To the Editor:
We are concerned about the alleged aetiological role of Legionella spp. in acute exacerbations of chronic obstructive pulmonary disease (AECOPD), as has been recently suggested by both Lieberman et al. 1 in their study published in the European Respiratory Journal, and Ewig 2 in an editorial in the same issue 2. To the best of our knowledge, the distinctive results of the study by Lieberman et al. 1 have not been reported in any other aetiological survey of AECOPD. A recent, in-depth review of bacterial infection in chronic obstructive pulmonary disease, by a worldwide expert on this subject, does not even mention any role for Legionella spp. in AECOPD 3. Although attachment to bronchial cells is the first step of Legionella spp. infection, it is well known that the primary pathogenic process takes place at the macrophage cell level. There are no consistent data that might favour a theoretical isolated infectious bronchitis without any pneumonic involvement 4. In fact, Legionella spp. isolation remains the gold standard for diagnosing any form of Legionella infection 5. An exhaustive review of AECOPD studies that have focused on the search of an aetiological agent by means of invasive procedures, such as bronchoscopic techniques, shows that Legionella spp. have never been identified until now. Moreover, the provided clinical data do not seem to correspond with the clinical syndrome of Pontiac fever.
Doubts about the validity of serological testing for a reliable diagnosis of Legionella infection can definitely be seen if the accumulated experience on these techniques during the last 25 yrs is reviewed. There are two well-known drawbacks of serology: low sensitivity, even for Legionella pneumophila serogroup 1 (<70%) 5, and a serious concern about specifity 6. Specificity (∼96–99%) has only been acceptably established for L. pneumophila serogroup 1. Cross-reactivity between L. pneumophila serogroup 1 and other serogroups and species has been consistently reported in the literature 7.
The risk of false-positive results should then not be neglected since the use of a completely nonspecific diagnostic method in a low prevalent disease, such as legionellosis, will, from a statistical point of view, undoubtedly increase the likelihood of false-positive reactions. This risk must be even higher when considering serology for non-L. pneumophila serogroup 1 infections and especially Legionella spp. other than L. pneumophila 8. In fact, as Ewig 2 points out, if only L. pneumophila serogroup 1 had been evaluated, the incidence of alleged Legionella infection would have decreased to 4%. The concept of unreliable specificity of immunofluorescent antibody (IFA) for diagnosing clinically relevant infections by Legionella is also supported by some studies. Andersen et al. 9 found a clinically silent four-fold indirect IFA seroconversion in their prospective study on annual (3–5 yrs) serum specimens among 52 children, more than 20-yrs ago. More recently, a four-fold IFA asymptomatic seroconversion has also occasionally been observed in high-risk populations such as adult renal transplantation patients 10. In a study by Dowling et al. 11, 7% of their 89 receptors seroconverted during the 6 months after transplantation, without any clinical evidence of pneumonia. In a recent cohort study, after an outbreak of travel-associated Legionnaires disease and Pontiac fever, 3–6% of patients who were not ill showed immunoglobulin (Ig)G seroconversion or IgM seropositivity, respectively 12.
A variety of possible serological cross-reactions have been reported in the literature. Among them, some Gram-negative bacteria and anaerobic microorganisms, including Pseudomonas, Proteus, Bordetella and Bacteroides fragilis, may sometimes be incriminated in AECOPD 6. Moreover, cross-reactions have also been reported with certain microorganisms that may be endemic or cause epidemic outbreaks in determined geographical areas, such as Ricketssia conorii, the agent of Mediterranean fever, R. typhii, the agent of murine typhus, Coxiella burnetii, and Campylobacter spp. 13–16. Synchronous serological studies would perhaps have been appropriate. We could also speculate on a possible cross-reactivity of Legionella spp. with other less studied common human pathogens that share some similarities. For example, serological cross-reaction between Capnocytophaga spp. and Legionella spp. has been reported 17 and these periodontal bacteria seem to be frequently identified by polymerase chain reaction (PCR) assay in apparently periodontally healthy subjects 18. In addition, Helicobacter pylori has shown cross-reactivity with L. micdadei 19. This finding could be of interest if further studies confirm that preliminary report, since H. pylori seropositivity has been reported to be common both in chronic bronchitis 20 and bronchiectasis 21. In fact, chronic bronchitis has been identified as a predictive factor of seropositivity for IgG antibodies to H. pylori 22.
In any case, if we accept that seroconversion means real infection, it is worth remembering that, in their outbreak case-control study, Boshuizen et al. 23 recently observed that control seroconvertors did not show any statistically significant clinical difference when compared to nonseroconverters. Their results would suggest then that Legionella infection could theoretically produce either pneumonia or just asymptomatic infection.
- © ERS Journals Ltd