OBJECTIVE

To determine whether presentation, progression, and genetic susceptibility of robustly defined adult-onset type 1 diabetes (T1D) are altered by diagnosis age.

RESEARCH DESIGN AND METHODS

We compared the relationship between diagnosis age and presentation, C-peptide loss (annual change in urine C-peptide–creatinine ratio [UCPCR]), and genetic susceptibility (T1D genetic risk score [GRS]) in adults with confirmed T1D in the prospective StartRight study, 1,798 adults with new-onset diabetes. T1D was defined in two ways: two or more positive islet autoantibodies (of GAD antibody, IA-2 antigen, and ZnT8 autoantibody) irrespective of clinical diagnosis (n = 385) or one positive islet autoantibody and a clinical diagnosis of T1D (n = 180).

RESULTS

In continuous analysis, age of diagnosis was not associated with C-peptide loss for either definition of T1D (P > 0.1), with mean (95% CI) annual C-peptide loss in those diagnosed before and after 35 years of age (median age of T1D defined by two or more positive autoantibodies): 39% (31–46) vs. 44% (38–50) with two or more positive islet autoantibodies and 43% (33–51) vs. 39% (31–46) with clinician diagnosis confirmed by one positive islet autoantibody (P > 0.1). Baseline C-peptide and T1D GRS were unaffected by age of diagnosis or T1D definition (P > 0.1). In T1D defined by two or more autoantibodies, presentation severity was similar in those diagnosed before and after 35 years of age: unintentional weight loss, 80% (95% CI 74–85) vs. 82% (76–87); ketoacidosis, 24% (18–30) vs. 19% (14–25); and presentation glucose, 21 mmol/L (19–22) vs. 21 mmol/L (20–22) (all P ≥ 0.1). Despite similar presentation, older adults were less likely to be diagnosed with T1D, insulin-treated, or admitted to hospital.

CONCLUSIONS

When adult-onset T1D is robustly defined, the presentation characteristics, progression, and T1D genetic susceptibility are not altered by age of diagnosis.

The impact of age on the presentation and progression of adult-onset type 1 diabetes (T1D) is unclear. It is commonly understood that T1D in older adults has a milder phenotype, with reduced rate of progression compared with those with young adult–onset disease (14). However, this understanding has been predominantly either extrapolated from findings in childhood and adolescence, taken from cross-sectional studies of clinically diagnosed T1D in older adults or extrapolated from studies of those with positive islet antibodies initially diagnosed as type 2 diabetes (T2D) (latent autoimmune diabetes in adults [LADA]) (2,5,6). A limitation of existing studies of older adults is the possibility findings are inadvertently affected by the inclusion of individuals who do not have T1D, “diluting” the observed phenotype. This is a major concern, as diagnosing T1D in older adults is challenging (710). The known overlap between features of T1D and T2D and very high prevalence of T2D means the predictive value of even classical clinical features of T1D, such as ketoacidosis or low BMI, is limited in adults (11). It is unsurprising that studies using robust approaches to diabetes classification in older adults suggest that 14–20% of those with a clinician diagnosis of T1D may be misclassified (8,10). While studies of LADA may support a changing phenotype of autoimmune diabetes with age, a combination of low prior prevalence and use of a biomarker with imperfect specificity means this phenotype may also be influenced by the inadvertent study of a mixed population of people with diabetes of autoimmune and nonautoimmune etiology (12,13).

To study the impact of age of diagnosis on phenotype in adult-onset T1D, diagnostic tools with very high specificity are needed to avoid inadvertent inclusion of other forms of diabetes. The optimal method will depend on the research question and disease duration. C-peptide has been recommended as a classification method that most closely relates to treatment requirements; however, this measure has limited utility close to diagnosis and precludes an unbiased study of disease progression (14,15). Multi-islet autoantibody positivity is highly specific for T1D but will identify only ∼60% of adults (10,16) with autoimmune diabetes, and it is unclear if autoantibody number directly alters phenotype (13,17,18). In the context of a clinical T1D diagnosis (and therefore high pretest probability of T1D), a single positive autoantibody will usually confirm T1D (10). However, clinical presentation cannot be studied, as these features will have influenced classification.

We aimed to determine whether the initial presentation and progression of adult-onset T1D defined using high-specificity definitions incorporating islet autoantibodies are altered by age of diagnosis.

We used longitudinal data from the prospective StartRight study (ClinicalTrials.gov NCT03737799) to evaluate the impact of age at diagnosis on the presentation characteristics and progression of adult-onset T1D in a U.K. population cohort defined by the presence of two or more positive islet autoantibodies (GAD, IA-2, and ZnT8). As a separate secondary comparison, progression was assessed when adult-onset T1D was defined by a single positive autoantibody in the context of a clinical diagnosis of T1D. The StartRight study was approved by the South West–Cornwall and Plymouth NHS Research Ethics Committee on 6 June 2016 (reference: 16/SW/0130).

Study Participants

The prospective StartRight study was a multicenter study across 55 sites in the U.K. that recruited 1,798 adults (≥18 years of age) with diabetes onset within the previous 12 months.

Exclusion criteria for the StartRight study included gestational and known secondary diabetes. For analysis, cases for which autoantibody results were missing were also excluded (n = 5). To ensure sufficient numbers of participants with late-onset T1D, the study was enriched for older adults receiving insulin treatment by aiming in those diagnosed after 50 years of age for equal recruitment at the site level of those treated with and without insulin. A study flow diagram is shown in Supplementary Fig. 1.

Diabetes Definitions

For the primary analysis, T1D was defined as the presence of two or more positive autoantibodies of GAD, ZnT8, or IA-2, regardless of clinical diagnosis. A secondary analysis was performed in participants with T1D defined by a clinical diagnosis of T1D (T1D reported as clinical diagnosis and insulin within 2 weeks of diagnosis) and positive for a single autoantibody (one of GAD, ZnT8, or IA-2).

A comparison population of T2D cases was defined by having all of: a self-reported clinical diagnosis of T2D, absence of insulin treatment within 2 weeks of diagnosis, and negative autoantibodies.

Data Collection

Presentation characteristics were self-reported at the baseline study visit (median duration 5 months), including: unintentional weight loss, osmotic symptoms (nocturia, polyuria, and thirst), hospitalization, and initial treatment. Presentation glycemia (glucose and HbA1c) and ketoacidosis were determined by reviewing participants’ medical notes and laboratory records. Diabetic ketoacidosis (DKA) was defined based on the Joint British Diabetes Societies guidelines (19): a pH <7.3 and either capillary β-hydroxybutyrate >3.0 mmol/L or ketonuria ≥2+ on standard urine sticks. In the absence of an available pH measurement, cases were included as DKA if DKA was recorded in the hospital notes alongside a supportive blood or urine ketone value as above. Participants not admitted to hospital were assumed not to have had DKA.

At the baseline visit, height and weight were assessed for BMI calculation, and a nonfasted (within 1–5 h of a meal) blood sample was collected for: DNA extraction, autoantibodies (GAD, IA-2, and ZnT8), plasma C-peptide, and paired glucose.

At each visit including baseline, participants collected a boric acid urine sample, within 1 and 5 h of a meal, for urine C-peptide–creatinine ratio (UCPCR) measurement (20,21). Samples were posted by participants directly to the Exeter Clinical Laboratory for analysis. Samples marked as not received within 7 days of collection were excluded from analysis (1.2% of all UCPCR samples [42 of 3,454]). UCPCR samples were aimed to be collected for year 1 follow-up between 10 and 16 months of baseline visit and 2-year follow-up between 22 and 28 months of baseline visit. Due to the coronavirus disease 2019 pandemic, 6% (63 of 1,136) of year 1 and 10% (93 of 896) of year 2 results were delayed and collected outside of these ranges. UCPCR time from recruitment was therefore calculated and evaluated in 6-month bins: 12 months (within range of 9 to <15 month), 18 months (15 to <21 months), 24 months (21 to <27 months), and 30 months (27 to <33 months). For all participants, median follow-up time from recruitment was 25 months (interquartile range 24–28 months).

Laboratory Analysis

Analysis of C-peptide and autoantibodies (GAD antibody [GADA], IA-2 antigen [IA-2A], and ZnT8 autoantibody [ZnT8A]) was performed by the academic Blood Sciences Department at the Royal Devon and Exeter Hospital. GAD, IA-2, and ZnT8 autoantibodies were measured using ELISAs (RSR Limited, Cardiff, U.K.) on a DYNEX DS2 Automated ELISA system (Launch Diagnostics, Longfield, U.K.). Autoantibodies were considered positive if ≥97.5th centile of 1,559 control subjects without diabetes (GAD, ≥11 World Health Organization units/mL; IA-2, ≥7.5 units/mL; and ZnT8, ≥65 and ≥10 units/mL, <30 and ≥30 years of age, respectively) (22). Specificity for all three assays was 99% in the 2020 International Islet Autoantibody Standardization Program Exeter Laboratory certification, with sensitivity of 74% for both GADA and ZnT8A and 72% for IA-2A.

C-peptide was measured using an electrochemiluminescence immunoassay on a Roche Diagnostics E170 analyzer (Roche, Mannheim, Germany; limit of detection 3.3 pmol/L; inter- and intra-assay coefficients of variation <4.5% and <3.3%, respectively). Blood C-peptide results were excluded if concurrent glucose was <4 mmol/mol (n = 46). Urine creatinine (for UCPCR) was analyzed using the Jaffe method on the Roche P800 modular analyzer.

Assessment of T1D Genetic Risk Score

A T1D genetic risk score (GRS) was calculated using 67 published variants known to be associated with T1D, as described in the Supplementary Materials and Methods and previously (23,24)

Statistical Analysis

The impact of age on T1D GRS and annual change in UCPCR was evaluated continuously using linear regression and mixed-effects models, respectively (see below). To further evaluate the impact of age on C-peptide loss and T1D genetic susceptibility and assess impact on presentation characteristics, all participants were split by the median age of diagnosis of those with two or more positive autoantibodies (young adult onset ≤35 years or older adult onset >35 years of age). Presentation features were not evaluated where clinical diagnosis was included in the definition of diabetes type, due to the likely impact of selection bias. The χ2 test was used to compare categorical variables between age-groups, and Student t test was used for continuous variables.

Continuous data were assessed visually for distribution and other than C-peptide were normally distributed. C-peptide and UCPCR were highly skewed and therefore log transformed in line with previous studies (3,25,26), with geometric mean and 95% CIs presented. Mixed-effects models were used to determine the percentage annual change in UCPCR with random effects at the patient level to allow each patient to contribute multiple C-peptide values at different 6-month time points (25). The impact of diagnosis age on the association of change in C-peptide over time was evaluated in T1D, in both continuous analysis and subgroups using an interaction term (diagnosis before and after 35 years of age as above). A random-intercept, random-slope model allows for variability between individuals in terms of both C-peptide at diagnosis (the intercept) and percentage change in C-peptide over time (the slope).

As the study enriched for initial insulin treatment in the group diagnosed at >50 years of age, we performed a sensitivity analysis in those with two or more positive autoantibodies, comparing progression and T1D GRS between participants receiving and not receiving insulin within 2 weeks of diagnosis.

All analyses were performed using Stata 16 (StataCorp LP, College Station, TX).

Data and Resource Availability

Data are available by contacting the corresponding author via e-mail.

Participant characteristics of those with T1D defined by two or more islet autoantibodies (n = 385) are shown in Table 1. In those positive for two autoantibodies diagnosed after (n = 193) or at and before (n = 192) the median onset age of 35 years, the mean age of diagnosis was 50 years (95% CI 49–51) and 26 years (25–27), respectively. For those diagnosed after and at and before 35 years of age with T1D defined by a clinical diagnosis and one positive autoantibody (n = 180), the mean age of onset was 50 years (48–52) and 28 years (27–29), respectively. The characteristics of all 1,793 participants from the StartRight study, including autoantibody results, are shown by age-group in Supplementary Table 1.

Table 1

Clinical characteristics of those defined as having T1D based on the presence of two or more positive islet autoantibodies split by median age of diagnosis

Diabetes diagnosed at ≤35 years of age (n = 192)Diabetes diagnosed at >35 years of age (n = 193)P
Baseline characteristics    
 Age at diagnosis (years), mean (95% CI) 25.9 (25.1–26.6) 49.9 (48.5–51.4) <0.0001 
 Duration at recruitment (months), mean (95% CI) 4.2 (3.7–4.8) 4.9 (4.4–5.5) 0.06 
 BMI (kg/m2), mean (95% CI) 24.8 (24.2–25.5) 25.6 (25.0–26.3) 0.08 
 Sex (male) 53 (46–60) 52 (45–59) 0.9 
 White European ethnicity 92 (88–96) 91 (87–95) 0.7 
Symptoms at presentation    
 Unintentional weight loss 80 (74–85) 82 (76–87) 0.6 
 DKA 24 (18–30) 19 (14–25) 0.3 
 Osmotic symptoms 96 (93–99) 92 (88–96) 0.09 
Biochemistry at presentation, mean (95% CI)    
 HbA1c at diagnosis (mmol/mol) 103.0 (99.0–107.0) 103.2 (99.5–106.9) 0.96 
 HbA1c at diagnosis (%) 11.6 (11.2–11.9) 11.6 (11.2–11.9) 0.96 
 Glucose at diagnosis (mmol/L) 20.6 (19.1–22.1) 20.9 (19.5–22.4) 0.8 
Management at presentation    
 Hospitalized at admission 60 (53–67) 40 (33–47) <0.0001 
 Initial insulin 93 (90–97) 73 (67–79) <0.0001 
 Initial tablets (with or without insulin) 9 (5–13) 31 (25–38) <0.0001 
Baseline visit characteristics    
 Stimulated geometric C-peptide (pmol/L), mean (95% CI) 434.5 (388.7–480.3) 430.1 (356.4–503.7) 0.9 
 Nonfasted UCPCR (nmol/mmol), mean (95% CI) 1.0 (0.8–1.2) 0.9 (0.8–1.1) 0.6 
 Severe insulin deficiency* 9 (5–13) 11 (6–15) 0.5 
 Reported type 1 96 (94–99) 87 (82–91) 0.001 
 Reported type 2 2 (0–3) 7 (3–10) 0.01 
 Insulin treatment 97 (95–100) 90 (86–94) <0.01 
 T1D GRS, mean (95% CI) 13.0 (12.8–13.3) 12.9 (12.7–13.2) 0.5 
 GADA 95 (92–98) 97 (94–99) 0.3 
 IA-2A 76 (69–82) 75 (69–81) 0.9 
 ZnT8 82 (77–88) 85 (80–90) 0.4 
Diabetes diagnosed at ≤35 years of age (n = 192)Diabetes diagnosed at >35 years of age (n = 193)P
Baseline characteristics    
 Age at diagnosis (years), mean (95% CI) 25.9 (25.1–26.6) 49.9 (48.5–51.4) <0.0001 
 Duration at recruitment (months), mean (95% CI) 4.2 (3.7–4.8) 4.9 (4.4–5.5) 0.06 
 BMI (kg/m2), mean (95% CI) 24.8 (24.2–25.5) 25.6 (25.0–26.3) 0.08 
 Sex (male) 53 (46–60) 52 (45–59) 0.9 
 White European ethnicity 92 (88–96) 91 (87–95) 0.7 
Symptoms at presentation    
 Unintentional weight loss 80 (74–85) 82 (76–87) 0.6 
 DKA 24 (18–30) 19 (14–25) 0.3 
 Osmotic symptoms 96 (93–99) 92 (88–96) 0.09 
Biochemistry at presentation, mean (95% CI)    
 HbA1c at diagnosis (mmol/mol) 103.0 (99.0–107.0) 103.2 (99.5–106.9) 0.96 
 HbA1c at diagnosis (%) 11.6 (11.2–11.9) 11.6 (11.2–11.9) 0.96 
 Glucose at diagnosis (mmol/L) 20.6 (19.1–22.1) 20.9 (19.5–22.4) 0.8 
Management at presentation    
 Hospitalized at admission 60 (53–67) 40 (33–47) <0.0001 
 Initial insulin 93 (90–97) 73 (67–79) <0.0001 
 Initial tablets (with or without insulin) 9 (5–13) 31 (25–38) <0.0001 
Baseline visit characteristics    
 Stimulated geometric C-peptide (pmol/L), mean (95% CI) 434.5 (388.7–480.3) 430.1 (356.4–503.7) 0.9 
 Nonfasted UCPCR (nmol/mmol), mean (95% CI) 1.0 (0.8–1.2) 0.9 (0.8–1.1) 0.6 
 Severe insulin deficiency* 9 (5–13) 11 (6–15) 0.5 
 Reported type 1 96 (94–99) 87 (82–91) 0.001 
 Reported type 2 2 (0–3) 7 (3–10) 0.01 
 Insulin treatment 97 (95–100) 90 (86–94) <0.01 
 T1D GRS, mean (95% CI) 13.0 (12.8–13.3) 12.9 (12.7–13.2) 0.5 
 GADA 95 (92–98) 97 (94–99) 0.3 
 IA-2A 76 (69–82) 75 (69–81) 0.9 
 ZnT8 82 (77–88) 85 (80–90) 0.4 

Data are percentage (95% CI) for binary outcomes unless otherwise indicated for continuous data.

*

Severe insulin deficiency defined as C-peptide <200 pmol/L.

Age of Onset Is Not Associated With Genetic Susceptibility (T1D GRS) to Adult-Onset T1D

In adult-onset T1D defined by multiautoantibody positivity, annual increase in onset age using linear regression had no effect on T1D GRS (β = −0.01 [95% CI −0.02 to 0.003]; P = 0.1). A lack of association between age of onset and T1D GRS was also seen when T1D was defined by clinical T1D diagnosis and one positive autoantibody (β = −0.02 [−0.04 to 0.008]; P = 0.2). Between participants diagnosed before and after 35 years of age, mean T1D GRS was similar irrespective of the definition of T1D used: multiautoantibody-positive, 13.0 (12.8–13.3) vs. 12.9 (12.7–13.2) (P = 0.5); and a clinical T1D diagnosis confirmed by a single autoantibody, 13.3 (12.8–13.8) (n = 83) and 13.1 (12.6–13.5) (n = 97) (P = 0.5) (Fig. 1A). T1D GRS was significantly higher in both age-groups than a comparison group with autoantibody-negative T2D, 10.1 (9.9–10.3), regardless of how T1D was defined (all P < 0.0001) (Fig. 1A).

Figure 1

Comparison of T1D genetic susceptibility (T1D GRS) (A) and recruitment blood C-peptide (B) in participants with T1D defined by both study definitions, aged ≤35 and >35 years at diabetes diagnosis. Horizontal lines represent the mean, and error bars represent 95% CIs. Comparisons by age-group are shown for each definition and across all definitions and age-groups.

Figure 1

Comparison of T1D genetic susceptibility (T1D GRS) (A) and recruitment blood C-peptide (B) in participants with T1D defined by both study definitions, aged ≤35 and >35 years at diabetes diagnosis. Horizontal lines represent the mean, and error bars represent 95% CIs. Comparisons by age-group are shown for each definition and across all definitions and age-groups.

Close modal

Age of Onset Is Not Associated With Progression of C-Peptide Loss in Adult-Onset T1D

Increasing onset age (years), evaluated continuously using mixed-effects models, had no effect on annual log C-peptide rate of decline in those with two or more positive antibodies (β = −0.002 [95% CI −0.005 to 0.001]; P = 0.3) or a clinical diagnosis confirmed by a single positive autoantibody (β = 0.002 [−0.003 to 0.007]; P = 0.4). Supplementary Fig. 2 shows the percentage annual change in C-peptide by decile of onset age in those classified as T1D.

For both T1D definitions, both age-groups had comparable baseline C-peptide. For multiautoantibody-positive cases (>35 years of age-groups shown first), geometric mean stimulated plasma C-peptide was 430 pmol/L (356–504) vs. 435 pmol/L (389–480) (Fig. 1B), and postmeal UCPCR was 0.9 nmol/mmol (0.8–1.1) vs. 1.0 nmol/mmol (0.8–1.2) (both P > 0.1). For clinician-diagnosed cases confirmed by a positive autoantibody, geometric mean stimulated plasma C-peptide was 430 pmol/L (332–526) vs. 414 pmol/L (335–493), and postmeal UCPCR was 1.0 nmol/mmol (0.7–1.3) vs. 0.9 nmol/mmol (0.6–1.2) (both P > 0.1).

The percentage annual decline in UCPCR in those aged ≤35 and >35 years at diagnosis was similar regardless of T1D definition. In those with multiautoantibody positivity diagnosed at >35 years, UCPCR declined by 44% (38–50) annually, compared with a 39% (31–46) annual decline in those diagnosed at ≤35 years (P = 0.2) (Fig. 2A). This corresponds to a half-life for C-peptide loss of 1.2 (1.0–1.4) and 1.4 years (1.1–1.9), respectively. In single autoantibody–positive cases with a clinician diagnosis of T1D, the percentage annual decline in UCPCR in those diagnosed at >35 years was 39% (31–46) and those diagnosed at ≤35 years was 43% (33–51) (P = 0.6) (Fig. 2B). In the T2D comparison group, the annual C-peptide decline was 6% (1–11), corresponding to a half-life of 11.1 years (5.9–98.3).

Figure 2

Progression of change in log UCPCR from mixed-effects models with age as an interaction term for T1D defined by multiantibody (more than two) positivity (A) and a single positive antibody (B) in the context of a clinical diagnosis of T1D. Error bars represent 95% CIs. Comparison T2D group shown for reference.

Figure 2

Progression of change in log UCPCR from mixed-effects models with age as an interaction term for T1D defined by multiantibody (more than two) positivity (A) and a single positive antibody (B) in the context of a clinical diagnosis of T1D. Error bars represent 95% CIs. Comparison T2D group shown for reference.

Close modal

Presentation of Adult-Onset T1D Defined by Multiautoantibody Positivity Is Similar at >35 and ≤35 Years of Age

In those positive for two or more autoantibodies, management at presentation was strikingly different between age-groups. Those diagnosed at >35 years (n = 193) were far less likely than those diagnosed at ≤35 years (n = 192) to report: being admitted at diagnosis, 40% (95% CI 33–47) vs. 60% (53–67); being treated with insulin at diagnosis, 73% (67–79) vs. 93% (90–97); or a diagnosis of T1D at recruitment, 87% (82–91) vs. 96% (94–99) (all P < 0.001) (Table 1). These differences in clinical management and diagnosis did not reflect differences in phenotype or clinical presentation, which were similar between age-groups. At presentation (>35 years shown first): HbA1c was 103 mmol/mol (95% CI 100–107) vs. 103 mmol/mol (99–107); glucose was 21 mmol/L (20–22) vs. 21 mmol/L (19–22); and recruitment BMI was 26 kg/m2 (25–26) vs. 25 kg/m2 (24–26) (all P > 0.05). Symptoms at presentation of those with two or more positive antibodies were also similar between those diagnosed at >35 and ≤35 years of age: reported unintentional weight loss, 82% (76–87) vs. 80% (74–85); DKA, 19% (14–25) vs. 24% (18–30); and osmotic symptoms, 92% (88–96) vs. 96% (93–99) (all P > 0.05). Characteristics of those with T1D (by either definition, n = 565) in comparison with antibody-negative T2D are shown in Supplementary Table 2. At 2 years’ follow-up, 88% (57 of 65) of multiautoantibody-positive cases not insulin treated at diagnosis had started insulin treatment vs. 3% (24 of 715) of those with T2D.

In Multiautoantibody-Positive Participants, Genetic Susceptibility and Progression Were Not Associated With Initial Insulin Treatment

In those with T1D defined by two or more positive autoantibodies who received insulin treatment within 2 weeks of diagnosis (n = 320) and those initially treated without insulin (n = 65), decline in UCPCR was near identical: in those insulin-treated at diagnosis, UCPCR declined by 44% (95% CI 37–50) yearly in comparison with 41% (26–53) (P = 0.7) in those not receiving insulin at diagnosis (Supplementary Fig. 3). This was despite recruitment C-peptide being significantly lower in those initially insulin-treated compared with those who were not: stimulated plasma C-peptide, 407 (366–447) vs. 581 pmol/L (390–772); and postmeal UCPCR, 0.9 (0.8–1.0) vs. 1.4 nmol/mmol (0.9–1.8) (both P < 0.05). T1D GRS was similar between the two treatment groups: insulin-treated at diagnosis, 13.0 (12.8–13.2) vs. no insulin at diagnosis, 12.9 (12.5–13.3) (P = 0.7) (Supplementary Fig. 4).

The Number of Positive Autoantibodies Does Not Alter the Progression or Genetic Characteristics of Robustly Defined Adult-Onset T1D

In all participants with a clinical diagnosis of T1D, we evaluated C-peptide progression and T1D GRS by antibody number. Neither differed between those positive for one autoantibody (n = 180) compared with those with two or more positive autoantibodies (n = 307): T1D GRS was 13.2 (95% CI 12.9–13.5) vs. 13.0 (12.8–13.2) (P = 0.3), and annual loss of C-peptide was 41% (32–48) vs. 44% (37–50) (P = 0.5) (Supplementary Fig. 5).

In Multiautoantibody-Positive Cases, BMI Was Not Associated With Progression or Genetic Characteristics

In T1D defined by multiple positive islet autoantibodies, BMI (assessed continuously) was not associated with presentation of DKA, osmotic symptoms, diagnosis of HbA1c, T1D GRS, or annual loss of C-peptide (all P > 0.1) (Supplementary Table 4). However, higher BMI was associated with higher C-peptide at diagnosis (P < 0.0001).

This study demonstrates that when adult-onset T1D is classified using a high-specificity definition, presentation, progression, and genetic predisposition for T1D are similar across all onset ages. Participants defined as having T1D with multiple positive autoantibodies, irrespective of reported clinical diagnosis, showed near-identical clinical characteristics and marked levels of dysglycemia when diagnosed at >35 and ≤35 years of age. However, despite similar presentation, there was substantial variation in initial clinical management: older patients less likely to be diagnosed with T1D, initially treated with insulin, or admitted to hospital. Regardless of T1D being defined by multiautoantibody positivity or a single positive autoantibody confirming a clinical T1D diagnosis, progression of C-peptide loss in adults was marked (∼40% annual C-peptide loss) and unaffected by diagnosis age.

A key strength of this study is that we used high-specificity biomarker-based definitions to evaluate adult-onset T1D clinical characteristics at presentation in a large mixed cohort and prospectively followed participants to evaluate early change in endogenous insulin secretion. Use of high-specificity definitions of T1D is important, as performing this analysis in those selected solely by clinical diagnosis could suffer from inadvertently including non-T1D. This would bias results toward falsely low progression and T1D genetic susceptibility in older adults, in whom misclassification is more common (8,9,13). The high specificity of T1D definitions in our study is supported by near-identical high genetic predisposition to T1D and progression irrespective of participants being defined by multiautoantibody positivity or a clinical diagnosis with a confirmatory single autoantibody. This is despite these definitions capturing entirely separate participants and 13% of older adults with multiautoantibody positivity not reporting a diagnosis of T1D at recruitment and 27% not receiving insulin within 2 weeks of diagnosis. The high specificity of our T1D definitions has also been previously demonstrated, with very low prevalence of multiple positive islet antibodies in populations without diabetes (27) and a single positive antibody in the context of a clinical diagnosis shown mathematically and using genetic approaches to confirm autoimmune diabetes in older adults (10,13,17).

A limitation of our study is that our study cohort is enriched for early insulin treatment in older participants. A total of 15% of participants diagnosed at >35 years of age were multiautoantibody positive, far higher than the reported ∼5% of this age-group with diabetes having T1D (28). Enriching for insulin treatment could have selected a more rapidly progressive older T1D cohort. Reassuringly, our sensitivity analysis within T1D cases defined by multiautoantibody positivity showed identical progression with and without initial insulin treatment. This enrichment also means that the 27% of older adults with two or more positive autoantibodies not initially insulin-treated will be an underestimate, consistent with higher proportions reported in other studies (9,29). We only assessed early progression of C-peptide loss, and further studies will be needed to determine whether long-term residual endogenous insulin secretion is altered by diagnosis age in robustly defined adult-onset T1D. Our specific definitions of T1D might pick out a more severe T1D phenotype in adults; however, reassuringly, progression and T1D GRS were identical irrespective of the number of autoantibodies included in the definition of T1D.

Our results are different from previous studies of adults using different approaches to defining diabetes of autoimmune etiology. When autoimmune diabetes is defined by a clinical diagnosis of T2D, lack of initial insulin treatment, and one or more positive autoantibodies (LADA), genetic characteristics and progression are, on average, intermediate between classical T1D and T2D (12,30). In adults with T1D defined by clinical diagnosis alone, there appears to be a modest reduction in both C-peptide (assessed cross-sectionally) (24) and T1D genetic susceptibility with increasing diagnosis age (2). This difference in phenotype, and progression of β-cell loss, observed by our group compared with previous studies can be explained by the specificity of T1D definition used (13). T2D is extremely common in older adults; this low prior prevalence means, in adults, a single positive antibody test or clinical diagnosis alone may not confirm diabetes of autoimmune etiology, resulting in the study of a mixed population of autoimmune and nonautoimmune diabetes, with a higher proportion of nonautoimmune diabetes as age increases (10,13,31). This is supported by previous research showing adults with autoantibody-negative clinician-diagnosed T1D have T1D genetic susceptibility and C-peptide loss intermediate to T1D and T2D and by research showing the relationship between characteristics (genotype and phenotype) and antibody titer and number (which impact test specificity) seen in LADA appear absent or modest in the setting of a high prior likelihood of T1D (10,18,32).

To date, studies of progression of C-peptide loss in T1D have predominantly focused on those with childhood onset. These have shown that within children, rates of C-peptide decline are fairly consistent across different onset ages, although C-peptide levels close to diagnosis are lower in younger children (3,5,25,26). In our adult study, C-peptide level close to diagnosis was unaffected by onset age, fitting with studies showing minimal differences in C-peptide close to diagnosis between older children (>10 years) and adults (5). In studies of progression, including robustly defined adult-onset T1D cases, progression is slower relative to childhood-onset cases, but the low number of adults included, mainly ≤45 years of age at onset, means evaluating the impact of age on progression within adult-onset T1D has not been possible (5,6,26). Larger studies in adults have assessed clinician-diagnosed T1D that may include nonautoimmune cases not associated with marked C-peptide loss (3). Childhood-onset cases were not recruited in the StartRight study, but the estimated ∼50% annual C-peptide loss seen in previous childhood-onset studies is only modestly higher than the ∼40% annual loss seen in our study (3,5,6,25,26). However, progression may be faster in very young children with histopathological studies, suggesting a more rapid loss of β-cells in those who developed T1D before 7 years of age (33,34), consistent with age-related immune differences at the level of the β-cell (35). Interestingly, in studies evaluating the development of T1D in multiantibody-positive children and young adults, in those developing T1D, age of diagnosis does not appear to affect prediagnosis progression rates (36,37).

Our results have implications for the clinical management and study of adult-onset T1D. We demonstrate that late-onset T1D remains rapidly progressive, even if insulin is not needed at diagnosis, highlighting that absence of initial insulin requirement does not exclude T1D. Our findings further emphasize the high prevalence of misdiagnosis of adult-onset T1D and support recent American Diabetes Association and European Association for the Study of Diabetes guidance advising a single positive antibody confirms T1D where this is clinically suspected (15). Older adults have largely been excluded from studies of interventions to preserve β-cell function. Our findings of ∼40% annual loss in C-peptide in adults irrespective of onset age support the potential inclusion of all adults in intervention studies in which T1D is robustly defined. Importantly, some people with true autoimmune diabetes may not meet the definitions of T1D used in this study. In those with uncertain diabetes type, further research (e.g., the use of advanced antibody assays or combining clinical, antibody, and genetic information) is required to help improve classification close to diagnosis (38,39).

In summary, our findings suggest that when adult-onset T1D is robustly defined, the presentation characteristics, progression, and T1D genetic susceptibility of adult-onset T1D are not altered by age of onset.

See accompanying article, p. 1135.

This article contains supplementary material online at https://doi.org/10.2337/figshare.21932247.

This article is featured in a podcast available at diabetesjournals.org/care/pages/diabetes_care_on_air.

*

A complete list of the members of the StartRight Study Group can be found in the supplementary material online.

Funding. The StartRight study is funded by Diabetes UK (17/0005624) and the National Institute for Health and Care Research (NIHR; CS-2015-15-018). Genotyping for generation of the T1D GRS was supported by the European Foundation for the Study of Diabetes (2016 Rising Star Fellowship). N.J.T. is funded by a Wellcome Trust–funded GW4 PhD. A.G.J. was supported by an NIHR Clinician Scientist award (CS-2015-15-018). A.V.H. is supported by the NIHR Exeter Clinical Research Facility and Diabetes UK. R.A.O. is supported by a Diabetes UK Harry Keen Fellowship (16/0005529).

This study was supported by the NIHR Exeter Biomedical Research Centre and NIHR Exeter Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

Duality of Interest. R.A.O. is a co-investigator on a Randox Laboratories R&D research grant. The study has received translational industry academic funding from Randox Laboratories R&D relating to autoimmune GRS for prediction and classification of disease. There are no established patents, loyalties, or licensing agreements relating to this grant. It is a 3-year grant (February 2022–2025). The approximate value is a £2.2 million program grant on GRS across autoimmune disease. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. N.J.T. and A.G.J. designed the study. A.V.H. and A.G.J. researched the data. N.J.T. and A.G.J. analyzed the data with assistance from B.M.S. N.J.T. wrote the first draft of the report. All authors provided helpful discussion and reviewed and edited the manuscript. A.G.J. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented at the Diabetes UK Professional Conference, London, U.K., 29 March–1 April 2022, and the 58th Annual Meeting of the European Association for the Study of Diabetes, Stockholm, Sweden, 19–23 September 2022.

1.
Leslie
RD
,
Evans-Molina
C
,
Freund-Brown
J
, et al
.
Adult-onset type 1 diabetes: current understanding and challenges
.
Diabetes Care
2021
;
44
:
2449
2456
2.
McKeigue
PM
,
Spiliopoulou
A
,
McGurnaghan
S
, et al
.
Persistent C-peptide secretion in type 1 diabetes and its relationship to the genetic architecture of diabetes
.
BMC Med
2019
;
17
:
165
3.
Barker
A
,
Lauria
A
,
Schloot
N
, et al
.
Age-dependent decline of β-cell function in type 1 diabetes after diagnosis: a multi-centre longitudinal study
.
Diabetes Obes Metab
2014
;
16
:
262
267
4.
Davis
AK
,
DuBose
SN
,
Haller
MJ
, et al.;
T1D Exchange Clinic Network
.
Prevalence of detectable C-peptide according to age at diagnosis and duration of type 1 diabetes
.
Diabetes Care
2015
;
38
:
476
481
5.
Greenbaum
CJ
,
Beam
CA
,
Boulware
D
, et al.;
Type 1 Diabetes TrialNet Study Group
.
Fall in C-peptide during first 2 years from diagnosis: evidence of at least two distinct phases from composite Type 1 Diabetes TrialNet data
.
Diabetes
2012
;
61
:
2066
2073
6.
Hao
W
,
Gitelman
S
,
DiMeglio
LA
,
Boulware
D
;
Type 1 Diabetes TrialNet Study Group
.
Fall in C-peptide during first 4 years from diagnosis of type 1 diabetes: variable relation to age, HbA1c, and insulin dose
.
Diabetes Care
2016
;
39
:
1664
1670
7.
Hope
SV
,
Wienand-Barnett
S
,
Shepherd
M
, et al
.
Practical Classification Guidelines for Diabetes in patients treated with insulin: a cross-sectional study of the accuracy of diabetes diagnosis
.
Br J Gen Pract
2016
;
66
:
e315
e322
8.
Foteinopoulou
E
,
Clarke
CAL
,
Pattenden
RJ
, et al
.
Impact of routine clinic measurement of serum C-peptide in people with a clinician-diagnosis of type 1 diabetes
.
Diabet Med
2021
;
38
:
e14449
9.
Thomas
NJ
,
Lynam
AL
,
Hill
AV
, et al
.
Type 1 diabetes defined by severe insulin deficiency occurs after 30 years of age and is commonly treated as type 2 diabetes
.
Diabetologia
2019
;
62
:
1167
1172
10.
Thomas
NJ
,
Walkey
HC
,
Kaur
A
, et al
.
The relationship between islet autoantibody status and the genetic risk of type 1 diabetes in adult-onset type 1 diabetes
.
Diabetalogia
2023
;
66
:
310
320
11.
Shields
BM
,
Peters
JL
,
Cooper
C
, et al
.
Can clinical features be used to differentiate type 1 from type 2 diabetes? A systematic review of the literature
.
BMJ Open
2015
;
5
:
e009088
12.
Buzzetti
R
,
Tuomi
T
,
Mauricio
D
, et al
.
Management of latent autoimmune diabetes in adults: a consensus statement from an international expert panel
.
Diabetes
2020
;
69
:
2037
2047
13.
Jones
AG
,
McDonald
TJ
,
Shields
BM
,
Hagopian
W
,
Hattersley
AT
.
Latent autoimmune diabetes of adults (LADA) is likely to represent a mixed population of autoimmune (type 1) and nonautoimmune (type 2) diabetes
.
Diabetes Care
2021
;
44
:
1243
1251
14.
Jones
AG
,
Hattersley
AT
.
The clinical utility of C-peptide measurement in the care of patients with diabetes
.
Diabet Med
2013
;
30
:
803
817
15.
Holt
RIG
,
DeVries
JH
,
Hess-Fischl
A
, et al
.
The management of type 1 diabetes in adults. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD)
.
Diabetes Care
2021
;
44
:
2589
2625
16.
Rogowicz-Frontczak
A
,
Pilacinski
S
,
Wyka
K
,
Wierusz-Wysocka
B
,
Zozulinska-Ziolkiewicz
D
.
Zinc transporter 8 autoantibodies (ZnT8-ab) are associated with higher prevalence of multiple diabetes-related autoantibodies in adults with type 1 diabetes
.
Diabetes Res Clin Pract
2018
;
146
:
313
320
17.
Bingley
PJ
.
Clinical applications of diabetes antibody testing
.
J Clin Endocrinol Metab
2010
;
95
:
25
33
18.
Redondo
MJ
,
Sosenko
J
,
Libman
I
, et al
.
Single islet autoantibody at diagnosis of clinical type 1 diabetes is associated with older age and insulin resistance
.
J Clin Endocrinol Metab
2020
;
105
:
1629
1640
19.
Savage
MW
,
Dhatariya
KK
,
Kilvert
A
, et al.;
Joint British Diabetes Societies
.
Joint British Diabetes Societies guideline for the management of diabetic ketoacidosis
.
Diabet Med
2011
;
28
:
508
515
20.
Besser
RE
,
Ludvigsson
J
,
Jones
AG
, et al
.
Urine C-peptide creatinine ratio is a noninvasive alternative to the mixed-meal tolerance test in children and adults with type 1 diabetes
.
Diabetes Care
2011
;
34
:
607
609
21.
Jones
AG
,
Besser
RE
,
McDonald
TJ
, et al
.
Urine C-peptide creatinine ratio is an alternative to stimulated serum C-peptide measurement in late-onset, insulin-treated diabetes
.
Diabet Med
2011
;
28
:
1034
1038
22.
Grace
SL
,
Cooper
A
,
Jones
AG
,
McDonald
TJ
.
Zinc transporter 8 autoantibody testing requires age-related cut-offs
.
BMJ Open Diabetes Res Care
2021
;
9
:
e002296
23.
Oram
RA
,
Sharp
SA
,
Pihoker
C
, et al
.
Utility of diabetes type-specific genetic risk scores for the classification of diabetes type among multiethnic youth
.
Diabetes Care
2022
;
45
:
1124
1131
24.
Sharp
SA
,
Rich
SS
,
Wood
AR
, et al
.
Development and standardization of an improved type 1 diabetes genetic risk score for use in newborn screening and incident diagnosis
.
Diabetes Care
2019
;
42
:
200
207
25.
Shields
BM
,
McDonald
TJ
,
Oram
R
, et al.;
TIGI Consortium
.
C-peptide decline in type 1 diabetes has two phases: an initial exponential fall and a subsequent stable phase
.
Diabetes Care
2018
;
41
:
1486
1492
26.
Dabelea
D
,
Mayer-Davis
EJ
,
Andrews
JS
, et al
.
Clinical evolution of beta cell function in youth with diabetes: the SEARCH for Diabetes in Youth study
.
Diabetologia
2012
;
55
:
3359
3368
27.
Vermeulen
I
,
Weets
I
,
Asanghanwa
M
, et al.;
Belgian Diabetes Registry
.
Contribution of antibodies against IA-2β and zinc transporter 8 to classification of diabetes diagnosed under 40 years of age
.
Diabetes Care
2011
;
34
:
1760
1765
28.
Scottish Diabetes Survey Group
.
Scottish Diabetes Survey 2020, 2020
.
29.
Munoz
C
,
Floreen
A
,
Garey
C
,
Karlya
T
,
Jelley
D
,
Alonso
GT
,
McAuliffe-Fogarty
A
.
Misdiagnosis and diabetic ketoacidosis at diagnosis of type 1 diabetes: patient and caregiver perspectives
.
Clin Diabetes
2019
;
37
:
276
281
30.
Mishra
R
,
Chesi
A
,
Cousminer
DL
, et al.;
Bone Mineral Density in Childhood Study
.
Relative contribution of type 1 and type 2 diabetes loci to the genetic etiology of adult-onset, non-insulin-requiring autoimmune diabetes
.
BMC Med
2017
;
15
:
88
31.
Steck
AK
,
Eisenbarth
GS
.
Genetic similarities between latent autoimmune diabetes and type 1 and type 2 diabetes
.
Diabetes
2008
;
57
:
1160
1162
32.
Eason
RJ
,
Thomas
NJ
,
Hill
AV
, et al.;
StartRight Study Group
;
StartRight Study Group
.
Routine islet autoantibody testing in clinically diagnosed adult-onset type 1 diabetes can help identify misclassification and the possibility of successful insulin cessation
.
Diabetes Care
2022
;
45
:
2844
2851
33.
Gepts
W
.
Pathologic anatomy of the pancreas in juvenile diabetes mellitus
.
Diabetes
1965
;
14
:
619
633
34.
Pipeleers
D
,
Ling
Z
.
Pancreatic beta cells in insulin-dependent diabetes
.
Diabetes Metab Rev
1992
;
8
:
209
227
35.
Leete
P
,
Willcox
A
,
Krogvold
L
, et al
.
Differential insulitic profiles determine the extent of β-cell destruction and the age at onset of type 1 diabetes
.
Diabetes
2016
;
65
:
1362
1369
36.
Gorus
FK
,
Balti
EV
,
Messaaoui
A
, et al.;
Belgian Diabetes Registry
.
Twenty-year progression rate to clinical onset according to autoantibody profile, age, and HLA-DQ genotype in a registry-based group of children and adults with a first-degree relative with type 1 diabetes
.
Diabetes Care
2017
;
40
:
1065
1072
37.
Ferrannini
E
,
Mari
A
,
Monaco
GSF
,
Skyler
JS
,
Evans-Molina
C
.
The effect of age on longitudinal measures of beta cell function and insulin sensitivity during the progression of early stage type 1 diabetes
.
Diabetologia
2023
;
66
:
508
519
38.
Achenbach
P
,
Hawa
MI
,
Krause
S
, et al.;
Action LADA consortium
.
Autoantibodies to N-terminally truncated GAD improve clinical phenotyping of individuals with adult-onset diabetes: Action LADA 12
.
Diabetologia
2018
;
61
:
1644
1649
39.
Lynam
A
,
McDonald
T
,
Hill
A
, et al
.
Development and validation of multivariable clinical diagnostic models to identify type 1 diabetes requiring rapid insulin therapy in adults aged 18–50 years
.
BMJ Open
2019
;
9
:
e031586
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.