To determine the therapeutic potential of orally delivered EDP1815 in Th1-driven inflammation a murine DTH was performed. Mice were sensitized subcutaneously with keyhole limpet hemocyanin (KLH) on the back and subsequently dosed with EDP1815 by oral gavage for 4 weeks and given an ear challenge with KLH on the day 29. 24hr later, ear swelling was measured as a marker of inflammation. Treatment with EDP1815 significantly reduced ear swelling compared with the vehicle-treated group (Figure 1A). Furthermore, in a shorter model, mice were dosed with EDP1815 by oral gavage for 8 days, given a KLH challenge on the ear on the ninth day and ear swelling was measured 24 h post challenge. EDP1815 was the most efficacious strain in lowering ear inflammation compared to strains of other closely related Prevotella species (Figure 1B).

EDP1815-102 was a phase 1b, single-center, randomized placebo-controlled study investigating the potential of EDP1815 to modulate Th1-driven inflammation in healthy human volunteers. Participants were dosed with EDP1815 for 28 days. Immunological sensitization to KLH was induced by intramuscular injection on day 3 of dosing followed by intradermal KLH challenge on day 26. Inflammation was assessed using specialized imaging techniques to measure antigen specific responses to KLH challenge comparing drug-treated and placebo participants, expressed in arbitrary units (AU) of Laser Speckle Contrast Imaging (LSCI) of basal flow and flare. Skin color and average redness was assessed by multispectral imaging. These quantitative endpoints were measured just before and then 2 days following the intradermal KLH challenge. Thirty two subjects were enrolled in 2 cohorts. In each cohort, 12 subjects received EDP1815, and 4 received matching placebo for 28 days. Active subjects in both cohorts were administered with 8.0 × 10¹¹ cells of EDP1815, once daily as either 10 (cohort 1) or 5 capsules (cohort 2). Participants in cohort 2 were fasted 2 h pre-dose.

Although the study was not powered to detect statistically significant differences across treatment groups, notable trends corresponding to a reduction in inflammation as measured by dermal imaging were observed in the groups treated with EDP1815 in comparison to placebo. This was observed for all measurements: LSCI basal flow and flare (Figure 2A), and multispectral imaging skin color and redness (Figure 2B). The effects were consistent across measures and reproduced in the two cohorts. Given subject numbers the effect size did not reach statistical significance.

EDP1815 was safe and well tolerated with no overall difference in adverse events from placebo. There were no severe adverse events in any participants (Table 1).

The vitamin D3 analog, MC903, can be used in mice to generate Th2-driven epidermal inflammation with increased dermal cell infiltrates consisting of eosinophils, T cells, neutrophils, and mast cells. Following application of MC903, skin shows increased levels of a range of Th2 cytokines including IL-4, IL-5, IL-13, and IL-31 (29). BALB/c mice were sensitized on the ears with MC903 for 14 days and dosed orally daily with EDP1815 for 14 days. Treatment with EDP1815 resulted in significantly lower ear inflammation in comparison with vehicle-treated animals, on par with tofacitinib, an oral JAK-1/3 inhibitor, and a systemic antibody blocking IL-4 (Figure 3A). In animals treated with EDP1815, ex vivo analysis revealed a reduction in levels of IL-4, a central Th2 cytokine, as well as of IL-31 (Figure 3B).

To test the potential of EDP1815 to treat Th2-driven inflammatory disease, it was evaluated in a phase 1b clinical study (EudraCT # 2018-002807-32). A cohort of 24 participants with mild and moderate atopic dermatitis was randomized 2:1 active: placebo. 8.0 × 10¹¹ total cells of EDP1815 was administered once daily for 56 days, with a follow-up visit after 14 days off drug on Day 70. The primary endpoint was safety and tolerability of EDP1815. Secondary endpoints included physician-rated scales of atopic dermatitis severity (EASI, IGA, BSA, IGA*BSA, and SCORAD); as well as patient-reported outcomes (Pruritus-NRS, DLQI, and POEM). Baseline mean EASI and IGA scores were 8.31 and 2.63 respectively, for the 16 patients receiving EDP1815, and 9.31 and 2.7, respectively for the 8 patients receiving placebo.

EDP1815 had a placebo like safety and tolerability profile with no treatment-related adverse events of moderate or severe intensity, and no serious adverse events (Table 1).

The differences in percentage decrease from baseline in EASI, IGA*BSA and SCORAD between the EDP1815 treated group and the placebo group were 52% (p = 0.062), 65% (p = 0.022), and 35% (p = 0.068), respectively (Figure 4A). 10 of 16 patients receiving EDP1815 saw improvements in their EASI score at day 56, compared to only 2 out of 8 patients receiving placebo (Figure 4B).

At the day 70 follow-up visit, further clinical improvements were observed in the EDP1815 treated group. The percentage of patients receiving EDP1815 achieving EASI50 was 44% compared with 0% in the placebo group (Figure 4C); and the proportion achieving an IGA score of 0 or 1 was 31%, with 0% again in the placebo group (Figure 4D). Figure 4E shows a representative clinical improvement of skin condition in a participant receiving EDP1815, and no other oral or topical treatments, for 56 days in this trial. This was an EASI50 response, with an improvement from an EASI score of 9.8 at baseline to 4.9 at day 56. In addition to the clinical improvements in the physician rating scales, this participant’s patient reported outcomes (PROs) also improved with the DLQI score moving from 13 (severe impact) to 1 (no impact), and the POEM score from 22 (self-rating of ‘severe eczema’) to 5 (‘mild eczema’) at day 56.

The mean individual PRO improvement from baseline in the DLQI (−3.6) and POEM (−4.1) in EDP1815-treated patients at day 56 exceeded the minimally clinically important difference thresholds (30, 31) and exceeded the placebo group changes (−0.3 and + 1.6, respectively). Mean improvements in itch and sleep were seen within all scales measuring these parameters (Pruritis-NRS, DLQI, SCORAD, and POEM) at the end of the treatment period.

These results provide proof of concept that EDP1815 can resolve Th2-driven inflammation with a placebo-like safety and tolerability profile in patients with atopic dermatitis.

A TLR7 agonist, imiquimod, induces clinical and histological changes characteristic of human psoriasis, including epidermal thickening, scaling and erythema (32). Mice were sensitized on the ear and back with imiquimod cream daily for 7 consecutive days and dosed daily with oral EDP1815, vehicle, dexamethasone, or anti-IL17A antibody. Ear thickness was a measure of inflammation. Treatment effects in animals dosed with EDP1815 were seen as early as 4 days after the start of IMQ application and were comparable to those observed in animals treated with dexamethasone or anti-IL-17A in reducing ear thickness as well as back inflammation (Figure 5A). At termination on day 8, IL-17A protein levels in the ear tissue were reduced by treatment with EDP1815 in comparison to vehicle (Figure 5B). IMQ is known to also induce an increase in IL-17A production in splenocytes (32). Ex vivo re-stimulation of splenocytes with PMA/Ionomycin showed decreased production of IL-17A in mice treated with EDP1815 (Figure 5B).

To determine the potential of EDP1815 to treat Th17-driven inflammatory disease it was evaluated in two parallel cohorts of a phase 1b clinical study in patients with psoriasis (EudraCT # 2018-002807-32). Adult patients with mild to moderate chronic plaque psoriasis were randomized 2:1 to receive EDP1815 or matching placebo capsules. Doses were 1.6 × 10¹¹ (cohort 3) and 8.0 × 10¹¹ (cohort 4) bacterial cells, once daily for 28 days, with follow-up after 14 days off treatment at day 42.12 patients were dosed with the lower dose, and 18 with the higher dose. Placebo subjects were pooled across both cohorts. The primary endpoint was safety and tolerability of EDP1815. Secondary endpoints included physician-rated scales of psoriasis: Psoriasis Area and Severity Index (PASI) and Lesion Severity Score (LSS). Baseline mean PASI scores were 9.5 (cohort A-3), 6.2 (cohort 4), and 6.7 (pooled placebo cohorts). Mean LSS scores at baseline were 8.1 (cohort 3), 7.8 (cohort 4), and 7.8 (pooled placebo cohorts).

The primary endpoint safety data showed EDP1815 to have a safety and tolerability profile comparable to placebo (Table 1). As for the other studies reported here, there were no serious adverse events, and no adverse events of severe intensity.

The PASI score is a composite measure of psoriasis plaque severity, and body coverage (33). Following 28 days of treatment, the mean percentage reduction in PASI for EDP1815 cohorts was 16%, compared to 0.1% for placebo. At day 42, the percentage improvement from baseline in cohort 2 active participants increased further to 21% (Figure 6A). In this cohort, 6 of the 12 patients achieved a 25% improvement in PASI or better at day 42, compared to 1 of 10 in placebo (Figure 6B).

LSS measures the severity of a target plaque using scaling, erythema, and plaque elevation, giving a maximum total score of 12. The mean percentage reductions in LSS scores at day 28 were 23 and 15% in the low-and high-dose cohorts respectively, compared to a 1% increase from baseline in the placebo group. Figure 6C shows the individual percentage changes in LSS from baseline at day 42, with 7 of the 12 patients in the high-dose cohort achieving 25% improvement or greater, compared to 1 of 10 in the placebo group.

These two comparable sets of clinical data provide proof of concept that EDP1815 can drive clinical improvements and resolve inflammation in the skin of patients with psoriasis with a placebo-like safety and tolerability profile. As seen in atopic dermatitis, responses were continuing to improve and had not reached peak effect at the end of the dosing period.

To determine the biodistribution of EDP1815 following oral dosing in mice, strain-specific primers were designed to differentiate EDP1815 from other P. histicola strains. This enabled sensitive tracking of EDP1815 in mouse experiments in the potential presence of alternate species of P. histicola in the background gastrointestinal microbiome.

Following oral administration of a single dose, EDP1815 was transiently detected in the GI tract and stool. EDP1815 was detected in the intestine and stool for up to 8 h and not at 16 h post-administration, showing that the lyophilized microbes did not colonize the gut. Importantly, EDP1815 was not detected outside of the GI tract at any time point. These data demonstrate that EDP1815 is luminally restricted with undetectable systemic exposure following oral dosing in mice (Supplementary Figure S1).

Evidence of systemic exposure of EDP1815 in humans was evaluated in the phase 1b clinical cohorts of psoriasis patients described previously. EDP1815 was not detected in blood by PCR or by culture at any time-point during the dosing period, through completion of the 28 days of dosing.

In these psoriasis cohorts EDP1815 was not detected by qPCR in stool samples taken 14 days after completion of dosing. This experiment was repeated in the healthy volunteer DTH study. Again fecal concentrations of EDP1815 were below the limit of quantification pre-dose and 5 days post last dose in all treatment groups, confirming a lack of gut colonization.

Finally, 16S ribosomal RNA sequencing of stool samples confirmed that the pharmacodynamic activity observed with EDP1815 was not due to secondary alterations in the colonic microbiome. Samples taken at baseline, on drug and at follow up showed no significant changes in the Shannon Index (diversity) or composition from baseline either during or following cessation of dosing (Figure 7). Furthermore, no significant changes in fecal microbe abundance at the genera level were detected for either placebo or EDP1815 dosed subjects when comparing between time points (Supplementary Figure S2). Microbes from the Bacteroides and Blautia genera were the most abundant in the fecal microbiome for all groups, and for all groups the 10 most abundant genera made up more than 90% of the microbiome by percent composition.

These data confirm that the resolution of inflammation observed with EP1815 is not due to systemic exposure, gut-colonization, nor indirect effects on the colonic microbiome.

This was a phase 1, randomized, placebo-controlled, double-blind, multiple dose study in thirty-two healthy volunteers performed at the Center for Human Drug Research (CHDR), Leiden, The Netherlands. The Declaration of Helsinki was the principle for trial execution. The independent Medical Ethics Committee “Medisch Ethische Toetsingscommissie van de Stichting Beoordeling Ethiek Biomedisch Onderzoek” (Assen, the Netherlands) approved the study prior to any clinical study activity. All subjects provided written informed consent before participation. The trial was registered on trialregister.nl (NL8676).

This clinical trial is a first-in-human study of EDP1815 in healthy volunteers, patients with psoriasis, and patients with atopic dermatitis. This randomized, double-blind, placebo-controlled study with dose escalations was designed to assess the safety, tolerability and pharmacodynamic effect of various doses and formulations of EDP1815. The primary objective was safety and tolerability of EDP1815 treatment in each cohort, the secondary objectives were clinical efficacy measures of either psoriasis or atopic dermatitis. As a phase 1 study investigating dose escalations and safety of the investigational medicinal product, the study was not powered for detection of statistical significance of clinical efficacy, but the sample size was selected to determine the initial safety profile of a range of doses of EDP1815, while informing sample size for a subsequent phase 2 study in both psoriasis and atopic dermatitis. Both participants and investigators were blinded to treatment allocation until study completion. 10 cohorts were assessed in this study: cohorts 1–4 and 7 assessed the enteric-coated capsule formulation. Cohorts 5–6 and 8–10 assessed alterations in drug substance or drug product, and therefore results of these cohorts are not presented in this manuscript. Cohorts 1 and 2 were performed in healthy human volunteers, Cohorts 3–4 in patients with mild and moderate psoriasis, and Cohort 7 in patients with mild and moderate atopic dermatitis. The data from these patient cohorts using enteric-coated capsules are presented in this manuscript.

Subjects were randomized to EDP1815 or placebo in a 2:1 ratio. Randomization for each cohort was created using a simple block design generated by an unblinded statistician who had no other involvement in the study. The randomization was administered centrally, with the next available randomization number used for each new participant. Investigators and participants were blinded to treatment assignment and all containers and study medication for placebo were identical to those for EDP1815. For Cohorts 3 and 4, a sentinel pair was used to dose one EDP1815 and one placebo participant; safety data for the first 3 days of multiple dosing for the sentinel pair was reviewed prior to the opening of the cohort for further participants. For Cohort 7, the same dose of EDP1815 as was used in Cohort 4 was administered and as such no sentinel dosing was required.

The sample sizes were chosen to explore the tolerability and safety of this new treatment and no formal power calculations were performed. All participants taking at least one dose of study medication were included in the safety analyses. For the efficacy analyses, all randomized participants were included. The protocol did pre-specify that any participants who had an important protocol deviation affecting psoriasis-related efficacy variables would be excluded, but no such deviations occurred.

For Cohorts 3 and 4, where a single dose was administered before being followed up with daily dosing on Day 3, Baseline for efficacy endpoints was assessed as the measurement taken at the Day 3 (start of daily dosing) visit. For Cohort 7, no sentinel dose was used and Baseline was assessed as the measurement taken on Day 1.

Incidences of AEs and SAEs were produced by treatment and severity with separate summaries of study drug-related events. For the continuous efficacy endpoints for skin assessment (LSS and PASI for psoriasis; EASI, IGA*BSA and SCORAD for atopic dermatitis), data was analyzed with a mixed model for repeated measures, including terms for treatment, visit, baseline score and treatment-by-visit interaction. Waterfall plots showing individual percentage changes from baseline were also produced. Patient reported outcomes (DLQI, POEM and pruritis numerical rating scales) were summarized using mean, median, standard deviation, and range. Responder endpoints were summarized using the number and percentage of participants to meet the relevant response definition.

Two cohorts of 12 and 18 patients with mild to moderate psoriasis, both randomized 2:1 active to matching placebo.

EDP1815 was administered as a single dose (day 1), and after confirming safety, as a once daily for 28 days (days 3–30), with follow-up at day 42. The dose was 1.6 × 10¹¹ (cohort 3) or 8.0 × 10¹¹ (cohort 4) bacterial cells per day. Placebo subjects were pooled across both cohorts in the analysis.

Stool samples were taken at three time points: baseline, day 28, and day 42. Stool was collected in DNA/RNA Shield Fecal Collection tube (Zymo Research) and stored at −80°C until processing. DNA extraction, qPCR and 16S sequencing were performed at and by Baseclear (Leiden, Netherlands) according to their SOPs.

For 16S sequencing, the V4 region of bacterial 16S ribosomal RNA were amplified using universal primer set 515F and 806R (41). Resulting products were sequenced through the Illumina MiSeq platform on a 2 × 250 paired-end run. Reads were demultiplexed and quality filtered before being uploaded to the One Codex platform (42). Paired-end reads were merged and then characterized using One Codex’s in-house Targeted Loci Database, a curated database of bacterial marker genes including 16S ribosomal RNA. Read count and relative abundance tables were calculated at the genus level and retrieved using custom Python scripts and the One Codex Python library.

To determine whether some genera were abundant in Placebo-vs EDP1815-treated individuals, read count tables were fed to ANCOM, a statistical framework for the analysis of microbiomes (43). Genera were determined to be significantly different between comparators if mean relative abundance was >1% in either comparator, or they passed the significance threshold identified by ANCOM.

16S sequencing reads were classified at the genera level using the One Codex Platform (42) Reads were grouped by subject ID, treatment and time point using custom Python scripts and the Pandas library (44). Figure was generated using custom python scripts and the python libraries Matplotlib and Seaborn (45, 46).

Twenty-four participants with mild and moderate atopic dermatitis were randomized 2:1 active to matching placebo. EDP1815 was administered as 8.0×10¹¹ bacterial cells once daily for 56 days, with follow-up at Day 70.