Original article:
UKPDS 25: autoantibodies to islet-cell cytoplasm and glutamic acid decarboxylase for prediction of insulin requirement in type 2 diabetes. Turner R, Stratton I, Horton V et al., for the UK Prospective Diabetes Study Group. Lancet 1997; 350: 1288-93.

Summary
In this study the authors focus on the presence of autoantibodies to islet cell cytoplasm (ICA) and glutamic acid decarboxylase (GADA) compared with clinical variables to predict insulin requirement within 6 years of diagnosis of Type 2 diabetes. ICA and GADA were measured in 3672 newly diagnosed Type 2 diabetic patients aged between 25 and 65 years, recruited from the UK Prospective Diabetes Study (UKPDS). The phenotype was assessed by age of onset of diabetes, BMI, HbA 1c and islet ß-cell function. A total of 1538 patients not assigned insulin at diagnosis and followed up for 6 years were investigated for insulin requirement within this period.
The proportion of patients with ICA positivity was approximately 6%, GADA 10%, either ICA or GADA 12%, and both antibodies 4%. These proportions decreased with increasing age at diagnosis (in patients aged 25-65 years and 55-65 years, 21 vs 4% for ICA and 34 vs 7% for GADA, respectively). Patients positive for both ICA and GADA had phenotypic similarities with classic early-onset Type 1 diabetes with lower ß-cell function, higher HbA 1c values and lower BMI than those without antibodies in each 10-year age group. Among patients below 35 years of age at diagnosis, 94% with ICA and 84% with GADA required insulin therapy by 6 years, compared with 14% of those without antibodies. Among patients older than 55 years at diagnosis, 44% with ICA, 34% with GADA and 5% with neither antibody required insulin therapy within 6 years of diagnosis.
In all age groups, particularly amongst those aged less than 45 years at diagnosis, the presence of GADA was a more sensitive predictor of insulin requirement than the presence of ICA, with little reduction in specificity or positive predictive value with increasing age. ICA positivity was a more specific marker than GADA positiv-ity; its specificity increased at concentrations above 20 JDF units. The presence of both ICA and GADA, particularly in the older age groups, improved the positive predictive value and decreased the sensitivity.

Comment
Patients with late-onset diabetes and early insulin requirement are considered by some authors to have a form of Type 1 diabetes (slowly progressive, late-onset Type 1 diabetes), and by others to have a form of Type 2 diabetes (autoimmune Type 2 diabetes, latent autoimmune diabetes in adults) or Type 1 ½ diabetes. Despite this confusion in the literature, it is almost generally agreed that this is a form of autoimmune diabetes with late onset because of the positivity of the immunological markers and the progressive destruction of the ß-cell. It should be considered as a different entity than classical Type 2 diabetes since the metabolic parameters are deteriorating and the complications develop early.
In this study, Turner et al. tried to find an answer to two important questions: (1) determination of the precise predictive criteria for the diagnosis of autoimmune diabetes, and (2) evaluation of autoimmune markers with regard to specificity and sensitivity. For identifying late-onset autoimmune diabetes, the most valuable criteria are considered to be the presence of autoimmune markers, chronological age, the reduction of ß-cell reserve and early requirement of insulin, BMI, and other metabolic parameters. GADA and ICA are the most sensitive and specific autoimmune markers. The sensitivity and specificity are directly correlated with the autoantibody titres. The positive predictive value for insulin requirement by 6 years in the whole group was 60.3% with ICA ³5 JDF units, 73% with ICA ³20 JDF units, 56.8% with GADA ³20 U/l and 66.5% with GADA ³60 U/l. There are studies showing the presence of antibodies other than ICA and GADA such as insulin, thyroid, gastric and mitochondrial anti-bodies [1, 2].
The most important criticism of this study is the absence of genetic analysis. Groop et al. [1] observed that Type 2 diabetic patients over the age of 35 years with HLA-DR3/DR4 showed significantly higher frequencies of ICA. Loh-mann et al. [2] demonstrated that in Type 1 diabetics with disease onset after the age of 40 years, the HLA-DRB1 and -DRQB1 distribution was different to that in patients below 40 years of age.
The most important contribution of immunological screening carried out in Type 2 diabetics with early insulin requirement is that it shows the development of autoimmune diabetes in each age group. Another important finding is that the speed of ß-cell destruction follows a slower clinical course than at an older age. In the UKPDS, the need for insulin therapy within 6 years of diagnosis decreased from 94% in patients below 35 years of age with GADA and ICA, to 77% in those above 55 years, whereas in almost all childhood-onset Type 1 patients the symptoms begin at the same time as the need for insulin.
The b-cell reserve in the group with auto-antibodies was half that of the group without auto-antibodies, and the HbA 1c values were higher and the BMI lower in the same group. These predictive parameters are concordant with those given in other reports [1-5].
The UKPDS shows that GADA positivity is more sensitive than ICA positivity in each age group, particularly in those younger than 35 years. Although other studies in the literature support this, particularly in patients aged over 45 years, the sensitivity was found to be less in the UKPDS than in the report by Niskanen et al., which is also a prominent contribution to this issue (35% and 64%, respectively) [3, 6, 7]. On the other hand, ICA positivity is more specific than GADA positivity. The presence of both antibodies or ICA titres ³20 JDF units has more specific value.
The UKPDS demonstrated the reduced positive predictive value amongst older patients (>45 years). Niskanen et al. [3] found the positive pre- dictive value to be several times higher in patients with a relative (C-peptide <0.7 nmol/l) than with an absolute insulin requirement. The negative predictive value was similar in both groups.
In conclusion, especially in non-obese Type 2 diabetic patients, GADA and ICA can be used as a positive screening test for early insulin requirement. Autoantibodies in high concentrations and/or the presence of both may be considered predictive of rapid ß-cell destruction. Decrease in ß-cell reserve, genetic susceptibility, low BMI and poor metabolic control are other predictors.

References
1. Groop L, Miettinen A, Groop P-H et al. Organ-specific autoimmunity and HLA-DR antigens as markers for b-cell destruction in patients with Type II diabetes. Diabetes 1988; 37: 99-103.
2. Lohmann T, Rotger J, Sessler J et al. Distinct genetic and immunological features in patients with onset of IDDM before and after age 40. Diabetes Care 1997; 20/4: 524-9.
3. Niskanen LK, Groop LC, Tuomi T et al. GAD anti-bodies in NIDDM. Diabetes Care 1995; 18/12: 1557-65.
4. Yilmaz MT, Dinççag N, Satman I . et al. Slowly progressive type 1 diabetes: a retrospective analysis of clinical and laboratory findings. Diabetologia 1992; 35 (suppl 1): A210.
5. Satman I . , Yilmaz MT, Karsidag K et al. Predictors of insulin requirement in patients with slowly progressive Type 1 diabetes. Diabetologia 1996; 39 (suppl 1): A90.
6. Hatziagelaki E, Jaeger C, Maeser E et al. GAD 65 anti-body but not ICA positivity in adult-onset diabetic patients is associated with early progression to clinical insulin dependency. Acta Diabetol 1996; 33/4: 291-4.
7. Vandewalle CL, Falorni A, Svanholm S et al., and the Belgian Diabetes Registry. High diagnostic sensitivity of glutamate decarboxylase autoantibodies in insulin-dependent diabetes mellitus with clinical onset between age 20 and 40 years. J Clin Endocrinol Metab 1995; 80: 846-51.