HLA genotyping by NGS in cases with delayed hypersensitivity and cases with immediate hypersensitivity to penicillins
The clinical characteristics and test results of the 24 patients with delayed hypersensitivity to penicillins who were subjected to HLA NGS/ NGS-based HLA typing are detailed in Supplemental Tables 1 and 2 . With the exception of patient 17 who had had a bullous exanthema with edema associated with bacampicillin treatment, none of the other patients in the cohort had a severe cutaneous adverse reaction (SCAR) as presentation. Twenty-two of the 24 patients had/displayed both positive patch test and delayed intradermal test responses, which indicate a delayed hypersensitivity mechanism Supplemental Table 2 . This pattern of positivity does not need to be supported by challenges, as previously reported.15,16 Of the remaining two patients with negative patch tests and positive delayed-reading intradermal tests, one (patient 7) underwent challenge with the amoxicillin experiencing a MPE and the other (patient 22) was not provoked as he had experienced mild urticaria a few minutes after benzylpenicillin intradermal tests (Supplemental Tables 1 and 2) . Taking into account the results of skin testing, subjects classified as group A (i.e., positive only to ampicillin and amoxicillin, n = 14) were more numerous than those with less selective responses classified as group B (i.e., positive to benzylpenicillin, ampicillin and amoxicillin, n = 6) and C (i.e., positive to penicillins and aminocephalosporins, n = 4). The clinical characteristics and allergy test results of the 20 subjects with IgE-mediated hypersensitivity are given in Supplemental Table 3. All patients were retained after the quality control step of NGS. On the HLA alleles retrieved in the 24 Italian patients, two HLA alleles,DRB3*02:02:01:02 and DRB3*02:02:01:01 , had a dramatic enrichment among the seven alleles with an allelic frequency >20%, compared to the control population of the Allele Frequency Net Database (Table 1 and Supplemental Tables 4 and 5 ). The alleles DRB3*02:02:01:02 and DRB3*02:02:01:01predicted a very high risk of delayed reaction, which could not be evaluated precisely because such alleles had not been previously reported in the general European population. 83% of the patients (20/24) harbored at least one of the two reported HLA-DRB3alleles, and 71% (11/24) were homozygous for one of the twoHLA-DRB3 alleles or had a compound heterozygous status (Table 2 and Supplemental Table 4 ). We observed a higher frequency of DRB3*02:02:01:02 and DRB3*02:02:01:01homozygous genotypes in group A (n=12, 69%) than in group B and C patients (n=3, 30%). However, no difference of allele frequency was observed among the 3 groups (Table 4 ).
The allele frequency of the DRB3*02:02 alleles (DRB3*02:02:01:02 and DRB3*02:02:01:01 ) was significantly higher in patients with delayed hypersensitivity than in those with immediate hypersensitivity to penicillins (77% vs. 28%, respectively). These figures corresponded to an increased risk of delayed reactions in penicillin hypersensitivity cases harboring DRB3*02:02 alleles, with an OR of 8.9 (95% CI, 3.4–23.3; P <0.0001) and an absolute risk difference of 49% (29%–64%; P<0.0001).
The HLA-DRB3 association did not appear to be affected by the severity of the initial delayed reaction. Indeed, patients with mild reactions still maintained this association.