Edited by: Matthew C. Wolfgang, The University of North Carolina at Chapel Hill, United States
Reviewed by: Sarah Rowe-Conlon, The University of North Carolina at Chapel Hill, United States; Dominique Limoli, The University of Iowa, United States
This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology
†ORCID: Paul Briaud
Sylvère Bastien
Laura Camus
Philippe Reix
François Vandenesch
Anne Doléans-Jordheim
Karen Moreau
‡These authors have contributed equally to this work. Author order was determined on the basis of lab position
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Cystic fibrosis (CF) is the most common genetic disease among the Caucasian population that affects multiple organs and causes various complications associated with patient death, such as cystic fibrosis liver disease (Debray et al.,
Although SA colonization decreases as patients age, SA infection concerns more than 30% of CF adults (Zolin et al.,
Few studies have investigated the impact of SA-PA co-infection and clinical outcomes. In addition, none of them considered the interaction state (competition vs. coexistence) between SA-PA. So far, available data have shown conflicting results on the link between SA-PA co-colonization and clinical outcomes. Ahlgren et al. did not find a significant clinical difference in adult patients co-colonized with SA and PA compared to patients colonized solely by PA (Ahlgren et al.,
The aim of this study was to better characterize (i) co-infected patients with demographic data (age, BMI, gender), and (ii) the interaction state (competition or coexistence) between the two bacterial species. We also examined the consequence of SA-PA co-infection and SA-PA interaction state (competition vs. coexistence) on pulmonary functions (FEV1%) and clinical outcomes (e.g., number of exacerbations, number of hospitalizations). By computing demographic and clinical data with pulmonary infectious status, we gain more information on the impact of SA and PA infections and interactions on CF patients' health.
The clinical and bacteriological data of CF patients supervised at the two CF Centers in Lyon, France (CRCM: Center de Ressources et de Compétences de la Mucoviscidose) were collected from February 2017 to August 2018.
The inclusion criterion was a stable microbiological status for SA and /or PA colonization, defined by the following criteria: (i) at least three respiratory samples collected for each patient throughout the period considered; (ii) at least 2 months between two successive samples; (iii) all the samples collected during the study for a patient had the same status with respect to the presence of SA and / or PA. The patients who did not match the criteria, or who were not co-colonized by SA or PA were excluded.
This study was submitted to the Ethics Committee of the Hospice Civil de Lyon (HCL) and registered under CNIL No 17-216. All the patients were informed of the study and did not oppose the use of their data.
Clinical data were extracted from computerized medical files (Easily®). The data collected were: gender, age at the time of the last sampling, CFTR genotype classified between severe and moderate genotype regardless of clinical severity (5, 19), pancreatic insufficiency defined by fecal elastase <200 μg/g, CF-related diabetes (CFRD) or a carbohydrate intolerance, cirrhosis, need for long term oral or enteral supplementation and/or undernourishment (defined as a body mass index (BMI) (weight/height2) lower than 17 kg/m2 in an adult and −2 standard deviations (SD) in a child according to gender and age). The pulmonary function was defined by the FEV1 expressed as a percentage of the predicted value (%pred). We also collected the number of hospitalizations, the length of hospital stays and the number of exacerbations during the 9 months preceding the last sample.
The microbiological composition of each respiratory sample was determined by the Institute for Infectious Agents, HCL. The interaction state (coexistence or competition) of SA-PA pairs was defined by agar competition assay as previously described (Briaud et al.,
Based on these microbiological analyses, the patients were categorized into four groups: (i) SA alone, (ii) PA alone, (iii) SA-PA in competition, and (iv) SA-PA in coexistence.
Two different analyses were performed using the same process: (i) SA vs. PA vs. SA+PA, and (ii) SA vs. PA vs. SA+PA in coexistence vs SA+PA in competition.
Factor Analysis of Mixed Data (FAMD) was used for initial data screening. Then, univariate analysis was performed to determine significant differences between groups. For continuous variables (age, BMI, FEV1, number of hospitalizations, length of hospitalization, and number of exacerbations), Kruskal-Wallis tests were used to identify whether one population was different from the others. Afterwards, to individually test each pair of populations with significant Kruskal-Wallis tests, a Mann-Whitney Wilcoxon tests was used. Fisher's exact tests were performed to compare categorical variables (CFTR genotype, gender, denutrition, pancreatic insufficiency, CFRD, liver cirrhosis, enteral nutrition, and oral food supply). For multiple comparisons, tests were corrected by a Bonferroni method and statistical significance was set with a q-value threshold at 0.05. Finally, a multinomial log-linear model [nnet package (William,
Of the 655 CF patients monitored in Lyon hospitals, we selected patients with at least three respiratory specimens with SA and/or PA during the study period. Two seventy six patients were excluded because of the absence of the three samples required. Sixty-five patients had neither SA nor PA infection. Finally, 48 and 54 patients were excluded because of unstable SA-PA co-colonization during the period or the inability to assess their pulmonary function (age <4 years). For co-infected patients, we excluded five patients because they presented several PA isolates with both competitive and coexisting interaction phenotypes with SA. Finally, 212 patients with CF were included in this study.
The cohort consisted of 124 adults and 88 children with a mean age of 21.70 years (range 4–69). The male-female ratio was homogeneous with ~51.42% (
The 212 patients were classified following their chronic bronchial colonization into 3 different groups: SA alone, PA alone and SA+PA. Co-infected patients were split into 2 subclasses (competition and coexistence) regarding the SA-PA interaction state determined by agar competition assay (
Clinical characteristics of CF patients according to their bacteriological status.
Number | 112/212 (52.83) | 48/212 (22.64) | 18/52 (34.61) | 34/52 (65.38) | 52/212 (24.53) | |
Age (years) | 16.49 ± 8.77 | 32.02 ± 13.82 | 23.67 ± 8.98 | 23.24 ± 10.52 | 23.38 ± 9.93 | <0.0001 |
≥18 years | 45/112 (40.18) | 42/48 (87.50) | 14/18 (77.78) | 23/34 (64.65) | 37/52 (71.15) | |
Sex, male | 63/112 (56.25) | 23/48 (47.92) | 10/18 (55.6) | 13/34 (38.24) | 23/52 (44.23) | ns |
Genotype | ||||||
Moderate | 26/112 (23.21) | 9/48 (18.75) | 3/18 (16.67) | 3/34 (8.82) | 6/52 (11.54) | ns |
Severe | 86/112 (76.79) | 39/48 (81.25) | 15/18 (83.33) | 31/34 (91.18) | 46/52 (88.46) | |
BMI (kg/m2) | 18.06 ± 2.68 | 20.74 ± 2.42 | 19.25 ± 2.56 | 19.71 ± 3.56 | 19.55 ± 3.23 | <0.0001 |
Undernourishment | 11/112 (9.82) | 1/48 (2.08) | 1/18 (5.56) | 3/34 (8.82) | 4/52 (7.69) | ns |
Oral food supplementation | 23/112 (20.54) | 13/48 (27.08) | 3/18 (16.67) | 15/34 (44.12) | 18/52 (34.62) | ns |
Enteral nutrition | 4/112 (3.57) | 0/48 (0) | 0/18 (0) | 4/34 (11.76) | 4/52 (7.69) | ns |
Pancreatic insufficiency | 101/112 (90.18) | 45/48 (93.75) | 17/18 (94.44) | 33/34 (97.06) | 50/52 (96.15) | ns |
CF-related diabetes | 9/112 (8.04) | 14/48 (29.17) | 3/18 (16.67) | 10/34 (29.41) | 13/52 (25.00) | 0.0007 |
Cirrhosis | 6/112 (5.36) | 2/48 (4.17) | 0/18 (0) | 1/34 (2.94) | 1/52 (1.92) | ns |
Hospitalizations | ||||||
Number | 17/112 (15.18) | 18/48 (37.50) | 6/18 (33.33) | 14/34 (41.18) | 20/52 (38.46) | 0.0003 |
Length | 8.00 ± 6.10 | 17.39 ± 14.39 | 7.33 ± 6.12 | 29.14 ± 27.39 | 22.60 ± 25.06 | 0.0002 |
Number of exacerbations | 0.25 ± 0.69 | 1.44 ± 1.37 | 1.28 ± 1.60 | 1.35 ± 1.50 | 1.33 ± 1.52 | <0.0001 |
FEV1 (% predicted) | 85.87 ± 22.39 | 55.09 ± 18.64 | 59.72 ± 18.73 | 64.59 ± 21.79 | 62.90 ± 20.73 | <0.0001 |
We proceeded to an FAMD analysis to decipher the similarity and heterogeneity between patient groups using both continuous and categorical variables. The first two axes retained accounted for 39.47% of the total variance of the data (
To define whether co-infection was associated with poorer clinical outcome, we compared three groups: patients infected with SA, patients infected with PA and patients co-infected with SA-PA (
Comparison between SA mono-infected (SA), PA mono-infected (PA) and co-infected groups in competition (Comp) or in coexistence (Coex) for age mean
P-values for continuous variable comparisons between SA mono-infected (SA), PA mono-infected (PA) and co-infected groups (SA+PA).
Age | <0.0001 | <0.0001 | 0.0012 |
BMI | <0.0001 | 0.0074 | 0.029 |
Number of hospitalizations | 0.0017 | 0.0013 | ns |
Length of hospitalization | 0.0011 | 0.0010 | ns |
Number of exacerbations | <0.0001 | <0.0001 | ns |
FEV1 | <0.0001 | <0.0001 | ns |
Since the PA group appears to have the worst clinical outcome but also corresponds to the oldest group of patients, it is questionable whether the deterioration of the clinical condition of the patients is linked to the age or presence of PA. To answer this question, we classified the patients into three age groups (4–14, 15–25, and 26–69) and analyzed the impact of PA infection within these different groups on continuous variables (
Comparison between SA mono-infected (SA) and PA mono-or co- infected (PA) groups within three age classes of patients. FEV1
To further investigate these results and consider the potential association between variables, multinomial analysis with infection type as outcome was performed using age, BMI, FEV1, number of exacerbations, number of hospitalizations, length of hospitalizations and CFRD (
Adjusted odds ratios of cystic fibrosis patients' infection status.
Age | – | 0.1354 | – | 0.6767 | 0.9477 (0.9017, 0.9961) | 0.0345 |
BMI | 1.2662 (1.0313, 1.5545) | 0.0242 | 1.2022 (1.0015, 1.4432) | 0.0481 | – | 0.5913 |
Number of exacerbations | 2.0282 (1.2559, 3.2755) | 0.0038 | 2.1460 (1.3514, 3.4078) | 0.0012 | – | 0.7298 |
FEV1 | 0.9612 (0.9355, 0.9877) | 0.0043 | 0.9701 (0.9478, 0.9930) | 0.0108 | – | 0.4947 |
In this context, variables about hospitalizations were not kept in the final model due to their correlation with the number of exacerbations [Cor (95% CI): 0.6648 (0.5822, 0.7338) (
The final model demonstrated that patients infected with PA were more likely to have a high BMI [OR (95% CI): 1.2662 (1.0313, 1.5545)], a higher number of exacerbations; [OR (95% CI): 2.0282 (1.2559, 3.2755)], and lower FEV1 [OR (95% CI): 0.9612 (0.9355, 0.9877)] compared to patients infected with SA (
The second objective of this study was to determine whether the interaction profile between SA and PA could affect patient clinical outcome. Thus, we compared the mono-infected groups (SA only and PA only) with two co-infected groups: SA plus PA in coexistence and SA plus PA in competition.
Using continuous variables, the comparison between coexistence and competition groups showed no significant difference (
Considering categorical variables, no characteristics were statistically significant between the two types of interaction (
Finally, multinomial analysis was performed using all the significant characteristics from the univariate tests. The final predictors retained after analyses of deviance through AIC criteria were BMI, need for oral food supplementation, number of exacerbations and FEV1. A significant difference was found between the two types of interaction in co-infected patients. The coexistence state group was more likely to need oral food supplementation than the competition group [OR (95% CI): 0.2262 (0.0536, 0.9541)] (
Adjusted odds ratios of cystic fibrosis patients' infection status.
BMI | 1.2047 (1.0210, 1.4213) | 0.0273 | – | 0.3199 | – | 0.4872 |
Oral food supplementation | – | 0.0801 | – | 0.3609 | 0.2262 (0.0536, 0.9541) | 0.0430 |
Number of exacerbations | 2.2896 (1.3982, 3.7494) | 0.0010 | 1.9090 (1.0956, 3.3262) | 0.0228 | – | 0.4472 |
FEV1 | – | 0.0861 | 0.9589 (0.9300, 0.9887) | 0.0072 | – | 0.1885 |
Additional multivariate analyses were performed on CFRD patients due to a statistically significant difference between groups when performing univariate analyses (
Adjusted odds ratios of cystic fibrosis-related diabetes (CFRD) patients' infection status.
Age | 1.1013 (1.0543, 1.1504) | <0.0001 |
Genotype | – | 0.1022 |
Oral food supplementation | 3.0259 (1.2297, 7.4459) | 0.0160 |
Pancreatic insufficiency | – | 0.8192 |
Number of hospitalizations | 2.1662 (1.5254, 3.0763) | <0.0001 |
The lungs of CF patients are colonized by multiple bacteria and several studies were conducted to decipher the impact of these colonisations on clinical outcomes. We focused our study on the two major pathogens responsible for chronic colonization:
The first point of our study was to compare the impact of chronic colonization by SA alone, PA alone and SA plus PA on the clinical outcome of CF patients. Indeed, being able to evaluate (or even predict) the risks that a patient incurs depending on the type of bacterium and/or on bacterial associations (in particular between PA and another pathogen) that colonize the lungs, remains one of the major challenges in the management of patients. Thus, identifying patients with harmful bacterial associations would be of clinical interest as they have increased risks of a worse clinical outcome. These bacterial associations could become a therapeutic priority and be eradicated by targeting, for example, one of the pathogens. Conversely, we could speculate that co-infection between PA and another bacterium represents a milder step in disease evolution, probably an intermediate stage between SA alone and PA alone.
Our results showed that SA chronic colonization led to a higher FEV1 (85.87 ± 22.39), fewer exacerbations (0.25 ± 0.69), a smaller number of hospitalizations (15.18% of SA patients) and shorter stays (8.00 ± 6.10) than SA plus PA co-infection or PA alone. We did not observe any differences regarding clinical status between patients co-infected by PA and SA and those infected by PA alone. This suggests that the SA plus PA combination does not lead to over or under morbidity compared to the presence of PA alone in the lungs, and that there is no synergism in term of pathogenicity. As soon as PA colonizes the CF patients' lungs, the clinical condition of the patients appears to deteriorate, whether SA is present or not. These results are in accordance with previous results from Ahlgren et al. that demonstrate by univariate analysis that there is no difference between PA mono-infected and PA plus SA co-infected patients (Ahlgren et al.,
Other studies analyzed the impact of co-infection based on the presence of methicillin-susceptible
CFRD is a severe complication of CF and its association with bacterial infection could be used as a predictive marker of severity. CFRD seemed to be overrepresented in mono-infected PA (29.17%) and co-infected (25%) patients and under-represented in mono-infected SA patients (8.04%). CFRD represented 16.98% of our total patients and was slightly higher than the CFRD rate in the European population (Kerem et al.,
The second objective of the present study was to evaluate the impact of the type of interaction between SA and PA (competition or coexistence) on clinical outcomes. Among the 52 patients co-colonized with SA and PA, 65.38% of strain pairs were in coexistent interaction and 34.61% in competition. These competitive and coexistence statuses have already been described (Baldan et al.,
It has long been recognized that PA could suppress SA growth by several mechanisms (Hotterbeekx et al.,
Nonetheless, in the present study, we did not observe a difference in age between patients colonized with coexisting isolates (23.24 ± 10.52) and patients colonized with competitive pairs (23.67 ± 8.98). This suggests that there may be no link between adapted late-infecting strains and coexistence. In order to clarify this point, longitudinal studies will be necessary to evaluate the link between bacteria interaction status and the evolution of isolates.
How PA tolerance toward SA may contribute to worse clinical outcomes is a point that has not been explored before. The analyses showed no significant difference in clinical outcomes between patients with coexisting pairs and patients with competitive pairs, except for the need of oral food supplementation observed with multinomial analysis.
Nevertheless, we highlighted several disparities between coexistence and competition groups when compared to SA mono-infected patients. The univariate analysis indicated that both groups were more likely to have a lower FEV1 and higher number of exacerbations. The multinomial analysis confirmed that patients with a competitive pair had a higher number of exacerbations and a lower FEV1, in comparison to SA group. However, we noticed that multinomial analyse did not confirm the lower FEV1 for the coexistence group in comparison to SA. Indeed, the number of exacerbations was correlated with FEV1. The correlation for the coexistence group [Cor (95% CI): −0.6159 (−0.7896, −0.3508) (
More importantly, univariate analysis demonstrated that only patients with a coexisting pair were more likely to have higher BMI, number and length of hospitalizations than SA mono-infected group. This increased BMI was confirmed by multinomial analysis and might be the results of the long term oral food supply, which was more frequent in coexistence group. Multinomial analysis also indicated that coexistence group presented more exacerbations than the SA group, consistent with previous results about hospitalizations.
Overall, these results suggest that patients with coexisting pair required food supply more frequently than patients with competitive pair. Moreover, when compared to SA mono-infection, co-infection with coexisting PA appears to be more severe than co-infection with competitive PA (higher number and length of hospitalizations and more exacerbations).
Several quorum-sensing dependent virulence factors are produced by competitive PA strains such as rhamnolipids and phenazines (Hotterbeekx et al.,
All datasets generated for this study are included in the article/
All the strains and clinical information used in this study were collected as part of the periodic monitoring of patients at the Hospices Civils de Lyon. As the study is retrospective and non-interventional neither ethics committee approval nor written informed consent were required under local regulations.
PB, AD-J, and KM contributed to the conception and design of the study. PR, CM, and FV contributed to design of the study. PB and LC conducted the experiments. SB conducted and analyzed all the statistics. MB and AD-J collected the data. All the authors contributed to the manuscript and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This work was funded by the associations Vaincre la mucoviscidose and Gregory Lemarchal. PB was funded by the French Ministry of Education and Research and L Camus by the Fondation pour la Recherche Médicale (grant number ECO20170637499). We thank Marie Verneret for her technical support.
The Supplementary Material for this article can be found online at: