Summary

Introduction

Enterotoxigenic Escherichia coli remains one of the most common bacterial pathogens causing diarrhoea, leading to substantial mortality and morbidity in children in low-income and middle-income countries (LMICs), and vaccine development remains a WHO priority.^{1, 2, 3, 4, 5, 6} Enterotoxigenic E coli causes disease by colonising the small intestine by expressing different colonisation factors on the bacterial surface and subsequently releasing heat-labile or heat-stable enterotoxins.^{7, 8, 9, 10} Immune protection is most likely provided by secretory IgA antibodies against colonisation factors and heat-labile toxin, which are produced in the intestine.^{1, 10}

On the basis of WHO recommendations,¹¹ we have developed an oral enterotoxigenic E coli vaccine (ETVAX) containing recombinant E coli strains overexpressing the most prevalent enterotoxigenic E coli colonisation factors (ie, CFA/I, CS3, CS5, and CS6) at substantially higher concentrations than expressed by clinical enterotoxigenic E coli isolates^{10, 12, 13} and combined with a heat-labile-like toxoid LCTBA (a hybrid molecule of cholera toxin B subunit [CTB] and heat-labile toxin B subunit [LTB]).¹⁴ ETVAX might be more immunogenic when used in combination with a mucosal adjuvant, the double-mutant heat-labile toxin (dmLT).^{13, 15}

When tested in Swedish adults, ETVAX, with or without dmLT, was safe and induced statistically significant faecal secretory IgA and IgA antibody in lymphocyte secretions (ALS) responses against all colonisation factors and LTB in most vaccinees.¹⁶ ETVAX also induced long-lasting immunological memory in Swedish adults.¹⁷ On the basis of these results, we designed a large, dose-escalation, age-descending phase 1/2 trial of ETVAX in Bangladesh.

First, we assessed the safety and immunogenicity of ETVAX in adults aged 18–45 years in Bangladesh.¹⁸ The results in adults indicated that ETVAX was safe and induced ALS responses against all five primary vaccine antigens in 100% of vaccinees and strong plasma immune responses in 60–100%.¹⁸ On the basis of the safety data in adults, we proceeded to assess the vaccine in children, with the primary objective to establish the largest tolerated dose of ETVAX, with or without dmLT. A secondary objective was to assess the potential of the vaccine to induce mucosal and systemic immune responses against the five primary vaccine antigens.

Materials and methods

Results

Between Dec 7, 2015, and Jan 10, 2017, we screened 1500 children across the three age groups, of whom 430 were enrolled and randomly assigned to the different treatment groups (130 children aged 24–59 months, 100 aged 12–23 months, and 200 aged 6–11 months; table 1; appendix pp 2–4). The sex, age, and nutritional status distributions of children were similar between the different cohorts in each age group (appendix p 5). All children were from a similar socioeconomic background.

Across all age groups, no solicited adverse events occurred that were greater than moderate in severity, and most were mild (table 2). The most common solicited adverse event was vomiting. Most episodes were mild, with moderate vomiting only in participants given the largest fractional doses. Vomiting was inversely related to age (p=0·004) and correlated to vaccine dose, with lower doses presenting less risk of vomiting (p=0·0076; appendix p 6). In children aged 24–59 months, moderate vomiting was observed in two of the three children receiving the full dose of vaccine. Because the study was not designed to test the full dose in children, although one of the goals of the study was to establish the highest tolerable dose, enrolment in this cohort was curtailed on the basis of vomiting events, identifying the half dose as the largest tolerated dose for children aged 24–59 months, as well as for children aged 12–23 months. Ten of 30 infants aged 6–11 months given the half dose had vomiting, thus the quarter dose was the largest tolerated dose to test in combination with dmLT in this age group. The addition of dmLT did not appear to modify the frequency or severity of vomiting (p=0·10; appendix p 6). Vomiting occurred exclusively in vaccine recipients, generally started within 1–8 h after administration, and was transient (in all cases it resolved without intervention within 1 day).

In all age groups, the only other solicited event that was ever greater than mild in severity was fever (table 2). Moderate fever occurred in at most one participant in any vaccine dose group, and the frequency of fever did not differ significantly between vaccine and placebo groups. 29 (7%; 20 children who received the vaccine and nine who received placebo) of the 430 participants had solicited fever, of which the majority occurred at least 24 h after vaccine administration and for a short duration. Only two participants (one in the age 6–11 months quarter dose and one in the age 6–11 months quarter dose with 2·5 μg dmLT group) had fever on the day of vaccination, and in both the fever was mild and resolved on the same day. The other solicited events were infrequent and mild (table 2).

As part of routine safety monitoring, blood tests were done on all participants 7 days after vaccination. The laboratory aberrations detected during monitoring that were assessed by investigators as treatment-related included seven events, most of which were increased white blood cell count. The investigators were unaware of treatment assignment when assessing relatedness to treatment. Of note, four of those seven events were in placebo recipients; given the ratio of vaccine to placebo recipients, the proportion of events in vaccine recipients was lower than in placebo recipients. Non-serious, unsolicited adverse events recorded during the short period of observation were infrequent and without discernible pattern in terms of nature of event or product received (appendix pp 7–9). The data from safety monitoring tests were unremarkable, without discernible patterns or differences between children receiving vaccine and those receiving placebo (appendix pp 7–9).

Three serious adverse events were observed in three children: one child aged 24–59 months who received placebo had fever followed by diarrhoea 13 days after the first dose and was admitted to hospital with enteric fever, one child aged 12–23 months who received a quarter dose of ETVAX was admitted to hospital for rickets on day 46 after first vaccination, and one infant aged 6–11 months who received a half dose of ETVAX was admitted to hospital for pneumonia on study day 15. No serious adverse event was considered related to the study drug.

Analysis of ALS specimens in children aged 24–59 months (excluding the three children who received the full vaccine dose) showed that two immunisations with ETVAX alone (quarter or half dose) or half doses with different doses of dmLT elicited high and significant (ie, at least two-times increase versus baseline) IgA responses against all four vaccine colonisation factors and LTB, as measured 7 days after the first dose or 5 days after the second dose; few responses were observed in placebo recipients (appendix p 10). Addition of dmLT to the vaccine did not have a significant effect on the ALS responses, and no significant differences in responses against any of the five vaccine antigens were noted among the different cohorts (appendix p 10). The magnitudes of responses in this age group (when combining responses in the five cohorts) were all highly significant and similar for CFA/I, CS3, CS5, and LTB (18–50-times increase), with somewhat lower responses against CS6 (eight-times increase; figure 1A). Responder frequencies for vaccine recipients in this age group ranged from 61 (81%) of 75 for CS6 to 75 (100%) for LTB (appendix p 11).

Open in a separate window

Figure 1

Geometric mean times increases in ALS (A) and plasma (B) IgA responses in children aged 12–59 months

Horizontal lines indicate the geometric mean response for the group, whereas the circles represent individual responses. Times increases were calculated as the post-immunisation response (the highest response on either day 7 after the first dose or day 5 after the second dose) divided by pre-immunisation antibody level. Geometric mean times increases in ALS concentrations were significantly different between vaccine and placebo recipients for all antigens in both age groups (all p<0·0001 with Student's t test and all p<0·0005 with Holm's-Bonferroni adjustment). In children aged 24–59 months, geometric mean times increases in plasma IgA concentrations were significantly different between vaccine (n=75) and placebo (n=50) recipients for all antigens (all p<0·0001 with Student's t test and all p=0·0005 with Holm's-Bonferroni adjustment); in children aged 12–23 months, the difference between vaccine (n=60) and placebo (n=40) recipients was significant only for CFA/I, CS3, and LTB (all p≤0·0061 with Student's t test and all p≤0·018 with Holm's-Bonferroni adjustment). ALS=antibodies in lymphocyte secretions.

Analysis of ALS specimens in the different cohorts of children aged 12–23 months showed that two immunisations with ETVAX alone (quarter or half dose) or half doses with 2·5 μg or 5·0 μg dmLT elicited significant increases in IgA antibodies in most vaccinees against all four vaccine colonisation factors and LTB (appendix p 12). These responses were of somewhat lower magnitudes than those observed in the older children. The magnitudes of responses in the different cohorts were similar against CFA/I, CS3, and LTB (six to 20-times increases), but somewhat lower against CS5 and CS6 (three to six times increases; appendix p 12). Addition of dmLT to the vaccine did not have any significant effect on the ALS responses (appendix p 12). Combined responses in all four cohorts showed that most vaccinees (ranging from 29 [51%] of 57 children for CS6 to 51 [95%] for LTB), but only a few placebo recipients (ranging from one [3%] of 40 children for CS5 to six [16%] for CS6), had at least two times increases in concentrations of antibodies to the vaccine antigens, and the magnitudes of responses and responder frequencies were significantly higher in vaccinees than in placebo recipients against each antigen (figure 1A; appendix p 11).

As in previous studies,¹⁰ ALS responses against the vaccine antigens in infants aged 6–11 months were low but significantly higher in vaccinees than placebo recipients for CFA/I (54 [38%] of 144 infants who received vaccine vs eight [16%] of 50 who received placebo; p=0·015), CS3 (58 [40%] vs four [8%]; p<0·0005), and LTB (107 [74%] vs 14 [28%]; p<0·0005; appendix p 13). Because immune responses did not significantly differ by vaccine dose among infants aged 6–11 months, an analysis of faecal immune responses as a more direct approach to assess intestinal immune responses was also done for all infants in this age group (pooled data from all cohorts) who provided faecal extracts that fulfilled inclusion criteria (ie, with concentrations of secretory IgA that were within detection limits and did not differ more than three-fold between different timepoints; n=177).

The magnitudes of faecal secretory IgA responses against all antigens in infants aged 6–11 months were significantly higher in vaccinees than placebo recipients (only against CFA/I, CS3, and LTB after Holm's Bonferroni adjustment), even in infants without a positive ALS response (figure 2A). Lactoferrin concentration was analysed in faecal specimens to exclude the possibility that responses were related to breast milk contamination. Few lactoferrin responses were observed in infants who received vaccine and responded to CFA/I and LTB (appendix p 14); results were similar for those responding to the other colonisation factor antigens and across the other infant cohorts (data not shown). The few infants with faecal IgA antibody responses to colonisation factors or LTB who also showed increases in their faecal lactoferrin concentrations had considerably lower increases in lactoferrin concentrations than in antibody responses (appendix p 14), suggesting that the noted antigen-specific increases resulted from the immunisation and were not related to the ingestion of breast milk antibodies. A secondary endpoint in the protocol was responses by either mucosal measure; this analysis showed that most of the vaccinated infants had significant mucosal immune responses (ALS IgA or faecal secretory IgA, or both) to the colonisation factors (from 70 [49%] of 144 for CS5 and CS6 to 89 [62%] for CS3) and LTB (116 [81%]; appendix p 15).

Open in a separate window

Figure 2

Geometric mean times increases in faecal secretory IgA (A) and plasma IgA (B) responses in infants aged 6–11 months

Horizontal lines indicate the geometric mean response for the group, whereas the circles represent individual responses. Times increases were calculated as the post-immunisation response (the highest response on either day 7 after the first dose or day 5 or 14 after the second dose) divided by pre-immunisation antibody levels. Faecal secretory IgA responses were the ratio between antigen-specific faecal secretory IgA antibody titres divided by total secretory IgA in preimmunisation and corresponding postimmunisation samples. Significant differences between vaccine (n=131) and placebo (n=46) recipients in faecal secretory IgA times increases were observed for CFA/I (p=0·0050), CS3 (p=0·0005), CS5 (p=0·026), CS6 (p=0·028), and LTB (p=0·0020) before Holm's-Bonferroni adjustment and for CFA/I (p=0·015), CS3 (p=0·0025), and LTB (p=0·0080) after adjustment. Significant differences in geometric mean times increases in plasma IgA between vaccine (n=148) and placebo (n=50) recipients were seen for CS3 (p=0·013) and LTB (p<0·0001) before Holm's-Bonferroni adjustment and for LTB (p<0·0001) after adjustment.

Colonisation factor-specific IgA and LTB-specific IgA and IgG responses in plasma were measured before immunisation and 7 days after the first and 5 days after the second dose (post-immunisation results given are the highest response on either of these days). In children aged 24–59 months, these analyses showed that two immunisations with quarter or half doses of ETVAX or half doses with different doses of dmLT elicited significant IgA responses against most vaccine antigens in most children (appendix p 16). IgG responses to LTB were also induced but of lower magnitude than LTB IgA responses. No apparent differences were observed between responses to the different doses of vaccine or to vaccine given with and without dmLT (appendix p 16). Few placebo recipients responded, and the frequencies and magnitudes (lowest for CS5 and CS6) of plasma responses were lower in vaccinees than observed for ALS responses (figure 1B; appendix pp 16, 19).

Colonisation factor-specific and LTB-specific IgA antibody responses in plasma from the children aged 12–23 months showed that two immunisations with quarter or half doses of ETVAX without dmLT or half doses with 2·5 μg or 5·0 μg dmLT elicited significant (at least two-times) increases in IgA antibody responses to most vaccine antigens, except CS5 and CS6, in most children (appendix p 17). Similar to the older children, IgG responses to LTB were induced in most vaccinees but with smaller increases than corresponding IgA responses. Plasma responses among vaccine recipients in this age group were significantly higher for CFA/I, CS3, and LTB, and significantly more frequent for CFA/I and LTB, than in placebo recipients (figure 1B; appendix p 19).

Colonisation factor-specific IgA and LTB-specific IgA and IgG responses in plasma specimens from infants aged 6–11 months showed that eighth, quarter, and half doses of ETVAX induced at least two-times increases in IgA and IgG immune responses against LTB when comparing antibody levels before and after vaccination. The frequency and magnitude of anti-LTB responses to all doses were similar. Addition of dmLT to the quarter dose did not have an apparent effect on immune responses to any antigen (appendix p 18). Many of the placebo recipients also had responses to the vaccine antigens, possibly because of asymptomatic enterotoxigenic E coli infections during the study period. Despite this finding, when combining results for all dose cohorts, antibody responses to all vaccine antigens tended to be higher in the vaccine group than the placebo group but were only significantly increased and more frequent for LTB (p=0·0005; figure 2B; appendix p 19).

Immune responses to the vaccine, with or without dmLT, were age related. The frequencies of ALS IgA responses against each colonisation factor declined with decreasing age of the vaccinees, whereas responses against LTB were almost similar across the age groups (figure 3; appendix pp 11, 13). Conversely, mucosal responses in the placebo recipients declined with increasing age (figure 3; appendix pp 11, 13, 19). Plasma responses followed a similar age-related pattern (figure 3; appendix p 19). The breadth of immune responses (ie, frequency of responders to different numbers of antigens) was greater in ALS (for children aged 12–59 months) and faecal samples (for infants aged 6–11 months) than in plasma samples and increased with age (table 3).

Open in a separate window

Figure 3

Frequencies of mucosal IgA (ALS or faecal, or both) and plasma IgA responses in all age groups

Frequencies of mucosal (A) and plasma (B) IgA antibody responses against the five primary vaccine antigens in different age groups in Bangladesh. Adult data from the same protocol included for comparison.¹⁸ All participants received two doses of ETVAX with or without dmLT or two doses of placebo. Mucosal responses were measured in ALS specimens for adults and children aged 12–59 months and in ALS or faecal specimens for infants aged 6–11 months. ALS=antibodies in lymphocyte secretions. ETVAX=enterotoxigenic Escherichia coli vaccine.

Comparison of ALS immune responses to the first and second vaccine doses showed that most of the children in the different age groups responded to the first dose. The only significantly higher responses to the second dose versus the first dose were against CS6 in children aged 24–59 months (p=0·015) and against CS3 (p=0·0041) and LTB (p<0·0001) in infants (appendix p 20).

Analysis of faecal immune responses in infants aged 6–11 months who were given a quarter dose of vaccine plus 2·5 μg of dmLT showed that dmLT significantly improved the faecal secretory IgA responses to three of the five antigens after the first immunisation (ie, day 7; appendix p 21). Two doses of vaccine without dmLT were needed to induce responses that were similar to those seen among infants receiving one dose of vaccine with dmLT. Furthermore, there was an increased frequency of mucosal antibody responses (ALS or faecal, or both) to three or more vaccine antigens in infants who received the vaccine with dmLT compared with infants receiving the same dose of vaccine alone (appendix p 22). Although the breadth of the response appeared greater in infants who received the adjuvanted vaccine than in those who received the vaccine without adjuvant, these differences were not significant. When comparing the breadth of response in the two vaccine groups with that in the placebo group, significantly more infants in the vaccine plus dmLT group than in the placebo group had antibody responses to three, four, or five antigens. In the vaccine-only group, antibody responses to multiple antigens were significant compared with placebo only for number of children responding to all five antigens (appendix p 22), although this response was of low frequency compared with infants given the vaccine with dmLT (seven [23%] of 30 without dmLT vs 12 [43%] of 28 with dmLT). When a more stringent measure (ie, at least a four-times increase in mucosal antibody concentration) was used, dmLT strengthened the response to all colonisation factor antigens in infants, but only the CFA/I response was significant (appendix p 23).

Discussion

To the best of our knowledge, this is the first clinical trial of ETVAX, as well as of the dmLT adjuvant, in children and infants. The primary objective was to ascertain safety data for ETVAX with or without dmLT and to identify the largest tolerable dose of vaccine and adjuvant in different age groups of Bangladeshi children. We show that two fractionated doses of ETVAX, when administered with or without different doses of dmLT, were safe and generally well tolerated in children. The most prevalent adverse events recorded were mild and, in a few cases, moderate vomiting, which is similar to what was observed in a previous study in young children in Bangladesh receiving similar amounts of inactivated enterotoxigenic or non-enterotoxigenic E coli bacteria.²⁴ At a quarter (2·5 × 10¹⁰ cells) of an adult dose, 13–17% of children in any of the three age cohorts had vomiting, which was mild in all cases. In infants aged 6–11 months, the dose was further reduced to an eighth of an adult dose, which resulted in a reduced frequency of vomiting (7% of infants). The doses of dmLT tested (2·5–10·0 μg) did not affect the safety or tolerability of the vaccine, as shown by multivariate analysis (appendix p 6).

Fractional doses of the vaccine reduced vomiting while remaining immunogenic. For example, in the youngest age group, immune responses at an eighth of an adult dose were similar to those in children given higher doses. These findings, together with the reduced vomiting at lower doses, indicate that infants might be safely immunised with low doses of ETVAX without unacceptable loss of immunogenicity.

High and frequent ALS immune responses to all five primary vaccine antigens were observed using the electrochemiluminiscence method in all vaccine groups except in the youngest age group. For the older age groups, the high magnitudes of ALS responses in some of the placebo recipients against CFA/I, CS3, CS6, and LTB might suggest asymptomatic infection with enterotoxigenic E coli or infection with vibrios expressing cholera toxin during the study period. ALS responses against the colonisation factors among vaccinated infants were infrequent (15–40%), which is in agreement with studies on oral cholera vaccine²¹ and rotavirus vaccine in Indian children.²⁹ The study on rotavirus vaccine²⁹ showed that only 68 (26%) of 258 of children who did not have increased rotavirus-specific IgA antibody-secreting cell levels at onset of vaccination developed significant IgA antibody-secreting cell responses to the rotavirus vaccine despite the same vaccine inducing protection in 45% of infants in similar populations in Bangladesh.³⁰ These results suggest that measuring circulating antibody-secreting cell or ALS responses is not optimal for assessment of mucosal immune responses to oral enteric vaccines in very young children. In this study, we show that secretory IgA anti-colonisation factor antibody responses in faecal extracts were considerably more frequent in infants than were ALS IgA responses to the same antigens. We identified frequent faecal immune responses in infant placebo recipients, which might suggest high incidence of asymptomatic infections in young children in high-endemic areas.²

The first generation of an oral inactivated ETEC vaccine consisted of a mixture of five different formalin-inactivated strains expressing CFA/I and CS1–5, but including considerably lower amounts of CFA/I, CS3, and CS5 than in ETVAX; CS6 was not present and the toxoid component was CTB and not LCTBA. It was tested in Swedish, US, and Egyptian adults, as well as Bangladeshi and Egyptian children and infants.¹⁰ Based on WHO recommendations, the initial formulation was improved upon with this version of ETVAX and the first generation was not assessed beyond the phase 2/3 trials in Bangladesh and Egypt and the phase 3 trials in Mexico and Guatemala (US travellers). An in-depth comparison of plasma and mucosal responses across studies of the first generation and second generation of ETVAX is not meaningful because of the different methods used, the different timepoints at which responses were assessed after the second dose, and the different spectrum and purity of vaccine antigens. However, a superficial comparison of the these vaccines in Bangladeshi infants and children suggests they induced similar frequencies and magnitudes of plasma IgA responses to the toxoid components.^{24, 31}

Our data suggest that any immunological effect of reduced doses of vaccine in infants might be mitigated by coadministration of the dmLT adjuvant. Furthermore, dmLT improved the kinetics of faecal secretory IgA responses—ie, responses to the adjuvanted vaccine were induced after the first dose, whereas a similar response to the non-adjuvanted vaccine was induced after only two doses. When a definition of an at least four-times increase for responses was applied, the apparent effect of dmLT coadministration with the vaccine was greater (appendix p 23).

Immune responses to ETVAX in infants might be further improved by administration of three rather than two doses. Beneficial effect of a third dose has been seen in a previous field trial of rotavirus vaccine in South Africa.³² In our study, we noted that responses to CS3 and LTB were significantly higher after the second dose than after the first dose in infants, suggesting that the effect of a third dose should be assessed in this age group in future studies. Encouragingly, responses to most antigens (other than CS3 and LTB) were remarkably robust, especially with the dmLT adjuvant, even after a single vaccine dose.

An important finding is that Bangladeshi infants receiving ETVAX mounted significant mucosal immune responses to all primary ETVAX antigens. Children in LMICs have been difficult to effectively immunise with other oral vaccines—eg, cholera and rotavirus vaccines.^{21, 29, 33} Based on the positive effect of dmLT in infants, further analysis of the adjuvanted vaccine is justified. Phase 1 and 2b trials based on the doses established in the youngest children in this study, an eighth or quarter dose of vaccine plus 2·5 μg dmLT, has been initiated in children in Zambia and will be followed by studies in The Gambia. Additional studies based on the work described here will not only be important for confirmation of ETVAX's public health benefit but could also enable more effective use of other oral inactivated vaccines in infants.

A strength of this study was the use of the electrochemiluminescence assay to assess mucosal immune responses against vaccine antigens in the small sample volumes that were available from children. The mucosal immune response data were strengthened by use of optimal timepoints^{21, 22} for assessing ALS responses and by analysing faecal secretory IgA immune responses in infants. We have few missing data, but a limitation of this study is our inability to analyse faecal immune responses in older children because faecal samples from children older than 1 year did not fulfil predefined criteria for analysis. Also, we were not able to test dmLT with the eighth dose. Furthermore, because this is the first study done in a cohort of Bangladeshi children, it is unknown whether the results can be generalised to other populations.

In conclusion, this study established the largest tolerated dose of ETVAX in different age groups of Bangladeshi children and the possible benefits of dmLT. This was a single-centre study in Bangladesh, and we have now initiated a phase 1 study of ETVAX in Zambia that will be followed by a phase 2b study in The Gambia to assess safety, immunogenicity, and protective efficacy of ETVAX in young children in different settings.

Data sharing

The data are not publicly available because of ethical restrictions on participant privacy, but data from this study will be made available to others in the scientific community upon request after the ongoing phase 2b study of ETVAX in Finnish travellers (NCT03729219) has been completed and the scientific data from the development of this candidate have been fully published. Standard criteria for making data available for valid research projects will be used, following application by suitably qualified researchers. All publications coming from the clinical and field evaluations of the ETVAX vaccine will appear in journals that have an open access format.

Acknowledgments

Contributors

Declaration of interests

References