A multicentre study to determine the value and safety of drug patch tests for the three main classes of severe cutaneous adverse drug reactions


  • Funding sources

  • Conflicts of interest
    None declared.

Annick Barbaud.
E-mail: a.barbaud@chu-nancy.fr


Background  Drug patch tests (PTs) can reproduce delayed hypersensitivity to drugs and entail a moderate re-exposure of patients to offending drugs.

Objectives  To determine the value of PTs for identifying the responsible drug in severe cutaneous adverse drug reactions (SCARs) such as acute generalized exanthematous pustulosis (AGEP), drug reaction with eosinophilia and systemic symptoms (DRESS) and Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN).

Methods  In a multicentre study, PTs were conducted on patients referred for DRESS, AGEP or SJS/TEN within 1 year of their SCAR. All drugs administered in the 2 months prior to and the week following the onset of the SCAR were tested.

Results  Among the 134 patients included (48 male, 86 female; mean age 51·7 years), positive drug PTs were obtained for 24 different drugs. These included positive tests for 64% (46/72) of patients with DRESS, 58% (26/45) of those with AGEP and 24% (4/17) of those with SJS/TEN, with only one relapse of AGEP. The value of PTs depended on the type of drug and the type of SCAR (e.g. carbamazepine was positive in 11/13 DRESS cases but none of the five SJS/TEN cases). PTs were frequently positive for beta lactams (22 cases), pristinamycin (11 cases) and in DRESS with pump proton inhibitors (five cases), but were usually negative for allopurinol and salazopyrin. Of 18 patients with DRESS, eight had virus reactivation and positive PTs. In DRESS, multiple drug reactivity was frequent (18% of cases), with patients remaining sensitized many years later.

Conclusions  PTs are useful and safe for identifying agents inducing SCAR.

Some severe cutaneous adverse drug reactions (SCARs) are severe, and when they occur in patients on a regimen of multiple drugs it can be difficult to identify the responsible agent based solely on chronological criteria. SCARs include acute generalized pustulosis (AGEP), drug reaction with eosinophilia and systemic symptoms (DRESS) and Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN). All three are due in part to delayed hypersensitivity to drugs.1 Unfortunately, in clinical practice, it is not possible to use an in vitro test to determine which drug was responsible for such reactions. Furthermore, re-exposure to each drug systemically would be dangerous. Therefore, patch tests (PTs) represent a low-risk method of diagnostic testing, as they can reproduce delayed hypersensitivity to drugs and entail only a moderate re-exposure of patients to offending drugs. From a previous update on drug skin testing2 we acknowledge that the value of drug PTs for SCAR is difficult to assess, and the results of published results are difficult to compare because the methods vary from one study to another. One recent study on DRESS in a Portuguese population reported low drug PT values (∼32%).3 Gex-Collet et al.4 reported that multiple drug reactivity (MDR) could be more frequent in cases of SCAR and that MDR can occur simultaneously during the same onset of SCAR. Therefore, the present study assessed the value and safety of drug PTs for identifying the responsible drug in SCAR. In addition, MDR was studied, and, as virus reactivation of the herpes virus family has been implicated in the DRESS pathomechanism,5 we also investigated the possibility of drug sensitization associated with virus reactivation in some DRESS cases.

Patients and methods

A prospective 3-year, multicentre study was conducted by the ‘Toxidermies’ group of the French Society of Dermatology on all immunocompetent patients referred for probable or definite DRESS (i.e. with a final score of 4 or more according to the criteria of Kardaun et al.),6 or with AGEP or SJS/TEN. The inclusion criteria are summarized in Table 1. All included patients had a specific file with their sex, age and clinical and biological features during the SCAR, a chronological schema of all drugs taken in parallel with the evolution of the SCAR, the delay between the SCAR and the date of skin testing, dilutions and vehicles used for PTs, and, in some cases, the results of other skin tests that had been done.

Table 1. Inclusion criteria: clinical features and biological findings in patients included in this study
Acute generalized exanthematous pustulosis
 Erythematous rash occurring in the 21 days following the beginning of a new drug regimen with many nonfollicular, sterile pustules of < 5 mm in diameter, and the development of wide, oedematous and erythematous plaques predominating on the face and/or folds
 AND a cutaneous biopsy sample with intraepidermal pustules
 AND a hyperleucocytosis with neutrophilic polymorphonuclear cells
Drug reaction with eosinophilia and systemic symptoms (DRESS)
 Rash with or without mucosal involvement occurring during the 2 months following the initiation of a new drug regimen
 AND at least one of the following affecting the viscera:
  Polyadenopathy of ≥ 1 cm in diameter at a minimum of two sites
  Hepatitis with aspartate aminotransferase and/or alanine aminotransferase more than twice the normal range
  Acute interstitial nephropathy: acute impaired renal function not completely explained by functional dysfunction and/or  a daily proteinuria of > 150 mg and/or haematuria and/or leucocyturia (urine cell count)
  Cardiopathy or other organ involvement
 AND at least one of the following biological abnormalities:
  Eosinophilic polymorphonuclear blood count > 1000 cells mm−3
  Mononucleosis syndrome (presence of hyperbasophilic lymphocytes)
  Hyperlymphocytosis > 5000 cells mm−3
 AND a score of 4 or more according to the DRESS criteria of Kardaun et al.6
Stevens–Johnson syndrome (SJS) or toxic epidermal necrolysis (TEN)
 Cutaneous exanthema that has occurred at least 48 h after the introduction of a drug AND blisters or erosions on < 10% of the cutaneous surface in cases of SJS, and on > 30% of the cutaneous surface in cases of TEN
 AND a cutaneous biopsy sample confirming the diagnosis of epidermal necrolysis

In the 12 months following the resolution of the SCAR in all centres, drug PTs were conducted and read according to the European Society of Contact Dermatitis guidelines for drug skin testing.7 As some patients received systemic corticosteroids for their SCAR, and as corticosteroids had to be stopped at least 1 month before testing, it was decided to include all cases in the 12 months following the SCAR, and not only 6 months as had been previously recommended. In order to study MDR, all drugs introduced within the 10 days prior to the onset of SCAR for AGEP, or within 2 months prior to onset for DRESS and SJS/TEN, as well as those introduced in the week following onset, were tested. Briefly, drug PTs were conducted using the commercialized form provided by the patients themselves, with each drug diluted to 30% (80% of cases) or 10% (20% of cases) in petrolatum. Moreover, when available, the material commercialized for drug PTs by Chemotechnique Diagnostics (Vellinge, Sweden), with drugs diluted to 10% in petrolatum, was used. For SJS/TEN or for testing pseudoephedrine, PTs were first performed diluted to 1%, and if the results were negative the tests were repeated with the drug diluted to 30% in petrolatum.

In DRESS, to determine whether positive PTs remained positive later, two patients were tested twice. One patient who had developed DRESS 11 years previously, and who had positive PTs for carbamazepine and beta-lactam antibiotics 6 months later, was referred again for another cutaneous disease during our study period. As she gave her informed consent to be retested, she was included and the results from her initial investigation were taken into account. Another patient tested 2·5 months after the onset of DRESS agreed to be retested 9 months later.

All drug PTs were conducted on the patients’ backs using either Finn Chambers® on Scanpor (Epitest Ltd Oy, Tuusula, Finland) or IQ-Ultra chambers (Chemotechnique Diagnostics). For SJS/TEN, drug PTs were also conducted at the sites of previous bullous lesions. The PTs were applied for 2 days, then read on day 2 and day 4.

For AGEP and DRESS, when PTs were negative, 22 patients were also given drug skin-prick tests on the forearm. These were read at 20 min and 1 day later, and their results were compared with those obtained with saline 0·9% and histamine (10 mg mL−1). In some patients with AGEP or DRESS, intradermal tests (IDTs) were also conducted when drug PTs were negative, by injecting 0·02 mL of the diluted drug. According to European guidelines on delayed readings at 24 h,7,8 IDTs were considered positive when there was an infiltrated and erythematous reaction.

In one centre, in some patients with DRESS the detection of cytomegalovirus (CMV), Epstein–Barr virus (EBV), human herpesvirus (HHV)-6 and HHV-7 was performed by whole-blood real-time quantitative polymerase chain reaction (PCR). After automated DNA extraction (Magna Pure LC System®, Roche, Meylan, France), DNA amplification was performed with the EBV R-gene™ and CMV HHV-6,7,8 R-gene™ kits (Argène, Varilhes, France), allowing quantitative detection of CMV, EBV and HHV-6, and qualitative detection of HHV-7.

To determine the absence of irritancy in positive PTs, results from patients tested with the same methods were identified for all agents that gave positive results, by searching the database Diamm-Toxiderm (Microsix Laboratory and University Hospital of Nancy, Vandoeuvre-lès-Nancy, France); this database includes the results of all patients tested with PTs (using the same methodology as the present study) at the University Hospital of Nancy since 2007 (Table 2).

Table 2. Characteristics of patients with positive patch tests (PTs)
CaseADRSexAge (years)Positive drug PTsNumber of negative PTs in databaseViral positive PCR
  1. ADR, adverse drug reaction; PCR, polymerase chain reaction; AGEP, acute generalized exanthematous pustulosis; M, male; F, female; Dpp, drug provided by the patient tested at 30% dilution in petrolatum (pet). These negative results were obtained from the Diamm-Toxiderm database for skin tests in Nancy, France; PPI, proton pump inhibitor; C, material commercialized for drug patch tests by Chemotechnique diagnostics with drug diluted at 10% in pet; DRESS, drug reaction with eosinophilia and systemic symptoms; HHV, human herpesvirus; EBV, Epstein–Barr virus; SJS/TEN, Stevens–Johnson syndrome/toxic epidermal necrolysis. aPatient retested 9 months after the occurrence of DRESS. bPatient retested 11 years after the occurrence of DRESS. cDelayed positive prick test with aciclovir, and delayed positive intradermal test (IDT) with imipenem–cilastatin. dDelayed positive IDT with glycopeptides (flare up). eDelayed prick test with vancomycin. fDelayed IDT with tazocillin.

1AGEPM56PristinamycinDpp: 15 
3AGEPM80Pristinamycin (flare up)  
8AGEPM63Fluindione, PPI, pristinamycinDpp: 8 
9AGEPM76AmoxicillinC: 180 
11AGEPM83Amoxicillin, enoxaparin  
13AGEPF50Amoxicillin–clavulanic acidDpp: 180 
14AGEPM65Amoxicillin–clavulanic acid  
15AGEPF64Amoxicillin, beclometasoneDpp: 177 
16AGEPF94CeftriaxoneDpp: 180 
17AGEPF85Paracetamol–dextropropoxypheneParacetamol, Dpp: 48 
19AGEPF29PrednisoneDpp: 9 
21AGEPF92IodixanolDpp: 245 
22AGEPM68IoversolDpp: 245 
24AGEPF51ClindamycinDpp: 6 
25AGEPF30TetrazepamDpp: 12 
26AGEPF51Pseudoephedrine (at 1%)C: 4 
27DRESSF76Amoxicillin–clavulanic acid  
28DRESSF28Amoxicillin HHV-7
29DRESSF75Amoxicillin, lansoprazoleDpp: 14 
34DRESSM41Amoxicillin and cephalosporins EBV, HHV-6
35DRESSF64Amoxicillin–clavulanic acid, esomeprazole, quinolonesaDpp: 32EBV
37DRESSF71Ceftriaxone, aciclovirDpp: 4 
38DRESSM34DicloxacillinDpp: 2EBV
39DRESSM54Imipenem–cilastatin, vancomycinDpp: 26 
40DRESSF51Carbamazepine, cloxacillinbDpp: 10, Dpp: 4 
49DRESSF65Carbamazepine, spironolactoneDpp: 5 
50DRESSM71Carbamazepine, spironolactone  
51DRESSM60EsomeprazoleDpp: 32HHV-6
52DRESSF31PantoprazoleDpp: 22HHV-6, HHV-7, EBV
53DRESSF51Vancomycin, pantoprazolec  
55DRESSF45Pristinamycin, vancomycin HHV-7, EBV
61DRESSF62AmikacineDpp: 15 
62DRESSF34CelecoxibDpp: 38 
63DRESSF47CitalopramDpp: 3 
64DRESSF32ClobazamDpp: 3 
65DRESSF68DiltiazemDpp: 12 
66DRESSF50EnoxaparinDpp: 33 (undiluted) 
67DRESSM49IodixanolfDpp: 245 (undiluted)EBV
68DRESSF40LamotrigineDpp: 6 
69DRESSM30OlanzapineDpp: 1
70DRESSF22PyrimethamineDpp: 1 
71DRESSF68TetrazepamDpp: 12 
72DRESSM56TixocortolC: 0·1% pet: 2755 
73SJS/TENF28Esomeprazole, vancomycin, amoxicillin  
75SJS/TENF41RamiprilDpp: 9 

The study concerning practice evaluation received the approval of the president of the ethical committee of Nancy.


In total 134 patients were included (48 male, 86 female; mean age 51·7 years; range 3–94 years): 72 with DRESS, 45 with AGEP and 17 with SJS/TEN. Among them, 76 (56·7%) had at least one positive drug PT (Table 2). The percentages of patients with positive PTs were 64% (46/72) among cases with DRESS, 58% (26/45) for AGEP and 24% (4/17) for SJS/TEN.

Of the 45 patients with AGEP (15 male, 30 female; mean age 56·8 years), 26 (58%) had positive drug PTs: eight for pristinamycin, eight for beta lactams, three for corticosteroids, two for radiocontrast media (RCM), two for dextropropoxyphene in combination with paracetamol (acetaminophen), and one each for fluindione, nonfractionated heparin, pseudoephedrine (at 1% in petrolatum), tetrazepam, clindamycin and varenicline. MDR to different drug classes occurred in three cases with amoxicillin and heparin or corticosteroids, and in another case with fluindione and pristinamycin. We observed only one relapse of AGEP requiring systemic corticosteroids following positive PTs with pristinamycin. None of the patients reported any late reaction (> 120 h).

Negative results were observed in 19 cases. The mean delay between the adverse reaction and the PTs was 3·7 months overall, and 3·6 months in patients with positive PTs and 4·3 months in those with negative results.

Beta-lactam antibiotics alone or associated with miscellaneous drugs could have been responsible in 11 cases, with well-tolerated IDTs and delayed positive results in four cases. In another case, pristinamycin was responsible, and delayed positive IDTs were observed with other injectable synergistin antibiotics. Considering both the patients with a delayed, positive IDT and those with positive PTs, a total of 31/45 cases (69%) had positive drug skin tests.

Among 72 cases with DRESS (26 male, 46 female; mean age 51·2 years), 46 (64%) had positive PTs (Table 2). These were observed in response to the following antimicrobials: beta lactams (14 cases), vancomycin (four cases), pristinamycin (three cases), quinolones (one case, with cross-reactions between ciprofloxacin, norfloxacin and pefloxacin), amikacin (one case), pyrimethamine (one case) and aciclovir (two cases). Positive PTs to nonantimicrobials included carbamazepine (11 positive PTs out of 13 suspected cases), proton pump inhibitors (PPIs; five cases), fluindione (two cases), spironolactone (two cases), celecoxib (one case at 30% in petrolatum), and olanzapine, citalopram, clobazam, corticosteroids, diltiazem, heparin, lamotrigine, RCM (iodixanol) and tetrazepam (one case each). MDR was observed in 13 cases, involving at least two different chemical classes; four of these were detected by skin-prick tests followed by IDT tests. One of these four patients had a relapse of maculopapular exanthema on her arms, without any biological disturbance (patient #59). One patient had positive delayed IDT only to beta lactams and glycopeptide antibiotics.

Overall, among 18 patients who had virus PCR analysis, virus reactivation was observed in 14 cases (78%), including 8/12 patients (8/11 with positive PTs) with positive drug skin tests and in 6/6 patients with negative ones (Table 3). Four of these patients had an MDR and a virus reactivation.

Table 3. Study of reactivation of herpes family viruses in 18 patients with drug reaction with eosinophilia and systemic symptoms, and the results of their drug skin tests
CaseDrug testsaPositive skin testsEBVHHV-6HHV-7CMV
  1. EBV, Epstein–Barr virus; HHV, human herpesvirus; CMV, cytomegalovirus; PT, patch test; PPI, proton pump inhibitor; IDTd, intradermal test with delayed positive results; RCM, radiocontrast media. aFor negative tests the suspected causative drug is given in brackets.

28+PT amoxicillin+
34+PT beta lactams++
35+PT beta lactams, PPIs, quinolones+
38+PT beta lactams+
48+PT carbamazepine
51+PT PPIs+
52+PT PPIs+++
55+PT pristinamycin, IDTd teicoplanin++
56+PT pristinamycin
67+PT RCM, IDTd beta lactams+
69+PT olanzapine
+IDTd beta lactams and teicoplanin
− (glycopeptides) +
− (allopurinol) +
– (allopurinol) +++
– (salazopyrin) +
– (salazopyrin) ++
– (minocycline) +

The mean delay between DRESS and drug PTs was 5·9 months. The one patient (case #40) retested 11 years after her original positive PTs for cloxacillin and carbamazepine remained strongly positive to both drugs. The patient (case #35) retested 9 months after the original onset of DRESS also remained positive.

A total of 26 DRESS cases had negative PTs. Intradermal tests with beta-lactam antibiotics were conducted in four of these patients, with positive delayed results in three cases, without any relapse of rash. Therefore, the total number of DRESS cases showing drug hypersensitivity increased from 46 to 49 patients (68%). In patients for whom only one drug was suspected, drug PTs were negative in seven cases for allopurinol, five cases for salazopyrin, and one case each for strontium ranelate, minocycline, co-trimoxazole, fluindione and olanzapine. In cases for which many different drugs could have been responsible, negative PTs were obtained in one patient for antituberculosis drugs and in four patients for PPIs in combination with more than three other drugs.

Among 17 cases of SJS/TEN (seven male, 10 female; mean age 39·7 years), four (24%) had positive PTs (Table 2). The tests were conducted with a mean delay of 6 months following the onset of the SCAR. Thirteen patients had negative PTs for miscellaneous drugs: nonsteroidal anti-inflammatory drugs (four cases), carbamazepine (five cases), and allopurinol, plant-derivative therapy, minocycline and co-trimoxazole (one case each).

The four positive PTs were obtained for a commercialized form of ramipril diluted to 30% in water (as it was negative in petrolatum), lamotrigine at 30% in water (as it was negative in petrolatum), tetrazepam (one case) and glycopeptide antibiotics and PPI (one case). We did not observe any differences in testing skin areas that were or were not previously involved.


The present study is the largest reported multicentre study to date to assess the positivity of PTs for identifying the responsible drug in SCAR. Our results emphasize the sensitivity of these tests, as they were positive in 56·7% of 134 cases.

By testing all possible agents, we were able to demonstrate the role of products that would normally not be suspected in SCAR, such as PPIs (five cases in DRESS), corticosteroids (four cases) and RCM (three cases).

For AGEP, drug PTs seem to be of value, as they were positive in 26/45 cases (58%) in the present study and in seven out of 14 cases in Wolkenstein et al.9 There have also been some individual cases of patients with AGEP and positive PTs for allylisopropylacetylurea,10 beta lactams, bleomycin, carbimazole, celecoxib,11 ciprofloxacin, clindamycin, corticosteroids, diltiazem, metamizole, methoxsalen, metronidazole, morphine, nimesulide, pseudoephedrine, ranitidine and tetrazepam.12

Delayed hypersensitivity is involved in AGEP, and drug PTs are able to reproduce this pathomechanism. In previous studies, from biopsies taken from the rash, as well as those taken from positive PT sites, T-cell lines and clones could be generated, some of which were drug-specific T cells.11 In four cases, positive IDTs were observed with a good tolerance.

In addition, our results suggest that drug PTs are safe, as they reinduced a rash in only one case. There is one previously reported similar case, possibly due to paracetamol but with doubtful PTs and inducing a symmetric vesiculopustular eruption.13

Our study population had a male/female ratio of 0·47, where it was 0·2 in Israel14 and 0·8 in the RegiSCAR study, which covered all of Europe.15 The drugs that most frequently induce AGEP in Europe are pristinamycin, beta-lactam antibiotics, quinolones, (hydroxy)chloroquine, sulfonamides, terbinafine and diltiazem.15 In our patients, beta-lactam antibiotics were responsible in 42% of the cases (with positive skin tests in 27%), pristinamycin in 20% of the cases and quinolones in one case.

Although it was not an objective of the present study, our results also show that IDTs with delayed readings are useful for determining the responsible drug in patients with AGEP and negative PTs, as five of six cases had delayed positive results.

In DRESS, drugs can enhance reactivation of the herpes virus family,16 but drug sensitization can also occur. We emphasize that hypersensitivity does exist in DRESS and that drug PTs are of value in DRESS, as they were positive in 64% of cases. This percentage is higher than that reported in Portugal (32%) among 56 patients with DRESS who were tested using the same method as ours.3 However, the present study included probable and definite DRESS induced by a large variety of drugs, whereas the previous study found positive PTs mainly with carbamazepine (13 with positive PTs out of 18 suspected cases), and reported patients with DRESS due to anticonvulsants in 59% of the cases and to allopurinol in 34%. Positive results on PTs could imply transient, nonspecific reactivity to drugs due to strong immunostimulation induced by virus reactivation,5 but this cannot explain why PTs would remain positive 11 years after the disappearance of DRESS (patient #40).

The value of patch testing varies according to the implicated drug. In our series, PT appeared to be unhelpful in cases of SCAR due to salazopyrin and allopurinol, not yielding any positive results. The finding of negative PT to allopurinol in cases of DRESS caused by this drug is corroborated by the study of Santiago et al.,3 who did not have any positive results among 19 such patients.

Many studies have reported positive PTs with carbamazepine at 10% or 30%, including 11 cases among 13 patients with DRESS possibly due to this anticonvulsant in the present study, seven out of 10 suspected cases in the study of Lin et al.,17 and 13/18 cases (at 20% in petrolatum) reported by Santiago et al.3

There are also some reports of positive drug PTs in patients with DRESS2 in response to abacavir,18 antituberculous agents,19 aspirin,20 carbamazepine,17,21 cyclins, beta-lactam antibiotics,9,22 celecoxib,19 esomeprazole,22 fluindione,23 lamotrigine,2 phenytoin,3 propylthiouracil,24 spironolactone and topiramate.3

In DRESS, herpes family virus reactivation has been reported in 76% of patients, and the main response may be related to an expansion of CD8+ T cells directed against EBV or HHV-6.3 Directly, without any sensitization, both sodium valproate and amoxicillin can induce virus reactivation in patients with DRESS.16 However, even if such reactivation is common in DRESS, our results demonstrate that delayed hypersensitivity to drugs is also involved.

It was not our objective to analyse in parallel the positivity of PTs and virus reactivation, but our data demonstrate that the virus reactivation observed in 78% of our 18 cases can be associated with drug hypersensitivity, as eight cases with positive drug skin tests also had virus reactivation during DRESS.

Regarding the five cases involving PPIs, all of them were introduced with other drugs, mainly antibiotics, more than 3 weeks before the occurrence of DRESS, and the condition began to improve only after the withdrawal of the PPI. These results emphasize that PPIs can induce DRESS, as reported in one previously published case.25

MDR to different classes of drugs administered at the onset of DRESS was frequent in DRESS (18% of our cases). This problem, unusual in other types of CADR, has to be kept in mind in the management of patients with DRESS. In nonsevere cutaneous adverse drug reactions, MDR was found in only seven of 1925 cases (0·3%).26

We propose that, through transient immunosuppression and the danger hypothesis,27 costimulatory signals provided by viral reactivation and/or first-drug sensitization could act as cofactors and enhance stimulation of an immune response to another drug–protein conjugate, and lead to the development of MDR to different drug classes. This MDR can be long lasting, as evidenced in the present study, which documented continued sensitization 11 years after the initial onset of DRESS. MDR has previously been reported in DRESS due to anticonvulsants,28 based on evidence from PTs, IDTs or drug readministration with the involvement of another anticonvulsant, or paracetamol or amoxicillin. We report 13 new cases, all with positive skin tests, in patients with probable or definite cases of DRESS, and with proven reactivity to two different drug classes, not only due to anticonvulsant drugs. These patients had no previous history of sensitization before the occurrence of DRESS.

In SJS/TENs, it has been demonstrated that the mediator for keratinocyte death is granulysin 15-kDa, released by cytotoxic T lymphocytes and natural killer cells29 acting through the caspase signalling pathway. However, it seems difficult to reproduce these mechanisms by drug PTs. In addition, drug PTs do not appear to be very sensitive for SJS/TEN, as only two of 22 patients tested by Wolkenstein et al.,9 and four of 17 tested in the present study had positive drug PTs. Nevertheless, positive PTs in SJS/TEN have been reported for beta-lactam antibiotics,2 carbamazepine,17 co-trimoxazole29 and pseudoephedrine.2 The sensitivity of PTs in this type of SCAR could be particularly dependent on the drug, as Lin et al.17 obtained 10 positive PTs for carbamazepine in 16 patients with SJS/TEN. Klein et al.30 obtained positive PTs when co-trimoxazole was tested on cutaneous sites previously affected by necrolysis, whereas drug PTs performed on other, less affected skin sites remained negative, but we did not observe any differences in testing skin areas that were or were not previously involved.

Some bias can be underlined. From a chronological point of view, the relevance of positive PTs was probable or possible in all cases. The specificity of positive reactions was difficult to assay for those drugs for which there were few or no controls. All participants had discussed each step of the protocol and received written detailed information in order to perform PTs, but no quality-control measures could be undertaken to ascertain that there were no slight differences from one centre to another.

In conclusion, PTs conducted with commercialized forms of some drugs diluted to 30% in petrolatum are of value and safe for investigating AGEP and DRESS. As any inflammatory stimulation could be responsible for virus reactivation in DRESS, we propose conducting PTs 6 months after the onset of rash in DRESS. Testing for all possible agents is necessary, as doing so can identify responsible products that are normally not suspected, such as corticosteroids in AGEP or PPIs in DRESS. Moreover, doing so allowed us to identify MDR in a patient with DRESS who remained sensitized many years later.