Fluoroquinolone treatments induce dysbiosis of the intestinal microbiota, resulting in loss of resistance to colonization by exogenous bacteria such as Clostridioides difficile that may cause severe diarrhea in humans and lethal infection in hamsters. We show here that DAV131A, a charcoal-based adsorbent, decreases the intestinal levels of the fluoroquinolone antibiotics levofloxacin and ciprofloxacin in hamsters, protects their intestinal microbiota, and prevents lethal infection by C. difficile.
KEYWORDS: Clostridioides difficile infection, antibiotics, dysbiosis, fluoroquinolones, hamster animal model, prevention
ABSTRACT
Fluoroquinolone treatments induce dysbiosis of the intestinal microbiota, resulting in loss of resistance to colonization by exogenous bacteria such as Clostridioides difficile that may cause severe diarrhea in humans and lethal infection in hamsters. We show here that DAV131A, a charcoal-based adsorbent, decreases the intestinal levels of the fluoroquinolone antibiotics levofloxacin and ciprofloxacin in hamsters, protects their intestinal microbiota, and prevents lethal infection by C. difficile.
INTRODUCTION
Fluoroquinolone (FQ) antibiotics have significant biliary excretion rates, resulting in high intestinal concentrations that impact the intestinal microbiota (1) and are thus risk factors for Clostridioides difficile infections (CDI) (2–5), the leading cause of health care-associated infections (6). To prevent this risk, Da Volterra has developed DAV132, an adsorbent delivered to the late ileum that sequesters antibiotic residues and preserves the microbiota, as demonstrated in human phase 1 clinical studies (7, 8). To assess whether it would prevent CDI, we initiated studies in hamsters, showing that DAV131A, a rodent-adapted version of DAV132, was able to prevent CDI in animals treated with clindamycin or the fluoroquinolone moxifloxacin (MXF) (9, 10). Here, we have extended MXF observations to levofloxacin (LVX) and ciprofloxacin (CIP), which are more frequently used than MXF (Da Volterra, unpublished data).
Male Golden Syrian hamsters were infected orally on day 3 (D3) with 104 spores of C. difficile strain UNT103-1 (VA-11, REA J strain obtained from Curtis Donskey, Ohio VA Medical Centre) and received 110 mg/kg LVX subcutaneously or 100 mg/kg CIP intraperitoneally daily from D1 to D5, the lowest doses resulting in 100% mortality (data not shown).
Groups of 10 LVX- or CIP-treated animals were randomly assigned to receive oral DAV131A twice daily from D1 to D8 at unit doses of 0 (placebo), 100, 200, 300, 600, 900, or 1,350 mg/kg (this last dose was only applied with CIP because free fecal concentrations were higher than with LVX and MXF [Table 1 and reference 10]). Further controls received antibiotic placebo (saline) together with DAV131A placebo, or either antibiotic (LVX or CIP) together with DAV131A placebo, but were not infected. Fecal free antibiotic concentrations were measured by bioassay (11) using the indicator strains K. pneumoniae ATCC 10031 for LVX and B. subtilis ATCC 6633 for CIP. Data below the limit of quantification (0.19 μg/g for LVX and 1.04 μg/g for CIP) were imputed to half of the limit of quantification. Fecal microbiota composition was analyzed by 16S rRNA genes sequencing as described previously (10), and sequence data were submitted to the NCBI database under accession number PRJNA478191.
TABLE 1.
Mortality rates and fecal concentrations of active antibiotic at D3 according to treatment groups in the levofloxacin and ciprofloxacin studiesa
| Antibiotic | Treatment group | n | Mortality |
Fecal free concn |
||
|---|---|---|---|---|---|---|
| No. (%) | P or q | Median concn in μg/g (min, max) | P or q | |||
| LVX | LVX/0 | 10 | 10 (100) | 82.5 (71.5, 269.1) | ||
| LVX/100 | 10 | 10 (100) | >0.99 | 78.5 (45.5, 120.8) | 0.19 | |
| LVX/200 | 10 | 3 (30) | 7.6 × 10–3 | 39.2 (18.5, 63.5) | 1.4 × 10–5 | |
| LVX/300 | 10 | 0 (0) | 5.0 × 10–5 | 20.5 (11.1, 51.0) | 1.4 × 10–5 | |
| LVX/600 | 10 | 0 (0) | 5.0 × 10–5 | 10.0 (2.5, 30.4) | 1.4 × 10–5 | |
| LVX/900 | 10 | 0 (0) | 5.0 × 10–5 | 4.2 (2.3, 9.2) | 1.4 × 10–5 | |
| All groups | 60 | 23 (38.3) | 3.1 × 10−13 | 29.3 (2.3, 269.12) | 6.9 × 10−10 | |
| CIP | CIP/0 | 10 | 10 (100) | 189.6 (86.7, 580.7) | ||
| CIP/200 | 10 | 4 (40) | 1.1 × 10–2 | 94.4 (26.6, 236.9) | 4.3 × 10–2 | |
| CIP/300 | 10 | 0 (0) | 1.4 × 10–5 | 64.4 (10.1, 151.5) | 2.2 × 10–4 | |
| CIP/600 | 9 | 0 (0) | 1.4 × 10–5 | 6.0 (0.5, 13.7) | 3.4 × 10–4 | |
| CIP/900 | 10 | 0 (0) | 1.4 × 10–5 | 3.1 (0.5, 21.2) | 3.0 × 10–4 | |
| CIP/1350 | 10 | 0 (0) | 1.4 × 10–5 | 1.1 (0.5, 27.45) | 3.0 × 10–4 | |
| All groups | 59 | 14 (23.7) | 3.6 × 10−10 | 25.6 (0.5, 580.7) | 6.6 × 10–9 | |
The data are presented as n (%) or median (min, max) as appropriate. q values refer to the comparison of the corresponding treatment group with the antibiotic/0 group (LVX/0 or CIP/0) after Benjamini-Hochberg correction. P values for comparisons of all treatment groups using the Fisher exact or Kruskal-Wallis tests are reported in the “All groups” row. For concentration data, only antibiotic-treated groups were included.
Mortality rates at D16 were compared using a global Fisher exact test, with the Benjamini-Hochberg’s correction method for multiple testing, when necessary. Fecal free antibiotic concentrations and microbiota diversity indices were compared using the Kruskal-Wallis test. For α-diversity indices, analyses were performed on changes between D3 and D0. In case of significant difference, post hoc comparisons were performed using nonparametric Wilcoxon tests with Benjamini-Hochberg’s correction.
The link between bacterial diversity and mortality by D16 was analyzed on the pooled data from the two studies presented here, together with those from a previously published study with the same design but exposing hamsters to 30 mg/kg MXF (10). Death probability was estimated for all diversity indices using the areas under the receiving operator curve (AUROCs) and the corresponding 95% confidence intervals.
Data are presented as the number of observations (n) or the median (minimum, maximum). All tests were two sided with a type I error of 0.05. All analyses were performed using R software v3.5.1.
LVX or CIP treatments resulted in 100% mortality at D16, after C. difficile inoculation (Fig. 1A and Table 1). Animals with LVX died within 3 days of inoculation and within 4 to 9 days with CIP, in parallel with high counts of fecal C. difficile (Fig. 1B). Protection by DAV131A was dose dependent (Table 1 and Fig. 1), following the dose-dependent decrease in fecal antibiotic concentrations (significant from 200 mg/kg DAV131A; Fig. 1C and Table 1).
FIG 1.
Survival rate (A), C. difficile counts (B), and free fecal antibiotic concentrations at D3 (C) in animals treated with LVX (left) and CIP (right) according to treatment group. In the “ABX/XXX” notation used in the legend, ABX is the antibiotic, and “XXX” is the unit dose of DAV131A (n = 10 for all groups except for CIP/600 group where n = 9). For the C. difficile counts, the mean and error bars representing the standard deviations are shown. For antibiotic concentrations, the boxes present the 25th and 75th percentiles, and the horizontal black bar reports the median value, while the whiskers indicate the 10th and 90th percentiles, and the outliers are also shown. Red asterisks represent adjusted P values of Wilcoxon tests comparing each group treated by antibiotic and DAV131A to the group treated by antibiotic alone (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).
Antibiotics induced a severe dysbiosis, more pronounced with LVX that CIP. Increasing amounts of DAV131A enabled a dose-dependent protection, as measured by changes in α- and β-diversity between D3 and D0 (Fig. 2 and Table 2; see Tables S1A and B in the supplemental material). However, protection of the microbiota was not complete, as further attested by a principal coordinate analysis of microbiota diversity evolution over time (Fig. S1). Surviving animals had a modest decrease in α-diversity and an increase in β-diversity. Strikingly, these changes were significantly larger in animals that developed CDI (Fig. 3; see Tables S2 and 3).
FIG 2.
Change of the Shannon index (α-diversity) (A) and unweighted UniFrac distance (β-diversity) (B) between D0 and D3 in animals treated with LVX (left) and CIP (right) according to the dose of DAV131A in each treatment group. The boxes present the 25th and 75th percentiles, and the horizontal black bar reports the median value, while the whiskers indicate the 10th and 90th percentiles, and the outliers are also shown. Asterisks represent adjusted P values of Wilcoxon tests comparing each group treated by antibiotic and DAV131A to the group treated by antibiotic alone (red stars) or to the untreated control group (gray star) (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).
TABLE 2.
Change between D0 and D3 of the Shannon index and unweighted UniFrac distances according to treatment groups in the LVX and CIP studiesa
| Antibiotic | Treatment group | n | Shannon index change |
Unweighted UniFrac distance |
||||
|---|---|---|---|---|---|---|---|---|
| Median (min, max) | P or q (i) | P or q (ii) | Median (min, max) | P or q (i) | P or q (ii) | |||
| LVX | Control | 10 | 0.0 (−0.3, 0.3) | 0.36 (0.33, 0.44) | ||||
| LVX/0 | 10 | –2.6 (−3.4, −2.1) | 0.69 (0.60, 0.72) | |||||
| LVX/100 | 10 | −1.9 (−3.4, −1.1) | 3.5 × 10–2 | 2.7 × 10–5 | 0.67 (0.60, 0.75) | 0.63 | 2.7 × 10–5 | |
| LVX/200 | 10 | −1.6 (−2.8, −0.3) | 1.7 × 10–3 | 2.7 × 10–5 | 0.62 (0.54, 0.67) | 2.3 × 10–3 | 2.7 × 10–5 | |
| LVX/300 | 10 | −1.1 (−2.2, −0.1) | 7.2 × 10–5 | 4.1 × 10–4 | 0.57 (0.51, 0.61) | 6.2 × 10–5 | 2.7 × 10–5 | |
| LVX/600 | 9 | −1.0 (−1.3, −0.3) | 4.3 × 10–5 | 1.2 × 10–4 | 0.59 (0.51, 0.61) | 1.1 × 10–4 | 3.6 × 10–5 | |
| LVX/900 | 10 | –0.7 (−1.4, −0.4) | 2.7 × 10–5 | 2.7 × 10–5 | 0.50 (0.45, 0.63) | 3.6 × 10–5 | 2.7 × 10–5 | |
| All groups | 69 | −1.1 (−3.4, 0.3) | 5.7 × 10–9 | 5.7 × 10–9 | 0.59 (0.33, 0.75) | 2.8 × 10–10 | 2.8 × 1010 | |
| CIP | Control | 10 | –0.8 (−1.3, −0.1) | 0.39 (0.36, 0.45) | ||||
| CIP/0 | 6 | –1.74 (−3.16, −1.3) | 0.58 (0.54, 0.66) | |||||
| CIP/200 | 7 | –1.3 (−1.5, −0.7) | 4.4 × 10–2 | 2.4 × 10–2 | 0.62 (0.57, 0.71) | 0.53 | 3.4 × 10–4 | |
| CIP/300 | 9 | –1.0 (−1.7, 0.4) | 1.4 × 10–2 | 0.30 | 0.56 (0.51, 0.60) | 9.8 × 10–2 | 2.1 × 10–4 | |
| CIP/600 | 8 | –0.9 (−1.7, −0.4) | 0.14 | 0.29 | 0.50 (0.48, 0.64) | 2.1 × 10–2 | 2.2 × 10–4 | |
| CIP/900 | 8 | –0.4 (−1.7, −0.4) | 1.6 × 10–2 | 0.57 | 0.51 (0.33, 0.61) | 4.2 × 10–2 | 5.4 × 10–2 | |
| CIP/1350 | 7 | –0.6 (−1.3, −0.2) | 1.4 × 10–2 | 0.96 | 0.46 (0.40, 0.49) | 2.9 × 10−3 | 9.2 × 10–3 | |
| All groups | 55 | −1.0 (−3.2, 0.4) | 2.1 × 10–3 | 2.1 × 10–3 | 0.52 (0.33, 0.71) | 1.1 × 10–6 | 1.1 × 10–6 | |
The data are presented as n (%) or median (min; max) as appropriate. q values refer to the comparison of the corresponding treatment group with the relevant antibiotic/0 group (LVX/0 or CIP/0) (i) or control group (ii) after Benjamini-Hochberg correction. The P values for the comparison of all treatment groups using Fisher exact or Kruskal-Wallis tests are reported in the “All groups” row.
FIG 3.
Change of the Shannon index (α-diversity) (A) and unweighted UniFrac distance (β-diversity) (B) between D0 and D3 in animals treated with LVX (left) and CIP (right) according to survival or death measured at the end of the study (D16). Colored circles indicate the dose of DAV131A received by each animal. The boxes present the 25th and 75th percentiles, and the horizontal black bar reports the median value, while the whiskers indicate the 10th and 90th percentiles. Red asterisks represent P values of the Wilcoxon tests for the comparison of animal fates (***, P < 0.001; ****, P < 0.0001).
To further assess how variations in diversity indices predicted D16 mortality, we pooled the data from these studies with those from a previous study (10) performed with MXF (Table 3 and see Table S4) and computed the AUROCs for each diversity index. For all, values close to or >0.8 were highly predictive of the outcome. Logistic models relating the mortality rate to the variations from D0 of the two most predictive diversity indices, the Shannon index, and unweighted UniFrac distances (Fig. 4), showed that, regardless of the FQs tested, changes from baseline in intestinal diversity around the time of C. difficile inoculation was a strong predictor of mortality.
TABLE 3.
Changes in microbiota diversity between D0 and D3a
| Diversity | Variable | Value (min, max) |
P | AUROC (95% CI) | |
|---|---|---|---|---|---|
| Died (n = 43) | Survived (n = 131) | ||||
| α-Diversity | Shannon index change | –2.0 (–3.4, −1.0) | –0.9 (−2.3, 0.4) | <2.2 × 10−16 | 0.93 (0.89−0.96) |
| β-Diversity | Unweighted UniFrac distance | 0.65 (0.54, 0.76) | 0.53 (0.33, 0.71) | <2.2 × 10−16 | 0.92 (0.88−0.96) |
According to the vital status at D16 in antibiotic-treated groups and their respective areas under the ROC curves for predicting occurrence of death by D16, after pooling data from antibiotic-treated animals in the two studies performed with LVX and CIP, together with a previously published study with MXF (10). The P values were determined using a nonparametric Wilcoxon test. 95% CI, 95% confidence interval.
FIG 4.
Logistic models of mortality according to the change of Shannon index (A) (P < 10−15) and unweighted UniFrac distance (B) (P < 10−15) between D0 and D3 after pooling data from antibiotic-treated animals in the two studies performed with LVX and CIP, together with a previously published study with MXF (10). Red bars represent the mortality rates and their 95% confidence intervals of deciles of the observed diversity indices. The shaded areas represent the 95% confidence interval of the predicted probability of death.
Altogether, the major result of this study is that DAV131A lowered the level of fecal antibiotics and spared the animals by limiting dysbiosis. Full protection from lethal CDI was obtained with all FQs at relatively close DAV131A doses (200 mg/kg for MXF [10] and 300 mg/kg for LVX and CIP [this study]), although fecal antibiotics were still very high (94.6 [43.0, 162.1], 20.5 [11.1, 51.0], and 64.4 [10.1, 151.5] μg/g feces for MXF, LVX, and CIP, respectively) and dysbiosis was still present (a drop of the Shannon index by approximately 1 U and an increase in the unweighted UniFrac index of 0.55 to 0.70 U). Thus, the second major result of this study is that full protection of the microbiota from FQ-induced dysbiosis was not necessary for the prevention of CDI in hamsters.
However, DAV131A maintained the balance between the main bacterial groups (such as Firmicutes and Bacteroidetes at the phylum level and Ruminococcaceae and Lachnospiraceae at the family level, Fig. S2). Possibly more importantly, DAV131A limited the loss of microbiota diversity during antibiotic treatments, supporting the hypothesis that CDI prevention was due more to the maintenance of complex bacterial communities than to the presence of particular bacterial groups.
That free fecal antibiotic concentrations in hamsters mimicked rather well those in humans (8, 12, 13) suggests the relevance of transposing the preventative effect of DAV131A on antibiotic-induced CDI in hamsters to humans treated with DAV132, the human-targeted version of DAV131A which is currently in clinical development (8).
Supplementary Material
ACKNOWLEDGMENTS
N.S.-L., F.S.-G., S.S.-J., A.A., F.M., and J.D.G. designed the research. N.S.-L., T.C., M.P., and W.W. performed the research. S.F., T.C., and A.N. performed the metagenomic analysis. A.N. performed the statistical analysis of the data. J.D.G., A.A., C.B., F.M., T.C., and N.S.-L. wrote the paper. All authors agreed on the final version of the manuscript.
N.S.-L., F.S.-G., T.C., A.N., and S.S.-J. are employees of the Da Volterra Company. C.B., A.A., F.M., and J.D.G. are consultants for the Da Volterra Company.
This study was supported by Da Volterra, Paris, France.
Footnotes
Supplemental material is available online only.
REFERENCES
- 1.de Lastours V, Fantin B. 2015. Impact of fluoroquinolones on human microbiota. Focus on the emergence of antibiotic resistance. Future Microbiol 10:1241–1255. doi: 10.2217/fmb.15.40. [DOI] [PubMed] [Google Scholar]
- 2.Brown KA, Khanafer N, Daneman N, Fisman DN. 2013. Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection. Antimicrob Agents Chemother 57:2326–2332. doi: 10.1128/AAC.02176-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Deshpande A, Pasupuleti V, Thota P, Pant C, Rolston DDK, Sferra TJ, Hernandez AV, Donskey CJ. 2013. Community-associated Clostridium difficile infection and antibiotics: a meta-analysis. J Antimicrob Chemother 68:1951–1961. doi: 10.1093/jac/dkt129. [DOI] [PubMed] [Google Scholar]
- 4.Slimings C, Riley TV. 2014. Antibiotics and hospital-acquired Clostridium difficile infection: update of systematic review and meta-analysis. J Antimicrob Chemother 69:881–891. doi: 10.1093/jac/dkt477. [DOI] [PubMed] [Google Scholar]
- 5.Vardakas KZ, Trigkidis KK, Boukouvala E, Falagas ME. 2016. Clostridium difficile infection following systemic antibiotic administration in randomised controlled trials: a systematic review and meta-analysis. Int J Antimicrob Agents 48:1. doi: 10.1016/j.ijantimicag.2016.03.008. [DOI] [PubMed] [Google Scholar]
- 6.Lessa FC, Mu Y, Bamberg WM, Beldavs ZG, Dumyati GK, Dunn JR, Farley MM, Holzbauer SM, Meek JI, Phipps EC, Wilson LE, Winston LG, Cohen JA, Limbago BM, Fridkin SK, Gerding DN, McDonald LC. 2015. Burden of Clostridium difficile infection in the United States. N Engl J Med 372:825–834. doi: 10.1056/NEJMoa1408913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.de Gunzburg J, Ducher A, Modess C, Wegner D, Oswald S, Dressman J, Augustin V, Feger C, Andremont A, Weitschies W, Siegmund W. 2015. Targeted adsorption of molecules in the colon with the novel adsorbent-based medicinal product, DAV132: a proof of concept study in healthy subjects. J Clin Pharmacol 55:10–16. doi: 10.1002/jcph.359. [DOI] [PubMed] [Google Scholar]
- 8.de Gunzburg J, Ghozlane A, Ducher A, Le Chatelier E, Duval X, Ruppé E, Armand-Lefevre L, Sablier-Gallis F, Burdet C, Alavoine L, Chachaty E, Augustin V, Varastet M, Levenez F, Kennedy S, Pons N, Mentré F, Andremont A. 2018. Protection of the human gut microbiome from antibiotics. J Infect Dis 217:628–636. doi: 10.1093/infdis/jix604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Burdet C, Sayah-Jeanne S, Nguyen TT, Miossec C, Saint-Lu N, Pulse M, Weiss W, Andremont A, Mentré F, de Gunzburg J. 2017. Protection of hamsters from mortality by reducing fecal moxifloxacin concentration with DAV131A in a model of moxifloxacin-induced Clostridium difficile colitis. Antimicrob Agents Chemother 61:e00543-17. doi: 10.1128/AAC.00543-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Burdet C, Sayah-Jeanne S, Nguyen TT, Hugon P, Sablier-Gallis F, Saint-Lu N, Corbel T, Ferreira S, Pulse M, Weiss W, Andremont A, Mentré F, de Gunzburg J. 2018. Antibiotic-induced dysbiosis predicts mortality in an animal model of Clostridium difficile infection. Antimicrob Agents Chemother 62. doi: 10.1128/AAC.00925-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kitzis MD. 2010. Antibiotic assay In Antibiogram, p 75–78. ASM Press, Washington, DC. [Google Scholar]
- 12.Edlund C, Sjöstedt S, Nord CE. 1997. Comparative effects of levofloxacin and ofloxacin on the normal oral and intestinal microflora. Scand J Infect Dis 29:383–386. doi: 10.3109/00365549709011835. [DOI] [PubMed] [Google Scholar]
- 13.Rashid M-U, Weintraub A, Nord CE. 2015. Development of antimicrobial resistance in the normal anaerobic microbiota during one year after administration of clindamycin or ciprofloxacin. Anaerobe 31:72–77. doi: 10.1016/j.anaerobe.2014.10.004. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.