**6. Community changes associated with treatment for acute pulmonary exacerbations**

Traditional dogma would dictate that intensive, targeted antimicrobial therapy with dual anti-pseudomonal agents will result in significant reductions in abundance of *P. aeruginosa*; however, in a similar vein to the findings from studies of the onset of exacerbations, microbiota responses to treatment for acute pulmonary exacerbations in CF have not aligned with this conventional understanding of infections.

One of the predominant themes that has emerged from studies of the respiratory microbiota response to acute antibiotics is that *P. aeruginosa* is not impacted to the same degree as other members of the community. For example, Daniels et al. studied 12 adult CF subjects across the cycle of an exacerbation and treatment and found that following initiation of anti-pseudomonal antimicrobials, the relative abundance of *P. aeruginosa* actually increased, alongside a reduction in the total number of species detected [45]. Cuthbertson et al. also found no evidence of reduced *P. aeruginosa* abundance in a study of 12 CF patients receiving treatment for pulmonary exacerbations [45]. Instead, reductions in *Streptococcus sanguinis*, *Prevotella* and *Porphyromonas* OTUs were observed [35]. Similarly, Li et al., in their analysis of over 700 sputum samples, found that antibiotic treatment had no effect on *Pseudomonas* abundance but did have significant effects on *Gemella*, *Staphylococcus*, *Actinomyces, Moraxellaceae* and *Fusobacterium* [37]. Further, Fodor et al. again found that dominant taxa such as *Pseudomonas* and *Burkholderia* were unchanged when compared at the beginning and end of an exacerbation but the relative abundances of *Gemella, Streptococcus* and a small number of other less abundant OTUs were all reduced [20].

In contrast, two studies have found reductions in *P. aeruginosa* following treatment. Firstly, Zemanick et al. investigated the association between inflammation and changes to the airway microbiota during treatment for exacerbations and found that although bacterial load did not change, the relative abundance of *P. aeruginosa* was observed to decrease and that these changes correlated with improved lung function [46]. A reduction in *P. aeruginosa* abundance following treatment was also reported by Smith et al., who noted rapid decreases in *P. aeruginosa* abundance and

an associated increase in diversity following the initiation of intravenous antibiotic treatment for pulmonary exacerbations in CF, although these changes were transient and returned to baseline following the completion of treatment [47].

There are a number of factors that may explain the differences between the studies mentioned in this section, and many of them apply to studies of the CF microbiome in general. The most obvious is the heterogeneous study designs, which are mostly retrospective and observational in nature and include a wide range of antibiotic regimens. For example, some authors such as Cuthbertson and Daniels included exacerbations treated with oral antibiotics as well as those requiring intravenous therapy [35, 45]. Milder exacerbations are often treated with oral antibiotics, and hence associated changes in the microbiota may also be expected to be more subtle. Even in those studies where only intravenous regimens were used, the antibiotic regimens or doses given are often not listed. The lack of a control or comparator group further makes interpreting results difficult [35, 45].

A further consideration is the sampling timeframes in each study, where again there exists a considerable variation that may have implications for interpreting results, particularly given that Smith et al. reported significant but transient reductions in *P. aeruginosa* abundance in the first few days of treatment [48].

Sample collection, storage, handling and DNA extraction techniques all also have the potential to impact on subsequent sequencing results. For example, multiple freeze–thaw cycles have been demonstrated to affect the results of microbiota analysis in respiratory samples [49]. Furthermore, different sequencing platforms can also produce different profiles [50].

There is no universally standardised protocol for the extraction of DNA from respiratory samples, and hence methods are often inconsistent between study groups. One obvious inconsistency is the use of propidium monoazide (PMA), a chemical compound that binds DNA in cells with damaged membranes and hence allows exclusion of non-viable DNA from sequencing. Excluding non-viable DNA has been suggested to be important for accurately identifying which members of the community are active at times of exacerbation and helps to avoid overestimation of viable microorganisms following treatment with antibiotics, but it is not utilised by all groups [51, 52]. There are concerns that PMA may incompletely penetrate sputum, hence only identifying a portion of non-viable cells. PMA is also known to stain viable cells of some species and stain dead cells in others [53]. In CF exacerbations, PMA treatment was not found to significantly alter the community as a whole, and only changes in low abundance 'satellite' taxa were apparent [51].

Overall, there is certainly evidence that acute antibiotic administration alters the respiratory microbiota; however, in the absence of prospective controlled trials, it is difficult to interpret these results given the confounders mentioned above. Indeed there have been calls for future clinical trials in CF to include biobanking of samples to allow a more rigorous scrutiny of the effect of antibiotic agents on the microbiome [54].

#### **7. Community changes associated with chronic suppressive antibiotics**

Inhaled antibiotics such as colistimethate (COL), tobramycin (TOB), aztreonam (AZLI) and levofloxacin (LIS) preparations are all licenced in the UK for the treatment of chronic *P. aeruginosa* infections and have, to varying degrees, demonstrated improvements in lung function and exacerbation rates as well as sputum density of *P. aeruginosa* [55–58]. However, despite the near ubiquitous use of these inhaled anti-pseudomonals in the chronic *P. aeruginosa* treatment in CF, the effect of these treatments on the microbiome remains poorly defined. Furthermore,

**9**

treatment.

*The Pulmonary Microbiome in Cystic Fibrosis DOI: http://dx.doi.org/10.5772/intechopen.91765*

many patients receive chronic macrolide therapy over many years, at least in part for its immunomodulatory effects, yet similarly little is known about the effects of this persistent selective pressure on the microbiota. When considering inhaled antibiotics, there is contrasting evidence as to their influence on the CF microbiome. For example, Kramer et al. [59] did not find any correlation between bacterial community structure and inhaled antibiotic treatments, although it is unclear which agents patients were using in that study. More recently Acosta et al. [24] utilised a prospectively collected Canadian sputum biobank primarily to investigate changes in CF cohort microbiota over time but also assessed whether different long-term antibiotics were associated with distinct microbiota. Eightytwo samples from 42 patients were sequenced, and those receiving long-term tobramycin and colistimethate harboured significantly different respiratory communities compared to those who were not, but interestingly no differences were seen in people receiving AZLI. The authors also assessed the impact of long-term oral azithromycin and nebulised dornase, but these agents were not associated with any differences. Perhaps intrigued by the failure of AZLI to be associated with microbiomic differences given its proven clinical benefits, the same group investigated its effects in more detail. In the only published study focussing on the microbiomic outcomes of a specific inhaled antibiotic, Heirali et al. [54] utilised the same Canadian biobank and sequenced 80 samples from 24 patients naive to AZLI and 82 samples from the same patients following initiation of AZLI. Overall no differences were observed in alpha or beta diversity measures, but at the OTU level, significantly lower relative abundances of *Prevotella* were seen following AZLI initiation. The authors then subclassified patients into AZLI 'responders' and 'non-responders' based on clinical outcomes and found 'non-responders' to have lower abundance of *Pseudomonas* and higher abundance of *Staphylococcus*. This novel approach raises the prospect of signatures in an individual's microbiota acting as a biomarker for response to antibiotics and may represent an important step in the march towards personalised precision medicine in CF. Further studies are required to explore the potential for an individual's microbiota to guide

**8. Community changes associated with CFTR modulators**

questions as to whether ivacaftor has an antimicrobial effect.

In the last 5 years, treatments targeted towards correcting the underlying defect in CF have become available. Ivacaftor, a cystic fibrosis transmembrane conductance regulator (CFTR) potentiator, is licenced specifically for the treatment of people with a G551D mutation and a number of other rare gating mutations, which together account for approximately 5–10% of the CF population in the UK [60]. In this subset of the CF population, ivacaftor use has been associated with improvements in lung function, reductions in exacerbations, reductions in sweat chloride, improved weight gain and improved quality of life [61, 62]. The restoration of CFTR activity by ivacaftor and the associated clinical benefits, in particular improved lung function and reductions in exacerbation, has inevitably raised

Theoretically, ivacaftor could have an antimicrobial effect in a number of ways. Firstly, the restoration of CFTR activity should result in a rehydrated airway surface layer, and this in turn will allow the mucociliary escalator to function physiologically. The improved clearance of airway secretions would then result in the elimination of bacteria. Secondly, the restoration of CFTR activity could result in a dramatic change in the local pulmonary microenvironment, turning a previously favourable environmental niche into an inhospitable one for resident

#### *The Pulmonary Microbiome in Cystic Fibrosis DOI: http://dx.doi.org/10.5772/intechopen.91765*

*Cystic Fibrosis - Heterogeneity and Personalized Treatment*

an associated increase in diversity following the initiation of intravenous antibiotic treatment for pulmonary exacerbations in CF, although these changes were transient and returned to baseline following the completion of treatment [47].

There are a number of factors that may explain the differences between the studies mentioned in this section, and many of them apply to studies of the CF microbiome in general. The most obvious is the heterogeneous study designs, which are mostly retrospective and observational in nature and include a wide range of antibiotic regimens. For example, some authors such as Cuthbertson and Daniels included exacerbations treated with oral antibiotics as well as those requiring intravenous therapy [35, 45]. Milder exacerbations are often treated with oral antibiotics, and hence associated changes in the microbiota may also be expected to be more subtle. Even in those studies where only intravenous regimens were used, the antibiotic regimens or doses given are often not listed. The lack of a control or

A further consideration is the sampling timeframes in each study, where again there exists a considerable variation that may have implications for interpreting results, particularly given that Smith et al. reported significant but transient reduc-

Sample collection, storage, handling and DNA extraction techniques all also have the potential to impact on subsequent sequencing results. For example, multiple freeze–thaw cycles have been demonstrated to affect the results of microbiota analysis in respiratory samples [49]. Furthermore, different sequencing platforms

There is no universally standardised protocol for the extraction of DNA from respiratory samples, and hence methods are often inconsistent between study groups. One obvious inconsistency is the use of propidium monoazide (PMA), a chemical compound that binds DNA in cells with damaged membranes and hence allows exclusion of non-viable DNA from sequencing. Excluding non-viable DNA has been suggested to be important for accurately identifying which members of the community are active at times of exacerbation and helps to avoid overestimation of viable microorganisms following treatment with antibiotics, but it is not utilised by all groups [51, 52]. There are concerns that PMA may incompletely penetrate sputum, hence only identifying a portion of non-viable cells. PMA is also known to stain viable cells of some species and stain dead cells in others [53]. In CF exacerbations, PMA treatment was not found to significantly alter the community as a whole, and only changes in low abundance 'satellite' taxa were apparent [51]. Overall, there is certainly evidence that acute antibiotic administration alters the respiratory microbiota; however, in the absence of prospective controlled trials, it is difficult to interpret these results given the confounders mentioned above. Indeed there have been calls for future clinical trials in CF to include biobanking of samples to allow a more rigorous scrutiny of the effect of antibiotic agents on the

**7. Community changes associated with chronic suppressive antibiotics**

Inhaled antibiotics such as colistimethate (COL), tobramycin (TOB), aztreonam (AZLI) and levofloxacin (LIS) preparations are all licenced in the UK for the treatment of chronic *P. aeruginosa* infections and have, to varying degrees, demonstrated improvements in lung function and exacerbation rates as well as sputum density of *P. aeruginosa* [55–58]. However, despite the near ubiquitous use of these inhaled anti-pseudomonals in the chronic *P. aeruginosa* treatment in CF, the effect of these treatments on the microbiome remains poorly defined. Furthermore,

comparator group further makes interpreting results difficult [35, 45].

tions in *P. aeruginosa* abundance in the first few days of treatment [48].

can also produce different profiles [50].

**8**

microbiome [54].

many patients receive chronic macrolide therapy over many years, at least in part for its immunomodulatory effects, yet similarly little is known about the effects of this persistent selective pressure on the microbiota. When considering inhaled antibiotics, there is contrasting evidence as to their influence on the CF microbiome. For example, Kramer et al. [59] did not find any correlation between bacterial community structure and inhaled antibiotic treatments, although it is unclear which agents patients were using in that study. More recently Acosta et al. [24] utilised a prospectively collected Canadian sputum biobank primarily to investigate changes in CF cohort microbiota over time but also assessed whether different long-term antibiotics were associated with distinct microbiota. Eightytwo samples from 42 patients were sequenced, and those receiving long-term tobramycin and colistimethate harboured significantly different respiratory communities compared to those who were not, but interestingly no differences were seen in people receiving AZLI. The authors also assessed the impact of long-term oral azithromycin and nebulised dornase, but these agents were not associated with any differences. Perhaps intrigued by the failure of AZLI to be associated with microbiomic differences given its proven clinical benefits, the same group investigated its effects in more detail. In the only published study focussing on the microbiomic outcomes of a specific inhaled antibiotic, Heirali et al. [54] utilised the same Canadian biobank and sequenced 80 samples from 24 patients naive to AZLI and 82 samples from the same patients following initiation of AZLI. Overall no differences were observed in alpha or beta diversity measures, but at the OTU level, significantly lower relative abundances of *Prevotella* were seen following AZLI initiation. The authors then subclassified patients into AZLI 'responders' and 'non-responders' based on clinical outcomes and found 'non-responders' to have lower abundance of *Pseudomonas* and higher abundance of *Staphylococcus*. This novel approach raises the prospect of signatures in an individual's microbiota acting as a biomarker for response to antibiotics and may represent an important step in the march towards personalised precision medicine in CF. Further studies are required to explore the potential for an individual's microbiota to guide treatment.
