If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Pediatrics, National Jewish Health, Denver, CO, United StatesDepartment of Medicine, National Jewish Health, Denver, CO, United StatesUniversity of Colorado School of Medicine, Aurora, CO, United States
Concomitant use of oral azithromycin and inhaled tobramycin occurs in approximately half of US cystic fibrosis (CF) patients. Recent data suggest that this combination may be antagonistic.
Methods
Test the hypothesis that azithromycin reduces the clinical benefits of tobramycin by analyses of clinical trial data, in vitro modeling of P. aeruginosa antibiotic killing, and regulation of the MexXY efflux pump.
Results
Ongoing administration of azithromycin associates with reduced ability of inhaled tobramycin, as compared with aztreonam, to improve lung function and quality of life in a completed clinical trial. In users of azithromycin FEV1 (L) increased 0.8% during a 4-week period of inhaled tobramycin and an additional 6.4% during a subsequent 4-week period of inhaled aztreonam (P < 0.005). CFQ-R respiratory symptom score decreased 1.8 points during inhaled tobramycin and increased 8.3 points during subsequent inhaled aztreonam (P < 0.001). A smaller number of trial participants not using azithromycin had similar improvement in lung function and quality of life scores during inhaled tobramycin and inhaled aztreonam. In vitro, azithromycin selectively reduced the bactericidal effects tobramycin in cultures of clinical strains of P. aeruginosa, while up regulating antibiotic resistance through MexXY efflux.
Conclusions
Azithromycin appears capable of reducing the antimicrobial benefits of tobramycin by inducing adaptive bacterial stress responses in P. aeruginosa, suggesting that these medications together may not be optimal chronic therapy for at least some patients.
In clinical trials testing suppressive treatment of chronic P. aeruginosa infections in people with cystic fibrosis (CF), both inhaled tobramycin and oral azithromycin demonstrate clear benefit to lung function and quality of life [
]. Concomitant use of the two agents is now common, but has not been tested prospectively. Several independent in vitro studies have reported antagonism between azithromycin and tobramycin against P. aeruginosa [
]. We have hypothesized that combined therapy with these two medications may unexpectedly provide less rather than more clinical improvement.
Herein we report the post-hoc analyses of a second CF clinical trial in which subjects received 4 weeks of inhaled tobramycin immediately preceding four weeks of inhaled aztreonam [
]. Focusing on outcomes within subjects reporting use of azithromycin, we observe a pattern of interaction in which concomitant azithromycin treatment was associated with poorer response to inhaled tobramycin compared with inhaled aztreonam. This differential pattern of effect between classes of inhaled antibiotics was not observed in a smaller group of subjects not using azithromycin, which is consistent with our recent post-hoc analyses of parallel subject groups exposed to inhaled tobramycin or aztreonam in a separate clinical trial [
]. To test directly if azithromycin may impair the response to tobramycin, as opposed to enhancing the response to aztreonam, we tested bacterial killing with these and other clinically relevant antibiotics using 30 CF clinical isolates of P. aeruginosa collected from unique individuals. Antimicrobial activity was selectively reduced when azithromycin was added to tobramycin and had either no effect or improved killing when added to several other anti-pseudomonal antibiotics used in CF clinical care. Importantly, we found that azithromycin, particularly when combined with tobramycin, greatly increased gene expression of a Pseudomonas efflux pump, MexXY, which is a central mechanism of inducible aminoglycoside resistance [
]. Both aminoglycoside and macrolide antibiotics target the bacterial ribosome. Overlap in the bacterial response to ribosomal perturbation from each antibiotic may help explain how chronic azithromycin can lessen the antimicrobial effectiveness of tobramycin [
]. Genetic removal of mexX in P. aeruginosa resulted in an additive rather than antagonistic antimicrobial pattern between azithromycin and tobramycin.
Our prior and current post-hoc analyses of CF clinical trials, combined with the complementary in vitro findings, provide additional evidence and biological plausibility indicating that in the setting of chronic P. aeruginosa infection, the common clinical practice of combining oral azithromycin with inhaled tobramycin as a maintenance therapy may, in some patients, be less effective than tobramycin alone. This contrasts with some earlier reports of synergy when combining azithromycin and tobramycin in vitro [
Synergistic activities of macrolide antibiotics against Pseudomonas aeruginosa, Burkholderia cepacia, Stenotrophomonas maltophilia, and Alcaligenes xylosoxidans isolated from patients with cystic fibrosis.
Macrolides decrease the minimal inhibitory concentration of anti-pseudomonal agents against Pseudomonas aeruginosa from cystic fibrosis patients in biofilm.
]. The methods used for multiple combined antibiotic testing vary and have produced, at times, inconsistent findings. Our approach to this question of drug interaction has begun with observations in clinical data sets, which inspired both microbiological and mechanistic research based on published work of others. Our bacterial culture methods are distinct from those used in prior testing and it is not known which in vitro model may best predict clinical response. The discrepant findings and potential clinical importance of such an adverse drug interaction indicates that confirmatory, prospective testing and additional mechanistic research are necessary; therefore, we are now conducting a dedicated, prospective clinical trial and continue to investigate how P. aeruginosa responds to this antibiotic combination. Some of these data have previously been reported in abstracts or oral presentation [
De-identified data from the completed AIR-CF2 clinical trial (Clinicaltrials.gov NCT00104520) were requested through an investigator initiated research process and provided by the sponsor, Gilead Sciences [
]. In this trial, subjects with CF and chronic P. aeruginosa all received 4 weeks of inhaled tobramycin immediately followed by 4 weeks of inhaled aztreonam or placebo. We obtained the following from all subjects who consented to share data: azithromycin use, gender, FEV1 (Liters and % predicted), scores on cystic fibrosis quality of life-revised respiratory symptom scale (CFQ-R RSS) [
], and sputum density of P. aeruginosa. At enrollment, 128 of 176 subjects (73%) reported concomitant azithromycin use. Baseline mean FEV1% predicted and gender did not significantly differ based on azithromycin use (FEV1 58% vs. 54%) or randomization to aztreonam vs. placebo (see Table S1). All subjects without missing data, including those randomized to placebo, were included in our analysis when applicable. This results in larger groups during the tobramycin study phase compared with aztreonam study phase. Subjects randomized to twice vs. thrice daily inhaled aztreonam demonstrated similar outcomes and were pooled for our analyses, as previously done by others in the original publication of this trial [
30 CF clinical isolates of P. aeruginosa were collected from 3 separate sources: 10 from Gilead Sciences banked during a clinical trial of CF subjects [
], 10 from National Jewish Health Microbiology Laboratory collected from adult CF patients in Denver, CO, and 10 randomly selected from the CF Isolate Core at Seattle Children's Research Institute. Additional details on this culture method are available in the on-line supplement and comprehensive characterization and comparison with planktonic cultures have previously been published [
]. In brief, bacteria were cultured with gentle, constant movement in low-nutrient media with human plasma and lysed human neutrophils added to form Pseudomonas aeruginosa aggregates over 48 h. Antibiotics were then added to cultures for an additional 16 h, at which point bacterial density was measured.
Gene expression was measured by qPCR at 5 time points between 5 and 240 min following addition of antibiotics to biofilm aggregates in a subset of the same clinical isolates of P. aeruginosa (N = 13). The following genes were tested for maximum expression in the 4 h following antibiotic challenge: PA5471, mexX, mexY, mexA, mexB. RNAprotect Bacteria Reagent (QIAGEN) was added to samples, which were stored at −80 °C. RNA was extracted with the QIAGEN QIACube and RNeasy kit (QIAGEN) using previously calculated amount of RNA, Random Hexamer (IDT 51–01–18-15), M-MLV Reverse Transcriptase (ThermoFisher Scientific 28,025–031) and 2.5 mM dNTPs (ThermoFisher Scientific AM8200) per manufacturer protocol. SYBR Green 2× (Maxima SYBR Green qPCR Master Mix ThermoFisher Scientific K0253) and 7.4ul nuclease-free water were used in a 20ul qPCR reaction, along with the primers provided in the online supplement (PA5471, mexX, mexY, mexA, mexB, IDT DNA). Relative gene expression was calculated using ∆CT values and internal control rpoD gene expression. Control cultures of each bacterial strain grown under identical conditions were used to determine the effect of antibiotic exposure on gene expression. Additional details are provided in the on-line supplement.
2.3 Disruption of mexX using a mobile group II intron
A unique plasmid, pCH1, was constructed using the previously generated targetron donor plasmid pBL1 (generous gift from Alan M. Lambowitz and Jun Yao, PMID:17322321) and pACDK4-C-loxP (Sigma). pCH1 was then electroporated into the pseudomonas laboratory strain PAO1. Targetron expression was induced using 2 mM m-toluic acid, and bacteria with disrupted mexX were selected for using kanamycin at a concentration tested to kill non-induced strain. The wild type PAO1 and mexX knockout strains were grown using the same aggregate model and antibiotic concentrations described above, and experiments were replicated 4 times.
2.4 Statistical analysis
FEV1 is reported as relative change in FEV1 volume (L, liters). Absolute change in FEV1 (L) is shown in the supplement and is consistent with relative change in FEV1 (L).
Statistical analyses were conducted using Prism (GraphPad, La Jolla, CA). Two-sided, unpaired T-tests with Welch's correction were used to analyze changes in FEV1 and CFQ-R Respiratory Symptom Score. These comparisons included all available complete data from subjects using azithromycin during the 4-weeks of inhaled tobramycin, including those randomized to subsequent placebo rather than aztreonam. Data from bacterial aggregate experiments were tested using two-sided, paired nonparametric test (Wilcoxon matched pairs signed rank) for comparison of concurrent cultures of each clinical isolate with addition of azithromycin vs. the same isolate with the anti-Pseudomonal antibiotic alone (e.g. tobramycin plus azithromycin vs. tobramycin alone). Gene expression was compared using two-sided, unpaired T-tests with Welch's correction, and correlation testing was made using Pearson correlation coefficient and two-tailed P values. The effect of interrupting the mexX gene on antibiotic sensitivity in the PAO1 bacterial strain was tested using a two-sided, unpaired, nonparametric T-test (Mann–Whitney test). Scatter plots or box and whisker graphs for data not presented in this way in the manuscript are shown in the on-line supplement, reflecting individual values and variability observed in both clinical analyses and lab experiments. Additional correlation coefficients and p values are also shown in the supplement.
3. Results
Data on concomitant medication use were collected during a clinical trial in CF subjects designed to compare twice vs. thrice daily aztreonam lysine by inhalation (Clinicaltrials.gov NCT00104520) [
]. All study subjects received 4 weeks of inhaled tobramycin solution twice daily immediately before 4 weeks of inhaled aztreonam lysine or placebo. Seventy-three percent of subjects in our dataset reported concomitant azithromycin use. As a group, users of azithromycin demonstrated little increase in lung function as measured by FEV1 over a 4-week period of inhaled tobramycin (Fig. 1a . mean change in FEV1 (L) 0.8%, N = 108). During the following 4-week aztreonam period, users of azithromycin showed significantly greater increase in FEV1 (mean 6.4%, N = 71, P < 0.005). Subjects not using azithromycin had similar increase in FEV1 during the 4-week periods of inhaled tobramycin and subsequent inhaled aztreonam (mean 2.6% and 3.6%, N = 40 and 22 respectively). Across the entire 8-week study period, users and non-users of azithromycin had similar improvement in FEV1 of approximately 6–8%; however, in those using azithromycin, the increase in lung function was largely observed during the period of aztreonam use (Fig. 1b).
Fig. 1Relative change in FEV1 (L) during 4-week use of inhaled tobramycin followed by 4-week use of inhaled aztreonam. Panel A: Mean (SEM) improvement in lung function for users of azithromycin (AZM, left) and non-users of azithromycin (No AZM, right). Panel B: Improvement in FEV1 (L) during the 8-week study period reflecting inhaled tobramycin (TOB) immediately followed by inhaled aztreonam (ATM) in subjects completing both inhaled antibiotic study periods. AZM users; N = 108 during tobramycin and 71 during aztreonam. AZM non users; N = 40 during tobramycin and 22 during aztreonam.
Determination of the minimal clinically important difference scores for the cystic fibrosis questionnaire-revised respiratory symptom scale in two populations of patients with cystic fibrosis and chronic Pseudomonas aeruginosa airway infection.
]. Consistent with the pattern of effect observed with FEV1, users of azithromycin had, on average, a decline in self-reported, disease-related quality of life during the 4-week period of inhaled tobramycin. This group reported a significant increase in this measure during the subsequent 4-week period of aztreonam (Fig. 2, TOB period mean − 1.8, N = 102, ATM period mean 8.3, N = 68. P < 0.001). Subjects not using azithromycin reported an increase in quality of life during the tobramycin period and a further increase with the subsequent aztreonam period of the study. When excluding subjects randomized to placebo rather than aztreonam, the mean total change in CFQ-R RSS from baseline for subjects completing the 8 weeks study period including both inhaled tobramycin and inhaled aztreonam was 4.6 points in users of azithromycin (N = 73) and 9.7 points in non-users of azithromycin (N = 23), a difference that was not statistically significant.
Fig. 2Absolute change in CFQ-R Respiratory Symptom Scale. Mean (SEM) improvement in self-reported, disease-related quality of life for users of azithromycin (AZM, left) and non-users of azithromycin (No AZM, right). The dashed line represents the threshold for clinically significant improvement. AZM users; N = 102 during tobramycin (TOB) and 68 during aztreonam (ATM). AZM non users; N = 39 during tobramycin and 22 during aztreonam.
Some subjects were unable to produce sputum at various time points during the study, reducing the available sample size for analysis of change in sputum P. aeruginosa bacterial density. Subjects in the non-azithromycin group had greater reduction in sputum P. aeruginosa density during 4-weeks of inhaled tobramycin than those in the azithromycin group, but this difference did not reach statistical significance. A further modest reduction in bacterial density during the subsequent 4-weeks of inhaled aztreonam was observed in subjects regardless of azithromycin use (Fig. 3).
Fig. 3Change in sputum P. aeruginosa density. Both users (AZM) and non-users (No AZM) of azithromycin demonstrated a reduction in sputum P. aeruginosa density over 4-weeks use of inhaled tobramycin (TOB). A further reduction in P. aeruginosa density was present in both users and non-users of azithromycin during 4-weeks use of inhaled aztreonam (TOB + ATM). AZM users; N = 85 during tobramycin and 53 during aztreonam. AZM non-users; N = 27 during tobramycin and 16 during aztreonam.
]. The P. aeruginosa aggregates produced by this model may better exhibit the morphology and antibiotic tolerance of bacteria living in the CF airway [
]. After 48 h of bacterial growth with gentle agitation, antibiotics were added at relatively high concentration for 16 h to model clinically relevant airway concentrations achieved in CF drug therapy. Tobramycin (100 μg/mL) resulted in significant bacterial killing of 4 log10 compared with no antibiotics. Azithromycin (20 μg/mL) showed notable but more modest effects with 2 log10 killing. Adding azithromycin to tobramycin at the same concentrations was significantly less effective than tobramycin alone (Fig. 4a , P < 0.0001).
Fig. 4P. aeruginosa density of biofilm aggregate cultures. Panel A: azithromycin (AZM), tobramycin (TOB); N = 30 isolates from unique persons with CF, each tested once; log10 CFU graphed as mean + SEM. Panel B: Difference in CFU density when azithromycin is added to comparative anti-pseudomonal antibiotics, including: amikacin (AMK), colistin (COL), levofloxacin (LVX), and tobramycin (TOB) (N = 30, log10 CFU graphed as mean + SEM). Statistical comparison is the change from zero X-axis, which represents the value from simultaneous cultures grown with the antibiotic of interest but without addition of azithromycin.
Antibiotics other than tobramycin that have been used as inhalational therapies in CF lung disease were also tested in the biofilm aggregate model (Fig. 4b). Consistent with our analyses of the clinical trial data, no evidence of antagonism was observed between azithromycin and aztreonam (ATM, Fig. 4b). Azithromycin did not alter the effect of amikacin (AMK), and increased bacterial killing was observed with addition of azithromycin to colistin (COL) and levofloxacin (LVX) (*P < 0.001).
The bacterial aggregate model was also used to test the effect of antibiotic exposure on known efflux pumps capable of reducing the intracellular concentration of aminoglycosides, including tobramycin. The PA5471 gene product responds to ribosomal perturbation and positively regulates the MexXY efflux pump [
]. PA5471, mexX, and mexY gene expression all increased within the 4 h following antibiotic challenge. Gene expression was greater in aggregates challenged with azithromycin or azithromycin and tobramycin than with tobramycin alone. (Fig. 5a-c , #P < 0.02, *P < 0.01)
Fig. 5Gene expression of PA5471, mexX, and mexY. Panel A: maximum PA5471 gene expression across all time points tested in clinical P. aeruginosa isolates (mean and each value shown). Data represent fold induction vs. control cultures with no antibiotic added. Panel B: mexX expression. Panel C: mexY expression.
Strong positive correlation occurred between PA5471 and mexX or mexY expression in pseudomonas aggregates challenged with azithromycin alone or when combined with tobramycin (Fig. 6A , Pearson r between 0.63–0.75, P < 0.01. Azithromycin shown in square, azithromycin plus tobramycin shown in circle).
Fig. 6Correlation in gene expression and P. aeruginosa killing after antibiotic challenge. Panel A: Correlation between PA5471 expression and mexX (left) or mexY (right) expression. Data represent fold induction vs. control cultures with no antibiotic added. P < 0.01 for correlation with both mexX and mexY. Azithromycin shown in square, azithromycin plus tobramycin shown in circle. Panel B: correlation in gene expression and bacterial killing with the combination of tobramycin and azithromycin.
Moderate negative correlation occurred between expression of PA5471, mexX, and mexY genes and bacterial killing in cultures challenged with combined azithromycin and tobramycin. Clinical isolates with greater expression of these genes in the MexXY adaptive resistance response generally showed less bacterial killing after challenge with azithromycin and tobramycin in combination. This reached statistical significance for mexX. Pearson r correlation coefficients and P values are shown below (Fig. 6B). Poor correlation existed between gene expression and bacterial killing in cultures treated with either azithromycin or tobramycin alone, suggesting that the combination is better able to induce MexXY and thereby reduce bacterial killing in our model system. Additional data available in on-line supplement.
Expression of mexA and mexB were tested under the same conditions and no significant induction was found (suppl Figure). This indicates that azithromycin may not significantly affect expression of the major efflux system reportedly capable of targeting aztreonam (mexAB) [
]. Whether or not other antibiotic resistance mechanisms are affected is unknown.
To test the importance of MexXY efflux response in this context, we inactivated the mexX gene in the laboratory strain PAO1 and compared the parent PAO1 and mexX inactivated strain in the aggregate model. As observed with the CF clinical bacterial strains, the addition of azithromycin impaired bacterial killing with tobramycin in the parent strain of PAO1. Conversely, azithromycin and tobramycin were additive in the mexX inactivated strain, suggesting that this bacterial adaptive resistance mechanism may explain much of the apparent antimicrobial antagonism (Fig. 7, P < 0.01).
Fig. 7PAO1 density of biofilm aggregate cultures with MexX disruption. PAO1 parent strain vs. PAO1 generated strain with mexX disruption. Both strains achieved similar bacterial density without antibiotic challenge. Azithromycin reduced bacterial killing when added to tobramycin in the parent strain and, conversely, improved bacterial killing when added to tobramycin in the mexX disrupted strain (P < 0.01, N = 4, log10 CFU graphed as mean + SEM). UNT: untreated, AZM: azithromycin 20 μg/mL, TOB: tobramycin (100 μg/mL).
The prevalence of chronic P. aeruginosa airway infection among people with CF increases with age and is present in ~70% of those over 18 years old receiving care in the US [
]. Azithromycin and inhaled tobramycin are two of the three most commonly prescribed chronic pulmonary medications for eligible patients, behind only dornase alfa [
]. Thus, we estimate that long-term, combined drug use occurs in half or more of all people with CF in the US over a lifetime. That estimate underscores the importance of determining whether or not this combination of therapies, as supported by current consensus guidelines, might be unexpectedly problematic [
Herein we show how CF subjects in a clinical trial of inhaled antibiotics who were using chronic oral azithromycin had lesser benefit during inhaled tobramycin vs. inhaled aztreonam. Our previous work highlighted this potential drug interaction between azithromycin and tobramycin by comparing separate parallel groups of CF subjects, either taking or not taking azithromycin, which were then randomized to inhaled tobramycin [
]. The current analyses focus on CF subjects who are all using chronic azithromycin and sequentially exposed to inhaled tobramycin followed by inhaled aztreonam. These additional data indicate that the pattern of drug interaction between azithromycin and inhaled antibiotics may not be due to confounding by an indication bias for clinical prescription of azithromycin.
There are clear limitations to such retrospective analysis, including: limited characterization of subjects (e.g. length of time using azithromycin prior to study enrollment, other airway pathogens, etc.), relatively small numbers of subjects in the defined subgroups, and unequal prior exposure to inhaled aztreonam vs. tobramycin. The impact of such limitations on our overall hypothesis may be partially mitigated by our parallel in vitro studies. We have tested the antibacterial effects ex vivo in 30 CF clinical isolates collected from geographically diverse patient cohorts spanning several years, employing a model system that may better represent the morphology and antibiotic resistance of bacteria in the CF airway [
]. We observe that, while tobramycin and azithromycin alone are antimicrobial, addition of azithromycin to tobramycin is not additive but rather significantly reduces tobramycin-mediated bacterial killing. In this same model azithromycin significantly improved P. aeruginosa killing with levofloxacin, and in particular, with colistin, which has recently been reported by others [
Azithromycin synergizes with cationic antimicrobial peptides to exert bactericidal and therapeutic activity against highly multidrug-resistant gram-negative bacterial pathogens.
Increased susceptibility of Pseudomonas aeruginosa to macrolides and ketolides in eukaryotic cell culture media and biological fluids due to decreased expression of oprM and increased outer-membrane permeability.
]. We observe no interaction between azithromycin and the monobactam, aztreonam. One could hypothesize that azithromycin improves the clinical response to inhaled aztreonam rather than impairing the response to inhaled tobramycin. Our in vitro studies do not support this theory; rather, we observe selective antagonism with tobramycin and no interaction with aztreonam.
The antibiotic concentrations used in our experiments were defined by initial kill curve testing of individual antibiotics to define parameters allowing us to measure the combined effect after adding azithromycin at concentrations that may be present in the CF airway [
]. This resulted in reasonably physiologic concentrations for the compound of greatest interest (i.e. tobramycin), but notably lower (e.g. colistin) or higher (e.g. aztreonam) than physiological concentrations for other compounds, which is also a limitation of this work. Some of these differences may be attributable to our eukaryotic media composition [
Azithromycin synergizes with cationic antimicrobial peptides to exert bactericidal and therapeutic activity against highly multidrug-resistant gram-negative bacterial pathogens.
Increased susceptibility of Pseudomonas aeruginosa to macrolides and ketolides in eukaryotic cell culture media and biological fluids due to decreased expression of oprM and increased outer-membrane permeability.
]. We recognize that comparative in vitro bacterial culture models have produced varying results and some have shown benefit when adding azithromycin to tobramycin [
Synergistic activities of macrolide antibiotics against Pseudomonas aeruginosa, Burkholderia cepacia, Stenotrophomonas maltophilia, and Alcaligenes xylosoxidans isolated from patients with cystic fibrosis.
]. To our knowledge, no in vitro model has thus far demonstrated an ability to accurately predict the most effective antibiotic regimen in CF clinical care and we do not know if our culture model will do so either [
Antibiotic susceptibility of multiply resistant Pseudomonas aeruginosa isolated from patients with cystic fibrosis, including candidates for transplantation.
Combination antibiotic susceptibility testing to treat exacerbations of cystic fibrosis associated with multiresistant bacteria: a randomised, double-blind, controlled clinical trial.
]. The consistency between our analyses of clinical trial data and in vitro results strengthens our confidence in these findings but we lack similar clinical trial data analysis for most of antibiotics tested in vitro and caution against over interpretation. The focus of this work is on tobramycin and azithromycin and, indeed, some of the comparative testing (e.g. lack of interaction between azithromycin and amikacin) may represent idiosyncrasies of our culture model rather than a prediction clinical effectiveness.
Investigators have shown that azithromycin and other macrolide antibiotics may be substrates for the MexXY efflux pump in P. aeruginosa [
]. MexXY is considered a critical mechanism of adaptive resistance to aminoglycosides (i.e. tobramycin) and may be induced by ribosomal perturbation occurring in response to antibiotics [
]. This overlap in bacterial response to azithromycin and aminoglycosides, combined with the importance of MexXY in inducible resistance to tobramycin, led to us focus on this pathway as a potential explanation for our findings. The data presented here show that azithromycin alone, and particularly when combined with tobramycin, significantly increased expression of mexX and mexY gene products in many of the CF clinical isolates tested. Azithromycin also induced PA5471, the positive regulator of MexXY and a bacterial response to ribosomal perturbation. Furthermore, we found strong correlation between maximum PA5471 and mexX or mexY gene induction following azithromycin/tobramycin challenge. This supports the theory that azithromycin, particularly when combined with tobramycin, can activate the MexXY efflux system by triggering a bacterial response to ribosomal targeting in a way that has been reported for aminoglycosides (i.e. tobramycin) alone [
]. Tobramycin, at concentrations achieving greater killing than azithromycin, induced less PA5471, mexX or mexY than azithromycin in our experimental model. This pattern is consistent with previous reports testing aminoglycoside and macrolide antibiotics with PAO1 [
]. We also observe at least moderate inverse correlation between gene expression and bacterial killing after challenge with azithromycin and tobramycin in combination. Such correlation found in a relatively small collection of diverse clinical isolates is not conclusive but strengthens our theory. This is additionally supported by the observation that genetically disrupting the MexXY pathway changes this interaction to an additive rather than antagonistic effect.
Aztreonam is a substrate for the MexAB efflux pump and investigators have reported that azithromycin may down regulate this efflux system [
]. We observed virtually no change in expression of mexAB genes in our clinical isolates following exposure to antibiotics (Fig. S2 online supplement). This is consistent with the lack of antagonism found in our microbiology and post-hoc clinical data analyses but does not eliminate a possibility of interaction in resistance pathways not tested.
Chronic use of macrolides, in particular azithromycin, have several proven benefits in subjects with CF. Published effects in clinical and laboratory research include: improved lung function, reduced frequency of exacerbations of CF lung disease, improved quality of life, antibacterial effects against many common respiratory bacterial pathogens, antiviral properties, and immunomodulation toward a less inflammatory state [
Randomized trial to evaluate azithromycin's effects on serum and upper airway IL-8 levels and recurrent wheezing in infants with respiratory syncytial virus bronchiolitis.
Azithromycin blocks quorum sensing and alginate polymer formation an dincreases the sensitivity to serum and stationary-growth-phase killing of Pseudomonas aeruginosa and attenuates chronic P. aeruginosa lung infection in cftr(−/−) mice.
Maintenance azithromycin treatment in pediatric patients with cystic fibrosis: long-term outcomes related to macrolide resistance and pulmonary function.
]. Thus, the potential adverse drug–drug interaction between azithromycin and tobramycin as reported by our group may be important but would not necessarily negate other benefits of azithromycin, and the net clinical effect of combined use remains uncertain. Given this, the burden of proof remains high for demonstrating an adverse drug interaction that would alter current clinical practice. It is also true that chronic use of medication by clinical prescription is difficult if not impossible to model in clinical trials testing efficacy; therefore, ongoing consideration of both benefits and potential liabilities of longstanding therapies is needed. Recent reports, which are also limited by post-hoc testing, suggest that the benefits of chronic oral azithromycin in CF patients may not persist beyond a year of use [
]. Whether or not cycled, intermittent use of macrolides would be more beneficial is entirely unknown and complicated by the fact that azithromycin may persist in the airway up to 30 days after discontinuing drug dosing [
In our analyses, a sizable minority of clinical trial participants and bacterial isolates do not demonstrate the suspected adverse drug interaction (see scatter plots, some of which are in on-line supplement). This phenomenon is currently unexplained and may suggest confounding variables or mechanisms beyond that explored to date. This could include inaccurate categorization based on self-report in our retrospective analysis, beneficial antibacterial or anti-inflammatory effects unrelated to pseudomonas, or biological functions beyond efflux pumps. The common use of azithromycin among subjects in this clinical trial dataset (73%) also resulted in a relatively small group of subjects without azithromycin use, precluding us from making robust comparison between these groups. We, therefore, cannot confidently state that non-users of azithromycin in this study behaved differently than those using azithromycin, and we have constrained most of our analyses to the group reporting azithromycin use. We retained full control of the analytical process and dissemination of results but also recognize that this and our prior post-hoc analysis were conducted using datasets provided to us by the manufacturer of a competing drug product, which may be seen to introduce bias. Ultimately, despite the perceived strengths of our collective findings to date, we recognize several caveats and many related unanswered questions. Available data are inadequate to guide clinical practice but are we are now conducting a dedicated, prospective clinical trial to test the impact of adding azithromycin to inhaled tobramycin in subjects with CF and chronic P. aeruginosa airway infection (Clintrials.gov NCT02677701). We believe that this trial and additional parallel work studying the mechanisms of drug interaction will provide important information to either support or bring into question a common approach to clinical care for CF patients with chronic P. aeruginosa.
Author contributions
DPN: involved in conception, hypothesis delineation, design, data acquisition, analysis, data interpretation, manuscript preparation and manuscript revision.
PEB, SMC, CLH, KCM, MTS, JLT: involved in design, data acquisition, analysis, and manuscript revision.
JFC, LS: involved in conception, hypothesis delineation, design, data interpretation, and manuscript revision.
JAN: involved in conception, hypothesis delineation, design, data interpretation, manuscript preparation and manuscript revision.
Acknowledgments
Supported by: Investigator initiated research support from Gilead Sciences (DPN), National Institutes of Health/National Heart Lung & Blood Institute, NIH 5R01 HL124053 (DPN), the Cystic Fibrosis Foundation Therapeutics NICHOL15 (DPN), the Cystic Fibrosis Foundation (DPN) and a kind gift from Susan and Jim Goodwin (JAN). We also acknowledge Drs. Alan M. Lambowitz and Jun Yao for the kind gift of the targetron donor plasmid pBL1, and Drs. Luke Hoffman and Rafael Hernandez for providing P. aeruginosa isolates from the CF Isolate Core at Seattle Children's Research Institute supported by NIH P30 DK089507. We thank Colin Manoil and Pradeep Singh for critical review in manuscript preparation. Some of these data have previously been reported in the form of an abstract [
Fig. S1. Absolute change in FEV1 (L). Data are shown as mean and individual data for the absolute change in FEV1 volume. The pattern is the same as reflected in the relative change in lung function and subjects using azithromycin show a statistically significant difference (Welch’s P = 0.023) in lung function improvement during inhaled aztreonam (AZTR; N=71, mean 0.088 liters) vs. inhaled tobramycin (TBRA; N=108, mean 0.012 liters).
Fig. S2. Time course gene expression of mexA and mexB. Relative expression of mexA and mexB following antibiotic exposure. Data represent mean of 10 unique CF clinical isolates of P. aeruginosa. No significant induction of these genes is observed in our culture model under the antibiotic conditions tested.
Fig. S3. Relative change in FEV1 (L) during 4-week use of inhaled tobramycin followed by 4-week use of inhaled aztreonam. Group means and individual values for change in lung function among users of azithromycin (AZM, left) and non-users of azithromycin (No AZM, right) during inhaled antibiotics. Subjects using azithromycin had little change in lung function during the initial 4-week period using inhaled tobramycin (TBRA; N=108, mean 0.78). Users of azithromycin had, on average, significantly greater increase improvement in FEV1 during the subsequent 4-week period while using inhaled aztreonam (AZTR, N=71, mean 6.35. Welch’s T-test P < 0.005). Those not using azithromycin had similar mean improvement during inhaled tobramycin (N=40, mean 2.55) and subsequent inhaled aztreonam lysine (N=22, mean 3.57). More subjects are included in the inhaled tobramycin groups than aztreonam groups because some subjects were randomized to receive inhaled placebo rather than aztreonam following inhaled tobramycin. Only subjects able to complete both phases of the study and with complete data for outcome were included for data analyses.
Figure S4. Absolute change in CFQ-R Respiratory Symptom Scale. Mean (SEM) improvement in self-reported, disease-related quality of life for users of azithromycin (AZM, left) and non-users of azithromycin (No AZM, right). A change of 5.5 points is the established threshold for clinically significant improvement. Subjects using azithromycin reported a non-significant decrease in quality of life during the initial 4-week period using inhaled tobramycin (TBRA; N=102, mean -1.78 points) and significantly greater increase during the subsequent 4-week period while using inhaled aztreonam (AZTR; N=68, mean 8.29 points. Difference 10.09, Welch’s T-test P < 0.001). Non-users of azithromycin reported a non-significant increase in quality of life during inhaled tobramycin and a further increase during the subsequent inhaled aztreonam period (TBRA; N=39, mean 3.13 points, AZTR; N=22, mean 3.03 points). More subjects are included in the inhaled tobramycin groups than aztreonam groups because some subjects were randomized to receive inhaled placebo rather than aztreonam following inhaled tobramycin. Only subjects able to complete both phases of the study and with complete data for outcome were included for data analyses.
Figure S5.P. aeruginosa density of bacterial aggregate cultures. Cultures exposed to azithromycin (AZM) in combination with tobramycin (TBRA) showed significantly greater bacterial density than tobramycin alone (Panel A: N = 30 unique isolates each tested once, log10 CFU graphed as mean + SEM, Wilcoxon matched pairs P < 0.0001). Panel B: Difference in CFU density when azithromycin is added to anti-pseudomonal antibiotics, including: amikacin (AMIK), colistin (COLI), levofloxacin (LEVO), and tobramycin (TBRA) (N = 30, log10 CFU graphed as median and quartiles (Box) plus 5-95% (whiskers)). The addition of azithromycin resulted on no significant change when added to aztreonam (AZTR) or amikacin (AMIK) and provided significantly improved bacterial killing when added to colistin (COLI) and levofloxacin (LEVO) (Wilcoxon test P < 0.001). Azithromycin significantly lessened bacterial killing only when added to tobramycin (Wilcoxon test P < 0.001). Zero x-axis represents CFU values of simultaneous cultures exposed to the anti-pseudomonal antibiotic without azithromycin added; N=30 unique clinical isolates tested without technical replication for each condition.
References
Mogayzel Jr., P.J.
Naureckas E.T.
Robinson K.A.
et al.
Cystic fibrosis pulmonary guidelines. Chronic medications for maintenance of lung health.
Synergistic activities of macrolide antibiotics against Pseudomonas aeruginosa, Burkholderia cepacia, Stenotrophomonas maltophilia, and Alcaligenes xylosoxidans isolated from patients with cystic fibrosis.
Macrolides decrease the minimal inhibitory concentration of anti-pseudomonal agents against Pseudomonas aeruginosa from cystic fibrosis patients in biofilm.
Determination of the minimal clinically important difference scores for the cystic fibrosis questionnaire-revised respiratory symptom scale in two populations of patients with cystic fibrosis and chronic Pseudomonas aeruginosa airway infection.
Azithromycin synergizes with cationic antimicrobial peptides to exert bactericidal and therapeutic activity against highly multidrug-resistant gram-negative bacterial pathogens.
Increased susceptibility of Pseudomonas aeruginosa to macrolides and ketolides in eukaryotic cell culture media and biological fluids due to decreased expression of oprM and increased outer-membrane permeability.
Antibiotic susceptibility of multiply resistant Pseudomonas aeruginosa isolated from patients with cystic fibrosis, including candidates for transplantation.
Combination antibiotic susceptibility testing to treat exacerbations of cystic fibrosis associated with multiresistant bacteria: a randomised, double-blind, controlled clinical trial.
Randomized trial to evaluate azithromycin's effects on serum and upper airway IL-8 levels and recurrent wheezing in infants with respiratory syncytial virus bronchiolitis.
Azithromycin blocks quorum sensing and alginate polymer formation an dincreases the sensitivity to serum and stationary-growth-phase killing of Pseudomonas aeruginosa and attenuates chronic P. aeruginosa lung infection in cftr(−/−) mice.
Maintenance azithromycin treatment in pediatric patients with cystic fibrosis: long-term outcomes related to macrolide resistance and pulmonary function.
30672-5&id=gr1.jpg){kind=link}
30672-5&id=gr2.jpg){kind=link}
30672-5&id=gr3.jpg){kind=link}
30672-5&id=gr4.jpg){kind=link}
30672-5&id=gr5.jpg){kind=link}
30672-5&id=gr6.jpg){kind=link}
30672-5&id=gr7.jpg){kind=link}