Article Text


  1. Immunopharmacology Group
  2. Southampton General Hospital
  3. Southampton SO16 6YD
  4. UK
  5. Department of Immunology & Rheumatology
  6. Harvard Medical School
  7. Boston
  8. Massachusetts
  9. USA
  10. Department of Medicine
  11. University School of Medicine
  12. Krakow
  13. Poland
  1. Unit of Critical Care Medicine
  2. Royal Brompton Hospital
  3. London SW3 6NP, UK
  4. Pharmacology Department
  5. Rhone-Poulenc Rorer
  6. Dagenham, Essex RM10 7XS and
  7. Department of Thoracic Medicine
  8. National Heart and Lung Institute
  9. London SW3 6LY, UK

Statistics from

In a recent editorial in ThoraxMitchell and Belvisi1 cite the findings of Sousa and colleagues2 to support a theory that aspirin induced adverse respiratory reactions are caused by excessive formation of 15-HETE and 15-epilipoxins due to the acetylation of cyclo-oxygenase 2 (COX-2) in the lungs of aspirin sensitive asthmatic (ASA) patients by aspirin. Sousa and colleagues reported greater immunoreactivity for the cytokine inducible isoform of COX-2 in the bronchial biopsy specimens of asthmatic patients compared with normal subjects, suggesting a role for COX-2 derived pro-inflammatory prostanoids in the asthmatic airway.2 This group also reported a greater proportion of COX-2 immunostaining localised to submucosal mast cells and eosinophils in bronchial biopsy specimens of ASA patients compared with those of non-aspirin sensitive asthmatic (NASA) patients.2

The paradigm put forward by Mitchell and Belvisi has a number of internal contradictions and is not supported by recent experimental evidence. Firstly, Sousa and colleagues found that the overall expression of COX-2 was, in fact, similar in ASA and NASA biopsy specimens,2 a finding supported by our group.3 Thus, there is no reason why aspirin induced synthesis of 15-HETE or 15-epilipoxins should be abnormal in the ASA lung. Secondly, 15-HETE produced by aspirin acetylation of COX-2 is predominantly or exclusively the 15R-HETE stereoisomer, which is biologically inactive.4 Thirdly, the active isomer 15S-HETE and its metabolites are extremely weak bronchoconstrictors, while it is well established that adverse respiratory reactions to aspirin and other NSAIDs are caused by the more potent bronchoconstrictors, cysteinyl leukotrienes (cys-LTs), as such reactions are blocked by specific leukotriene synthesis inhibitors and cys-LT receptor antagonists.5 Fourthly, and most critically, among the NSAIDs only aspirin covalently acetylates COX-2 and generates 15-HETE, yet patients with ASA show extensive cross sensitivity to other NSAIDs which do not share this property.6 The mechanism proposed by Mitchell and Belvisi1 can therefore play only a negligible pathobiological role in adverse respiratory reactions to aspirin itself, and would not appear to be relevant to similar reactions to other NSAIDs.

The currently accepted model of ASA instead suggests that non-selective NSAIDs including aspirin trigger adverse reactions by releasing the cys-LT synthetic pathway from the suppressive effects of endogenous prostaglandin (PG)E2, probably synthesised by constitutive COX-1, not by inducible COX-2.6-8 It is agreed that there is no difference in COX-1 expression between ASA, NASA, and normal bronchial biopsy specimens.2 3 9Furthermore, we have recently described a mechanism3 which resolves the question of why only patients with ASA, but not those with NASA or normal subjects, produce large amounts of cys-LTs when the PGE2 “brake” is removed by NSAIDs.8 10 It also explains why these patients have chronically elevated cys-LT production associated with chronic severe asthma, even in the absence of exposure to NSAIDs.10

The unique terminal enzyme for cys-LT production is LTC4synthase, and cloning of the cDNA followed by mutagenic analysis has delineated the amino acid residues central to LTC4biosynthesis.11 12 The gene for LTC4 synthase has been localised to chromosome 5q, telomeric to other candidate asthma genes.13 We have found that the frequency of cells expressing LTC4 synthase was five times higher in bronchial biopsy specimens from patients with ASA than in those with NASA, and 19 times higher than in normal biopsy specimens, irrespective of aspirin exposure.3 In contrast, expression of other leukotriene pathway enzymes (5-lipoxygenase and FLAP) and of COX-1 and COX-2 was similar in all subject groups. Increased levels of cys-LTs in the bronchoalveolar lavage (BAL) fluid of unchallenged ASA patients correlated exclusively with the increased number of cells expressing LTC4 synthase in the bronchial mucosa.3Following segmental lysine-aspirin challenge, the greater numbers of cells with LTC4 synthase expression in patients with ASA were associated with rapid increases in cys-LTs in the BAL fluid while the response in patients with NASA was negligible. Increased LTC4 synthase expression in biopsy specimens from patients with ASA was the only enzyme or cell marker which correlated significantly with bronchial hyperresponsiveness to inhaled lysine-aspirin,3 a definitive clinical measure of aspirin sensitivity. A polymorphism in the LTC4 synthase gene promoter has since been identified.14 The A-444C mutation creates an additional regulatory sequence for the nuclear transcription factor AP-2, perhaps leading to increased expression of LTC4 synthase protein. By RFLP analysis the polymorphism was shown to be twice as frequent in patients with ASA as in those with NASA or normal subjects, and represents a significant risk factor for aspirin sensitive asthma (odds ratio 3.89, 95% CI 1.57 to 8.98).14

Together these studies suggest that genetic polymorphisms leading to overexpression of LTC4 synthase in the bronchial wall of patients with ASA cause chronic overproduction of cys-LTs, which is exacerbated when non-selective NSAIDs suppress the endogenous PGE2 “brake” generated by COX-1 activity. In contrast to the ASA lung, relatively low expression of LTC4 synthase precludes clinically significant reactions to NSAIDs in the airways of patients with NASA and those of normal subjects. It is possible that further studies may lead to genetic approaches to identify individuals at risk of aspirin induced reactions, which in turn would aid the targeting of anti-leukotriene drug therapy to this patient group.


authors’ reply We are very pleased that four such eminent scientists as Professors Sampson, Holgate, Austen and Szczeklik took the time to respond to our recent editorial inThorax. Our editorial1-1 was written to accompany a paper by Sousa and colleagues1-2showing that COX-2 expression is increased in airway biopsy specimens from asthmatic patients—both aspirin sensitive and aspirin tolerant—when compared with levels in samples from healthy volunteers. Moreover, Sousa et al showed that a much greater proportion of masts cells contained COX-2 in tissue from aspirin sensitive patients than from aspirin tolerant patients. This work suggests that COX-2 is a feature of asthma, an observation which confirms the hypothesis made by our group in several papers.1-1 1-3 1-4 Moreover, the differential expression of COX-2 reported by Sousa suggests that this form may have relevance for aspirin sensitivity. In our editorial we have highlighted a phenomenon which occurs when COX-2 is exposed to aspirin—the shunting of products to 15-HETE and further, via 5-LO, to 15-epilipoxins. However, Professor Sampson and colleagues clearly do not agree that this pathway has any relevance for aspirin induced asthma. We would like to take this opportunity to clarify some of the points we have made and to address directly the comments made by Sampson et al.

Firstly, it is suggested that since Sousa and colleagues found that the overall expression of COX-2 was the same for aspirin sensitive and aspirin tolerant asthmatics, COX-2 can have no role in aspirin sensitive asthma, a notion that is supported by a recently published study by Sampson and colleges.1-5 This is true; however, Sousa showed a sixfold increase in the number of mast cells expressing COX-2 in aspirin sensitive compared with aspirin tolerant patients. Since cellular location of COX enzymes dictates the products produced and overall effect, we submit, along with Sousa, that this finding is of significant importance. Unfortunately there are clear differences in the results presented by Sousa and those recently published by Sampsonet al.1-5 Perhaps we must wait for further studies on this subject to have a full appreciation of the differences in COX-2 expression in the airways of asthmatic and normal individuals. It is also suggested that the aspirin induced formation of 15-HETE is predominately in the R form, which is biologically inactive. This point is also true, but aspirin generated 15-HETE is further metabolised by 5-LO to a novel group of metabolites—the 15 epilipoxins1-6—the biological activity of which is not known. Indeed, our editorial clearly showed that it is the involvement of the 5-LO pathway that justifies a role for aspirin generated products in asthmatic symptoms. The final point made by Sampson and colleagues was particularly important. They illustrate the fact that aspirin modification of COX-2 to form 15-HETE is a feature of aspirin and not of other non-steroidal anti-inflammatory drugs (NSAIDs).1-7 Aspirin sensitive patients are thought to respond to all NSAIDs,1-8 a fact that we have freely stated throws doubt on the “COX-2–15HETE” hypothesis. However, it remains the case that the vast majority of studies (including the recent study by Sampson and colleagues) of the severity and mechanisms of aspirin induced asthma use aspirin and not other NSAIDs as a challenge. This approach would be improved if less complex and more potent NSAIDs such as ibuprofen, indomethacin, or diclofenac were used in such studies, at least in addition to aspirin.

In the letter by Sampson and colleagues a new hypothesis is put forward for the mechanisms involved in aspirin sensitive asthma. These authors have recently published new and exciting observations showing that aspirin sensitive individuals express elevated levels of LTC4 synthase in airway cells. They hypothesise that aspirin sensitive patients express high levels of LTC4synthase and that, when the “PGE2 brake” is removed by NSAIDs, LTC4 is produced in abundance. It is most unfortunate that this paper was not published at the time we wrote our editorial as we would, of course, have discussed this new hypothesis in the light of data presented by Sousa. However, there are still important questions that require attention before we can be confident that the question of aspirin sensitive asthma is fully addressed. For example, Sampson et al show in their study that aspirin sensitive asthmatics express five times higher levels of LTC4 synthase than aspirin tolerant asthmatics, and 20 times higher levels than normal individuals. For this new hypothesis to have importance we would assume that LTC4 is rate limiting in the production of adenyl leukotrienes in tolerant asthmatics. This may well be the case; however, the levels of urinary LTE4excreted by asthmatics (aspirin sensitivity status not addressed) challenged with allergen (150–1816 pg/ml of creatine)1-9are in line with those observed in aspirin sensitive patients challenged with aspirin (300–1500 pg/mg creatine).1-10Perhaps the asthmatic patients who develop aspirin sensitivity not only have defects in the expression of LTC4 synthase but also in the number of braking mechanisms available to suppress its release.

Clearly, the past and current contributions of Professors Sampsonet al are of the greatest importance and we are pleased that our editorial has initiated debate on the subject.


  1. 1-1.
  2. 1-2.
  3. 1-3.
  4. 1-4.
  5. 1-5.
  6. 1-6.
  7. 1-7.
  8. 1-8.
  9. 1-9.
  10. 1-10.
View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.