Safety, tolerability and immunogenicity of PRV-101, a multivalent vaccine targeting coxsackie B viruses (CVBs) associated with type 1 diabetes: a double-blind randomised placebo-controlled Phase I trial

Coxsackie B viruses (CVBs) are common human viruses causing a significant disease burden ranging from mild respiratory symptoms to more severe diseases including myocarditis, pericarditis, meningitis, encephalitis, herpangina, pancreatitis and severe multiorgan infections in infants [1, 2]. Among the more than 110 human enterovirus types, the six CVB types are among the 15 most frequent serotypes reported in healthcare patients in the USA [3, 4].

There is a substantial body of evidence associating CVB infections with type 1 diabetes [2, 5‐7]. In addition, prospective studies have shown an association between enterovirus infections, most specifically CVB infections, and the initiation of the beta cell-damaging process in genetically susceptible individuals [5]. Immunohistochemical analysis has revealed the presence of enterovirus capsid protein in pancreatic beta cells in the majority of individuals with type 1 diabetes [8]. This has been confirmed by the detection of enterovirus RNA in the pancreases of such individuals [9‐13]. Additional evidence connecting CVBs to type 1 diabetes comes from the observation that beta cells strongly express the cell surface coxsackie and adenovirus receptor (CAR) and that certain CAR gene polymorphisms are associated with type 1 diabetes [5]. CVBs are the only enteroviruses known to use this receptor for cell entry, providing one possible mechanism mediating the tropism of CVBs to human pancreatic beta cells [14].

The exact mechanisms by which CVBs could cause beta cell damage are not known. The detection of viral proteins in beta cells suggests that such cells may be directly infected, with cell damage and inflammation and subsequent initiation of islet autoimmunity [8]. On the other hand, molecular mimicry between viral and beta cell proteins may also play a role [15, 16]. CVBs can also cause persistent infections leading to chronic diseases. For example, acute CVB myocarditis can progress to chronic cardiomyopathy, in which persisting low-grade CVB infection of cardiac myocytes is considered to be the underlying mechanism [17]. CVBs cause persistent infection in the pancreas in mouse models [18], and signs of viral persistence have been observed in type 1 diabetes [5, 11, 19]. CVBs have also been linked to certain other autoimmune disorders such as coeliac disease [13, 20‐22].

Enterovirus infections, particularly CVB infections, are currently considered to be prime candidates for environmental triggers of type 1 diabetes, and the need for intervention studies with a CVB vaccine is widely acknowledged [23]. Preclinical studies have demonstrated that both CVB vaccines [24, 25] and anti-CVB antibodies [26] are capable of providing protection against experimental CVB infections and the subsequent development of diabetes. There are precedents for effective formalin-inactivated vaccines against another subgroup of enteroviruses, polioviruses, providing a path for the development of novel CVB vaccines. This, together with the significant disease burden of acute CVB infections and their association with chronic diseases other than diabetes, has prompted the development of human CVB vaccines.

Here, we present the results of the first-in-human trial testing the safety and immunogenicity of a multivalent formalin-inactivated CVB vaccine candidate, PRV-101. PRV-101 is being developed for the prevention of acute CVB infections and the CVB-triggered autoimmune damage to pancreatic beta cells that often progresses to type 1 diabetes [5] and damage to intestinal cells potentially leading to coeliac disease [6, 20].

All methods were carried out in accordance with CONSORT guidelines [27].

A total of 44 prospective study participants were assessed for eligibility, 32 of whom met all eligibility criteria and were enrolled in the study. These 32 participants were randomised between 14 December 2020 and 10 April 2021 as follows: eight participants received placebo, 12 received the lower dose of PRV-101 in Cohort 1 and 12 received the higher dose of PRV-101 in Cohort 2 (Fig. 1). All study participants completed the study and attended the EOS visit. One participant in Cohort 2 received only two injections of PRV-101 due to family relocation, while the other participants received all three injections according to the study protocol. All 32 participants were assessed for the primary outcome and all scheduled blood and nasal swab samples were obtained from them. Their baseline characteristics are summarised in Table 1. The markers of glucose metabolism (fasting C-peptide and HbA_1c) were similar in all three groups. Eleven of the study participants carried HLA genotypes conferring high risk for type 1 diabetes and four carried HLA genotypes conferring moderately increased risk, as defined previously [29]. Fifteen participants carried coeliac-disease-associated HLA genotypes [30].

There were no SAEs or adverse events leading to study treatment discontinuation. All participants in the placebo group and all but one of the participants in the PRV-101 groups had at least one TEAE. TEAEs occurring in three or more participants in any group are listed in Table 2 by Medical Dictionary for Regulatory Activities (MedDRA) preferred term (https://​www.​meddra.​org/​ accessed 20 December 2021). TEAEs comprised all adverse events including those characterised as vaccine reactions and injection site reactions of sufficient severity in the opinion of the investigator. Headache and injection site pain were the most common (Table 2). Based on investigator-reported causality evaluation, those TEAEs that were considered to be related to the study drug (PRV-101 or placebo) occurred in 37.5% of the participants in the placebo group and 62.5% in the PRV-101 group. Among the PRV-101-vaccinated study participants, the most common of these causally evaluated TEAEs were injection site pain, headache, injection site discomfort and injection site pruritus, observed in 33.3%, 20.8%, 16.7% and 12.5% of participants, respectively. One participant in the placebo group reported a severe TEAE of neck pain whereas no severe TEAEs were reported in the PRV-101 groups.

Local injection site reactions were reported separately, in three (37.5%) participants receiving placebo and in ten (41.7%) participants receiving PRV-101. These included injection site pain, discomfort, pruritus, paraesthesia and/or erythema. No apparent trends were observed in the safety laboratory values. Markers of glucose homeostasis remained within the normal range in all study participants. One participant in the high-dose PRV-101 group showed IAA seroconversion from negative at baseline to low positive at week 12, and remained IAA-positive at week 32. However, this participant was IAA negative in all samples when tested at a speciality reference laboratory using a similar radiobinding assay (RBA) as well as with an electrochemiluminescence (ECL)-based assay for IAA. None of the trial participants experienced seroconversion to positivity for coeliac disease-associated autoantibodies.

PRV-101 is primarily intended to induce protective immunity in CVB seronegative individuals. Therefore, data from the individuals negative for CVB antibodies at baseline are the most relevant for this goal. Among the participants seronegative at baseline, a clear dose–response relationship was observed in neutralising CVB antibody titres, with higher antibody titres being observed in the high-dose group than in the low-dose group (Fig. 2). Both high- and low-dose PRV-101 groups showed increasing neutralising antibody titres during the vaccination period for all five CVB serotypes. The magnitudes of the responses were different for the five different CVB serotypes but all (100%) of the initially seronegative participants turned seropositive at least at one time point during the study in both PRV-101 groups (ESM Fig. 1). In the low-dose PRV-101 group, 100%, 75%, 100%, 100% and 100% of the initially seronegative participants developed the presumably protective titre ≥8 for CVB1, CVB2, CVB3, CVB4 and CVB5, respectively. In the high-dose group, all participants developed titres ≥8 for CVB1–5. The peak antibody titres showed dose proportionality for each of the five CVB serotypes (ESM Table 1). Seven of the 32 participants were initially seronegative for all five CVB serotypes at baseline. Three of four such participants in the low-dose PRV-101 group (75%) and all such participants in the high-dose group (n=3) developed protective neutralising titres (titre ≥8) against all five CVB serotypes, a response that was still seen at the EOS/week 32 visit. In the placebo group, the participants initially classified as seronegative remained seronegative throughout the study, except for one participant who developed low levels of antibodies (titre of 8) against CVB1 (ESM Fig. 1). This participant had low CVB1 antibody titres already in the screening visit sample, suggesting that this apparent seroconversion might have reflected assay variation or natural fluctuation of low antibody titres during the trial from a titre of 4 at the screening visit to titres 2, 4, 6, 8 and 8 at the later visits. None of the study participants was positive for enterovirus RNA in nasal swab samples.

When data from all participants (those initially seropositive and seronegative) were analysed as one group, all 12 participants in the high-dose group and eight (67%) of the 12 participants in the low-dose group met the definition of a responder for each of the five serotypes (seroconversion in initially seronegative participants or a fourfold or greater increase in neutralising antibody titres in initially seropositive participants at any time during the study) (ESM Table 2). The median peak titres against the five CVB serotypes ranged from 30 to 768 in the low-dose group and from 128 to 2048 in the high-dose group. Peak titres were usually reached after the third dose, at or before the week 12 visit, similarly in both PRV-101 groups. Among those participants who had been seropositive at baseline, the majority reached peak titres already after the first dose of PRV-101, suggesting that the vaccine served as a booster of the pre-existing immunity.

Antibodies against CVBs belonging to classes IgG, IgM and IgA were analysed using an ELISA method that was not serotype specific but assessed the overall immunogenicity of PRV-101. IgG levels increased markedly within 4 weeks after the first PRV-101 dose and remained elevated at the later sampling time points, showing slight decreases towards the end of the study (ESM Figs 2, 3). The magnitude of this response was higher in the high-dose group than in the low-dose group (mean 11.2-fold vs 5.6-fold increase, respectively, when compared with baseline). At the EOS/week 32 visit, antibody levels still remained 5.4-fold and 3.8-fold higher compared with baseline, respectively. IgM responses varied considerably between individuals in both dosing groups. The mean IgM levels peaked within 4 weeks, being 11.4-fold higher when compared with baseline in the high-dose group and 11.0-fold higher in the low-dose group, with levels gradually declining at subsequent time points (ESM Figs 2, 3). PRV-101 also induced IgA class CVB antibodies, peaking at week 4, with a 3.7-fold mean increase in the high-dose group and a 4.1-fold increase in the low-dose group compared with baseline levels, varying considerably between individuals. IgG, IgM and IgA levels remained unchanged in the placebo group. ELISA antibody responses were similar in participants who were negative for neutralising antibodies against all tested CVB serotypes at baseline (n=7) and those who had neutralising antibodies against at least one CVB type (n=17) (ESM Fig. 4).

PRV-101 met the primary safety endpoint of the PROVENT trial, providing initial evidence of acceptable safety and tolerability of the product in healthy adult volunteers. No treatment-emergent SAEs, or adverse events leading to study drug discontinuation or study withdrawal were observed. All TEAEs in the PRV-101 groups were mild to moderate in severity. In addition, the vaccine-induced robust antibody responses to all five CVB serotypes included in the product. Altogether, these findings create a solid basis for the future development of the PRV-101 vaccine candidate.

The observation of robust neutralising antibody responses against the five CVB serotypes is highly important for the future development of PRV-101. Neutralising antibodies are specific for individual CVB serotypes and this finding therefore indicates that each inactivated virus type was immunogenic in the vaccine. Of note, previous studies with the poliovirus vaccine have shown that neutralising antibodies mediate protection against the virus [31, 32].

PRV-101 induced a clear increase in neutralising CVB antibody titres in both initially CVB seropositive and seronegative participants. In the initially seronegative participants, a clear dose–response relationship was observed, as the higher PRV-101 dose induced higher antibody levels than the lower dose. The results also showed the durability of the antibody response as all participants in the high-dose group maintained antibody titres of 8 or higher until the end of the study, with the exception of one participant whose CVB2 antibodies decreased to a titre of <4. The cut-off titre 8 corresponds to antibody levels that have been considered protective against another enterovirus, the poliovirus, measured as protection against paralysis as a consequence of blocking the spread of the virus to the central nervous system [33]. The high seroconversion rate among initially seronegative participants (100% across all CVB serotypes in both dosing groups) is another indicator of relevant immunogenicity. The finding that peak neutralising antibody titres occurred after the third vaccination suggests a booster effect of repeated vaccinations.

At the end of the study, 6 months after the third dose of the vaccine, >90% of participants had titres of 8 or higher against all five serotypes. The neutralising antibody levels were comparable with the neutralising poliovirus antibody levels previously seen in children who had received three or four doses of an inactivated poliovirus vaccine a few months before sampling, when using the same plaque reduction assay as in the current study [34]. The magnitude of the PRV-101-induced CVB antibody responses also compares well with those induced by prototype CVB vaccines in our preclinical studies in mice and rhesus macaques [24, 25]. These preclinical studies were carried out using a formalin-inactivated multivalent CVB vaccine similar to PRV-101, and antibodies were analysed using the same plaque reduction assay. The vaccinated animals were efficiently protected against experimental CVB infection.

In the current study, PRV-101 also induced high levels of IgG class CVB antibodies as measured with ELISA. Dose-dependent IgG responses developed rapidly, were already apparent 1 month after the first vaccine injection and were long-lasting, as the IgG levels remained elevated at the end of the study. PRV-101 also induced IgM and, somewhat more variably, IgA class antibody responses but these were less robust than the IgG responses. The IgA responses suggest that PRV-101 may also have the potential to induce mucosal immunity, even if such responses may be weak and vary from one individual to another.

The neutralising antibody response to CVB2 was somewhat weaker than the responses to the other CVB serotypes. This may be due to differences in the proportions of the individual CVB serotype components in the PRV-101 product, as both the protein concentration and the viral particle number of the CVB2 component were lower than those of the other components. In our preclinical animal studies with similarly inactivated CVB vaccines, CVB2 was equally immunogenic as the other CVBs, suggesting that the immunogenicity of CVB2 per se does not markedly differ from the other CVBs. Together with the clear dose–response pattern seen in the high-dose and low-dose PRV-101 groups, these findings suggest that the immunogenicity of each CVB component of PRV-101 may be optimised by adjusting their relative proportions in the vaccine.

PRV-101 was well tolerated in this trial and only relatively mild events, including headache, injection site pain or discomfort and itching were associated with dosing of PRV-101, being in line with the experience from the inactivated poliovirus vaccine. From this point of view, the results from this first-in-human trial support the feasibility of continuing this development programme. Biological proof-of-concept of preventing CVB-induced diabetes by a CVB vaccine has recently been obtained from preclinical mouse studies where the prototype CVB vaccine efficiently protected against experimental CVB infections and against beta cell damage and diabetes that might otherwise occur after a CVB infection [24, 25].

The mechanisms by which CVB infections can cause beta cell damage and clinical diabetes are not fully understood. The prevailing hypothesis is that CVBs infect insulin-producing beta cells, leading to cell damage, local inflammation in pancreatic islets, loss of immune tolerance in a genetically predisposed host and initiation of an autoimmune process. Since PRV-101 is an inactivated vaccine, it cannot cause type 1 diabetes by such a mechanism. However, molecular mimicry between CVBs and host proteins might induce cross-reactive immune responses that could lead to cell damage. In fact, one mimicry epitope has been discovered in the non-structural viral protein 2C and the GAD65 autoantigen expressed in beta cells, and this epitope has been shown to be recognised by the immune system [15, 16]. Importantly, PRV-101 does not contain this epitope since it only consists of structural virus proteins. Nevertheless, some other cross-reactive epitopes may potentially still exist in CVBs, so the trial participants were carefully monitored for signs of type 1 diabetes. None of them developed type 1 diabetes, signs of subclinical beta cell dysfunction or clinically relevant levels of diabetes-associated autoantibodies. One trial participant turned weakly positive for IAA during the study but all follow-up samples from this participant were found to be IAA negative in another laboratory using an RBA similar to that employed in the trial. High-affinity IAAs are predictive of progression to type 1 diabetes. An additional, ECL-based assay for detecting high-affinity IAAs [35, 36] was also negative for this trial participant. The IAA-positive participant also remained negative for all other islet autoantibodies. Altogether, we conclude that this participant had no signs of clinically relevant islet autoimmunity. The low-titre and low-affinity IAAs detected by one laboratory but not confirmed by another laboratory may be considered to lie within the background variation and to have no predictive value for the development of type 1 diabetes. These findings are in line with the safety results from toxicological studies with PRV-101, including a 4 week repeat-dose study in mice (data summary available on request via the corresponding author). In addition, other preclinical studies with similar CVB vaccines have indicated no induction of diabetes or IAA in different mouse models or in rhesus macaques. It is also important to note that PRV-101 did not induce autoantibodies against tissue transglutaminase, which would be predictive for coeliac disease, another autoimmune disease that has been linked to CVB infections. Overall, there was no evidence to suggest that PRV-101 would promote autoimmunity, even in genetically predisposed individuals (almost half of the study participants carried HLA markers predisposing to type 1 diabetes and/or coeliac disease).

This study has some limitations. Evaluation of long-term safety and immunogenicity of PRV-101 is needed, as well as studies of T cell-mediated immune responses to CVB that may help to maintain long-term immunity. Inactivated poliovirus vaccination schedules include later booster vaccinations after the primary vaccination regimen and it is possible that this also holds for PRV-101. Studies evaluating PRV-101-induced T cell responses as well as later-stage CVB antibody levels are currently in progress. Another limitation is that the current study only included adults. Therefore, we cannot exclude the possibility that some of the CVB seronegative individuals had been exposed to CVB in the past and that their antibodies had decreased to undetectable levels by the initiation of the current study. In such cases, the pre-existing memory T cells could have facilitated the immune responses induced by PRV-101. Therefore, subsequent studies in young, exposure-naive children are needed. A critical next step would be to initiate a Phase Ib study of PRV-101 in children, particularly infants, who are the future target population of the vaccine.

In conclusion, this first-in-human, proof-of-mechanism study has demonstrated, for the first time, that a multivalent formalin-inactivated CVB vaccine was both well tolerated and immunogenic in humans. The vaccine induced robust and dose-dependent immune responses, in both male and female participants, towards all five CVB serotypes included in the vaccine. Future studies are planned to progressively increase age and ethnic and regional diversity of the programme. The results of this randomised, placebo-controlled trial support further development of this first-in-class vaccine to prevent CVB infections and several CVB-associated diseases, and potentially also ultimately decrease the global incidence and disease burden of type 1 diabetes and coeliac disease.