Respiratory tract infection and inflammation are well known to affect respiratory function, particularly in exacerbations of asthma and COPD, but there have been few studies into the action of infection on cough. We have not been able to show, based on our limited sample size, any data which might suggest an ongoing or latent respiratory infection in these patients, irrespective of the presence or absence of lymphocytic inflammation. We have also shown that BAL lymphocytosis does not appear to be related to cough reflex sensitivity in this study group.
Pathogens detected in the study subjects did not cause symptomatic infection, since this would either have explained their intractable cough or prevented them from undergoing a bronchoscopy. Although there have been many epidemiological studies investigating symptomatic respiratory infection, the incidence of asymptomatic carriage is less well known, although detection of persistent adenovirus and EBV has been documented in the lung [28, 29]. Nevertheless, the pathogens detected in our study are reflected in other publications [30–33] and low rates of PCR detected coincident infection have also been reported by other authors . For example, asymptomatic carriage of RSV and hMPV is rare. One study reported only one case of RSV and no cases of hMPV in 158 control patients , so it is unsurprising that these viruses were not detected in our subjects. In contrast, Chlamydophila pneumoniae frequently causes asymptomatic or low grade infection . Serum microimmunofluroscence tests to Chlamydophila sp. report 64.3% seropositivity in healthy volunteers  and serological studies of Chlamydophila pneumoniae infection in COPD have found a significantly increased positivity (33%) when compared to healthy controls (7%) . Overall, our study detected Chlamydophila sp. in 16.6% of subjects, 30% of controls but 10% of chronic cough patients, suggesting that prevalence might be lower in patients with chronic cough. There tended to be fewer detected pathogens in chronic cough group as a whole, compared with the healthy volunteer group, 20% vs 50% respectively but given the small sample size this was not significantly different. The protective effect of coughing, increased mucous production or a heightened immune state, might all account for this observation.
The numbers of detected pathogens were not different in lymphocytic and non-lymphocytic groups. The types of pathogens varied but the level of EBV infection in the lymphocytic population was very similar to that previously found in healthy volunteers by PCR . Although high levels of EBV infection have been reported in patients with COPD , the numbers of patients investigated here was not sufficient to infer that our finding was more than a chance observation.
Although an association has been suggested between BAL lymphocytosis and autoimmune disease in some idiopathic chronic cough patients, to our knowledge the possibility that infection could be responsible for inflammation and cough reflex sensitivity has not previously been investigated. We could not find any association between the level of BAL lymphocytosis and the number of detected pathogens or cough reflex sensitivity. In keeping with this finding, there was no difference in the objective cough counts between lymphocytic and non-lymphocytic groups [see Additional file 2, implying that neither local infection nor the inflammatory process is obviously linked to cough severity. It is known that the long-term, predominantly monocytic, pathology of chronic infection differs markedly from that of the predominantly neutrophilic acute phase , suggestive of distinct infective activities and host responses. Indeed the mechanisms which promote cough in URTI, might be very different from those evident in chronic cough. For example, patients with chronic airway diseases might react differently to non-isomolar solutions than those with URTI  supporting the notion that distinct neuronal mechanisms might be important in chronic cough.
It appears that an inflammatory process may be present within the lymphocytic patient population, since neutrophils and eosinophils are also raised in this group. The presence of such inflammation is common in a proportion of chronic cough patients [21, 41], although it is difficult to draw the conclusion from the data collected in our study that this inflammation is related to an infective or autoimmune process. Nonetheless, the association between lymphocytosis, auto-immune disease and the modulation of neuronal function is well established in conditions such as IBD, neuropathic pain syndromes and many other neurological conditions [42–45]. It must be noted though, that the symptoms in these diseases are often attributed to auto-immune mediated damage of peripheral nerves where some modalities of sensation can be exaggerated whilst others are lost. In IBD patients, and multiple rodent models of colitis, T-cell mediated, neutrophilic and eosinophilic, inflammation results in necrosis of enteric axons and significant neuroplasticity . The subsequent expansion of surviving neurons, observed in these models , probably accounts for the increased numbers of TRPV1 expressing neurons . A similar mechanism might account for increased TRPV1 expression in chronic cough . Moreover, some post-herpetic neuralgia (PHN) patients report pain associated with healing onset rather than the appearance of lesions . The mechanism by which the well documented nerve damage caused by autoimmune diseases might contribute to cough is not clear, although a form of infective or autoimmune ganglionitis is an intriguing possibility. Future studies should address this question.
To investigate the current airway pathogenic load we used PCR analysis of biopsy samples since serology documents past, but does not always indicate ongoing infection. An involvement of respiratory viruses or consequent autoimmune lymphocytosis in chronic cough cannot be completely ruled out. Firstly, this study was not powered to detect differences between controls and chronic cough patients for organisms with a high prevalence in the general population. Secondly, the effects of infection can outlast the pathogen  and auto-immunity itself may only become apparent after a substantial period of time. In addition, we cannot preclude the existence of a current but undetected/undetectable infection, in the lung or at a distal site not sampled. In the case of PHN, Zoster infection is thought to lead to T-cell infiltration to selective dorsal root ganglia. This inflammation is associated with significant loss of myelinated neurons and atrophy of the dorsal horn . Whilst VZV might still be detected in lymphocytic infiltrates in a number of ganglia, the affected peripheral tissue is not thought to be a viral reservoir .