In this retrospective study a relative high percentage of 21% COX-2 hypersensitivity was found in patients with non-selective NSAID hypersensitivity. However, in COX-2 hypersensitive patient’s tolerance to another COX-2 inhibitor was found in the majority (86%) of patients.
There is a large variation (0% to 33%) in the percentage of COX-2 inhibitor hypersensitivity in patients with NSAID hypersensitivity (17). Most studies describe low percentages [15, 21, 23–37]. One study associated a high percentages (25%) of hypersensitivity to etoricoxib with paracetamol intolerance (compared to 6% in multiple NSAID hypersensitivity without paracetamol intolerance) . In contrast, COX-2 inhibitor hypersensitivity was not associated with paracetamol hypersensitivity in our patient population.
Differences in study design may explain the variability in the different studies. Many studies did not perform oral challenges to confirm non selective NSAID hypersensitivity [21, 24, 26, 27, 29, 34, 35, 38]. This may have induced a higher percentage of incorrect diagnosis of non selective NSAID hypersensitivity, especially since some studies included patients with a long time interval, up to 72 hours, between intake of the culprit non selective NSAID and onset of symptoms [26, 29, 33]. Incorrect diagnosis of non selective NSAID hypersensitivity may consequently lead to a lower estimate of the percentage of COX-2 inhibitor hypersensitivity. The majority of patients in our study were also included based on a suggestive history. However, we used strict criteria: only patients with objective symptoms and within 5 hours after intake were included. Furthermore, 53 (58%) patients had re-intake of the same culprit drug followed by a similar hypersensitivity reaction, which strongly supports the diagnosis. 44 patients with a repeated reaction did not underwent a challenge with the culprit drug. In addition, the percentages of COX-2 hypersensitivity did not differ between the patients diagnosed by careful history only versus those in which the diagnosis was confirmed by challenge.
Provocation with COX-2 inhibitors was open and not placebo controlled. However, the challenges were only considered positive if the patient developed objective symptoms, in agreement with other studies [21, 36]. Since we performed this study, the challenged drug dosages have been revised; the lowest dosage of 0.1 mg has appeared to be unnecessary and has been removed from the protocol.
Another important factor that might influence the rate of COX-2 inhibitor hypersensitivity is patient selection. Patients in our population who previously reacted to numerous non-selective NSAIDs had a higher rate (38%) of reaction to COX-2 inhibitors than patients who reacted to only one NSAID (16%). This is in line with Dona et al., analyzing patients with multiple NSAID hypersensitivity, showing a generally high percentage of hypersensitivity to etoricoxib . Multiple NSAID reactivity was not known in all patients. Therefore, the current classification of EAACI/ENDA group  could not be applied to subdivide the study population.
Another important factor could be the specific culprit drug, since acetylsalicylic acid (ASA) was the most frequently reported culprit drug in many other studies (21 - 100%) [14, 21, 23, 25–27, 29, 30, 33–36, 39],. In our study population, diclofenac, ibuprofen and naproxen represented the most common culprit drugs with 46% , 15% and 9% , respectively. Only 4 patients (4%) reported a hypersensitivity reaction to ASA only. None of them showed a reaction to a COX-2 inhibitor. Further studies are needed to investigate the relation between the specific culprit drug in non-selective NSAID hypersensitive patients and hypersensitivity to COX-2 inhibitors.
The frequency of chronic spontaneous urticaria (12%) in our study group was comparable with previous literature (2.7-11%) [21, 23, 29, 41] and cannot explain the high percentage of COX- hypersensitivity. The frequency of asthma in our study population was rather low, compared to the literature [14, 21, 23, 27, 29, 39]. Specific studies are needed to investigate whether hypersensitivity to COX-2 inhibitors is dependent on these factors.
Interestingly, a second COX-2 inhibitor was tolerated in the majority of challenged patients (86%). Of the patients with a second COX-2 inhibitor challenge, six had a history of reactions to multiple NSAIDs. However, five of them tolerated the challenge with a second COX-2 inhibitor. This shows that patients who react to a particular COX-2 inhibitor, may be tolerant to another one. This was also described in two small case-series [19, 42]. Cimbollek et al. described two patients with multiple NSAID hypersensitivity reacting to one COX-2 inhibitor but tolerating another. Quinones-Estevez described 8 patients with multiple NSAID hypersensitivity of which five were challenged with both celecoxib and etoricoxib. Three patients reacted to only one of the drugs, while two to both. In our study, etoricoxib was mainly used as a secondly challenged alternative COX-2 inhibitor, known to have a much higher COX-2 selectivity than celecoxib . This might explain the higher percentage of tolerance to etoricoxib. However, advantage of etoricoxib over celecoxib was not observed in case-series [19, 20] or a recent review analyzing COX-2 inhibitor hypersensitivity .
Our results in a large group of patients with NSAID hypersensitivity and reaction to a specific COX-2 inhibitor show that an alternative COX-2 inhibitor can be tolerated.Challenge with a second COX-2 inhibitor can be recommended in these patients. So far, there seems no preference which COX-2 inhibitor, celecoxib or etoricoxib, should be challenged first and which second.