The Current Landscape for Screening and Monitoring of Early‐Stage Type 1 Diabetes

ABSTRACT

Type 1 diabetes (T1D) has two pre‐symptomatic phases (stages 1 and 2) with progressive destruction of beta cells which have been identified through longitudinal cohort studies in recent decades. The definition of T1D, with hyperglycaemia that may or may not be symptomatic, is now defined as stage 3. There is growing evidence that screening for stages 1 and 2 reduces rates of diabetic ketoacidosis and prevents long‐term complications. These stages can be defined by the presence of islet autoantibodies which are markers of autoimmune beta cell damage. Furthermore, genetic risk scores, which combine a variety of single nucleotide polymorphisms, identify people at high genetic risk of future T1D. Thus, they provide an opportunity to select high‐risk individuals for islet autoantibody testing. Individuals identified as having stage 1 or 2 T1D require ongoing monitoring to detect hyperglycaemia and the need for insulin replacement. These individuals may also be eligible for emerging immunotherapies in future to delay progression to stage 3. This review article explores the current evidence for screening and summarises the recommended clinical care for early‐stage T1D.

Summary.

• Stage 1 and 2 T1D are pre‐symptomatic phases that can be detected by islet autoantibody (IA) testing.

• Genetic risk scores can help identify individuals at high risk for T1D.

• Individuals with stages 1 and 2 T1D require regular monitoring for the early detection of hyperglycaemia and the prevention of complications including diabetic ketoacidosis.

1. Introduction

Type 1 diabetes (T1D) is a lifelong autoimmune condition that causes destruction of pancreatic beta cells and need for insulin replacement therapy. The pathogenesis of T1D is a complex interaction between genetic susceptibility and environmental factors leading to immune attack of the pancreatic β cells [1].

T1D affects ~134 000 Australians and approximately 85% of children with T1D do not have a family history [2]. The highest incidence occurs in children 10–14 years of age [3] and at least 30% present with diabetic ketoacidosis (DKA) at diagnosis [4]. In addition to the acute metabolic compromise caused by DKA, there are potential long‐term detrimental effects on cognitive function and achievement of glycaemic targets [5, 6]. Elevated HbA1c levels over time are associated with nephropathy, neuropathy, retinopathy (vision loss is most commonly due to diabetes in working age adults) and cardiovascular disease [7]. Rates of DKA in screened populations are reported to be around 5% [8] compared to background rates of between 15% and 80% depending on the population studied [9].

T1D is defined as random plasma glucose levels of ≥ 11.1 mmol/L or a fasting plasma glucose of ≥ 7.0 mmol/L [10]. This diagnosis point is now considered stage 3 T1D. Pre‐symptomatic phases (stages 1 and 2) indicating progressive destruction of beta cells, have now been established [11] and offer an opportunity for early diagnosis. These stages may be present for months or years prior to symptom onset and are denoted by the presence of islet autoantibodies (IA). These IAs (IAA, GAD, IA‐2, ZnT3) are considered markers rather than mediators of islet autoimmunity [1]. This staging system (Figure 1) classifies the presence of

• Two or more autoantibodies with normoglycaemia as stage 1;

• Two or more autoantibodies with dysglycaemia as stage 2, and

• Two or more autoantibodies with hyperglycaemia, with or without symptoms, as stage 3 [11]. Upcoming guidelines from the International Society for Paediatric and Adolescent Diabetes (ISPAD) will further classify stage 3 into asymptomatic (3a) and symptomatic (3b).

FIGURE 1.

Stages of T1D. Reproduced with permission from the ‘ISPAD clinical practice consensus guidelines 2022: Stages of type 1 diabetes in children and adolescents’. DOI: 10.1111/pedi.13410.

Individuals in stage 1 have a 44% 5‐year risk of stage 3 T1D compared to a 75% 5‐year risk in stage 2 [2].

Recently published international consensus guidelines emphasise the importance of early screening in order to prevent DKA and subsequent complications and potentially delay progression to stage 3 T1D through emerging immunotherapies [12]. This review article will summarise the rationale for screening and highlight the recommendations from the guidelines.

2. Disease Progression

2.1. Genetic Risk

Genetic predisposition to T1D is largely conferred by polymorphisms within the HLA Class II region on chromosome 6, and to a lesser extent by over 50 polymorphisms in non‐HLA loci [13]. Within the HLA region, particular haplotypes in the HLA‐DR and HLA‐DQ loci confer approximately 40% of the risk [14], especially the combination of DR3‐DQ2 and DR4‐DQ8 [1]. High‐risk HLA haplotypes have been used in population screening studies, such as the Finnish Diabetes Prediction and Prevention (DIPP) and The Environmental Determinants of Diabetes in the Young (TEDDY) studies, to identify and follow high‐risk infants for the development of T1D IAs [15, 16]. In these cohorts, 27.5% had developed IA by the end of follow‐up period of 15 years, and 3.5% had stage 3 T1D [17].

As other non‐HLA loci can also predispose to T1D, multiple genetic risk scores (GRS) have been developed, which combine multiple risk loci, including HLA, to provide a composite score reflecting an individual's genetic susceptibility to T1D [18]. The GRS2 score incorporates 67 single nucleotide polymorphisms (SNPs); 14 of which fall in the DR‐DQ region, 21 in other HLA regions and 32 in non‐HLA loci [13]. Using HLA and non‐HLA loci provides higher sensitivity and specificity than using HLA or non‐HLA loci separately. Whilst the GRS2 is most useful in those of European ancestry, who are also the highest‐risk ancestry population, GRS2 scores still perform well in other ancestries [19]. In addition, ancestry‐specific GRS have been developed, including for people of South Asian and African American ancestries [19].

The GRS2 has a diagnostic accuracy (receiver operating characteristic area under the curve, ROC AUC) of 0.92 to differentiate children and young people at risk of T1D [13]. Those with a GRS2 > 90^th percentile have a 2.4% risk of T1D and represent 77% of future T1D cases [13]. Those with a GRS2 score > 99^th percentile have a > 20% chance of developing T1D in future but this group reflects approximately 7% of future cases (Figure 2) [13]. Hence, GRS are powerful tools to select at‐risk individuals for targeted IA testing and the positivity threshold can be set to reflect the purpose (e.g. population screening to capture most cases vs. identifying highest‐risk individuals for clinical trials) [13].

FIGURE 2.

Background risk of T1D is 0.1% in those with a GRS < 90^th centile. 77% of all cases of T1D in those with a GRS ≥ 90^th centile, 7% of all cases in those with a GRS ≥ 99^th centile [13].

2.2. Environmental Factors

Genetic susceptibility confers approximately 50% of the risk of T1D [20]. The impact of environmental factors on progression is significant although not fully understood. The incidence of T1D globally has increased by 2%–3.5% per year since the 1990s [21], although more recent data suggest a plateau in T1D incidence [3, 21]. Nevertheless, the rising incidence cannot be explained by genetic risk alone. Migration from low‐incidence to high‐incidence countries leads to emigrants having a higher incidence of T1D, particularly if children relocated at a younger age [22].

Viral infections in childhood, particularly enteroviral infections, are considered a potential trigger. A recent meta‐analysis showed that the odds ratio for the detection of enteroviruses in people with IA was 2.07 compared with controls and 16.22 for those who developed T1D [23]. Other environmental factors such as cow's milk intake, Vitamin D levels and gluten have not shown conclusive associations [24]. A meta‐analysis suggested breastfeeding may have a weak protective effect on T1D risk although the studies were heterogeneous and had a high risk of bias [25]. Increased Body Mass Index (BMI), both in children at risk and parents, potentially impacts progression due to the increased burden on beta cells [24], and chronic inflammation [26]. The Environmental Determinants of Islet Autoimmunity (ENDIA) study in Australia is investigating key environmental factors such as the microbiome, virome and pregnancy‐related factors and their associations with T1D development [27].

2.3. Islet Autoantibodies

Four main autoantibodies are hallmarks of islet autoimmunity, namely, insulin antibodies (IAA), glutamic acid decarboxylase 65 antibodies (GADA), tyrosine phosphatase‐related antigen 2 (IA‐2A) and zinc transporter 8 antibodies (ZnT8A). These IA are thought to be markers rather than mediators of islet cell inflammation [1]. Although the autoimmune process in T1D is mediated by T cells [1], there are no assays to detect T cell autoreactivity to islet antigens [11]. However, there are sensitive and specific assays for the detection of IAA, IA‐2A, GADA and ZnT8A. IAA are measured via radioimmunassay (RIA) whilst IA2‐A, GADA and ZnT8A can also be measured by enzyme‐linked immunoassays (ELISA) on venous samples. A newer assay, antibody detection by agglutination‐PCR (ADAP) requires much less blood and allows for multiple antibodies to be detected via capillary blood spot samples [28, 29].

Natural history studies have demonstrated that the first appearance of IA in children occurs around or before 2 years of age [30, 31, 32] although rarely before 6 months of age [31]. Seroconversion to IAA or GADA tends to precede seroconversion to IA‐2A and ZnT8 [33]. Progression from single to multiple antibodies is more common in children under the age of 5 years, and generally occurs within 2 years of the initial IA [11].

The risk of progression to stage 3 has been examined by multiple cohort studies around the world, following infants and children genetically predisposed to T1D. A combined analysis of children in Germany, Finland and Colorado, USA, demonstrated that the 10‐year risk of progression to stage 3 was 44% with multiple IA (stages 1 or 2), compared with 15% in those with a single antibody [30]. This risk increased to almost 100% in those with multiple IA by 20 years of follow‐up. The type of antibody also appears to have an impact on progression with the presence of IA‐2A showing greater risk of progression than GADA or IAA [8, 30]. Seroconversion under the age of 6 years leads to a more rapid diagnosis of stage 3 than those in those who seroconvert after 6 years [34]. The appearance of multiple antibodies at the first test also increases the risk of progression [35].

The severity of disease and the rapidity of progression seen in young children under 10 years compared to adolescents may be related to heterogeneous pathogenesis now defined as endotypes 1 and 2, respectively [36]. Endotype 1 (T1DE1) is characterised by greater destruction of β cells and evidence of inflammation in remaining islet cells (insulitis). In contrast, endotype 2 (T1DE2) may show more insulin‐containing islets and a less aggressive inflammatory profile [36]. IAA and IA‐2A antibodies are more likely to manifest first in T1DE1 whereas GADA are often the first appearing IA in T1DED2 [36]. Thus, the heterogeneity and multifactorial nature of T1D complicate the prediction of disease progression.

3. Screening Programs

Interest in routine population screening for pre‐symptomatic T1D has gained momentum internationally. In Italy, a public health screening program for T1D and coeliac disease has recently been legislated [37].

This review will describe a few illustrative examples of screening programs, including from Australia.

The Trialnet Pathway to Prevention (Trialnet) screens relatives (although the general population is now included) aged 3–45 years for IA. Of the more than 200 000 relatives screened, 5% had at least one IA and approximately 2.5% had multiple IA [38, 39]. Type1Screen in Australia/NZ also screens relatives and has had a similar IA detection rate [39].

General population screening programs have either stratified based on genetic susceptibility or screened directly for IA. GRS‐based approaches have the benefit of targeting a high‐risk population for further screening and may be more cost‐effective than screening all children for IA irrespective of GRS. The DIPP cohort in Finland and the TEDDY study both screened infants with high‐risk HLA haplotypes, as discussed above, and found that approximately a quarter of participants developed IA and over 3% developed T1D [17]. The Global Platform for the Prevention of Autoimmune Diabetes (GPPAD) trial demonstrated that those with a high genetic risk of T1D had > 10% chance of IA by the age of 6 years. Methods of screening could include alignment with newborn screening or infant primary care visits. Identification of increased risk during infancy could allow for early diagnosis in toddler/pre‐school‐aged children, who are most at risk of progression [34].

Alternatively, an IA population‐based approach identifies early‐stage T1D, rather than genetic risk. The Fr1da study in Bavaria, Germany, screened pre‐school children attending primary care paediatrician visits through capillary antibody tests. Positive capillary samples were confirmed with venous testing. This study found a prevalence of 0.31% for stages 1 and 2 in pre‐school children from the general population [8]. The incidence of DKA in the screened population was significantly lower at 3.2% compared with a background rate of 20% [8]. Similarly, the Autoimmunity Screening for Kids (ASK) program in Colorado screened children and youth aged 2–17 years for IA with a prevalence of approximately 0.5% for early‐stage T1D [40]. Economic modelling from ASK showed that screening would be cost‐effective if DKA rates were reduced by 20% and HbA1c reduced by 0.1% over the lifetime [40].

In Australia, the Type 1 Diabetes National Screening Pilot (T1D NSP) is comparing three screening approaches [41]. This pilot commenced in 2022 and completed recruitment at the end of 2023. It screened three cohorts of children: newborns, infants 6–12 months and pre‐school‐/school‐aged children (Figure 3). Neonates were tested for genetic susceptibility using the GRS2 alongside routine newborn screening dried bloodspot sampling. Infants were offered the same GRS2 test via a saliva swab collected at home or at primary care visits. Neonates or infants found to be at risk of T1D were then offered annual antibody testing for 5 years using a fingerprick dried bloodspot kit. Pre‐school and school‐aged children (2, 6 or 10 years old) were screened for IA by fingerprick dried bloodspot sampling, which was then confirmed on serum IA testing for those that were positive on the initial screen [41]. This approach seeks to identify the most feasible and acceptable option for routine general population screening. Results of this trial are pending.

FIGURE 3.

T1D NSP cohort diagram. Reproduced with permission from the T1D NSP.

4. Clinical Care for Early‐Stage T1D

The JDRF consensus guidelines provide a framework for monitoring children, young people and adults with early‐stage T1D, i.e. with positive IA, in the context of current clinical knowledge [12]. They also provide guidance on education and psychological support and for monitoring both children and adults with early‐stage T1D. Similar monitoring guidelines for early‐stage T1D have also been provided in the ISPAD [2] and American Diabetes Association (ADA) guidelines [42].

4.1. Monitoring Methods

The ideal test to predict progression to stage 3 over time is still under debate. OGTTs are the current gold standard and may still have the most predictive value in distinguishing progressors vs. non‐progressors [43] however, these are onerous and invasive to perform, especially for young children. A fingerprick glucose is a simpler and cheaper test to monitor for dysglycaemia, defined as a fasting blood glucose (FBG) between 5.6 and 6.9 mmol/L. Other indices based on a single OGTT time point as well as C‐peptide levels, age, sex HbA1c and antibody status, such as M60 and M120 [44], have been developed to predict risk of progression to stage 3. These scores predicted progression in 65% of participants who scored above the median and may prove less onerous as they involve a single blood test [44]. More recently, HbA1c levels have been used to predict progression. A HbA1c of over 5.7% [45] or a 10% rise from baseline in those with stages 1–2 T1D has been associated with an increased risk of stage 3 [46].

Continuous glucose monitoring (CGM) may prove to be a more feasible and less invasive method of monitoring for dysglycaemia. CGM is currently subsidised in Australia for people with stage 3 T1D but not for those with stages 1 and 2. Youth with early‐stage T1D have a higher mean CGM glucose level as well as a greater proportion of time spent with values ≥ 7.8 mmol/L compared to those with a genetic risk for T1D but without IA [45]. Recent data from the ENDIA study confirm these findings in young children [47]. This is in contrast to normative data in healthy children who spend 96% of their time between 3.9 and 7.8 mmol/L [48]. Thus dysglycaemia or stage 2 has been defined as CGM values more than 7.8 mmol/L for more than 10% of the time over 10 days along with one other dysglycaemic measurement such as FPG or OGTT levels [12].

All studies to date on CGM in early‐stage T1D have used professional or ‘blinded’ CGM where real‐time glucose data is not visible to the user but is reviewed by the clinicians at the end of the monitoring period [45, 47, 49]. However, commercially available CGM allow users to view glucose levels in real time to modify behaviours or treat them with insulin. The impact on glucose levels of potential behaviour modification by individuals with early‐stage T1D viewing data in real time is unknown. At present, blinded CGM are only available for research purposes whereas clinical use of CGM for monitoring would necessitate the use of commercially available devices. The consensus guidelines support blinded CGM however, further research on unblinded CGM in this population, as well as what metrics may be predictive of progression, is necessary to recommend its use in clinical practice.

4.2. Children and Young People (CYP)

The guidelines separate monitoring regimens based on single and multiple IA status, given the greater risk of progression in the latter group [12]. The frequency of monitoring is also dependent on the age of seroconversion as children ≤ 3 years are at the highest risk of rapid progression [39]. The key aspects of monitoring CYP with IA are summarised in Tables 1 (single IA) and 2 (multiple IA).

TABLE 1.

Monitoring single IA children as recommended by the JDRF Consensus guidelines [12].

Age	Antibody testing	Blood glucose and HbA1c	Duration
≤ 3 years	6 monthly	6 monthly	3 years then follow protocol for ages > 3 years
> 3 years	Annually	Annually	3 years then cease if no progression

TABLE 2.

Monitoring multiple IA children as recommended by the JDRF Consensus guidelines [12].

Age	Blood glucose levels	HbA1c	CGM a
≤ 3 years OR Stage 2 T1D (any age)	After recent confirmation of IA status—check BGL on 2 different days over a 2‐ week period then every 1–3 months	3 monthly	3 monthly
3– ≤ 9 years		6 monthly	6 monthly
> 9 years		Annually	Annually

^a CGM can be offered and must be interpreted by trained health care professionals, with education provided for the individual and their family.

4.3. Adults

The incidence of T1D in adults is less clear due to possible misdiagnosis as type 2 diabetes, and varies by region [50]. However, approximately 50% of individuals with T1D are diagnosed as adults. As with childhood T1D, the lowest incidence is in East Asia, and the highest in Western Europe and could be up to 32/100 000 person years [50]. The risk of progression to stage 3 T1D is less rapid than in children. However, these individuals still need monitoring. In general, the guidelines advise annual glucose and HbA1c monitoring in adults with IA, with alterations in frequency based on stage and other risk factors [12]. The Type 1 Diabetes Risk in Adults (T1DRA) study in the UK is currently screening adults aged 18–70 years from the general population for IA to better understand aetiology and progression in the adult population (https://t1dra.bristol.ac.uk).

5. The Nuances of Screening and Monitoring

5.1. What is the Optimal Model?

Population screening for a disease needs a simple, accessible, cost‐effective and acceptable test for screening as well as an accepted management pathway for those identified at risk [39].

IA screening would potentially identify early‐stage T1D. Screening at a single time point, e.g. in preschoolers, would help identify those at greatest risk of rapid progression and be more cost‐effective. However, it would miss children who had already progressed to stage 3 or who seroconverted at a later age. Similarly, if only single IA positivity was identified (10%–15% risk of progressing to stage 3) [30], then repeat screening would be required for multiple IA detection. The TrialNet group postulated that in paediatric multiple IA individuals, 14 monitoring visits between the ages of 1 and 18 years, with greater frequency in the early paediatric years compared to adolescence, would result in an undiagnosed period of 6 months, balanced against too onerous a clinical burden [51].

Italy is adopting the IA model with testing occurring at ages 2–2.9 years, 6–6.9 years and 10–10.9 years with repeat screening if previously negative at one of the earlier age groups. Those with single or multiple IA will have monitoring consistent with the consensus guidelines at specialist T1D centres [52]. In contrast to Australia/NZ, primary care visits for children, and hence IA screening, are conducted by paediatricians in Italy. A different model of care would be required for Australia/NZ, with particular consideration as to its implementation in regional and rural areas.

The improvement in IA detection via self‐collected capillary blood spots would make it a relatively accessible and specific test which could be posted out to families across the country [29]. However, the cost‐effectiveness of general population screening through this method to detect a relatively small number of individuals is unclear. Economic modelling from the ASK study suggested that screening could be cost‐effective in the US if DKA rates were reduced by 20% and HbA1c reduced by 0.1% over the lifetime [40]. The benefits in a public health care system such as in Australia and NZ are not yet known. A JDRF Australia commissioned report by Accenture suggested that the average lifetime cost of T1D is approximately $500 000. This could be reduced by about $33 000 if there was no DKA at diagnosis and by approximately $88 000 if diagnosis could be delayed by 5 years [53].

Alternatively, the GRS could be used to identify high‐risk individuals who would benefit most from serial IA testing rather than screening the general population. This approach could be more cost‐effective; however, unlike other childhood screening programs, individuals could be accorded a high‐risk score and yet never manifest the disease. Furthermore, the validity of the GRS2 in multiethnic populations such as Australia/NZ needs to be evaluated.

5.2. Psychosocial Support and Education

The diagnosis of early‐stage T1D may be a source of distress to families [8] with elevated levels of parental anxiety as demonstrated by the State Anxiety Inventory (SAI) [54]. The Fr1da study demonstrated that although parental anxiety was initially elevated at the time of seroconversion, overall anxiety improved with time and was less than that felt by parents of children diagnosed with DKA or T1D without prior monitoring [8].

The ASK study found that whilst parental anxiety about IA positivity, measured by the SAI, reduced over time, they still remained above the threshold of 40, indicating high anxiety. Higher anxiety levels were seen in those from a non‐white ethnicity or with lower educational attainments [54]. If population screening for T1D is to be implemented, psychosocial support and multidisciplinary care will need to be an integrated part of screening and monitoring visits [12].

Less is known about the potential for anxiety following a high genetic risk score. One relatively old study from New Zealand reviewed parental perceptions 12 years after their newborns were screened for T1D risk. This suggested that families were not particularly anxious about the risk score although did not perceive much benefit from the knowledge if no intervention could be offered [55]. However, with immunotherapies on the horizon, one strong argument for T1D screening would be to identify those who could be eligible for intervention.

6. Disease Modifying Therapies

Emerging disease‐modifying therapies to preserve β‐cell function and potentially delay progression to stage 3 T1D present a paradigm shift in the management of T1D. A recent review article has described the various agents and trials in detail [56] and is beyond the scope of this paper. Teplizumab, a monoclonal antibody against CD3, is currently the only FDA‐approved agent for use in stage 2 T1D. It was demonstrated to delay the onset of stage 3 by close to 5 years compared to 2 years in the placebo group [57]. There appeared to be a reversal in the decline of C‐peptide levels on teplizumab compared to placebo, possibly correlated with an increased number of partially exhausted CD8 T cells. Teplizumab is a 14‐day course of infusions which is not yet approved by the Therapeutic Goods Administration (TGA) in Australia nor by Medsafe in NZ. Thus in Australia/NZ, interventions are currently only available through clinical trials although this is likely to change over time.

A variety of therapies have shown benefit in newly diagnosed stage 3 individuals. One example was the Australian‐run BANDIT (Baricitinib and β‐cell Function in Patients with New‐Onset Type 1 Diabetes) trial which showed higher C‐peptide levels and lower HbA1c and insulin doses with the orally administered janus kinase (JAK) inhibitor baricitinib, in those with newly diagnosed stage 3 T1D [58]. It is known that interferon activation of JAK pathways and induction of signal transducer and activator of transcription (STAT) factors lead to overexpression of HLA‐1 and pro‐inflammatory effects in T1D [59]. Further research is needed to ascertain if JAK inhibitors could be used in early‐stage T1D.

In addition to immunotherapies, intrinsic properties of the β‐cell, which may predispose it to immune attack, are also being studied, such as overexpression of certain proteins leading to apoptosis [1]. Verapamil, a well‐known calcium channel blocker, may reduce the expression of some of these proteins and was recently shown in a small RCT to preserve C‐peptide at 52 weeks in newly diagnosed stage 3 CYP [60]. It may be that multiple types of agents may need to be used for disease modification in T1D and further research in the early‐stage population is needed.

7. Unmet Needs and Future Directions

T1D poses a significant burden to individuals and families across the lifetime. The cost‐benefit of screening and monitoring needs to be ascertained. Once early‐stage individuals are identified, a collaborative approach will be needed incorporating education, diabetes technology and immunotherapies. Diabetes education so far has focussed on insulin administration and glycaemic management however, for early‐stage individuals not on insulin, education will need to focus on an understanding of progressive dysglycaemia, signs of DKA and the benefits of any emerging immunotherapies. Models of care will likely need to be tailored to the unique demographic characteristics of health districts within Australia and NZ. Pilot studies like the Australian T1D NSP will help to ascertain the appropriate clinical models.

8. Conclusions

The importance of screening and monitoring of children with early‐stage T1D is gaining international recognition, particularly for reducing rates of DKA and subsequent complications. Early, easily identifiable stages exist for a condition that has a significant long‐term impact on CYP, as well as their families. A clinical framework to guide the management of people with early‐stage T1D now exists and is predicated on age and IA positivity. Understanding the feasibility and acceptability of screening and monitoring to families, health care professionals and the broader community will be crucial to guide practical implementation. The possibility of a paradigm shift in the management of T1D toward delaying progression means that identification of individuals with a predisposition to or early‐stage disease is essential.

Conflicts of Interest

The authors declare no conflicts of interest.