Article Text


Effect of salmeterol on airway eosinophils
  1. Firestone Regional Chest & Allergy Unit
  2. St Joseph's Hospital, 50 Charlton Avenue East
  3. Hamilton, Ontario, Canada L8N 4A6
    1. Cardio-Thoracic Department, Pneumology Section
    2. University of Pisa, Pisa, Italy

      Statistics from

      Dente et al 1used the examination of induced sputum cell counts (as well as sputum eosinophil cationic protein (ECP), blood eosinophils and serum ECP) to examine the inhibitory effects of a single dose of placebo and salmeterol on allergen induced inflammatory (as well as asthmatic) responses. They performed a crossover randomised study in 11 subjects who had two allergen challenges four weeks apart and found that salmeterol inhibited the allergen induced increases in sputum eosinophils but had no effect on the other inflammatory parameters.

      The results differ from the negative results reported by us2 and Dante et al 1 attribute this to differences in study design. We agree with this explanation but not for the reasons given. Their study design has a major flaw in the use of only one baseline measurement to investigate two distinct interventions four weeks apart. The time of baseline measurements in relation to the randomised crossover design study was not given; we presume that it was at some appropriate point before the first allergen challenge. It will therefore be relevant only for half of the subjects. This design therefore ignores a basic rule of the randomised crossover trial—namely, to provide evidence that the baseline measurements of the outcome of interest before each challenge are similar. The absence of a baseline measurement is even more crucial in their study because there was a wide variation in the baseline proportion of sputum eosinophils and the randomisation of subjects was not performed after stratification for this. Therefore the claim cannot be made that salmeterol prevents sputum eosinophilia because there were no baseline measurements in at least half of the subjects to prove or disprove this assumption.

      The authors suggested that differences between their results and ours may be due to the fact that we performed five allergen challenges and four hypertonic saline inductions for each allergen and say that “this could have resulted in a progressive increase in airway inflammation in each subject during the progression of the study, leading to a more persistent eosinophilic inflammation and consequently to the low repeatability reported by these authors in sputum eosinophil percentages measured before each allergen challenge”.1Although a small change in airway inflammation can be induced by repeated hypertonic saline challenges,3 this statement has three inaccuracies. Firstly, in our study the repeated allergen challenges did not lead to a progressive sputum eosinophilia as can be clearly seen in fig 4 which shows individual values of sputum eosinophils before and after each intervention. Secondly, the authors incorrectly translated the baseline period variations in sputum eosinophils as “low repeatability”. This was not a repeatability study but an intervention study. Repeatability refers to, and reflects, the amount of error, both random and systematic, inherent in any measurement.4 Our study shows that the method of sputum examination we used is responsive to longitudinal changes, whether occurring by regression to the mean or after an intervention. Finally, the study by Holz3 refers to the effect of repeated inductions on neutrophils and not eosinophils, a point which is irrelevant to the interpretation of the results of our study.


      authors' reply In their letter Dr Pizzichini and colleagues disagree with our conclusion that salmeterol prevents allergen induced sputum eosinophilia. We do not agree with their opinion.

      In the results section of our study it is clearly stated that comparison of sputum eosinophils, as well as other parameters, was made between measurements performed after allergen challenge following pretreatment with placebo or salmeterol. Our conclusions were made on the basis of this comparison and not by evaluating differences from baseline as is commonly used in other models in other studies.1-1 Only one baseline measurement was performed for several reasons: (1) it showed good reproducibility for sputum cell counts in two samples collected under similar conditions1-2; this assumption is used to consider unchanged cell counts in sputum in two tests with hypertonic saline performed under the same conditions except for allergen challenge or premedication as intervention; (2) it is a good rule to perform only a few challenge tests as it is less likely that the characteristics of the subjects will change if they are examined over a short period of time than when they have to perform many tests over a longer period of time. We have previously reported an influence of the shortness of the time interval between two subsequent allergen challenges.1-3 Moreover, the number of subjects in our study was adequate to study a difference in sputum eosinophils due to an intervention, as calculated by power analysis, but the number was too small to permit stratification for sputum eosinophils in the baseline evaluation.

      In the second part of their letter Pizzichini et al quote from our paper that differences between their results and ours may be due to the large number of challenges that each patient performed in their study. Our sentence refers mostly to the large number of allergen challenges. In fact, if a lot of allergen challenges are performed over a short time there will probably be a worsening of asthma symptoms and late asthmatic response, and consequently an increase in the airway inflammation. If the large number of allergen challenges is performed over a longer period of time with an adequate interval between subsequent allergen challenges, there is a greater probability that different conditions will occur as a result, for example, of respiratory infections or allergen exposure, and consequently a poorer repeatability is likely.

      Finally, we think that the calculation of repeatability evaluating the baseline series of data in the study by Pizzichiniet al is correct. In fact, if similar conditions are maintained, having two or more different series of data of the same parameter is good for performing repeatability tests (as performed in our study).


      1. 1-1.
      2. 1-2.
      3. 1-3.
      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.