The DAV132 colon-targeted adsorbent does not interfere with plasma concentrations of antibiotics but prevents antibiotic-related dysbiosis: a randomized phase I trial in healthy volunteers

Abstract

The deleterious impact of antibiotics (ATB) on the microbiome negatively influences immune checkpoint inhibitors (ICI) response in patients with cancer. We conducted a randomized phase I study (EudraCT:2019-A00240-57) with 148 healthy volunteers (HV) to test two doses of DAV132, a colon-targeted adsorbent, alongside intravenous ceftazidime-avibactam (CZA), piperacillin-tazobactam (PTZ) or ceftriaxone (CRO) and a group without ATB. The primary objective of the study was to assess the effect of DAV132 on ATB plasma concentrations and both doses of DAV132 did not alter ATB levels. Secondary objectives included safety, darkening of the feces, and fecal ATB concentrations. DAV132 was well tolerated, with no severe toxicity and similar darkening at both DAV132 doses. DAV132 led to significant decrease in CZA or PTZ feces concentration. When co-administered with CZA or PTZ, DAV132 preserved microbiome diversity, accelerated recovery to baseline composition and protected key commensals. Fecal microbiota transplantation (FMT) in preclinical cancer models in female mice from HV treated with CZA or PTZ alone inhibited anti-PD-1 response, while transplanted samples from HV treated with ATB + DAV132 circumvented resistance to anti-PD-1. This effect was linked to activated CD8⁺ T cell populations in the tumor microenvironment. DAV132 represents a promising strategy for overcoming ATB-related dysbiosis and further studies are warranted to evaluate its efficacy in cancer patients.

Fig. 1. Clinical trial design and pharmacokinetics of antibiotics in the plasma and feces of healthy volunteers (HV).

A Prospective randomized trial design of 148 HV that included in one of the 12 groups: no ATB control (CTRL) without and with DAV132 alone, CZA or PTZ or CRO alone, or CZA or PTZ or CRO in combination with DAV132. ATB were delivered intravenously for 5 days, while DAV132 was administered orally at two doses either 7.5 g po tid or 12 g po tid for 7 days. B Mean plasma concentration (mg/L) of ceftazidime (left panel) and piperacillin (right panel) measured over 18 h after D5. τ was defined as the time interval between two administrations (8 h for PTZ, 8 h for CZA, 24 h for CRO). C Mean fecal concentration (mg/kg) of ceftazidime (left panel) and piperacillin (right panel) measured over 9 days. AUC of antibiotics concentrations were compared between arms using an ANOVA, followed by pairwise comparisons with a Tukey’s adjustment, n = 11–13 HV per group. ATB antibiotic, AUC area under the curve, CZA ceftazidime-avibactam, D day, po orally, PTZ piperacillin-tazobactam, CRO ceftriaxone, q8h every 8 h, q24hr every 24 hour, rRNA ribosomal RNA, tid three times a day, hashtag 16S rRNA sequencing performed, black diamond, metagenomic sequencing performed. (ns non-significant, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 2. Impact of DAV132 12 g on the gut microbiome in healthy volunteers (HV).

A 16S rRNA microbiota profiling: mean change in α-diversity from baseline as measured by the Shannon Index in HV treated in CTRL, CZA, and CZA + DAV132 (left panel), and CTRL, PTZ, and PTZ + DAV132 (right panel) groups. AUC of change from baseline of the Shannon index until D16 were compared between arms using an ANOVA, followed by pairwise comparisons with a Tukey’s adjustment. B Bray–Curtis representation for beta diversity (β-diversity) between HV treated with CTRL, CZA, and CZA + DAV132 (upper panel) at D1, D6, D25, and D37, and in CTRL, PTZ, and PTZ + DAV132 (lower panel) at D1, D6, D16, and D37. Global and pairwise PERMANOVA were used to compare differences in β-diversity between groups. C Bray–Curtis distance measure between control and CZA or CZA + DAV132 at D1, D6, and D16 (upper panel) and between control and PTZ or PTZ + DAV132 (lower panel) at D1, D6, and D25. AUC area under the curve, CTRL control, CZA ceftazidime-avibactam, D day, ns not significant, PTZ piperacillin-tazobactam, DAV132, DAV132 12 g po tid. (ns non-significant, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 3. Impact of antibiotics alone or in combination with DAV132 12 g on microbiome composition in healthy volunteers (HV).

A Heatmap comparing the global 16S rRNA intestinal microbiota of CZA vs. CZA + DAV132 over time from D1 to D16. Each column represents one HV included in the clinical trial. B Metagenomics microbiome profiling with volcano plot representation of the adjusted p-value versus the magnitude of change (fold change) of the Metagenomic Species relative abundance observed between CZA and CZA + DAV132 at D1, and at (C) D6 or (D) between PTZ and PTZ + DAV132 at D6. Dotted red line represents the significance threshold 0.05 used in the differential analysis. E. Common bacterial species found to be significantly different between CZA or PTZ and CZA or PTZ + DAV132. CZA ceftazidime-avibactam, D day, PTZ piperacillin-tazobactam, DAV132 DAV132 12 g po tid. Source data are provided as a Source Data file.

Fig. 4. Correlation between qPCR panel and metagenomics on the microbiome evolution of healthy volunteers (HV) on the trial.

A Heatmap aligning matching metagenomics and qPCR bacteria n = 84 of CZA D1, D6 and CZA + DAV132 D1 and D6. Kendal score represents the commonality between both sequencing strategies. Each column represents one HV. B qPCR normalized absolute representation of 6/12 common bacterial species found to be significantly different between CZA and PTZ vs CZA-DAV132 and PTZ + DAV132. Statistical test only performed between ATB vs ATB + DAV132 with a non-parametric Wilcoxon rank sum test. ATB antibiotic, CZA ceftazidime-avibactam, D day, DAV132 DAV132 12 g po tid. (ns non-significant, ****p < 0.0001). Source data are provided as a Source Data file.

Fig. 5. Murine recolonization with feces from healthy volunteers (HV) on DAV132 protects αPD-1 response during CZA and PTZ treatment.

A Experimental design of avatar mice experiments. B Representative MCA-205 tumor growth curve in germ-free mice after FMT from HV, in the CZA group (HV1) at baseline D1 (upper panel) or D6 (lower panel), or from HV in the CZA + DAV132 group (HV4) at baseline D1 (upper panel) or D6 (lower panel), treated with Iso-PD-1 control or αPD-1 n = 5 mice/group. C Mean tumor size at sacrifice in MCA-205–inoculated germ-free mice at sacrifice following FMT from three HV on CZA (D6) and three HV on CZA + DAV132 (D6). D Pooled MCA-205 tumour growth curves in ATB-treated mice post-FMT samples obtained at D6 in 2 HV who received CZA (n = 10 mice/group) or CZA + DAV132 in duplicate HV treated with Iso-PD-1 control or αPD-1 (n = 20 mice/group). E Mean tumor size in MCA-205–inoculated in ATB-treated mice at sacrifice following FMT from duplicated two HV on CZA (D6) and duplicated two HV on CZA + DAV132 (D6). F Mean tumor size in B16-OVA–inoculated in ATB-treated mice at sacrifice following FMT from two HV on CZA (D6) and two HV on CZA + DAV132 (D6). G Pooled MCA-205 tumour growth curves in ATB-treated mice post-FMT samples obtained at D6 in 3 HV who received PTZ or 3 HV PTZ + DAV132 treated with Iso-PD-1 control or αPD-1 n = 15 mice/group. (Data are presented as mean ± SEM, the Mann–Whitney U test was used to determine significant differences among the different groups at the time of sacrifice. For C–F, G, each dot represents one mouse and each shape represents one HV. (ns non-significant; ****p < 0.0001). ATB antibiotic, CZA ceftazidime-avibactam, D day, FMT fecal microbiota transplantation, HV healthy volunteer, ns not significant, PTZ piperacillin-tazobactam, DAV132 DAV132 12 g po tid. Source data are provided as a Source Data file.

Fig. 6. DAV132 maintains an effective immune response during ATB treatment in mice.

A Flow cytometry analysis of the total CD8⁺Tcells and the ratio of CD8⁺ T cells/Treg population in MCA-205 tumor infiltrating lymphocytes (TILs) from ATB-treated mice recolonized with FMT from D6 samples in four HV (1, 2, 4, and 5) who received CZA alone or CZA + DAV132 at D6 (Fig. 5D, E). Data are presented as mean ± SEM, the Mann–Whitney U test was used to determine significant differences among the different groups. Each dot represents one mouse. B From experiment presented in Fig. 5D, E, UMAP visualization of tumor-infiltrating T cells in the group FMT CZA + DAV132 (Iso-PD-1 or αPD-1) C Heatmap representation of the 26 populations visualized in the UMAP with their respective normalized mean fluorescence intensity of the depicted markers. In grey, the frequency of pooled population for each group and the statistical analysis were done based on the percentage per mouse. D RNAseq of the bulk tumor representing the FMT from CZA-DAV132 + αPD-1 compared to the 3 other groups. n = 5mice/group, samples depicted in first two principal components space. For each of the 2 axes, values within parentheses correspond to explained variance. Ninety-five percent confidence ellipses were added. E ImmuCellAI-mouse extrapolation of central memory CD8⁺ T cells. The center line indicates the median value, lower and upper hinges represent the 25th and 75th percentiles, respectively, and whiskers denote minimum and maximum. F RNAseq heatmap representing expression of genes differentially expressed on MCA-205 CD8⁺ T cells sorted n = 5mice/group between CZA-DAV132 + Iso-PD-1 and CZA-DAV132 + αPD-1 and other groups (FDR adjusted p < 0.01). Rows: genes were clustered based on Euclidean distance and « complete » method. Columns: samples were ordered given group membership. G qPCR of ifng gene and the ifng/il-10 ratio in the mesenteric lymph nodes and draining lymph node from the 4 experimental groups, Data are presented as mean ± SEM, the Mann–Whitney U test was used to determine significant differences among the different groups. CZA ceftazidime-avibactam, FMT fecal microbiota transplantation, infg interferon gamma, il10 interleukine-10, PD-1 programmed cell death protein 1, PTZ piperacillin-tazobactam, DAV132 DAV132 12 g po tid. (ns non-significant).

Fig. 7. Correlation of the murine microbiota post-FMT.

A Compositional histograms of two representative examples of the gut microbiota composition determined with 16S rRNA sequencing for the HV donor collected at D1 or D6 of CZA vs CZA + DVA132 compared to the gut microbiota mice post-FMT with these feces samples at different timepoints; T0 (tumour inoculation), T1 (before start of the treatment) T5 (after four αPD-1 or isotype control cycles). Left panel represents on top the HV1 microbiota composition before (D1) and after (D6) CZA treatment and below, recolonized murine fecal samples collected during experiment. Right panel represents HV5 before (D1) and after (D6) CZA + DAV132 treatment. In the lower panels, each column represents one fecal sample. B Principal coordinates analysis (PCoA) analysis based on Bray–Curtis distances of HV donor and matching murine fecal samples collected after four cycles of αPD-1 or isotype control (D17 on the Fig. 5A). Each color corresponds to one treatment at one timepoint and each shape to one HV donor. A PERMANOVA test was applied to test statistical differences between groups. C Heatmap comparing fecal samples from germ-free mice after two αPD-1 administrations (D12) following FMT from HV donors who were on CZA alone or CZA + DAV132 obtained at D1 or D6. Each column represents one mouse. CZA ceftazidime-avibactam, D day, FMT fecal microbiota transplantation, HV healthy volunteer, PD-1 programmed cell death protein 1, DAV132 DAV132 12 g po tid. Source data are provided as a Source Data file.