Results

Patients

In 796/932 cases (85.4%) the primary suspected drug was a penicillin, in 136 (14.6%) a cephalosporin. As observed in many allergy studies before, the sex ratio was balanced in children of up to 10 years, while the rate of female patients steadily increased among the teenagers to finally increase to a ratio of 3 females per 1 male in the adults through all older age groups (Table 1).
A DHR could already be excluded by history in 135/932 patients (14.5%) before even entering the allergy workup. In most cases, the patients had already unintentionally but safely taken a β-lactam again e.g. in the form of a generic drug under a different brand name.
Two patients had a history of severe cutaneous adverse reactions (SJS/TEN) of whom one was patch tested negatively; patients with a clear history for an immediate DHR (< 1h after drug intake): 243/932 (26.1%); for a delayed DHR (> 24h after drug intake) 365/932 (39.2%).

Specific IgE

Drug-specific IgE was determined in all study patients and 40/932 (4.3%) had at least one positive test above the standard cut off threshold level ≥0.35kU/l (Table 2). In more detail, there were 83/2227 (3.7%) positive sIgE tests. Surprisingly, Pencillin V and not Amoxicillin was the most often positive drug (24/479=5.0%; Table 2 and Figure 2a). Measuring sIgE to minor determinants mixture MDM or Cefaclor added little value in the diagnosis of type I allergy to β-lactams.
It is known that the specificity of sIgE to penicillin decreases with higher total IgE levels while sensitivity decreases with very low total IgE levels (23, 24). The calculation of the ratio drug-specific IgE/total IgE had been introduced to overcome the problem with high total IgE. Vultaggio et al. published a threshold of > 0.0022 with a sensitivity of 43%, a specificity of 95%, and a positive predictive value of 93% for the β-lactam ImmunoCAP® assays (25). We found elevated total IgE of ≥100kU/l in 244/917 (26.6%) sera. Hence, we also calculated sIgE/total IgE ratios. Applying the hypothetical 0.0022 cut off would have largely increased the positivity rates to Amoxicilloyl (from 2.8% to 16.3%), Penicilloyl V (from 5.0% to 11.7%), Ampicillin (from 4.0% to 7.5%) and Cefaclor (1.0% to 6.3%) but not to Penicilloyl G and MDM ( grey columns in Supplementary Figure 3).
Another published recommendation is lowering the cut off from ≥0.35kU/l to ≥0.1kU/l (16) to expand the sensitivity of sIgE especially in patients with low total IgE levels. This lower threshold level would double the overall positivity rates from 83/2227 (3.7%) to 171/2227 (7.8%), double all penicillins and septuple cefaclor but leave MDM sensitivity unchanged (Table 2 and purple columns in Supplementary Figure 3). For the further per-protocol-analysis we decided to exclude both possibilities to maintain comparability with previous studies.
Interestingly, 7/365 (1.9%) of the patients with a history suggestive for a delayed type-response had drug-specific IgE suggesting rather an immediate than a delayed mechanism. Considering, that there were only 40 patients with a positive sIgE test, the history concerning the timing and type of DHR was not reliable in 17.5% (7/40), which can have consequences when assigning safety precautions during DPTs.

Skin tests

Skin tests were positive in 61/787 (7.8%) patients. IDTs were the most sensitive tests for both immediate (= blue) and delayed reactions ( red) (Table 2 and Figure 2c). Contrary to the in vitro tests, Aminopenicillins were more often positive than Penicillin G in early and late readings. This was true for SPT and IDT (Figure 2b and 2c). The relatively highest rates of positive SPTs were observed with cephalosporins (Figure 2b). In our algorithm, a positive SPT in the immediate reading abrogated a continuation with IDTs (see Methods). That is why cephalosporins had lower rates of positive IDTs than aminopenicillins (Figure 2c).
Overall, Ampicillin had the highest percentage of positive IDT reactions (Figure 2c). This reflects the recent trend to turn away from prescribing Amoxicillin/Clavulanic acid back to the stronger allergen Ampicillin/Sulbactam due to the high liver-toxicity (26).
PT showed low additive value to late readings of IDTs. There were only two cases, each with a history of maculopapular rash, where the PT rendered positive with an at the same time negative late reading of the IDT.

Safety

Safety is always a matter of concern when doing skin tests in patients with a history of DHR. In 5050 skin test (1718 SPT, 1697 IDT, 1635 PT) we did not observe a single systemic reaction in any patient.

Cross-reactivity of type I responses

Cross-reactivity within β-lactam antibiotics is caused by structural similarities (Supplementary Figure 1) (22). That is why many patients had more than one positive test. The 83 positive in vitro tests represented only 40 patients and the 120 positive skin tests only 61 patients.
There were only 2 patients with concomitant positive reactions to penicillins and cephalosporins and that is why we could not investigate them thoroughly (sIgE to Amoxicillin & immediate IDT to Cefuroxim; positive immediate IDT to Amoxillin & Cefuroxim).

The amount of solved cases

The primary endpoint of this study consisted in the number of solved cases without performing a DPT: Of the 932 patients in the per-protocol-analysis, 135 (14.5%) had not been submitted to tests because the DHR had already been excluded by history (Figure 3). In another 115 patients (12.3%), a DHR could be ruled out because a very improbable history was further underlined by negative in vitroand in vivo tests. Drug allergy was confirmed by positivein vitro and/or in vivo test in 96 (10.3%) of the patients and the patients received an allergy passport, which is the usual way of handling this situation in central European countries (27). Summing up, 346 (37.1%) cases were solved while 586 (62.9%) remained unresolved. Histories of DHR to cephalosporin were more often solved than the ones to penicillin (54% vs 34% Figure 3).
A successful diagnostic procedure depended a lot on the patients’ compliance. Although we tried to make the workup as convenient as possible, DHR cases could usually not be resolved at the first visit by obtaining the drug-specific history only. The intention-to-treat population also included the patients not showing up for their scheduled skin tests (Figure 1). Including these patients markedly increased the numbers of unsolved cases from 62.9% to 76.4% (Penicillin 65.7% to 78.5% and Cephalosporin from 46.3% to 62.2%).
Unsolved cases were referred to 4 different hospital-based dermatological and 2 paediatric departments in Vienna that perform DPTs. However, due to their limited capacities (see Discussion), only some patients ended up in a DPT (personal communication with the aforementioned departments). Very strict national data protection laws inhibited a structured follow up of these routine patients after leaving our allergy centre. Only a single 52-year-old woman returned to us reporting about her negative DPT, why we also included her into the ‘solved negative’ category.

Risk Factors for positive reactions

Finally, we looked for risk factors for a confirmation or disapproval of the suspected DHR (Figure 4). For this analysis, we only considered patients where the algorithm resulted in a clear yes/no situation (according to Figure 3). Of the several possible risk factors, we identified two that were significant:
1. Male (40%) vs. female sex (29.5%) (p = 0.0012, Χ² test). This was remarkable as the number of female patients with a referral history for DHR was 2.54 times higher than the one of males (Table 1).
2. Age >10: Younger children carried a lower risk for a confirmation of the DHR, than the older study population <10 years (p = 0.0411, Χ²-test).
All other risk factors shown in Figure 4 such as a parental history for a DHR in childhood, an underlying atopy (defined as a positive allergy test to inhalative or nutritive allergens, a history of atopic dermatitis, allergic rhinoconjunctivitis or bronchial asthma), underlying chronic urticaria, a history suggestive of an immediate or a delayed reaction pattern of the DHR did not differ significantly between confirmed and disapproved cases.
In a previous study, we had reported elevated serum tryptase levels as a risk factor for severe DHR (28). In contrast, in the present study elevated serum tryptase of ≥11.4 ng/ml was only detectable in 17/447 (3.8%) and only one of these patients had a confirmed reaction.