Article Text
Statistics from Altmetric.com
We read with interest the paper by Dawson et al1 on the prevalence of fibrosing alveolitis (FA) diagnosed by HRCT scanning in rheumatoid arthritis (RA). This well designed cross sectional study estimates the prevalence of FA at 19% in patients with RA irrespective of respiratory symptoms. This is in keeping with current literature and our earlier report of 20% in unselected patients with RA not suspected of having interstitial lung disease (ILD).2
However, neither of these studies has been sufficiently powered to assess a possible association of smoking with ILD. Smoking may adversely affect the outcome of ILD in RA and Saag et al3 suggested that smoking was the most consistent independent predictor of ILD patterns in lung function tests and chest radiographs in RA. One of our previous studies4 reported a prevalence of ILD of only 5% on HRCT scanning in a cohort of 20 never smokers with RA, while Dawson et al reported a prevalence of 11% in never smokers compared with 22% in smokers. There is therefore evidence of a trend towards an association between ILD and smoking which could be explored in a larger study. However, a sample size of 450 patients would be needed to test the hypothesis that smokers are twice as likely to develop ILD in RA than never smokers (95% confidence; power = 80%; smoker/never smoker ratio 2:1).
We agree with the authors that further work on the natural progression of FA diagnosed by HRCT scanning in RA is due. We have commenced a longitudinal prospective study of 18 RA patients with ILD diagnosed by HRCT scanning compared with a cohort of patients with cryptogenic fibrosing alveolitis (CFA) matched for age, sex, smoking, and respiratory symptoms.5 There are significant baseline differences in clinical and radiological features between these two groups. Clubbing and honeycomb appearance on the HRCT scan is more common in patients with CFA while ground glass appearance is more common in RA patients with ILD. The presence of rheumatoid factor appears to be protective against honeycombing in both groups. These differences in clinical and HRCT features may be important predictors of outcome.
References
Authors' reply
We are pleased to receive the letter from Saravanan and Kelly in response to our recent publication in Thorax.1 The relationship between smoking and RA associated FA is an interesting one. There is no consistent finding in the literature of smoking and RA associated FA and, as far as we are aware, no prospective HRCT based study has shown a statistically significant association between RA associated FA and smoking. In the study by Cortet et al2 68 patients with RA were prospectively studied with HRCT scanning. Cigarette smoking was less prevalent than in the North of England and the ratio of smokers to non-smokers was 1:3. No statistical association was seen linking smoking and interstitial lung disease (ILD) and a prevalence of 20% of ILD (17% ground glass pattern and 2.9% reticular pattern) was still found. It is true that in our study the absolute risk of ever smoking cannot be excluded as a risk factor for FA as the number of lifelong non-smokers is small; however, the pack year data are adequately powered to show no statistically significant difference.
With regard to the paper by Rajasekaran et al,3 we feel it necessary to point out that the patients in their study with FA and RA had the diagnosis confirmed by HRCT scanning and, in addition, were symptomatic with dyspnoea, bibasal crackles, restrictive pulmonary function tests, and chest radiographic changes of FA. We are sure this will provide very valuable information about the progression of FA in patients with RA but it will not add to our knowledge on the outcome of HRCT changes detected at a subclinical stage.
Rajasekaran et al found honeycombing on the HRCT scan in three of 18 patients with RA associated ILD and in four of 18 patients with CFA; this difference is not statistically significant.3 None of these patients was rheumatoid factor positive, which has led the authors to postulate that rheumatoid factor may be protective against honeycombing in ILD. These findings are in direct contrast to those of Muller-Leisse et al4 who found higher levels of rheumatoid factor to be associated with ground glass changes and honeycombing on the HRCT scan, and also to McDonagh et al5 who reported that at least five of 16 patients (31%) had honeycombing and were rheumatoid factor positive. This finding is particularly interesting given that there is evidence in the literature of smoking being associated with seropositivity for rheumatoid factor in patients with6 and without RA.7,8 We would suggest that larger studies need to be undertaken and explored for confounding factors such as smoking before a statement can be made that rheumatoid factor is protective against honeycombing.