Editor,—We thank Dr Briars for his recent comments and are aware of his opinions regarding the potential source of the intestinal cytokines that we discussed in paper, including reference to his previous paper.1

We do not agree that our data is dependant upon IL-8 alone. We have shown statistically significant differences for a whole range of proteins and types of assays. Due to the large number of proteins and types of assays that we have performed, we have not carried out the extensive experiments for IL-8, as reported by Dr Briars. We do know that the polyethylene glycol, a key constituent of the lavage fluid, does not affect the IL-8 assay. There are two reasons why variable recovery is unlikely to be a major factor in our results. Firstly, by collecting whole gut lavage, any intestinal secretions present, including bile, are extremely dilute. Substances found in faeces (for example, stercobilin) are effectively absent. Secondly, whole gut lavage is a perfusion system found to be equivalent to balloon perfusion systems.2 ,3 Thus, the dilution of any interfering factors would be very similar between the subjects and controls. Using whole gut lavage minimises any interference from intestinal material as much as is feasible in vivo.

Assuming the worst case scenario from Dr Briar's data (that is, a two fold overestimate of IL-8 in the cystic fibrosis patients, which is not found in the controls), this still shows significantly increased IL-8 output in the cystic fibrosis patients (p<0.0001) and unfeasible volumes of sputum would still be required.

For these, and reasons detailed in our paper and previous correspondence,4 we do not believe that sputum is the primary cause of the abnormalities found. Our observations concerning the increase in intestinal inflammatory markers in the whole gut lavage of cystic fibrosis patients have now been supported by a study which investigates intestinal inflammation within mucosal biopsy samples.5 This provides additional support to the hypothesis that the basic defect of cyctic fibrosis transmembrane regulator can be proinflammatory.

Dr Eisenberg correctly points out the potential influence of pancreatic enzymes and degradation. The results we found for α1 antitrypsin were unexpected, given differences for albumin and IgG. Some discordance in data has been found previously in whole gut lavage from subjects with active inflammatory bowel disease6 who are pancreatic sufficient and who also can have raised intestinal permeability.7

However, our data that showed raised albumin and IgG are consistent with well established data showing raised intestinal permeability in children with cystic fibrosis.8 As we discussed, it has been found that protein outputs from balloon perfusion experiments (which exclude upper intestinal secretions) are similar to those found in whole gut lavage, which suggests that any potential effect of degradation from pancreatic enzymes is minimal.2 ,3 We also showed eosinophilic cationic protein to be raised in children with cystic fibrosis. As with α1 antitrypsin, this is relatively stable in faeces at room temperature (approx 21 % loss over 24 hours9). This loss would be considerably lower during whole gut lavage. Thus, degradation would be unlikely to explain this difference.