Influence of age at diagnosis and duration of diabetes on the positivity of glutamic acid decarboxylase antibody in South-Brazilian type 1 diabetes mellitus

Abstract

Aim

To evaluate the influence of age of onset and duration of diabetes on the positivity of glutamic acid decarboxylase antibody (GADA) in South-Brazilian type 1 diabetes mellitus (DM) patients.

Methods

GADA was evaluated in 92 patients with type 1 diabetes, in 147 gestational DM patients, and in 59 subjects with normal glucose tolerance.

Results

Type 1 patients with positive GADA (N = 44, 48%) were older at the onset of diabetes (22 ± 9 versus 18 ± 10 y, P = 0.043) and had a shorter DM duration (12 ± 8 versus 19 ± 9 y, P < 0.001), as compared with negative GADA patients. A logistic regression with antibody positivity as the dependent variable and diabetes duration as the independent variable, showed that the shorter diabetes duration was related to the presence of the antibody with an odds ratio (OR) = 5.6; (CI 95% = 2.1–14.6); P < 0.001. Another model, with age at diagnosis as the independent variable, did not show any association with antibody positivity. However, analysing only men, a shorter DM duration (OR = 6.5; CI = 1.7–24.0; P = 0.006), and also a higher age at diagnosis (OR = 5.5; CI = 1.5–21.0; P = 0.01) were significantly related to the antibody positivity. The performance of GADA was similar in up to 15 y of duration of DM (P = 0.78), but significantly diminished with higher duration (P = 0.001).

Conclusion

GADA testing is a helpful tool in the diagnosis of type 1 DM starting in young adults and older individuals. Even though the positivity rate declines along the course of disease, it still provides useful information up to 15 y after the diabetes diagnosis.

Introduction

Diabetes mellitus (DM) classification is currently divided into four categories: type 1 DM, type 2 DM, other types (which includes several subgroups with different pathogenesis) and the category of gestational DM.^1,2 However, sometimes it is difficult to distinguish one type of diabetes from another based only on clinical information. Overlapping clinical features frequently occur in early onset type 2 diabetes, maturity onset diabetes of the young (MODY) and recently diagnosed type 1 DM with preserved pancreas function. In these situations, the presence of autoimmunity against the pancreatic β-cell is decisive information to establish the diagnosis of type 1A diabetes.^2–5

Autoantibodies to a number of islet β-cell constituents, including glutamic acid decarboxylase (GADA), islet antigen-2, insulin and zinc transporter-8 are characteristic of patients with type 1 DM. Only 2–4% of these patients are autoantibody negative, fewer than 10% have only one marker, and around 70% have three or four markers.⁴ GADA has a sensitivity of 75–84%,^6–9 and specificity of 95–98.8% for the diagnosis of type 1 DM.^10–12 A harmonization programme was recently established by international consortia studies to increase comparability of the antibody results between laboratories.^4,13

GADA persist for longer in patients than other islet autoantibodies after the onset of type 1 diabetes.¹⁴ However, its accuracy can differ among different ethnicities, age of onset of diabetes and diabetes duration.^9,15,16 Therefore, the purpose of this study was to evaluate the performance of GADA in the diagnosis of type 1 diabetes in a South-Brazilian population, taking into account the duration and the age of onset of diabetes.

Patients and methods

Patients

This study included 298 individuals: 92 patients with an established diagnosis of type 1 DM, 59 healthy individuals and 147 women with gestational diabetes, evaluated between January 2010 and July 2011. The patients with DM were consecutively recruited from the Endocrinology Outpatient Clinic at Hospital de Clínicas de Porto Alegre (HCPA).

The clinical diagnosis of type 1 DM was based on a history of documented diabetic ketoacidosis, insulin use within up to three years after diagnosis and fasting baseline levels of C-peptide lower than 0.3 nmol/L.¹⁷ Fifty-nine healthy individuals matched by age, with no history of DM in family members and a normal oral glucose tolerance test (OGTT) according to ADA criteria,^1,3 were included to assess antibody quality control and provide cut-off values of normality. Gestational DM was diagnosed based on the Guidelines of the Second Meeting of The Diabetes and Pregnancy Task Force in Brazil, with diagnostic cut-off points of a fasting plasma glucose ≥6.1 mmol/L or a 2 h plasma glucose ≥7.77 mmol/L.¹⁸

The study was approved by local ethical committee.

Methods

A cross-sectional study design was employed, and a complete medical history and physical examination were performed. Patients’ weight was assessed in light clothes without shoes. Body mass index (BMI) was calculated from weight in kilograms (kg) divided by height in square meters (kg/m²). The waist was measured as recommended by World Health Organization. Blood pressure was measured after a five-min rest, using a calibrated mercury sphygmomanometer.

OGTT was performed by measurement of fasting and two hours glucose after the intake of 75 g of glucose, with an enzymatic method. A fasting glucose below 5.55 mmol/L and a two hours value below 7.77 mmol/L were considered normal.^1,3 GADA was evaluated by radioimmunoassay, RSR LIMITED^® Cardiff UK (inter-assay variation coefficient of 4.9% and 7.0% for values of 6.1 and 42.9 U/mL, and intra-assay of 3.6% and 3.7% for values of 6.4 and 42.7 U/mL, respectively).¹⁹ GADA values above the 95th percentile value of controls were considered positive. C-peptide was measured by chemoluminescence using an IMMULITE ^® kit (inter-assay variation coefficient of 6.3% and 4.4% for values of 0.30 and 1.47 nmol/L, respectively, and limit of detection of 0.10 nmol/L).²⁰

Variables without Gaussian distribution were log transformed before analyses. Logistic regressions were performed taking the positivity of the GADA as the dependent variable, and gender, age at diagnosis and duration of DM as independent variables.

Sensitivity, specificity, positive predictive value, negative predictive value and accuracy of GADA were calculated. The significance level was set at 5%. The analyses were performed using statistical package SPSS 16 for Windows.

Results

Table 1 shows the clinical and laboratory characteristics of patients with type 1 DM, healthy individuals and gestational DM. GADA was considered positive if values were above the 95th percentile of the control group (above 0.9 U/mL), and only 1 out of 59 controls was positive. In the group of women with gestational diabetes, only one had positive GADA – a 39-year-old woman with a previous diagnosis of Graves’ disease, whose hyperglycaemia was controlled by diet alone and with after labour fasting glucose of 4.77 mmol/L.

Table 1

Clinical and laboratory characteristics of patients with type 1 diabetes, healthy individuals and gestational diabetes mellitus

	Type 1 DM(N = 92)	Healthy individuals (N = 59)	GestationalDM (N = 147)
Age (years)	35 ± 10	30 ± 7	32 ± 7
Male gender	49 (53%)	33 (56%)	–-
Ethnic group (W/B)	85/7	57/2	112/35
Age at diagnosis (years)	20 ± 9	–-	–-
Duration of DM (years)	16 ± 9	–-	–-
Smoker	24 (32%)	5 (8%)	23 (16%)
BMI (kg/m²)	24 ± 3	23 ± 3	30 ± 7
C-peptide (nmol/L)	0.17 ± 0.03	0.60 ± 0.27	0.73 ± 0.06

W, white; B, black; BMI, body mass index; DM, diabetes mellitus

Data presented as mean ± SD and number of cases (%)

In patients with type 1 DM (n = 92), the GADA was positive in 44 out of the 92 cases, with a sensitivity of 47.5% (95% confidence interval [CI]: 37.4–58.4), specificity of 98.3% (CI: 89.7–99.9), positive predictive value (PPV) of 97.8% (CI: 86.8–99.9), negative predictive value (NPV) of 54.7% (CI: 44.8–64.3) and accuracy of 67.5%. The patients with type 1 DM were classified according the positivity of GADA for further analyses (Table 2).

Table 2

Clinical and laboratory characteristics of positive versus negative GADA type 1 diabetes patients

	GADA+(n = 44)	GADA−(n = 48)	P value
Age (years)	34 ± 10	37 ± 11	0.22
Male gender	23 (52%)	26 (54%)	0.511
Ethnic group (W/B)	41/3	44/4	0.549
Arterial hypertension	7 (16%)	12 (24%)	0.207
Age at diagnosis (years)	22 ± 9	18 ± 10	0.043
DM duration (y)	12 ± 8	19 ± 9	0.001
BMI (kg/m²)	24 ± 3	24 ± 3	0.664
C-peptide (nmol/L)	0.17 ± 0.003	0.17 ± 0.13	0.572
Fasting glycaemia (mmol/L)	8.4 ± 4.1	9.3 ± 5.1	0.41
A1c test (%)	8.5 ± 1.9	8.3 ± 1.8	0.497

W, white; B, black; BMI, body mass index; DM, diabetes mellitus; A1c, forms of glycosylated haemoglobin

Data presented as mean ± SD, number of cases (%)

Diabetes duration was stratified into five-year periods to evaluate the diagnostic performance of GADA in each range of duration. The performance was similar in up to 15 y of duration of DM (P = 0.78), but significantly diminishing with higher duration (P = 0.001).

In order to evaluate the specific influence of age at onset of DM on the accuracy of test, the different sensitivities and specificities were calculated according to age of diagnosis. The diagnostic performance of GADA in patients with DM onset at age ≥16 years (33 positive tests in 56 patients) was significantly superior to <16 years, and showed sensitivity of 58.9% (CI: 45–71.6), specificity of 98.3% (CI: 89.7–99.9), PPV of 97.1% (CI: 82.9–99.8), NPV of 71.6% (CI: 60.3–80.8) and accuracy of 79.1% (Figure 1). On the other hand, in patients in whom type 1 DM began before the age of 16 years (11 positive tests in 36 patients), GADA had a sensitivity of 30.6% (CI:16.9–48.3), specificity of 98.3% (CI:89.7–99.9), PPV of 91.7% (CI:59.8–99.8), NPV of 69.9% (CI:58.7–79.1) and accuracy of 72.6%.

Figure 1

Distribution of GADA titres according to age at diagnosis of DM. GAD, glutamic acid decarboxylase; DM, diabetes mellitus

A logistic regression with antibody positivity as the dependent variable and diabetes duration (≥15 years versus <15 years) as the independent variable, showed that the shorter diabetes duration was related to the presence of the antibody, with an odds ratio (OR) = 5.6; (CI 95% = 2.1–14.6); P < 0.001. Another model, with age at diagnosis as the independent variable, did not show any association with antibody positivity. In the subset of women, results were similar. However, when analysing only men, a shorter DM duration (OR = 6.5; CI = 1.7–24.0; P = 0.006), and also a higher age at diagnosis (OR = 5.5; CI = 1.5–21.0; P = 0.01) were significantly related to the antibody positivity (Table 3).

Table 3

Univariate logistic regressions performed with GADA positivity as the dependent variable and diabetes duration and age at diagnosis of diabetes as the independent variable in each model

Dependent variable: GADA positivity	OR	CI	P value
Independent variable: Diabetes duration (<15 years)
Whole group	5.6	2.1–15	<0.0001
Females	6.7	1.8–25	0.005
Males	6.5	1.7–24	0.006
Independent variable: Age at diagnosis (>16 years)
Whole group	2.4	0.9–6	0.07
Females	2.0	0.6–7	0.27
Males	5.5	1.5–21	0.01

OR, odds ratio; CI, confidence interval

Evaluating the diagnostic performance of GADA in the subsets of age at diagnosis and DM duration simultaneously, the best performance was in patients with a diagnosis of DM over 16 years of age, and duration of the disease less than 15 years simultaneously (28 positives tests out of 37) with a sensitivity of 75.7% (CI: 58.4–87.6), specificity of 98.3% (CI:89.7–99.9), PPV of 96.6% (CI:80.4–99.8), NPV of 86.6% (CI: 75.5–93.3) and accuracy of 89.5%.

Discussion

In our study, the best diagnostic performance of GADA test was in patients with the duration of the disease of less than 15 years, and in the subset of men, also with the onset of DM after 16 years of age.

Diagnosis of the correct type of diabetes is usually made based on clinical information, but the use of autoimmunity markers is sometimes necessary. Therefore, the accuracy and the limitations of these antibodies must be studied in detail. Taking the whole group of patients, the specificity of GADA found in our study is practically identical to that of previous studies in other populations – around 98% – although our sensitivity was lower.^10,11 This is probably due to the fact that about half of the patients had a DM duration over 15 years in our sample. When analysing the patients with diabetes duration of less than 15 years, a best diagnostic performance of GADA was demonstrated, with a sensitivity similar to that reported by other studies, that ranged from 75 to 84%.^6–9,21

Age at diagnosis of type 1 diabetes is an important characteristic for the interpretation of the GADA result. Our data is in agreement with international studies that show that sensitivity of the GADA is greater in older individuals at diagnosis.^{8,9,15,21–25} GADA is considered the most prevalent antibody in type 1 DM with age of onset above 20 years,⁹ and its rate of positivity is constant with ages in the range of 10–40 years in some series.^9,26 GADA sensitivity is also greater than the anti-islet and the anti-insulin antibodies in adults, and it is, therefore, the most sensitive autoimmunity marker in this age range.^9,14,21,22

We have considered the period of less than three years to start insulin as a criteria for type 1 diabetes, and though it is quite a long period for patients with classical type 1 diabetes, it has been described that patients with LADA (latent autoimmune diabetes in adults) patients may present a highly variable median period to insulin initiation, sometimes up to 30 months after diagnosis or longer.²⁷

The longer the duration of DM, the lower the positivity of GADA. In the recent study of McDonald et al,⁵ patients with type 1 diabetes duration of less than six months had GADA and/or islet antigen-2 antibody in 82% of cases. The positivity is reported to decrease with time, but more than 50% of the patients with type 1 DM have positive GADA after 10-year duration of the disease, and it is the antibody that remains positive the longest after diabetes onset.^9,15,23,26 Our study reproduced this finding in the Brazilian population, but expanded the range of satisfactory test performance up to 15 years of diabetes.

GADA sensitivity also varies with gender.^8,21,24 In women, the sensitivity is reported to be around 80%, without variation according to age at diagnosis; and, in men, it is 50–60% below the age of 10 years and 75–90% above this age.^8,9,21,24 However, no gender difference has been shown above the age of 20 years.⁹ In our study, both age at diagnosis and duration of DM determined GADA positivity in men, while in women only the duration of diabetes influenced the positivity. This finding suggests that the stimuli to autoimmunity may operate differently at various ages, and they may also be sex-related. Any time gender is an issue, it raises the differences between men and women regarding lifestyle, and occupation, as well as influence of sex hormones and sex chromosomes. This highlights the complexity of the mechanisms.²⁸ Environmental factors predisposing to diabetes might also be different for men and women before and after puberty. For example, there is an increased risk of diabetes in boys with any neonatal infections and a decreased risk in girls, indicating a gender difference in the impact of neonatal infections on the risk of type 1 diabetes in children.²⁹

In the Brazilian population, the study of Pardini et al.³⁰ also reported a negative correlation of type 1 DM duration with GADA positivity: GADA positivity of 80% in recent-onset (<12 months) and 54% in longer duration diabetes (>12 months). A recent study in Southern Brazil found 63% of GADA positivity in patients with type 1 DM duration between three and five years.³¹ On the other hand, the study of Rodacki et al.³² described GADA positivity prevalence of 45.8% of type 1 diabetes patients with a mean duration of disease of 10 y, not affected by disease duration or age at diagnosis. The authors speculate that this low positivity may be due to the fact that only patients with more than one-year of diabetes duration were included in the analysis. So, the differences in GADA prevalence in our country appear to be influenced mainly by the subjects characteristics, such as diabetes duration or by the use of different assays.

Racial and ethnical differences have also been shown to influence the prevalence of GADA.¹⁶ Prevalence of GADA positivity in other countries is variable, reported to be around 45.8% in Chinese population,²⁵ 23% in Mexicans³³ and, in European patients, it varies from 39% in longstanding patients with age at diabetes diagnosis below 15 years, to 97% in patients recently diagnosed and with age at diagnosis older than 15 years.¹⁵ Therefore, the ethnic background should be taken into account to predict the performance of the test. Furthermore, the exact prevalence of type 1 diabetes (insulin dependent not caused by autoimmunity) is not well described in these populations, and is probably responsible for a subset of cases, a phenomenon that remains to be further investigated.

The clinical importance of antibodies evaluation for the classification of DM was recently shown in the so called Aβ classification. This classification evaluates the presence of autoimmunity (A = antibodies) and the function of the β cell (β = C-peptide level) and so far presented the higher diagnostic accuracy in comparison with the traditional clinical classifications.^2,34,35 Therefore, it is important to take into account the effect of age of onset and diabetes duration upon the GADA test interpretation in order to more properly use this tool.

Our study presents some limitations. The most important is the cross-sectional design employed. Obviously, a longitudinal design would allow the observation of the evolution and influence of several factors on the antibody course, and not merely an association between putative variables. Moreover, the sample size was relatively small, limiting conclusions to a similar population.

In conclusion, GADA testing is a helpful tool in the diagnosis of type 1 DM starting in young adults and older individuals. Even though the positivity rate declines along the course of disease, it still provides useful information up to 15 years after the diabetes diagnosis. The interpretation of the GADA results in children and in patients with long-lasting diabetes should be made with caution.

Footnotes

DECLARATION

Competing interest: None.

Funding: FIPE

Ethical approval: Ethics Commitee of Hospital de Clínicas de Porto Alegre, RS, Brazil

Guarantor: SPS

Contributorship: JFM researched data, made statistic analysis and wrote the manuscript. LSW reviewed/edited the manuscript. NM and VW researched data and reviewed the manuscript. TCR contributed to the design of the study and reviewed the manuscript. ERR contributed to laboratory analysis and reviewed the manuscript. SPS contributed to the design of the study and reviewed/edited the manuscript.

Acknowledgements: This study was supported by Fundo de Incentivo à Pesquisa (FIPE) at Hospital de Clínicas de Porto Alegre. We thank Dr Joíza Lins Camargo from Unidade de Bioquímica e Imunoensaios.

References

Standards of medical care in diabetes – 2012. Diabetes Care 2012;35(Suppl. 1):S11–63

Maraschin

Murussi

Witter

Silveiro

. Classificação do Diabete Melito. Arquivos Brasileiros de Cardiologia 2010;95:40–7

Diagnosis and classification of diabetes mellitus. Diabetes Care 2012;35(Suppl. 1):S64–71

Bingley

. Clinical applications of diabetes antibody testing. J Clin Endocrinol Metab [Evaluation Studies Review] 2010;95:25–33

McDonald

Colclough

Brown

Islet autoantibodies can discriminate maturity-onset diabetes of the young (MODY) from type 1 diabetes. Diab Med. [Research Support, Non-US Gov't] 2011;28:1028–33

Winter

Harris

Schatz

. Immunological markers in the diagnosis and prediction of autoimmune type 1a diabetes. Clin Diab 2002;20:183–91

Bingley

Bonifacio

Mueller

. Diabetes Antibody Standardization Program: first assay proficiency evaluation. Diabetes 2003;52:1128–36

Notkins

Lernmark

. Autoimmune type 1 diabetes: resolved and unresolved issues. J Clin Invest 2001;108:1247–52

Vandewalle

Falorni

Svanholm

Lernmark

Pipeleers

Gorus

. High diagnostic sensitivity of glutamate decarboxylase autoantibodies in insulin-dependent diabetes mellitus with clinical onset between age 20 and 40 years. The Belgian Diabetes Registry. J Clin Endocrinol Metab 1995;80:846–51

10.

Falorni

Grubin

Takei

Radioimmunoassay detects the frequent occurrence of autoantibodies to the Mr 65,000 isoform of glutamic acid decarboxylase in Japanese insulin-dependent diabetes. Autoimmunity. [Research Support, Non-US Gov't Research Support, US Gov't, PHS] 1994;19:113–25

11.

Falorni

Ortqvist

Persson

Lernmark

. Radioimmunoassays for glutamic acid decarboxylase (GAD65) and GAD65 autoantibodies using 35S or 3H recombinant human ligands. J Immunol Methods [Research Support, US Govt, PHS] 1995;186:89–99

12.

Bingley

Williams

Colman

Measurement of islet cell antibodies in the Type 1 Diabetes Genetics Consortium: efforts to harmonize procedures among the laboratories. Clin Trials [Comparative Study Research Support, N.H, Extramural] 2010;7:S56–64

13.

Bonifacio

Williams

Harmonization of glutamic acid decarboxylase and islet antigen-2 autoantibody assays for national institute of diabetes and digestive and kidney diseases consortia. J Clin Endocrinol Metab 2010;95:3360–7

14.

Borg

Gottsater

Fernlund

Sundkvist

. A 12-year prospective study of the relationship between islet antibodies and beta-cell function at and after the diagnosis in patients with adult-onset diabetes. Diabetes 2002;51:1754–62

15.

Hermitte

Atlan-Gepner

Mattei

Diverging evolution of anti-GAD and anti-IA-2 antibodies in long-standing diabetes mellitus as a function of age at onset: no association with complications. Diabet Med 1998;15:586–91

16.

Tuomi

Zimmet

Rowley

Differing frequency of autoantibodies to glutamic acid decarboxylase among Koreans, Thais, and Australians with diabetes mellitus. Clin Immunol Immunopathol 1995;74:202–6

17.

Faber

Binder

. C-peptide: an index of insulin secretion. Diabetes Metab Rev [Review] 1986;2:331–45

18.

Reichelt

Opperman

Schmidt

. Recomendações da 2a. Reunião do Grupo de Trabalho em Diabetes e Gravidez. Arq Bras Endocrinol Metab 2002;46:574–81

19.

Powell

Prentice

Asawa

Glutamic acid decarboxylase autoantibody assay using 125I-labelled recombinant GAD65 produced in yeast. Clin Chim Acta 1996;256:175–88

20.

Baekkeskov

Nielsen

Marner

Bilde

Ludvigsson

Lernmark

. Autoantibodies in newly diagnosed diabetic children immunoprecipitate human pancreatic islet cell proteins. Nature 1982;298:167–9

21.

Graham

Hagopian

Kockum

Genetic effects on age-dependent onset and islet cell autoantibody markers in type 1 diabetes. Diabetes [Research Support, Non-US Gov't Research Support, US Gov't, PHS] 2002;51:1346–55

22.

Hathout

Thomas

El-Shahawy

Nahab

Mace

. Diabetic autoimmune markers in children and adolescents with type 2 diabetes. Pediatrics [Comparative Study] 2001;107:E102

23.

Scholin

Bjorklund

Borg

Islet antibodies and remaining beta-cell function 8 years after diagnosis of diabetes in young adults: a prospective follow-up of the nationwide Diabetes Incidence Study in Sweden. J Intern Med 2004;255:384–91

24.

Verge

Howard

Rowley

Anti-glutamate decarboxylase and other antibodies at the onset of childhood IDDM: a population-based study. Diabetologia 1994;37:1113–20

25.

Shiau

Tsai

Hwang

Chang

. Relationship between autoantibodies against glutamic acid decarboxylase, thyroglobulin/thyroid microsome and DNA topoisomerase II in the clinical manifestation of patients with type 1 diabetes mellitus in Taiwan. Eur J Endocrinol [Comparative Study] 2000;142:577–85

26.

Aanstoot

Sigurdsson

Jaffe

Value of antibodies to GAD65 combined with islet cell cytoplasmic antibodies for predicting IDDM in a childhood population. Diabetologia [Clinical Trial Comparative Study] 1994;37:917–24

27.

Brophy

Yderstraede

Mauricio

Time to insulin initiation cannot be used in defining latent autoimmune diabetes in adults. Diabetes Care 2008;31:439–41

28.

Shoenfeld

Tincani

Gershwin

. Sex gender and autoimmunity. J Autoimmun [Editorial Introductory] 2012;38:J71–3

29.

Svensson

Carstensen

Mortensen

Borch-Johnsen

. Gender-associated differences in Type 1 diabetes risk factors? Diabetologia 2003;46:442–3

30.

Pardini

Mourao

Nascimento

Vivolo

Ferreira

Pardini

. Frequency of islet cell autoantibodies (IA-2 and GAD) in young Brazilian type 1 diabetes patients. Braz J Med Biol Res. [Research Support, Non-US Govt] 1999;32:1195–8

31.

Moreira

Lara

Linden

Frequency of the anti-glutamic acid decarboxylase immunological marker in patients with diabetes duration longer than three years in southern Brazil. Sao Paulo Med J 2011;129:130–3

32.

Rodacki

Zajdenverg

Albernaz

Bencke-Goncalves

Milech

Oliveira

. Relationship between the prevalence of anti-glutamic acid decarboxylase autoantibodies and duration of type 1 diabetes mellitus in Brazilian patients. Braz J Med Biol Res 2004;37:1645–50

33.

Becerril Angeles

Garcia Lara

Gonzalez Barcena

Ibarra Olmos

Torres Ambriz

Angeles Garay

. [Autoantibodies in GAD65 in Mexican adults with diabetes types 1 and 2 and their siblings]. Rev Alerg Mex 2010;57:190–5

34.

Balasubramanyam

Garza

Rodriguez

Accuracy and predictive value of classification schemes for ketosis-prone diabetes. Diab Care 2006;29:2575–9

35.

Maldonado

Hampe

Gaur

Ketosis-prone diabetes: dissection of a heterogeneous syndrome using an immunogenetic and beta-cell functional classification, prospective analysis, and clinical outcomes. J Clin Endocrinol Metab 2003;88:5090–8