39% | 37.21% | 89.01% | 61.54% | 75.00% | 0.744(0.661 to 0.815) |
f13 | 64 | 18 | 15 | 6 | 25 | 67.19% | 54.55% | 80.65% | 75.00% | 62.50% | 0.731(0.606 to 0.834) |
f24 | 216 | 36 | 127 | 1 | 52 | 40.74% | 22.09% | 98.11% | 97.30% | 29.05% | 0.613(0.545 to 0.678) |
d1-Der. p1, d2- Der. f1, e1-Cat dander, f2-Milk, f13-Peanut, f24-Shrimp. CAP-ImmunoCAP, +-positive, –-negative, SE-sensitivity, SP-specificity, PPV-positive predictive value, NPV-negative predictive value, AUC-area under the ROC curve. For AUC values, the 95% interval value (95%, CI) is also shown in the table. |
Table 2: Clinical performance between the two systems. d1 - D. pteronyssinus, d2 – D. farina, e1 – cat dander, f2 – milk, f13 – peanut, and f24 – shrimp. CAP – ImmunoCAP, + – positive, – – negative, SE – sensitivity, SP – specificity, PPV – positive predictive value, NPV – negative predictive value, AUC – area under the ROC curve. For the AUC values, the 95% interval value (95%, CI) is also shown in the table.
Kappa(95%,CI) | Spearman'rho(95%,CI) | |
d1 | 0.727(0.617 to 0.838) | 0.896(0.869 to 0.918) |
d2 | 0.783(0.617 to 0.948) | 0.731(0.631 to 0.807) |
e1 | 0.876(0.786 to 0.965) | 0.927(0.900 to 0.947) |
f2 | 0.293(0.122 to 0.463) | 0.681(0.579 to 0.763) |
f13 | 0.349(0.129 to 0.569) | 0.969(0.949 to 0.981) |
f24 | 0.112(0.062 to 0.162) | 0.833(0.788 to 0.870) |
Table 3: Correlation and agreement between the two systems. d1 - D. pteronyssinus, d2 – D. farina, e1 – cat dander, f2 – milk, f13 – peanut, and f24 – shrimp. For kappa and Spearman’s Rho values, the 95% interval value (95%, CI) is also shown in the table.
Similar to the results from many other studies15,16,17, the results from the microfluidic system based on sera from 293 allergic patients showed that house dust mites (including D. pteronyssinus, D. farinae, and B. tropicalis) are the main inhalant allergens leading to allergic diseases in southern China, whereas for food, milk, peanut, shrimp, and crab are the commonest allergens that cause allergic symptoms. With regard to the reproducibility study done on three allergens, all of them showed good results, with an overall repetition rate of 88.89%, which means it met the maximum allowable error.
Using System 1 as reference, the current study evaluated the clinical diagnosis efficacy of the System 2. With a serum sIgE level of 0.35 IU/mL as cutoff2, a sample with sIgE > 0.35 IU/mL implies that the patient is sensitive to the allergen, and the higher the titer, the better correlation to the patient’s symptoms18. The results show that the concordance rate of the three inhalant allergens were all more than 90%. In addition to the results for food allergens, which had a concordance of ~40.74%-72.39%, the total concordance was 81.75% (757/926). The kappa value of D. pteronyssinus, D. farina, and cat dander were 0.778, 0.663, and 0.860 (p < 0.001), respectively. The kappa values for food allergens were all below 0.4. For the three major inhalant and food allergens, a significant correlation of quantitative results was seen between the two systems (rSpearman ≈ 0.681-0.969, p < 0.01).
It was noticed that while the rS coefficient for peanut was 0.969, the kappa value of the consistency evaluation index was only 0.349 (95% CI, 0.129-0.569). Such discrepancy might be due to the low prevalence of peanut sensitivity in the region and, hence, most of the recruited sera were negative for that specific IgE. Many studies have indicated that a significant discrepancy could be seen between sIgE titer and clinical symptoms of food allergens. The use of different assay systems for food-specific IgE determination could also create big variations19. This might be due to the fact it is not the raw ingested food which triggers the allergic symptoms, but the modified components generated during cooking or digestion. The use of different raw materials by different manufacturers to make the allergens can also contribute to the result discrepancy20.
The microfluidic cartridge is composed of five major parts: five storage tanks, five reagent delivery channels, five unidirectional pumps, a single reaction zone in which allergen extracts can be immobilized, and a waste tank to collect all reaction by-products. Based on the assay need, up to 40 allergen extracts can be dotted on the reaction zone. Controlled by the PC, the five unidirectional pumps guide and coordinate the flow of serum sample, washing buffer, blocking reagent, conjugates, and subtracts, to finish a two-step enzyme-linked immunosorbent assay. After the reaction is completed, the chemiluminescence reaction images are captured by a low-resolution cooled CCD camera and the signals are processed by the PC to establish the calibration curve and to calculate quantitative and semiquantitative sIgE results.
The current study shows that the two systems demonstrate good consistency. However, compared with the System 1, the System 2 is easier to use and has a lower demand for operator training. Since each microfluidic cartridge has its own quality control curve, the reliability of the system is greatly enhanced. Other advantages of the system include a light and small footprint, an expandable modular setup, and the ease with which it is connected to a PC for operation control. All these advantages greatly reduce the setup and running cost, and at the same time, they do not jeopardize the accuracy and speed requirements in daily clinical practice, which makes the system particularly suitable for allergy screening in primary care hospitals in China. Nevertheless, one major drawback of the microfluidic system is that it is not a fully automatic system, and frequent intervention is needed during the operation. It still cannot replace the systems that need to process a big number of samples daily.
Due to the lack of enough positive sera for certain allergens, this study did not cover all the 19 allergens available in the microfluidic cartridge, but just six common ones available in southern China. More study is needed to elaborate on whether the evaluation is also applicable to other allergens.
The authors have nothing to disclose.
The authors thank Professor Mei Jiang for her help in the statistical analysis and Mr. Hammer Tsui in the preparation of the manuscript. This study was supported by the Guangzhou Science and Technology Foundation (201804020043) and the National Natural Science Foundation of China (NSFC 81572063 and NSFC 81802076). The funding groups agreed with the study design, data analysis, manuscript preparation, and decision to publish. No other funding was received for this study.
Agnitio BioIC Analyzer | Agnitio Science & Technology(Taiwan, China) | BA-G2000 | |
BioIC Allergen specific-IgE Detection Kit | Agnitio Science & Technology(Taiwan, China) | DR17A12 | |
BioIC Cartrideg Placement plate | Agnitio Science & Technology(Taiwan, China) | T20SET | |
BioIC Reagent Dispenser | Agnitio Science & Technology(Taiwan, China) | DS-1 | |
Image two-dimensional barcode machine | Agnitio Science & Technology(Taiwan, China) | NLS-HR200 | |
Software Package, LabIT | Agnitio Science & Technology(Taiwan, China) | Version 2.4.12 | |
VORTEX-5 Vortex Mixer | Haimen Kylinbell Lab Lastruments Co., Ltd. | VORTEX-5 |