Type 1 diabetes (T1D) is characterised by the chronic immune-mediated destruction of pancreatic β-cells, with affected individuals requiring lifelong exogenous insulin ^[1][2][1,2]. Globally, over 1.1 million children and adolescents under the age of 20 are estimated to have T1D, with approximately 128,900 new cases diagnosed each year. In children 0–14 years, India and USA currently have the highest prevalence of T1D (95.6 and 94.2 thousand cases, respectively) [3]. In Australia, a recent study of T1D incidence in children 0–14 years from 2002 to 2017 found a mean incidence of 25.0 per 100,000, additionally revealing a sinusoidal pattern in incidence over time represented by 5-yearly cycles. Mean incidence also increased with age, with the highest incidence in 10–14-year-olds (224% higher than 0–4-year-olds). Wide geographical variation in the mean incidence of T1D has been described, with incidence increases of up to 6.6% per year in Poland, a levelling off reported in populations such as Finland and Sweden, and a slight decreasing trend in Australia over recent years, particularly in 0–4-year-olds. This variation both between and within countries and different ethnic populations is suggested to reflect geographical differences in genetic susceptibility and environmental risk in addition to disparities in diagnostic criteria including islet autoantibody testing requirements ^[2][4][2,4].

T1D is categorised into four stages: (1) presymptomatic T1D with the presence of multiple islet autoantibodies (type 1a) but normoglycemia; (2) presymptomatic T1D with progression to dysglycaemia; (3) dysglycaemia and clinical symptoms such as polyuria, polydipsia, polyphagia, weight loss, fatigue and diabetic ketoacidosis (DKA)); (4) long-standing T1D ^[5][6][5,6]. Acute and long-term complications of T1D include severe hypoglycaemia, DKA, vascular disease, nephropathy, retinopathy and neuropathy; with lifespan also reduced ^[7][8][9][7,8,9].

Most T1D is preceded by the development of islet autoimmunity (IA), serologically confirmed by the presence of at least one diabetes-associated islet autoantibody to insulin (IAA), glutamic acid decarboxylase 65 (GADA), protein tyrosine kinase-related islet antigen 2 (IA-2A) and zinc transporter 8 (ZnT8A). IA can appear from around six months of age with incidence peaking prior to two years of age in the genetically at risk but will be generally present months to years before symptomatic onset, reinforcing the need for early-stage interventions and increased monitoring of presymptomatic T1D. These autoantibodies typically appear sequentially rather than simultaneously, making it unclear whether multiple or single events precipitate seroconversion and eventual T1D development ^[10][11][10,11]. The risk of developing T1D increases as additional autoantibodies are detected ^[12][13][14][12,13,14], with the presence of a single autoantibody (‘early’ IA) conferring a 15% risk of progression to T1D [15], whereas two or more antibodies (‘established’ IA) are associated with an 80% risk of progression to T1D ^{[16][17][18][19][20]}[16,17,18,19,20]. Early seroconversion and increased autoantibody concentrations can be observed in a high proportion of at-risk children, with over 80% of children who developed T1D seroconverting before three years of age [21]. The first-appearing or primary antibody has been proposed to represent two major IA phenotypes representing early or late diagnosis of IA ^[18][22][18,22]. Increasing IAA concentrations have been used to predict progression to overt T1D, with proinsulin highlighted as an important autoantigen in T1D diagnosed in early childhood. Conversely, the appearance of GADA as the primary autoantigen may result in progression at a later age, affecting the design of early interventions ^[21][23][24][21,23,24].

T1D pathogenesis is marked by selective destruction of insulin-producing cells by effector autoreactive and bystander CD8⁺ T cells, directly contrasted by the action of regulatory T cells. Dendritic cells and even mast cells (although limited data are available) have also been implicated in T1D pathogenesis, as they present islet autoantigens to autoreactive T-cells, resulting in dysregulated peripheral immune tolerance [25]. However, the occurrence of the resulting islet infiltration by autoantibodies (insulitis) is heterogenous amongst islets both within lobules of a single pancreas and between individuals, following a relapsing–remitting nature during early disease and perhaps reflecting the highly variable asymptomatic period in preclinical T1D [26]. Approximately 70–95% of β-cells are usually lost at the onset of symptoms, resulting in a reduced pancreas size, although in some individuals a 40% reduction is adequate to elicit symptoms ^[2][16][27][2,16,27]. Efforts to preserve any residual β-cell function (measured by C-peptide production) using immune intervention therapies have had limited success ^{[28][29][30][31][32]}[28,29,30,31,32]. Notably, only 15% of children displaying single IA positivity progress to T1D [33], and conversely, only 10% of individuals with T1D display single IA positivity [16]. Therefore, there is an increasing focus on the prevention of T1D progression from the early stages of non-clinical disease. Development of more economical and efficient assays of islet autoantibody detection may allow for more widespread employment of IA screening and potential for use in the general population, enabling earlier diagnosis and intervention ^[17][34][17,34].

Comprehensive genome-wide association studies have identified over 60 genetic loci associated with increased T1D risk, with approximately half of the genetic risk attributed to the human leukocyte antigen (HLA) genotype, with notable contributions also arising from the INS, PTPN22, CTLA4 and IL2RA genes [35]. HLA-class II DR and DQ allele haplotypes DRB1*03:01-DQA1*05: 01-DQB1*02:01 and DRB1*04-DQA1*03:01-DQB1*03:02 show the highest risk, with DR3/DR4 (DQ2/DQ8) heterozygotes displaying a 30-fold increased risk of IA and T1D in the general population. Whilst a combination of islet autoantibodies has been previously used to predict increased risk of progression to T1D in first-degree relatives [36], we can now use genetic risk scores (GRS) to predict progression to T1D in IA positive children [37]. GRS are calculated using a combination of HLA and non-HLA genes, with weighted values given to both high-risk HLA class II genotypes plus a weighted value assigned to each susceptible allele of HLA class I and non-HLA single nucleotide polymorphisms (SNPs). Individuals with lower GRS experience slower progression to IA, and slower development from both single and multiple IA to T1D in The Environmental Determinants of Diabetes in the Young (TEDDY) cohort [38]. Recent improvements in T1D GRS algorithms have led to the development of T1D GRS2 for standardised use with greater predictive power [39].

Although individuals with a first-degree relative with T1D are at approximately 15-fold increased relative lifetime risk for T1D compared to the general population, over 85% of diagnosed children have no family history, highlighting the major contribution of environmental factors in the aetiology of T1D ^[5][40][5,40]. T1D heritability varies depending on which family member has T1D, with the risk of T1D in the offspring higher with a T1D-affected father (~6%) compared to a T1D-affected mother (~2%). Furthermore, having a dizygotic twin imposes a slightly higher risk of T1D (~10%) compared to a non-twin sibling (~6%), highlighting the role of the intrauterine environment on T1D-risk. [41]. Interestingly, the proportion of individuals with the highest risk genotype DR3-DQ2/DR4-DQ8 has been shown to decrease over time in multiple populations in the United Kingdom, Finland and United States [2].

An interplay between genetics and environmental factors such as the virome, microbiome and metabolome is suggested to regulate immune tolerance, with the introduction of environmental, lifestyle or dietary exposures currently being investigated as either accelerating or protective [42]. A range of potential environmental triggers has been proposed, including viruses. The hypothesised role of viral infections in the initiation of IA and the progression to T1D is supported by a large body of epidemiological and animal model-based evidence ^[43][44][43,44]. Multiple viruses have been associated with IA/T1D to date, including enterovirus (EV) ^{[45][46][47][48][49][50][51][52][53][54]}[45,46,47,48,49,50,51,52,53,54], rotavirus ^{[55][56][57][58][59][60][61]}[55,56,57,58,59,60,61], cytomegalovirus ^{[62][63][64][65][66][67][68]}[62,63,64,65,66,67,68], Epstein-Barr virus ^[63][69][70][63,69,70], parechovirus ^[71][72][73][71,72,73], influenza ^[74][75][76][74,75,76], parvovirus ^[77][78][77,78], mumps ^[79][80][81][79,80,81], rubella ^{[80][81][82][83][84][85]}[80,81,82,83,84,85] and human endogenous retrovirus ^{[86][87][88][89]}[86,87,88,89]. By far, the strongest supporting evidence exists for EVs. Our previous meta-analysis of 26 molecular studies and >4400 participants revealed EV infection was 10 times greater at the onset of T1D compared to healthy controls [90]. Furthermore, T1D-specific risk alleles contained within genes involved in immune function have been shown to alter susceptibility to viral infection or affect the extent of the host antiviral response [91]. The rs1990760 SNP within IFIH1 has been associated with increased detection of EV RNA in blood [92] and separately with severe EV-A71 infection [93]. The rs2476601 SNP within PTPN22 has been associated with lower IFN production by macrophages in response to TLR ligand stimulation (as would occur during viral infection) [94], and additionally it has been suggested that PTPN22 could suppress the function of effector T cells, diminishing their response to viral infection and allowing the establishment of persistent infection ^[91][95][91,95].

The timing of environmental triggers is likely to be critical, with environmental influences potentially commencing in utero and within the first year of life, emphasising the importance of longitudinal prospective cohort studies that follow at-risk children from pregnancy, such as the Environmental Determinants of Islet Autoimmunity (ENDIA) and Type 1 Diabetes Prediction and Prevention (DIPP)-novum studies ^[96][97][96,97]. Our recent meta-analysis of observational studies revealed maternal viral infections during pregnancy resulted in offspring that were twice more likely to develop T1D (OR 2.16, 95% CI 1.22–3.80; p = 0.008), highlighting the need to measure infections in utero as well as during early life. The adoption of large, national or international prospective birth cohort studies allows for the examination of any temporal links between infection in utero and the eventual development of IA or T1D in the offspring [46].

EVs are non-enveloped, single-stranded icosahedral RNA viruses classified within the Picornaviridae family that primarily display faecal–oral transmission, within occasional cases of vertical and respiratory transmission also possible [98]. Human EVs are ubiquitous and responsible for serious diseases such as poliomyelitis, myocarditis and aseptic meningitis [99]. However, many EV infections cause subclinical or mild disease and are thus underreported, with a small proportion proceeding to clinical identification [100]. More severe EV infection is typically seen in children and neonates, with proposed intrinsic immunity in the adult mature gut moderating the course of infection and preventing viraemia [101].

There are more than 100 characterised genotypes of human EV, classified into four species: EV-A to -D. Also included within the EV genus are rhinoviruses, which predominately cause upper respiratory tract infections and distinct clinical presentation ^{[102][103][104]}[102,103,104]. The linear EV RNA genome spans 7.2–8.5 kb in length, consisting of four structural (P1) capsid proteins and seven non-structural (P2 and P3) proteins, forming a single polypeptide which is cleaved by a viral 3C and 2A proteases ^[105][106][105,106]. The EV 5′-untranslated region (UTR) contains internal ribosome entry sites which allow for ribosome recruitment during cap-independent translation of EVs’ single polypeptide. Self-assembly of VP1-VP4 capsid proteins into empty capsid particles and transcription of the positive-strand RNA genome facilitated by non-structural proteins is followed by RNA encapsidation and formation of infectious virions. The mechanism of release is still unconfirmed but is proposed to involve changes to cell membrane integrity, lysis and apoptosis [107]. The 5′ and 3′ UTRs are highly conserved amongst all EV species and have historically formed the basis of primer and probe designs used in molecular diagnostics of general EV infection [108]. The highly variable major capsid protein VP1 codes for genotype-specific determinants of neutralisation and antigenic sites and is therefore typically used for EV genotypic classification ^[109][110][109,110].

EVs invade host cells primarily via the coxsackievirus and adenovirus receptor (CAR), expressed in both α- and β-cells, with entry of genomic RNA post adsorption followed by translation and replication of sense viral RNA in the cytosol in a cap-independent manner [111]. A specific isoform of CAR with a terminal SIV motif and a unique PDZ-binding domain at the C-terminal (CAR-SIV) has been shown to be highly and selectively expressed within β-cells and is localised mainly to insulin secretory granules, which may further contribute to the sensitivity of human β-cells to EV infection [112]. Secretory granule proteins are proposed to be hijacked during exocytosis, allowing internalisation of virus particles by existing endocytic machinery. This is further supported by the identification of viral replication complexes around insulin granule membranes in coxsackievirus B (CVB)-infected human islets using electron microscopy ^[113][114][113,114].