Breathing Hope into Directed Therapy for Pulmonary Infections

Empiric therapy for respiratory infections, including pneumonia and exacerbations of chronic obstructive pulmonary disease (COPD), has remained the norm despite decades of promise that new diagnostic techniques and platforms would deliver clinicians accurate, timely, and affordable information on the pathogen(s). Although microbiological tests have been part of our standard of care for patients at risk of unusual or antibiotic-resistant pathogens, they have had little to no impact on initial therapy. As each successive pneumonia guideline has pointed out (1, 2), there is no evidence that traditional diagnostic offerings from the laboratory have any meaningful impact on patient outcomes or clinician behavior in usual settings. Equally, in the setting of acute exacerbations of COPD, it is well recognized that although viral pathogens are extremely frequent, there is substantial overuse of antibiotics owing to clinical uncertainty over the pathogen(s). Although recently some small gains had been made, such as the use of rapid-diagnostic platforms to screen for methicillin-resistant Staphylococcus aureus (3), the sense of promise molecular methods engendered in the 1990s still remained to be realized.

Then, along came coronavirus disease (COVID-19) and changed the world’s perspective on the importance of having the ability to rapidly determine the pathogen(s) in play. In the last 18 months, we have seen a massive uplift in the capability of “ordinary” hospitals to rapidly process respiratory samples, driven by clinical need and loosening of the normally tight fiscal controls over new technology. A host of platforms are now in widespread use, and when COVID-19 finally retreats into being just one of the usual pathogens we need to deal with each winter, there is likely to be an explosion of knowledge arising from this vast increase in availability and use of the latest generation of diagnostic tools.

However, if we are to truly drop concepts like “community-acquired pneumonia” or “acute exacerbation of COPD” and move to pathogen-specific therapeutic approaches, a number of hurdles need to be crossed. Although the designs of many of the new platforms are amenable to point-of-care use, in most settings, they remain based in traditional laboratories, adding complexity and time to a time-critical decision process. Breaking down traditional fiefdoms and developing greater trust in their use as true point of care devices will take time. The challenge of altering clinician behavior, particularly in using diagnostic data to reduce the use of broad-spectrum antimicrobials, should not be underestimated, as it has been a struggle to achieve this with conventional microbiological tests (4–7). We will need new trials to tell us what the significance of multiple pathogens being present means: are these sequential or concurrent infections, and does the presence of one or more additional pathogens alter the treatment or prognosis?

Perhaps the biggest barrier, however, remains getting adequate specimens from patients for analysis. Obtaining an adequate sputum specimen is a challenge in somewhere between one-third and one-half of patients with community-acquired pneumonia (7, 8) and even more in nonventilated hospital-acquired pneumonia (9). In acute exacerbations of COPD, obtaining sputum is easier but still not achievable in a significant minority of patients (10). This is a fundamental limitation for molecular techniques. Nasopharyngeal aspirates have their own technical challenges, and, although now well accepted for many viral infections and not harmful, they remain highly questionable for bacterial infections and can be unpleasant. Saliva or oral wash samplings have proven useful in COVID-19 (11) as they have Pneumocystis (12), but, again, their utility in bacterial infection is problematic and unlikely to be high. How, then, do we move forward if our current generation platforms are not going to help us in a sizeable proportion of patients with lower respiratory tract infection?

In this issue of the Journal, Kamal and colleagues (pp. 1075–1085) show one possible way to progress (13). In their study of infected airway epithelial cell cultures, exhaled breath samples from healthy subjects challenged with rhinovirus and patients with COPD, Kamal and colleagues (13) used use highly sensitive gas-chromatography mass spectrometry to detect volatile organic compounds (VOC) as diagnostic signatures of infection. VOC are expelled in the airways in response to infection. Bacteria release VOC as part of their metabolic process, and many have characteristic signatures (14). In both bacterial and viral infections, VOC are released as part of the inflammatory process from airway epithelial cells (15), and whether these have diagnostic signatures was addressed by Kamal and colleagues (13).

Through screening a variety of VOC, they showed that in tissue culture, rhinovirus infection increased the production of decane, but Streptococcus pneumonia and Haemophilus influenza did not, suggesting this maybe a useful marker for differentiating viral from bacterial infection. They next demonstrated that infecting healthy volunteers with rhinovirus induced the production of decane and other long-chain alkanes in proportion to other inflammatory responses and to viral load. Finally, in a cohort of 139 patients with COPD, they assessed VOC in exhaled air while stable and again during exacerbation in the 98 who did so during the period of the study. In viral exacerbations, the long-chain alkane 2,9 methyl undecane was significantly increased and correlated with the severity of exacerbation. In patients with bacterial, mixed viral/bacterial, or no pathogen identified, 2,9 methyl undecane was not elevated, suggesting that it may be a useful marker of pure viral infection and therefore potentially a useful tool for limiting antibiotic use in the setting of lower respiratory tract infections.

Clearly the work by Kamal and colleagues is not yet a point-of-care diagnostic box and has many steps to go before reaching that point. However, the appeal of a diagnostic tool patients simply breath into that can rapidly determine whether an infection is viral or bacterial, and if so, which bacteria, is immensely appealing. Such an “electronic nose” may also have applications well beyond infection, as characteristic VOC signatures have been shown to be present in many systemic disease processes, including a variety of cancers (16). Thanks to COVID-19, enthusiasm to progress such technology is likely to be high, and, hopefully, unlike molecular diagnostic tools, we will not have to wait three decades to see a fully functional product in our hospitals.