Pseudomonas aeruginosa in the Cystic Fibrosis Airway: Does It Deserve Its Reputation as a Predatory “Bully”?

To the Editor:

Despite similarities in global standards of care in cystic fibrosis (CF), one long-standing area of regional disagreement is prevention of Staphylococcus aureus (Sa). UK clinics advocate prophylaxis; U.S. centers do not. However, cross-sectional data from both national registries show Sa being common in childhood with Pseudomonas aeruginosa (Pa) dominating later (1, 2). This is often ascribed to virulent Pa “outcompeting” Sa (3–5). Conversely, cooperative behaviors have been described (5).

Fischer and colleagues recently analyzed a longitudinal data set from a U.S. CF center (6). They described long-term, sustained coinfection with Pa and Sa (many methicillin resistant). The accompanying editorial (7) highlighted the provocative nature of these findings, which seemingly undermine a role for competitive inhibition in population trends. Both the original authors and editorial questioned the generalizability of the findings, suggesting that replication in other cohorts, differentiated by age or geography, would be useful.

Given the regional differences in Sa epidemiology, prevention, and treatment, we considered that we could substantially add to this debate. Our anonymized data set from the Royal Brompton Hospital was compiled to explore interorganism competition. Pa and Sa interaction was not the original focus of this work, but the data provided us with an opportunity to address some of the editorial’s questions. All patients had consented to data entry into the national CF patient registry; approval was issued by the registry for this study. Data are median (interquartile range).

The data set, spanning 6 years (2012–2017), comprised >16,000 positive bacterial or fungal growths from sputum or BAL (1,001 patients); cough/throat swab data were not collected (poor sensitivity for fungal detection). Here, we analyze results from 248 patients who contributed ≥10 samples and isolated Sa and Pa (alone or together) at least once during this period. We hypothesized that inhibitory competition would be evidenced by 1) copositive samples being uncommon and 2) temporal and age-related skewing away from copositivity (Pa+ and Sa+) and changing frequencies of the organisms as Pa gained its advantage.

A total of 8,040 samples were included, 42 (26–58) per patient over 5.3 (3.6–5.6) years. The age at sampling midpoint was 28.8 (21.4–38.8, range 6.3–72.0) years (Figure 1). Across all samples, coinfection was very common: Pa with Sa 30%, Pa without Sa 61.5%, and Sa without Pa 3.5%. On a per-patient basis, the median Pa and Sa copositive rate was 24% (7–66%). The majority of patients (79% [196/248]) only isolated Sa in samples copositive for Pa; only 21% (52/248) had any cultures of Sa alone. In this group of 52, Pa and Sa copositivity (28% [13–47%]) was seen significantly more often (P < 0.02) than was Sa alone (9% [5–38%]).

Figure 1.

Here, we address the question of whether sample copositivity is more common later in the disease course. On a group basis (n = 248), we explore the frequency of samples copositive for both Pa and Sa as a function of an individual’s age. Patients contributed samples for a median (interquartile range) of 5.3 (3.6–5.6) years, so for the purposes of standardization, they are plotted here against their age at study midpoint. The percentage of individuals’ copositive samples throughout the study period ranged from 0% to 100% (median, 24%), but no relationship was seen with age. Pa = Pseudomonas aeruginosa; Sa = Staphylococcus aureus.

“Early” Pa isolates are more virulent than those adopting a chronic lifestyle (8–10), suggesting that the former may be better equipped to compete. We could not accurately determine Pa chronicity owing to data collection methods, so we used the surrogate of mucoidy. When comparing samples with only nonmucoid or mucoid Pa (ignoring mixed growths), proportions of Pa and Sa copositivity were no different. Reflecting known global differences, our UK sample set included fewer methicillin-resistant Sa (MRSA) isolates (10.9%) than that of Fischer (59.0%); we found no difference between MRSA and methicillin-sensitive Sa on the prevalence of Pa coinfection.

When assessing changes over time on the entire sample set, we saw no shifts in copositive samples: frequency of Pa and Sa copositivity showed no relationship with patient age (Figure 1) or year of data capture. Recognizing the relatively short duration of the study, we next compared extremes of our period: first three and last three cultures for each subject (Figure 2). Here, we observed significant (chi² P < 0.001) temporal changes: decreasing prevalence of Sa (both alone and in copositive samples) with simultaneous increasing prevalence of Pa alone. In the 52 subjects with any Sa-alone samples, the prevalence of these was significantly (P < 0.01) lower in their last three samples compared with their first.

Figure 2.

Subjects were categorized based on results of their first three samples and last three samples as 1) neither Pa nor Sa on any sample; 2) Pa only; 3) Sa only; 4) Pa and Sa, cultured together from same sample at least once; or 5) Pa and Sa both cultured, but never from the same sample. The majority of patients initially had Pa and Sa together (57.6%) or Pa alone (33.9%). Most of the change between the first three and last three samples occurred between these groups (hashed areas); significantly fewer people grew Pa and Sa together (P = 0.003), and significantly more fell into the Pa-only group (P = 0.003). Pa = Pseudomonas aeruginosa; Sa = Staphylococcus aureus.

Thus, with this UK data set, we confirm a primary finding of Fischer and colleagues that coinfection is common and sustained into older ages. However, with careful scrutiny, in contrast to their findings, we could detect subtle temporal shifts toward Pa dominance; this could support the concept of competitive inhibition, or it could reflect the evolution of an airway environment more favorable for Pa growth as disease progresses. The absence of this in the Fischer study may reflect regional management differences, approach to analysis, or biological differences between methicillin-sensitive Staphylococcus aureus and MRSA lending the latter a survival advantage.

There are limitations to our study. Cough swabs were not included, excluding the youngest patients and the time of highest Sa prevalence. The requirement for both Sa and Pa at least once likely further reduced the number of people with Sa-alone samples. Conventional cultures were assessed, not molecular analyses; interspecies competition ex vivo leading to false-negative cultures could have misled. The short per-subject time period limits longitudinal analysis of any transition. Mitigating this, the very large sample set and wide age range of our subjects means that data were captured from across the life course from preschool to old age.

Our data demonstrate that sustained coexistence of Pa and Sa is possible and is common; whether this reflects “cooperation” or “indifference” cannot be determined. In vivo interspecies competition may also occur, mirroring in vitro experimental findings; however, despite our very large study size, the impact of this was subtle, only apparent when comparing temporal extremes. It would thus appear to be inadequate as a primary explanation of the very large age-related cross-sectional trends in infection status seen in registries. Importantly, neither we nor Fischer present genetic or expression analyses of these organisms. There is substantial variability between strains; we consider it likely that some exhibit competitive or predatory behaviors whereas others favor coexistence. This would support research in a more “personalized” fashion, to enable an understanding of the dynamics within an individual’s lung. This is unlikely to be possible on retrospective data sets.

This story therefore remains incompletely understood and is likely to change further as we enter the era of highly effective modulator therapies for most patients. We strongly endorse the proposal from Wolter and Ramsey that the issue is further studied in a prospective, international, longitudinal study of airway infection from a wide age range of patients. We would recommend that both culture and molecular techniques are used and that evolving patterns are fully explored on a genetic/transcriptome level, to fully inform the relevance of this work to clinical outcomes.