Participant disposition and baseline characteristics

The study was conducted between October 2018 and October 2021. Participant enrolment in Phase 1b (AG019 monotherapy) and Phase 2a (AG019/teplizumab combination therapy) is shown in Fig. 1. Sixty participants were screened, 18 of whom were excluded from participation in the study. Twenty-four participants were enrolled and treated in the AG019 monotherapy cohorts; two participants voluntarily withdrew after the 3 month visit. In the AG019/teplizumab combination therapy cohorts, 18 participants were enrolled and treated. All participants completed the study PP, except for one AG019/teplizumab adolescent (open-label) who did not start teplizumab infusions due to a non-TEAE and one placebo adolescent who was lost to follow-up after the 9 month visit.

Fig. 1

Participant disposition. aA total of three single-dose participants were re-enrolled as repeat-dose participants in the different monotherapy cohorts. bSingle-dose participants were not included in the ITT and PP analysis sets. FU, follow-up; PD, protocol deviation; RD, repeat dose; SAF, safety analysis set; SD, single dose

All groups were equally balanced for age, sex, autoantibody positivity and ethnicity (Table 1), and the mean time from diagnosis to treatment start was 102.2 days. Participants’ baseline characteristics were similar in each study arm. All but one of the participants were taking insulin at study entry.

Table 1 Demographics and baseline characteristics Safety and tolerability of AG019

No serious adverse events, deaths or TEAEs leading to discontinuation of AG019 treatment were reported.

TEAEs according to severity and to system organ class are summarised in ESM Tables 3 and 4, respectively. All TEAEs in Phase 1b were of Common Terminology Criteria for Adverse Events (CTCAE) grade 1 or 2; no severe TEAEs (grade ≥3) were reported. The majority of TEAEs (89.7%) were ‘not reasonably related to AG019’. There was no evidence of an effect of dose or age on incidence of adverse events. In Phase 2a, 95.3% of TEAEs were of CTCAE grade 1 or 2. Nine TEAEs of grade 3 or higher were reported in six AG019/teplizumab-treated participants (none in placebo). TEAEs reported as reasonably related to AG019 were mostly gastrointestinal (GI) disorders (diarrhoea and vomiting). One AG019/teplizumab-treated adult reported two grade 3 TEAEs considered reasonably related to AG019 and teplizumab (diarrhoea and vomiting) but AG019 or teplizumab treatment was not discontinued. In line with protocol-defined infusion-withholding criteria, teplizumab treatment was discontinued in five participants due to TEAEs (ESM Table 5); all participants continued AG019 treatment and completed the study. Transient changes in laboratory safety variables were observed in AG019/teplizumab-treated participants (including increases in liver function tests and decreases in lymphocyte, leucocyte and platelet counts), which were considered clinically significant and were reported as TEAEs in nine participants.

AG019 PK analysis

AG019 bacteria were not detected in blood by plating or by qPCR, nor were there indications of AG019-related hPINS or hIL-10 in plasma measured by ELISA, either during treatment or 1 month after the last dose of AG019 (day 90).

Twenty-five participants (nine high-dose AG019-treated, 13 AG019/teplizumab-treated and three placebo-treated) provided faecal samples. AG019 bacteria were detected in faecal samples of 18/22 (82%) repeat-dose participants treated with AG019 high-dose monotherapy or AG019/teplizumab combination therapy at one or more post-screening sampling timepoints, indicating GI exposure to AG019 after oral dosing (Fig. 2). Of the four participants who lacked faecal bacterial recovery, two were excluded from the PP analysis set due to a major protocol deviation (the use of antibiotics, etc.) or AG019 compliance <75%. No AG019 bacteria were detected in the faecal samples from placebo-treated participants.

Fig. 2

Concentrations of AG019 bacteria in faecal samples collected after the last day of AG019 dosing (day 56) from participants treated with (a) AG019 monotherapy or (b) AG019/teplizumab combination therapy. LLeq, L. lactis equivalents; grey area is below the limit of detection (LOD) (i.e. 9×103 LLeq/g). Bacterial concentrations are expressed as LLeq/g and were log-transformed. Bars indicate means ± SEM. Faecal samples from 25 participants were analysed: nine AG019 monotherapy participants (five adults, four adolescents, all high-dose treated), 13 AG019/teplizumab combination therapy participants (ten adults, three adolescents) and three placebo participants (all negative; data not shown). Participants excluded from the PP analysis set are indicated in open symbols

On day 64 (1 week after last AG019 dosing), AG019 was undetectable in 11/15 (73%) of the participants who provided a sample and had detectable levels of AG019 bacteria at one or more earlier timepoints. AG019 levels were detectable but strongly decreased (mean 2.9 log decrease from peak levels) in the other four participants.

Effects of treatments on metabolic responses

The C-peptide analysis results are shown in Fig. 3 and ESM Fig. 3. In the adult monotherapy cohort, the mean 2 h C-peptide AUC at 6 months was not significantly changed from baseline (85%) but declined at 12 months (60% of baseline, p=0.03 on absolute values). Among adolescents, the mean 2 h C-peptide AUC declined at 6 months (70% of baseline, p=0.044 on absolute values) and 12 months (66% of baseline, p=0.07 on absolute values). In the adult combination therapy cohort, the C-peptide response increased (112%) at 6 months and was unchanged (100%) at 12 months compared with declines in the placebo-treated group (73% and 54% of baseline, n=2). Similarly in adolescents, the C-peptide increased to 124% of baseline levels at 6 months (p=0.007) and 108% at 12 months vs 77% in the placebo-treated adolescent at 6 months (no data at 12 months).

Fig. 3

C-peptide mean 2 h AUC over time. (a) Absolute values in participants treated with AG019 monotherapy. (b) Percentage change from baseline in participants treated with AG019 monotherapy. (c) Absolute values in participants treated with AG019/teplizumab combination therapy and placebo. (d) Percentage change from baseline in participants treated with AG019/teplizumab combination therapy and placebo. Data are based on the PP analysis set and are means ± SEM. Pairwise comparison vs baseline (post hoc t test): *p<0.05 (in adolescents), **p<0.01 (in adolescents), †p<0.05 (in adults). Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period

In the adult monotherapy cohort, the mean levels of HbA1c declined during AG019 treatment and were significantly decreased as compared with baseline (p=0.036 and p=0.044 at 3 and 6 months, Fig. 4). In the adolescent monotherapy cohort, HbA1c was not significantly changed from baseline. Daily insulin use increased from baseline by 0.17 IU kg−1 day−1 in AG019 monotherapy-treated adults and by 0.02 IU kg−1 day−1 in adolescents at 12 months. In AG019/teplizumab-treated adults and adolescents, there was a decrease in HbA1c from baseline to month 12. In adults receiving the combination therapy, the levels were significantly lower at months 2 (p=0.009) and 3 (p=0.038). At 12 months, the mean daily insulin use decreased as compared with baseline (by −0.03 IU kg−1 day−1) in AG019/teplizumab-treated adults and increased by 0.03 U kg−1 day−1 in adolescents.

Fig. 4

HbA1c (a, c) and insulin use (b, d) over time in participants treated with (a, b) AG019 monotherapy or (c, d) AG019/teplizumab combination therapy and placebo. Data are based on the PP analysis set. Horizontal line for HbA1c represents the target for glycaemic control (53 mmol/mol). Pairwise comparison vs baseline (post hoc t test): *p<0.05, **p<0.01 in adults. Data are means ± SEM. Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period

Effects of treatment on total and antigen-specific CD4+ and CD8+ T cells

We previously reported that teplizumab induced CD8+ T cells that express the KLRG1, TIGIT and EOMES genes, and found that higher levels after treatment were associated with clinical responses to the drug [13, 26]. Consistent with past results, these CD8+ T cells were increased in the AG019/teplizumab combination therapy group compared with baseline at month 6 (p=0.013), but not in the participants treated with AG019 alone or placebo (Fig. 5).

Fig. 5

Frequency of partially exhausted (EOMES+TIGIT+) CD8+ T cells over time in participants treated with AG019 monotherapy, AG019/teplizumab combination therapy and placebo. (a) Data per treatment group. (b) Individual data. Data are based on a subset of the PP analysis set; adults and adolescents are included together. Percentages were normalised to baseline using the log fold change and data are means ± SEM. Mixed effect analysis vs baseline: †p<0.05; unpaired t test vs AG019: **p<0.01. Blue lines, monotherapy; red lines, combination therapy; black lines, placebo. Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period

Based on preclinical data in NOD mice, we analysed the frequencies of antigen-specific CD8+ T cells and two populations of regulatory CD4+ antigen-specific T cells. The frequency of PPI-specific CD8+ T cells significantly decreased from baseline by 22.5% (p=0.016) at 3 months in AG019 monotherapy participants and by 21.6% (p=0.035) at 6 months in the combination group (Fig. 6). In the placebo group there was an average 12.5% increase at 3 months and a 17.1% reduction at 6 months. There was no significant change in the frequency of cytomegalovirus/Epstein–Barr virus viral-specific CD8+ T cells in either treatment group (Fig. 6), and no correlation was found between the change in PPI-specific CD8+ T cell frequency and age.

Fig. 6

Frequency of (a) PPI-specific CD8+ T cells and (b) CMV/EBV-specific CD8+ T cells over time in participants treated with AG019 monotherapy and AG019/teplizumab combination therapy. Data are based on the PP analysis set. Percentages were normalised to baseline using the log fold change and data are means ± SEM. Mixed effect analysis vs baseline: *p<0.05 (AG019 monotherapy), †p<0.05 (AG019/teplizumab combination therapy). Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period. CMV/EBV, cytomegalovirus/Epstein–Barr virus

There was a modest increase in the frequency of PPI-specific IL-10+ regulatory type 1 T cells (Tr1s) in adults treated with monotherapy (0% at baseline, 2.7% at 3 months) and with combination therapy (2.3% at baseline, 4.1% at 3 months, Fig. 7a), but not in placebo-treated participants. Responses to control (viral and bacterial) antigens were unchanged. The PPI-reactive Tr1s were below the limit of detection in adolescent participants. The frequency of PPI-specific CD4+ memory Tregs showed a similar increase in adult participants treated with monotherapy and with combination therapy (Fig. 7b), whereas this was not seen in adolescents.

Fig. 7

Frequency of PPI-specific (a) Tr1s and (b) memory Tregs (mTregs) in adults treated with AG019 monotherapy or AG019/teplizumab combination therapy. (a) The frequency of PPI-specific IL-10+ Tr1s in adults treated with AG019 monotherapy (n=3) or AG019/teplizumab combination therapy (n=5) at baseline and at month 3. Thick lines represent the mean frequency and thin lines depict the frequency of cells for each participant. (b) The frequency of PPI-specific mTregs in adults treated with monotherapy (n=4) or combination therapy (n=7) at baseline, 3 months and 6 months as described in (a). Two adult participants treated with placebo are depicted in black. The parent population for Tr1s and mTregs is total PPI-specific CD4+ T cells. Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period

Association of T cell changes with C-peptide preservation

We designated post hoc clinical responders as participants who lost ≤9.7% of their baseline C-peptide at month 6 based on published criteria and examined a correlation with immunological results. Seven of 16 (44%) participants (5/9 adults and 2/7 adolescents) in the AG019 monotherapy group were classified as responders at 6 months. In AG019/teplizumab combination therapy, there were 11/14 (79%) responders (7/10 adults and 4/4 adolescents) at month 6 and 0/3 in the placebo group (ESM Table 6, ESM Fig. 4). In AG019 monotherapy, the frequency of PPI-specific CD8+ T cells was lower in the responders vs non-responders at 3 months (mean decrease from baseline of 33% in responders vs 15% in non-responders). In AG019/teplizumab combination therapy, antigen-specific T cells showed a 27% reduction in the frequency of PPI-specific CD8+ T cells in the responders compared with an increase of 6% in the non-responders at 6 months (Fig. 8).

Fig. 8

PPI-specific CD8+ T cells over time in clinical responders and non-responders at 6 months in (a) AG019 monotherapy and (b) AG019/teplizumab combination therapy. Data are based on a subset of the PP analysis set and are presented as means ± SEM. AG019 monotherapy responders (n=3, blue); AG019 monotherapy non-responders (n=4, grey); AG019/teplizumab combination therapy responders (n=11, red); AG019/teplizumab combination therapy non-responders (n=2, grey). Grey shading indicates the AG019 treatment period; blue shading indicates the AG019+teplizumab treatment period