A systematic search strategy was adopted, in order to summarize the currently available literature. PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) searches were carried out using search terms basophil activation in titles and/or abstracts, for the period from January 1990 to August 2013. A manual search, using the same keywords, in Google Scholar (http://scholar.google.com/) was performed to identify additional papers. The search process followed the recommendations of the PRISMA statement (Fig. 1) [8], and was confined to articles with full-text accessibility. The present review includes analyses from 74 relevant papers, including original articles and case reports.

Conventionally, diagnoses of beta-lactam antibiotic hypersensitivities have been based on patient's clinical history and positive skin tests, or specific IgE antibody measurements [9]. To date, nine studies [10-18] have described the utility of BAT for diagnoses of beta-lactam allergies (Table 1). The sensitivities ranged from 28.6% to 55%; however, several large-scale studies have consistently demonstrated the sensitivity to be approximately 50%, in patients with positive clinical history and skin tests. Interestingly, the sensitivity of BAT was approximately 10% higher than that of the commercial specific IgE tests [14, 17, 18], and the specificity was more than 90%, clearly indicating that a positive BAT result was clinically significant. Importantly, BAT was positive in 25% of patients with positive provocation test and negative for specific IgE [17], and in 37% of patients with positive clinical history but negative skin tests [14]. These results suggest that BAT should be administered in cases where the diagnosis of drug allergy is highly suspected but is not supported by results of skin testing or in vitro IgE measurements. Because specific IgE tests are not available for most cephalosporins, BAT can be developed further for diagnosing allergies to a wider range of beta-lactams [9].

Currently, data for evaluating BAT results from patients with a history of perioperative hypersensitivity are available from seven clinical trials [19-25]. The sensitivity of BAT varied from 36.1% to 91.7% (Table 2); however, there was considerable heterogeneity in the inclusion criteria and cutoff levels. In patients with proven NMBA anaphylaxis, the BAT sensitivity was primarily 36.1%, which increased to 85.7% when allergies with an onset of less than 3 years were separately considered [21]. In the same patients, BAT showed high correlations with skin prick tests [20, 23, 26], better sensitivity [23], and higher specificity (range, 93% to 100%). Therefore, the time elapsed between the anaphylaxis and in vitro basophil activation [21] is a significant parameter for analyzing BAT sensitivity. In addition, BAT also plays an important complementary role in identifying cross-reactivity and safe alternatives in these patients [19-21, 23, 27].

Aspirin or NSAIDs hypersensitivity is a heterogeneous disorder, encompassing IgE-mediated allergic reactions and non-immunological intolerances. The results with BAT on aspirin/NSAIDs hypersensitivity are conflicting or inconclusive (Table 3) [28-40]. Aspirin intolerance is mediated by the pharmacological effects on cyclooxygenase enzyme inhibition; therefore, it may not be a usual indication for BAT. It was discovered that BAT was not useful in patients with mild or cutaneous reactions, but it could only be indicated for severe reactions [30, 31]. In patients with aspirin intolerance, the combination of CD63 and CD203c measurements did not enhance the test sensitivity, which remained at 33.3% [35]. De Weck et al. [41] have questioned the proper interpretation on two earlier positive reports [38, 39]. Release of tryptase and histamine in response to oral challenges with aspirin suggested that circulating basophils play a role in aspirin intolerance [42]. However, these relationships are dose-dependent and likely to be mediated by the pharmacological inhibition of synthesis of prostaglandin E2, a natural inhibitor of basophil activation [41]. Therefore, BAT in aspirin intolerance may have to be sophisticated further to enhance the differences in dose responses between patients and controls. As diclofenac and naproxen have stronger in vitro pharmacological activity than aspirin, their inclusion has been suggested for enhancing the sensitivity of BAT [41].

Specific allergy to dipyrone has been evaluated by BAT [28, 33, 40]. Sensitivity and specificity ranged from 42.3% to 70% and 85.7% to 100%, respectively, depending on the cutoff values. A propyphenazone allergy case, which was diagnosed by BAT after human serum albumin (HSA) conjugation, has been previously reported [43]. However, in patients with selective diclofenac allergies, either diclofenac- or HSA-conjugated metabolites did not trigger CD63 expression [32].

Fluoroquinolones, in addition to beta-lactams, cause one of the most common antibiotic allergies, and this hypersensitivity has become increasingly common with increased prescription rates of the drug [44]. BAT has gained considerable interest for testing fluoroquinolone hypersensitivities because the diagnostic utility of skin tests is very limited due to its skin-irritation properties in intradermal tests (88% false positives) [45]. To date, seven studies [46-52] have investigated the diagnostic utility of BAT (Table 4). The first study reported no positive BAT results in four DPT-proven patients [52]. Similarly, negative findings were reported in another study (n = 4, 0% positivity) [50]; however, larger scale studies performed later contradicted these findings. Another group discovered 70%-83% up-regulation of CD203c upon drug stimulation in all five participants with a history of anaphylaxis [51]. Other studies confirmed these findings by showing 71.1% sensitivity in 38 patients [49], and 36% sensitivity in 66 patients [47]. The excellent negative predictive value for DPT outcomes advocates the high utility of BAT in patients with suspected history of fluoroquinolone hypersensitivity [48].

RCM hypersensitivity is a commonly encountered adverse drug reactions, and is the most common cause for anaphylaxis at a referral hospital in Korea [53]. Despite the introduction of non-ionic contrast media, the incidence of immediate hypersensitivity and severe reactions still appear as frequent as 2.1% and 0.01% per exposure, respectively [54]. Although skin testing is a relevant diagnostic method to determine the cause of hypersensitivity, it was meaningful only among patients with a history of moderate to severe hypersensitivity (40% positive in intradermal tests) [55]. Moreover, skin testing cannot detect non-IgE mediated RCM reactions.

Several studies [56-59] so far have analyzed the diagnostic value of RCM BAT (Table 5). Initial studies by Pinnobphun et al. [57] found the sensitivity to be 46.2%-61.5% and specificity 88.4%-100%, depending on the cutoff values. Recent studies reported the BAT sensitivity to be 62.5% compared to the outcome from intravenous challenges (n = 8), thereby confirming previous findings [56]. Interestingly, the skin test positivity did not correlate with BAT results, and BAT positivity did not correlate with the severity of reactions [57]. These findings suggest complementary roles for BAT in the diagnosis of RCM hypersensitivity. Further studies are necessary to understand its negative predictive values and to identify the precise mechanism for predicting safe alternative RCM in high-risk patients.

Recent studies [60-64] examined the outcome of BAT in patients with hypersensitivities to antineoplastic, biologic agents, or other drugs (Table 6). L-Asparaginase allergies were assessed using CD203c expression and were found to have high sensitivity (75%) and negative predictive value (96%) [60]. One case study also reported the potential utility of BAT in cisplatin hypersensitivity [65]. Because patients with malignancies may frequently have comorbidities or conditions that hamper skin testing, administering BAT will be advantageous in these cases.

Hypersensitivity to other biologic agents such as rituximab [61] or infliximab [66] were examined by BAT, although the results warrant further confirmation. Among corticosteroids, methylprednisolone [62, 67] and succinylated corticosteroids [68-70] have been tested. Hypersensitivity to anti-histamines such as cetirizine, desloratadine, ebastine, fexofenadine, or dexchlorpheniramine was also assessed by BAT [71-74]. Other reports included testing for pholcodine [75], glatiramer [63], gelofusine [64], amidotrizoate [76], pristinamycin [77], enoxaparin [78], heparin [79], afloqualone [80], cremophor EL [81], hydrochlorothiazide [82, 83], polyoxyethylene (20) sorbitan monooleate [84, 85], chlorhexidine [86], ophthalmic atropine [87], and carboxymethylcellulose [88] in allergic or non-immunologic adverse reactions (Table 7) [13, 27, 61, 66-93]. Further studies are required for determining the causal relationships and identifying safe alternatives in patients with hypersensitivity to drugs that are not evaluated until date.

The preparation of drugs and their dose determination is one of the most challenging steps of BAT because they have a narrower range of testing concentrations than inhalant or food allergens [98]. Several varieties of drug allergens are commercially available, but they are expensive, and selection is frequently a limiting factor. In the case of drugs that are not commercially available, dose response curve analyses and cytotoxicity assays are mandatory steps for determination of optimal concentrations [100]. In this section, we have summarized the methods and results from previous studies, as a reference point. Higher drug concentrations can be used for diagnostic purposes since they provide enhanced sensitivity; however, they should be tested in tolerant controls due to the risk of cellular toxicity and nonspecific basophil activation.

Previous dose-response and cytotoxicity studies provided a range of drug concentrations that can be used for stimulation. Beta-lactams, in general, were reconstituted at 0.01, 0.1, and 1 mg/mL in the dilution buffer [15]; and specifically, benzylpenicillin at 0.4 and 2 mg/mL; penicilloyl-polylysine at 0.005 and 0.025 mg/mL; penicillin minor determinant mixture at 0.1 and 0.5 mg/mL; ampicillin at 0.25 and 1.25 mg/mL [14, 16]; clavulanic acid at 0.156 and 0.625 mg/mL [90]; cefuroxime at 0.83 and 1.2 mg/mL; and cefazolin at 0.16 and 0.4 mg/mL [18]. In the case of amoxicillin, 1.25 mg/mL and a range of 0.25-0.31 mg/mL final concentrations were utilized [11, 13, 14].

Several studies successfully tested varying concentrations of NMBAs, ranging from 1:1000 to 1:10 dilutions [21, 23, 25]. At a dilution of 1:10,000, no significant basophil activation was observed [21]. Other studies have reported 5 × 10² µg/mL NMBA concentration as optimal [20]. However, it should be noted that there might be different optimal concentrations required for stimulation [20, 25].

Aspirin intolerance is usually dose dependent; therefore, the dose determination in this case is extremely critical. According to some functional cytotoxicity studies, only diclofenac showed in vitro cytotoxicity at levels higher than 1.25 mg/mL [38]. The concentrations recommended for stimulation are as follows: aspirin at 0.3, 1.25, and 5 mg/mL; paracetamol at 0.3, 1.25, and 5 mg/mL; dipyrone at 0.6, 5, and 20 mg/mL; and diclofenac at 0.08 and 0.3 mg/mL. Interestingly, high concentrations of aspirin (5 mg/mL) enhanced the sensitivity of the test but also lowered its specificity (to 89.5%). Diclofenac at a high concentration (1.25 mg/mL) resulted in false-positive reactions in 36.8% of controls, but it gave acceptable results at lower concentrations. Naproxen at 5 mg/mL resulted in up-regulation of CD63 in controls, giving rise to increased false positives (85.2%); therefore, it was not routinely recommended for use in BAT. The test concentrations determined by other researchers were sometimes quite low [36] but mostly within the range as for aspirin [29-31, 35].

Fluoroquinolones are known to have skin-irritating properties [45]. Recent studies have reported contrasting but interesting results. Two studies have shown negative BAT results in patients. In the first study, 4 patients were administered 1:10, 1:100, and 1:1,000 dilutions of levofloxacin, moxifloxacin, or ciprofloxacin, ranging from 1.6 to 5 mg/mL parenteral preparations [52]; and in the second study, 6 patients were administered ciprofloxacin at 0.05-0.1 mg/mL, levofloxacin at 0.05-0.1 mg/mL, and moxifloxacin at 0.125-0.25 mg/mL [50]. Aranda et al. [49] were the first to report the dose response analyses for fluoroquinolones in a large group of patients (n = 38), and they have provided an optimal stimulation concentration range (ciprofloxacin at 0.2-2 mg/mL; moxifloxacin at 0.1-0.2 mg/mL; and levofloxacin at 2-4 mg/mL) in their subsequent studies [46, 47].

One important point to note is the potential difference in immunogenicity between fluoroquinolones. Researchers have found that in patients with moxifloxacin hypersensitivity, moxifloxacin was the most frequent culprit drug in vivo [47, 49], but it had a lower sensitivity than ciprofloxacin in inducing basophil activation in vitro [49]. These results demonstrated the cross-reactive nature of fluoroquinolone hypersensitivity, and highlighted the involvement of specific critical factors related to in vitro moxifloxacin allergenicity. Recently it was discovered that moxifloxacin underwent photo-degradation, which critically decreased in vitro basophil responses, thus resulting in lower BAT positivity under light (17.9%) than under dark (35.7%) conditions [46]. In contrast, ciprofloxacin did not have different outcomes between light and dark conditions (both 46.4%). It is not confirmed whether these observations are applicable to other kinds of drugs, but they emphasize the importance of accurate drug preparations for conducting in vitro drug assays.

The effects of a wide range of RCM concentrations, from 10⁰ to 10⁵ dilutions, were first tested on 3 × 10⁵ peripheral blood mononuclear cells [57]. Cell viability was measured by staining for annexin-V, and the optimal dilution of RCM was determined to be 1:10 and 1:100. Later studies confirmed the optimal dilutions for RCM at 1:10 [56].

A sufficient number of well-defined cases and controls are necessary for determining appropriate cutoff points for each drug. Based on these, researchers perform receiver operating characteristic (ROC) curve analyses to locate optimal points. However, the prevalence of drug allergies is low, and drug allergens are more varied than inhalant or food allergens.

The cutoff points are usually based on the percentage of activated basophils, e.g., > 15% above background for inhalant or food allergens, and > 10% above background for latex or hymenoptera venoms [41]. However, in the case of drug allergens, the basophil response is usually lower than that of inhalant or food allergens; therefore, the cutoff is set at > 5% above background, or determined specifically for individual drugs. The stimulation index (defined as the percentage of activated basophils after allergen stimulation per negative control stimulation) of ≥ 2 is additionally adopted, to decrease the chances of false positivity resulting from the low cutoff levels.

Leysen et al. [99] recently summarized several factors that should be considered while carrying out the drug BAT. The maximum recommended time interval between the anaphylactic reaction and its testing was 12 months. Effects of medications such as antihistamines and corticosteroids on in vitro basophil reactivity warranted further studies and should be taken into account while testing. Oral intake of 10 mg desloratadine, an antihistamine, did not influence CD63 expression in basophils upon anti-IgE stimulation, even after 3 h. However, a 30-min in vitro pretreatment of basophils with dimethindene (antihistamine) or prednisolone significantly influenced their activation at concentrations 50-fold higher than the therapeutic level, but not at 10-fold higher concentrations [96].