Islet Autoantibody Screening Throughout Australia Using In-Home Blood Spot Sampling: 2-Year Outcomes of Type1Screen Free

Early diagnosis of type 1 diabetes (T1D) through detection of circulating islet autoantibodies, combined with glycemic monitoring, prevents hospitalization from diabetic ketoacidosis (DKA) (1,2). In addition, identifying children at an earlier stage of disease, who have significant residual β-cell function, is likely to favor successful outcomes of disease-modifying immunotherapies (3). Given these benefits, Italy has committed to screen children for T1D (4), and general population screening programs are being established in other countries (5–7). This raises the logistical challenge of population-wide screening. Key characteristics of a successful screening test will be low cost, simplicity, accuracy, and high acceptability to both children and their caregivers.

Most T1D screening programs rely on blood sampling by a trained health professional. This imposes additional cost, can be inconvenient, and is inaccessible for low-resource populations, including regional and remote communities. To address this, some programs offer sample self-collection. The T1Detect study, launched in 2020, used the antibody detection by agglutination-PCR (ADAP) assay (8) to screen self-collected capillary blood spot samples from people living in the U.S. However, low assay specificity proved problematic (9). Swedish participants of a general population screening study called TRIAD collected capillary blood samples at home and mailed them to a central laboratory for serum extraction and ADAP testing for islet, celiac, and thyroid autoantibodies (10). TRIAD identified the expected prevalence of autoimmunity with good specificity but was challenged by a high rate of unsuccessful sample collections.

We recently developed self-collected blood spot screening methodology suitable for detection of islet autoantibodies against insulin (IAA), glutamic acid decarboxylase (GADA), insulinoma antigen-2 (IA-2A), and zinc transporter 8 (ZNT8A) by ADAP assay. A validation study involving research participants with and without islet autoimmunity showed capillary blood spot collection was highly acceptable and preferred over venipuncture and accurately diagnosed islet autoimmunity (11).

Australian research programs operating since 2007 have demonstrated national capacity to detect preclinical T1D and prevent serious illness in the event of disease progression (2). In 2019, Type1Screen was established to offer islet autoantibody screening to families living with T1D and monitoring for those who test positive. Online registration and self-collected capillary blood spot screening was implemented in July 2022. In this analysis we describe screening, monitoring, and clinical outcomes over the subsequent 2 years and characteristics and outcomes of people who collected a blood spot at home are compared with those who underwent venipuncture at a community collection center.

Type1Screen is an Australian observational cohort study coordinated by Royal Melbourne Hospital and involving five pediatric diabetes centers located in each mainland state. The protocol, registered as ACTRN12620000510943, was approved by the Melbourne Health Human Research Ethics Committee. The primary outcome is the confirmed presence of one or more islet autoantibodies.

Key inclusion criteria are Australian residence, age >2 years, family (but not personal) history of T1D or no family history of T1D, and prior detection of islet autoantibodies by an external laboratory. Participants were recruited during interactions in clinical settings or through mail, e-mail, and social media communications. All participants consented and registered through an online portal (www.type1screen.org) and selected either a capillary blood spot collection in the home mailed to the Endocrine Laboratory at Royal Melbourne Hospital or venipuncture and sample centrifugation at a local pathology collection center with courier transfer to the same laboratory. Written instructions for blood spot collection (Supplementary File) were mailed to participants and provided online alongside an instructional video (available at https://vimeo.com/739896967). All participants whose screening test result was positive were asked to provide a second venipuncture sample to confirm the result.

We included participants who registered between 1 July 2022 and 30 June 2024. All participants who screened positive were contacted by telephone and asked to arrange a venous sample for confirmatory antibody testing as well as HbA_1c and random glucose at their local laboratory. If these tests were not performed within 3 months of the initial telephone contact, additional attempts to arrange them by e-mail and telephone were made on at least two separate occasions. In October 2024, e-mail and text message surveys (Supplementary File) were sent to all participants who had registered but not undertaken screening to ask why they had not returned a sample.

Capillary blood spot samples were mailed to the laboratory in airtight bags containing desiccant and then stored at −20°C for up to 4 months before testing. The median (quartile 1, quartile 3 [Q1, Q3]) postal time was 7 (4, 9) days (range 0–17 days).

Blood spot and serum assays using ADAP (for IAA, GADA, IA-2A, and ZNT8A), radioimmunoassay (RIA; for IAA) and ELISA (for GADA, IA-2A, and ZNT8A) have been described previously (11). The ADAP assay for ZNT8A was introduced in May 2023, 10 months after blood spot screening had been implemented.

Serum screening used IAA RIA and the multiplex 3Screen ELISA to test for the presence of GADA, IA-2A, and/or ZNT8A, with sera positive by 3Screen then tested using individual ELISAs for GADA, IA-2A, and ZNT8A. The RIA and ELISA assays have demonstrated high accuracy in the Islet Autoantibody Standardization Program (12). Because our RIA/ELISA assays predict clinical disease in Australian relatives (2), they were used to define final autoantibody status using confirmatory serum samples. Serum ADAP assays were performed when results of blood spot ADAP and serum RIA/ELISA were discordant.

Blood spot autoantibody concentrations measured by ADAP were normalized to the lowest detectable titer of an in-house standard prepared by diluting extracts from a single T1D donor into extracts from a healthy donor. All blood spot ADAP tests were performed in singlicate. A positive result required a sample to test positive on the initial screening test and again in a second ADAP assay performed on a different day. Samples that were positive on the first and negative on the second ADAP assay were classified as negative screening tests.

All participants with confirmed islet autoimmunity were e-mailed a summary of their test results and information about the value of monitoring and about symptoms of diabetes. They were asked to have glycemic monitoring tests at a local pathology collection center every 6–12 months using an oral glucose tolerance test (OGTT), random glucose, HbA_1c, and continuous glucose monitoring (CGM), according to their preference. Monitoring tests were arranged by the participant’s usual health service where possible or virtually by a study nurse located in Melbourne. Dysglycemia was defined as one or more of HbA_1c 39–47 mmol/mol (5.7–6.4%), fasting glucose 5.6–6.9 mmol/L (100–125 mg/dL), 60-min OGTT glucose ≥11.1 mmol/L (≥200 mg/dL), 120-min OGTT glucose 7.8–11.0 mmol/L (140–199 mg/dL), or CGM values >7.8 mmol/L (>140 mg/dL) for >10% of time over ≥10 days’ continuous wear (13,14). Stage 3 T1D was defined by American Diabetes Association criteria (15). Glycemia data collected up to 30 September 2024 were included in outcome analyses. The diabetes status of all registered participants was sought in October 2024 by e-mail and text message (Supplementary File).

Analyses used Prism Software 10.3.0 (GraphPad Software, Boston, MA). Categorical data were compared using the Fisher exact test and χ² test for trend. Continuous data were compared using Mann-Whitney test. A P value <0.05 was considered statistically significant. No corrections for multiple comparisons were applied.

Over the 2 years following introduction of blood spot screening, 2,064 people registered with Type1Screen. Of these, 1,507 (73%) requested in-home capillary blood spot sampling (Fig. 1A). Children comprised most of the participants who requested blood spot collection (1,065 of 1,507 [71%]) compared with approximately half who requested venipuncture (290 of 557 [52%]). Within the group that requested blood spot screening, those who did not action a request for blood spot screening were younger than those who did (median [Q1, Q3] age 9 [4, 17] vs. 12 [6, 26] years; P < 0.0001) and were more likely to be male (53% vs. 44%; P = 0.0014).

A testable sample was returned by 1,285 participants (63% of the registered population). The return rate was higher for participants who requested blood spot screening (986 of 1,507 [65%]) compared with participants who requested venipuncture (299 of 557 [54%]). Unsuccessful sample collections were reported for one child who chose blood spot and five children who chose venipuncture. Only 1 of the 967 blood spot samples received was insufficient for testing. A second collection kit was mailed to this child, and a testable blood sample was returned.

By 30 June 2024, 967 blood spot and 276 venous samples had been tested for islet autoantibodies. In this group, participants who chose blood spot screening were significantly younger (Table 1) and more likely to live outside a major Australian city (39% vs. 29%, P = 0.0037) (Fig. 1B). The sex ratio and genetic risk according to T1D proband were similar between the groups (Table 1).

Positive screening test results occurred in a lower proportion of blood spot compared with venous samples (5.1% vs. 8.3%; P = 0.0561) (Table 1). However, 6 of the 23 positive venous samples (3 with a single and 3 with multiple islet autoantibodies) had been found to have islet autoimmunity by an external laboratory prior to joining Type1Screen. After these participants were excluded, positive screening tests were observed in 5.1% of blood spot and 6.1% of venous samples (P = 0.4503).

Each participant with a positive screening test and no prior history of islet autoimmunity (n = 66) was asked to provide a second serum sample for confirmatory testing. Only 45 participants (33 of 49 tested using blood spot and 12 of 17 tested using venipuncture) undertook this at a median (Q1, Q3) of 84 (44, 177) days after the initial test. Reasons for not having a confirmatory islet autoantibody test included the participant’s decision not to repeat it (n = 6), pregnancy (n = 2), progression to stage 3 T1D (n = 3), inability to visit a phlebotomist prior to the censor date of 30 September 2024 (n = 5), and loss to follow-up (n = 5).

Confirmatory testing of serum samples from 12 positive venous screens (10 with single and 2 with multiple autoantibodies) returned identical antibody profiles for 8 single- and both multiple-autoantibody participants. Two participants with a single low-titer GADA (12 and 29 units/mL; reference range <5 units/mL) tested negative to each of the four autoantibody specificities on their confirmatory serum test.

Confirmatory testing of the serum samples from 33 positive blood spot screens (21 with single and 12 with multiple autoantibodies) returned identical autoantibody specificities for 5 single- and 10 multiple-autoantibody-positive screens. Two participants who had multiple autoantibodies on blood spot testing had a single autoantibody on serum testing (both had false-positive IAA) (Fig. 2, triangles). Ten participants who had a single autoantibody were found to have multiple autoantibodies on the confirmatory serum test. One single-antibody participant screened positive for GADA by blood spot with a subsequent confirmatory serum sample testing positive by 3Screen ELISA and GADA ADAP but negative by GADA ELISA (Fig. 2, square). The remaining five participants whose blood spots screened positive to a single autoantibody were found to have no autoantibodies when serum samples were tested by RIA, ELISA, and ADAP assays (Fig. 2, open circles).

One blood spot participant without a T1D family history (Table 1) had a positive test for IA-2A and ZNT8A in an external laboratory. The blood spot screen was negative. Tests of a subsequent venous sample were also negative, suggesting a false-positive result from the external laboratory.

After the six venous sample participants with known islet autoimmunity and the seven participants (five blood spot and two venous) who had a positive screening test but no detectable autoantibodies in their confirmatory serum sample were excluded, the overall incidence of islet autoimmunity decreased from 5.9 to 4.8% (Table 1). Accordingly, there were corresponding decreases in the prevalence of positive screens to 4.6% in blood spot and 5.6% in venipuncture participants (P = 0.5182). In addition, confirmatory venous testing increased the number of blood spot participants with multiple autoantibodies from 15 to 20, thereby increasing the prevalence of preclinical T1D in blood spot participants to 2.1% and decreasing their prevalence of single autoantibody status from 3.5 to 2.5%. The corresponding final prevalences of multi- and single-autoantibody participants screened by venipuncture were 1.5% (4 of 270) and 4.1% (11 of 270), respectively.

Correlations between blood spot ADAP and serum RIA/ELISA autoantibody concentrations are presented in Fig. 2. Blood spot and venous tests showed significant correlations, with R values for IAA, GADA, IA-2A, and ZNT8A of 0.544, 0.878, 0.648, and 0.759, respectively (P < 0.005 for all). For each islet autoantibody there was reduced sensitivity of ADAP compared with RIA and ELISA for a few samples (upper left quadrants of correlation graphs). On the basis of available paired blood spot and serum results, the sensitivities of blood spot ADAP for IAA, GADA, IA-2A, and ZNT8A were 57%, 92%, 80%, and 70%, respectively.

For each of the 72 participants who had positive autoantibody screening results, we sought to determine glycemic status using CGM and venous blood tests for HbA_1c and glucose (Fig. 3). Twenty participants did not obtain monitoring tests, of whom, two developed symptomatic diabetes and commenced insulin. Of the 52 participants who had glycemic tests, 5 arranged them with their local doctor and 47 with the assistance of a study nurse based in Melbourne. Results showed 6 participants had silent (stage 3a) diabetes, 5 had dysglycemia, and 41 had normal test results (Fig. 3). Only six participants undertook OGTT for longer-term glycemic monitoring. Results revealed stage 3 T1D in one child and normoglycemia in five adults.

By censure date of 30 September 2024, 12 participants who had screened positive for islet autoantibodies (7 children and 5 adults) had progressed to stage 3 T1D and commenced insulin. None developed DKA, and six (two children and four adults) started insulin without requiring hospital admission. No participants who screened negative reported developing clinical diabetes. A text message and e-mail survey was sent to all registered participants in October 2024. This identified an additional child with stage 3 T1D who had registered but had not provided a screening sample. They noted diabetes symptoms and presented to the hospital where hyperglycemia and ketosis without acidosis was diagnosed, requiring an overnight admission to commence insulin.

We demonstrate that Type1Screen has national reach and identifies relatives at risk for developing T1D. In-home blood spot sampling enabled detection of preclinical T1D, defined by the presence of multiple islet autoantibodies (16), at the expected frequency. In addition, our glycemic monitoring program reliably identified disease progression to stage 3 T1D and enabled timely commencement of insulin in 12 antibody-positive participants before DKA developed.

To our knowledge, this is the first report of successful implementation of self-collected blood spot screening for islet autoantibodies. The 5.9% positive screen rate in Type1Screen is comparable to rates observed in other family screening programs such as TrialNet (5.5%) (17), Bart’s-Oxford Family Study (7%) (18), and Innovative approach towards understanding and arresting type 1 diabetes (INNODIA; 6%) (9). The only other study to report outcomes of islet autoantibody screening using sample self-collection is the TRIAD study (10), which required participants to collect >250 µL capillary blood for subsequent serum extraction and testing. Although the TRIAD methodology detected the expected prevalence of islet, celiac, and thyroid autoimmunity in Swedish children, sample collection proved too difficult for more than one-third, many of whom made multiple attempts. Our finding that only one blood spot participant reported unsuccessful sample collection indicates our capillary blood spot collection method is much more feasible.

A significant proportion (40%) of participants enrolled in Type1Screen did not submit samples for testing. This was partly explained by participants who had not yet had time to collect a sample. Failure to action screening might also be explained by our two-step process of consent/registration followed later by sample collection as well as resource limitations that prevented us from individually contacting participants to encourage them to submit a sample. In addition, our unpublished feedback from T1D families suggests many decide not to proceed after reconsidering the potential harm of screening positive without access to disease-modifying therapy. It will be of interest to determine whether screening rates increase following implementation of planned prevention trials in 2025.

Our successful deployment of blood spot screening contrasts with outcomes of the T1Detect general population screening program, which used similar collection and testing methodology but found that most positive screens could not be confirmed when subsequent serum samples were tested using RIA and ELISA autoantibody assays (9). Potential reasons for improved assay specificity in our study include the higher risk of islet autoimmunity in T1D relatives compared with the general population (9), ADAP assay modifications that decreased liquid transfer steps and enhanced DNA amplification from whole blood (11), and our requirement to confirm all positive blood spot screens in a second independent assay. The Bart’s-Oxford Family Study previously implemented capillary blood self-collection (19) and reported high feasibility, acceptability, and accuracy in pilot studies (20,21) but has not described autoantibody outcomes for prospective blood spot screening. More recently, the EarLy Surveillance for Autoimmune diabetes (ELSA) study used self-collected capillary blood spots to screen >10,000 U.K. children from the general population using 3Screen assay for GADA, IA-2A, and ZNT8A (6). A detailed report of ELSA screening outcomes is awaited.

Type1Screen participants preferred blood spot sampling over venipuncture, consistent with findings from our validation study (11). Participants who chose blood spot sampling were younger than those who chose venipuncture and were less likely to experience collection failure. These findings suggest blood spot collection is particularly well suited to young children, for whom T1D screening appears to be most cost-effective (22). In addition, a substantial proportion (39%) of blood spot participants lived in regional and remote communities, compared with 29% of venous participants and 28% of the Australian population (23). The ability of blood spot screening to reach beyond major cities should help address health inequalities in underserved communities, including in regional Australia, where limited access to primary health care contributes to poor health outcomes (24).

The prevalence of single-antibody participants in the blood spot group was lower than that observed in the venous group. Correlation analyses (Fig. 2) indicate this was probably due to lower sensitivity of blood spot ADAP compared with serum RIA/ELISA, consistent with findings of our earlier validation study (11). It is likely that the multiplex nature of ADAP permits identification of the expected number of multiantibody participants within our population, but lower sensitivity may fail to identify single-antibody participants, particularly when autoantibody titers are low (Fig. 2). In the current environment, where disease-modifying immunotherapy is reserved for people with multiple islet autoantibodies, these performance characteristics are probably acceptable. However, low blood spot ADAP sensitivity for detecting single islet autoantibody status might prove problematic for very young children, who often develop single IAA initially and then progress from single to multiple antibody status (6,25). On the other hand, given the positive association between islet autoantibody concentration and risk of disease progression (26), low ADAP blood spot sensitivity might improve specificity for predicting clinical disease, thereby improving precision of preventative treatment.

More than one-quarter of the 72 participants who screened positive did not undertake recommended glycemic monitoring tests, in line with prior findings from TrialNet (27). This high rate of nonadherence may relate to our reliance on telephone and e-mail communications to deliver monitoring services in Type1Screen. In rural Australia, telehealth is most effective when there is a preexisting relationship between the patient and the health care provider (28), which we had not established. To address missed opportunities to screen and monitor for T1D, we are now offering Type1Screen participation to families during in-person consultations at each of our affiliated pediatric diabetes services.

While the absence of DKA following disease progression suggests that our “light touch,” predominantly virtual monitoring program is safe, further follow-up of larger numbers of at-risk participants will be needed to confirm this. The relatively low prevalence of stage 2 T1D in our population is also notable. This is probably explained by low uptake of OGTTs, particularly by children. On the basis of outcomes of screening programs in which OGTT is routine (7,16), it is likely that at least six of our antibody-positive participants had stage 2 T1D despite normal HbA_1c and glucose. OGTT monitoring has not been universally accepted by participants of other screening programs (7,29), perhaps because they do not consider OGTT necessary to prevent DKA. The advent of teplizumab to prevent progression from stage 2 to stage 3 T1D highlights a need to increase OGTT acceptability and feasibility, particularly for children.

Our study has some limitations. Although Type1Screen uses a dedicated telephone line and e-mail address to help participants communicate with us, we did not personally contact those who did not return screening samples and those who screened negative to inquire about their experiences and their health. Accordingly, we may have underreported the difficulty of sample collection and the rates of stage 3 T1D and DKA in the entire study population. In addition, our study design did not permit determination of overall ADAP blood spot assay accuracy in our population. Furthermore, we did not assess 3Screen blood spot testing, which we (11) and others (30) have shown can discriminate between people with and without T1D. Future prospective study of both ADAP and 3Screen blood spot assays across a larger number of samples will be required to determine whether one of these assays, or their combination, is most accurate. Finally, the success of blood spot screening in this study may not translate to the general population for three reasons. First, our recruitment strategy was poorly suited to people with limited access to the internet or low self-efficacy, who were probably underrepresented. Second, family members usually have background knowledge of T1D screening and capillary blood collection. Lower awareness in the general population might make it more challenging to recruit into a universal screening program and might also increase the rate of unsuccessful blood spot collections. Third, we did not obtain information regarding socioeconomic status, ethnicity, or linguistic diversity to confirm representation of minority populations.

In summary, Type1Screen is a truly national islet autoantibody screening program for relatives of people living with T1D, including those living in regional and remote communities. Blood spot ADAP identified the expected frequency of multiantibody participants, and a centralized monitoring program enabled timely commencement of insulin for those who progressed to stage 3 T1D. Type1Screen is a model of T1D screening and monitoring that could be scaled up to meet the needs of the general population to maximize the benefits of early diagnosis and disease-modifying immunotherapy.

Acknowledgments. The authors are grateful to their research participants, many anonymous donors, the Eccles family, and the Type 1 Foundation for their contributions to the success of Type1Screen.

Funding. This study was funded by the Medical Research Future Fund (RARUR000103) and by JDRF Australia (2-SRA-2022-1282-M-X and 4-SRA-2022-1246-M-N) through funding from the Medical Research Future Fund Accelerated Research Funding Program administered by the Australian Government Department of Health.

Duality of Interest. J.M.W. and P.G.C. direct Type1Screen and serve alongside J.J.C., T.H., E.A.D., and M.E.C. on the Steering Committee of the National Screening Program. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.M.W. drafted the manuscript. J.M.W., A.B.E.S., G.N., D.H., C.C., A.G., K.W., and L.C.H. performed laboratory assays and analyzed the data. J.M.W., E.A.-Y., B.M., K.M., L.R., R.K., A.H., F.H., J.J.C., C.H., T.H., E.A.D., M.E.C., F.J.C., J.J.C., and P.G.C. recruited and cared for participants. J.M.W., T.W.K., and P.G.C. obtained funding. J.M.W. and P.G.C. established and direct Type1Screen. J.D.B. performed data analyses. All authors edited and approved the manuscript. J.M.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 20th Immunology of Diabetes Society (IDS) Congress, Bruges, Belgium, 4–8 November 2024.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Steven E. Kahn and Michael J. Haller.