Severe insulin deficiency occurs in 21% of insulin-treated patients diagnosed after 30 years of age and has similar clinical characteristics to young-onset type 1 diabetes
Twenty-one per cent (123/583) of insulin-treated participants diagnosed with diabetes after 30 years of age met the study criteria for type 1 diabetes and had severe endogenous insulin deficiency (insulin treatment within 3 years and C-peptide <200 pmol/l) (ESM Fig. 1).
The characteristics of participants with late-onset (>30 years of age) type 1 diabetes, young-onset (≤30 years of age) type 1 diabetes and late-onset type 2 diabetes (retained endogenous insulin secretion) are shown in Table 1. Participants with late-onset type 1 diabetes defined by development of severe insulin deficiency had broadly similar characteristics to those with young-onset type 1 diabetes: BMI, insulin dose and HbA1c did not differ. However, those with later onset type 1 diabetes had a modestly lower T1DGRS (0.268 vs 0.279 [expected type 2 diabetes population median 0.231 ]), higher islet autoantibody prevalence (78% vs 62%, at 13 vs 26 years duration) and were more likely to be treated as, and identify as having, type 2 diabetes (oral glucose-lowering agent use, 15% vs 5%; insulin at diagnosis, 62% vs 96%; self-reported type 2 diabetes 20% vs 0%).
Classical clinical criteria cannot reliably identify individuals with late-onset type 1 diabetes leading to severe insulin deficiency
Despite similar clinical features to young-onset type 1 diabetes, classical clinical criteria could not robustly identify late-onset type 1 diabetes. Only 41% had a BMI <25 and 28% of participants with a BMI <25 had type 2 diabetes. UK National Institute for Health and Care Excellence guidance for type 1 diabetes identification (age of diagnosis <50 or BMI <25 kg/m2) identified 81% of type 1 diabetes cases but had very low specificity (41%). The specificity of these criteria would be far lower had non-insulin treated individuals been included in this cohort.
Thirty-eight per cent of participants with late-onset type 1 diabetes did not receive insulin at diagnosis
Participants meeting criteria for type 1 diabetes after age 30 who did not receive insulin at diagnosis (38%) commenced insulin a median of 12 months from diagnosis, and had similar characteristics to those commencing insulin at diagnosis (ESM Table 1). However, 47% of those who had delayed insulin treatment reported a diagnosis of type 2 diabetes, and 30% received co-treatment with oral glucose-lowering therapy; this is in marked contrast with the 7% who received oral glucose-lowering therapy in the group who received insulin from diagnosis (p < 0.01).
Early progression to insulin is a strong predictor of future severe insulin deficiency
Eighty-five per cent (104/123) of all participants meeting criteria for type 1 diabetes (C-peptide <200 pmol/l and insulin therapy within 3 years) in the cohort diagnosed >30 were treated with insulin within 1 year of diagnosis (78% [104/133] for those defined by low C-peptide only) vs 18% (55/306) of those meeting criteria for type 2 diabetes (Fig. 1). Thirty-two per cent (37/115) of those with an intermediate C-peptide (≥200–<600 pmol/l) required insulin within 1 year of diagnosis. Of all those progressing to insulin within 3 years of diagnosis, 47% (123/264) met the study criteria for type 1 diabetes and 18% (48/264) had intermediate C-peptide (≥200–<600 pmol/l). Severe insulin deficiency was rare in those progressing to insulin after 3 years, occurring in only 10 of 231 participants (4%; Fig. 1).
Thirty per cent of those diagnosed >30 years of age and administered insulin at diagnosis had type 2 diabetes according to C-peptide criteria
Despite clear clinical, biochemical and genetic characteristics of type 2 diabetes (ESM Table 2), 30% of those meeting study criteria for type 2 diabetes were commenced on insulin at diagnosis. In the group meeting study criteria for type 2 diabetes and treated with insulin at diagnosis, 25% reported a diagnosis of type 1 diabetes, and only 59% received concurrent oral glucose-lowering therapy at a median diabetes duration of 10 years (ESM Table 2).