Introduction

The average life expectancy of patients with type 1 diabetes (T1DM) has increased significantly over the last decades.1 However, recent studies have reported an estimated loss in life expectancy of 11 to 13 years in patients with T1DM as compared with the general population.2,3 Moreover, cardiovascular disease remains the most common cause of death in these patients.4-6 Risk factors for cardiovascular disease in this population include microvascular complications, age, diabetes duration, body mass index (BMI), hemoglobin A1c (HbA1c), hypertension, and dyslipidemia.7,8 The number and severity of microvascular complications were shown to be associated with an increased rate of all-cause mortality and cardiovascular events. Among these complications, chronic kidney disease seems to be the greatest risk factor for excess mortality.9,10

Although the available evidence supports potential benefits from using new insulins and technologies in diabetes management, numerous patients with T1DM still experience chronic microvascular complications that adversely affect their life expectancy and quality of life.11,12 However, not all patients with long-term T1DM develop advanced microangiopathy, as reported in the Joslin 50-years Medalist Study and Golden Years Cohort.13,14 This suggests that there may be some genetic and environmental factors that protect patients against microvascular complications. One of the possible factors is an elevated level of high-density lipoprotein (HDL) cholesterol,15,16 while the others include normal BMI, lack of hypertension, low insulin requirement, HbA1c near the treatment target (about 7%), and a family history of longevity.15 Another possible protective factor is preserved insulin secretion. It was also hypothesized that the population of patients with long-term T1DM without advanced complications may include individuals with monogenic diabetes, especially maturity-onset diabetes of the young (MODY),16 misdiagnosed as T1DM. The percentage of undiagnosed MODY in this specific group of patients is higher than in the general population of patients with T1DM.16 In support of this hypothesis, published estimates showed that monogenic diabetes was misdiagnosed as T1DM or type 2 diabetes (T2DM) in the vast majority of cases (90%).17 Both patients with monogenic diabetes and those with T1DM are usually slim and young at the time of diagnosis. Another supporting piece of evidence is that chronic complications are almost absent in glucokinase MODY (GCK-MODY).18,19 They are also less prevalent and less severe in the other forms of MODY, such as the most frequent MODY3 caused by a mutation in the hepatocyte nuclear factor-1α gene, HNF1A.18,19 In fact, in the Medalist Study, almost 8% of T1DM patients with a disease duration of 50 years were suspected to have monogenic diabetes.16

The aim of the present study was to determine the clinical characteristics of patients with long-term T1DM without advanced microvascular complications, with a particular focus on individuals with well-preserved kidney function. Additionally, patients were screened for mutations within a set of monogenic diabetes genes.

Patients and methods

Patients

Patients diagnosed with T1DM were recruited at 2 Polish university hospitals in Kraków and Poznań. The inclusion criteria were as follows: T1DM duration of at least 40 years and absence of advanced complications defined as chronic kidney disease with an estimated glomerular filtration rate (eGFR) lower than 60 ml/min/1.73 m2, overt proteinuria or previous kidney transplant, blindness in at least 1 eye, and diabetic foot syndrome (currently or in the past). After completing a standard questionnaire, all patients underwent physical examination. We collected data on sex, age at examination, age of diagnosis, weight, height, waist-to-hip ratio, blood pressure, daily dose of insulin, medication use, family history of diabetes, presence of chronic microvascular and macrovascular complications as well as comorbidities, and history of smoking. Fasting blood and first-pass urine samples were obtained for laboratory tests, including the measurement of urinary albumin-to-creatinine ratio and the levels of HbA1c, C-peptide, creatinine, lipids, and high-sensitivity C-reactive protein.

Clinical retinal examination was performed by a trained ophthalmologist. Peripheral polyneuropathy was assessed using a 10-gram monofilament for tactile sensation, 128-Hz tuning forks for vibration sensation, and a rod with 2 different ends for temperature sensation. Polyneuropathy was diagnosed if 2 or more of the following criteria were met: the presence of symptoms, lack of the ankle reflex, and impaired sensation of touch, temperature, and / or vibration.20

Genetic testing

Next-generation sequencing was used for detecting mutations in a set of selected monogenic diabetes genes.21 Genomic DNA was extracted, libraries prepared, and data processed as described in detail previously.21 We evaluated 7 genes that are the most frequent causes of monogenic diabetes (GCK, HNF1A, HNF4A, HNF1B, ABCC8, KCNJ11, and INS) for potentially pathogenic variants, in line with a recent French study.22 Variant scoring was based on the American College of Medical Genetics and Genomics (ACMG) guidelines. To predict the pathogenicity of the variants, the VarSome engine was used.23,24

Ethical approval

The study was approved by the Bioethics Committee of Jagiellonian University Medical College in Kraków, Poland, and conducted in accordance with the 1975 Declaration of Helsinki, with subsequent revisions. All patients gave written informed consent to participate in the study.

Statistical analysis

The parametric t test or the nonparametric U test was performed, as applicable, to describe the clinical characteristics of patients and differences between individuals with or without diagnosed proliferative retinopathy. For nominal variables, the Fisher exact test was used. A multivariable logistic regression analysis was performed to identify factors associated with the presence of proliferative retinopathy and / or albuminuria. The parameters used to build the multivariable model included sex, age at onset, duration of diabetes, BMI, daily insulin dose, HbA1c, hypertension, smoking, family history of diabetes, and the levels of C-peptide, low-density lipoprotein (LDL) cholesterol, HDL cholesterol, and triglycerides. A separate analysis was performed to examine the factors associated with macrovascular complications, such as previous myocardial infarction and / or stroke. In addition to the parameters listed above, the multivariable model in this analysis included also a number of microvascular complications (proliferative retinopathy, albuminuria, and peripheral polyneuropathy). Statistical analysis was performed using Statistica, version 13 (TIBCO Software Inc, Palo Alto, California, United States). A P value of less than 0.05 was considered significant.

Results

The study included 45 patients with T1DM (29 women and 16 men) with a mean (SD) age at examination of 59.2 (8.0) years and a mean (SD) age at diabetes onset of 14.6 (6.7) years. The mean (SD) BMI in the study group was 26.4 (5.0) kg/m2. Moreover, patients had good glycemic control with a mean (SD) HbA1c level of 7.6% (1.4%) (mean [SD], 59.3 [15.1] mmol/mol) and a mean (SD) daily insulin dose of 0.48 (0.17) units/kg. The mean (SD) eGFR was 82.2 (12.1) ml/min/1.73 m2. Albuminuria was reported in 7 patients. There were no cases of overt proteinuria. Retinopathy was found in 39 participants (nonproliferative in 7 and proliferative in 32). Consistent with the inclusion criteria, there were no cases of blindness. Peripheral polyneuropathy was present in 24 participants. Cardiovascular disease, defined as coronary artery disease, stroke, or peripheral artery disease, was diagnosed in 20 individuals based on medical records. The clinical and biochemical characteristics of patients are shown in Table 1. The independent risk factors for proliferative retinopathy and / or albuminuria, identified by a backward stepwise elimination procedure, included T1DM duration (odds ratio [OR], 1.25; 95% CI, 1.02–1.53) and LDL cholesterol levels (OR, 2.83; 95% CI, 1.05–7.65). The only independent factor associated with myocardial infarction and / or stroke was smoking (OR, 2.83; 95% CI: 1.05–7.65).

Table 1. Clinical and biochemical characteristics of the study group

Parameter

Study group (n = 45)

Retinopathy

P value

No / Nonproliferative (n = 13)

Proliferative (n = 32)

Sex, n (%)

Male

16 (35.6)

4 (8.9)

12 (26.7)

0.74

Female

29 (64.4)

9 (20.0)

20 (44.4)

Age, y

59.2 (8.0)

58.7 (7.2)

59.4 (8.5)

0.78

Age of diabetes onset, y

14.6 (6.7)

16.0 (5.3)

14.1 (7.1)

0.38

Diabetes duration, y

44.5 (41.0–47.0)

42.7 (40.0–45.0)

45.4 (41.5–47.5)

0.12a

Family history of diabetes, n (%)

Yes

16 (35.5)

2 (4.4)

14 (31.1)

0.09

No

29 (64.4)

11 (24.4)

18 (40.0)

BMI, kg/m2

26.4 (5.0)

24.9 (3.9)

27.1 (5.4)

0.19

Waist-to-hip ratio

Male

0.94 (0.07)

0.90 (0.07)

0.95 (0.06)

0.21

Female

0.85 (0.09)

0.84 (0.05)

0.86 (0.10)

0.73

HbA1c, %

7.3 (6.7–8.4)

7.1 (6.7–8.2)

7.4 (6.7–8.5)

0.80a

HbA1c, mmol/mol

56.8 (50.0–67.8)

54.1 (49.7–66.1)

57.4 (50.8–68.3)

0.80a

DDI, U

33.0 (13.5)

32.5 (12.4)

34.6 (14.0)

0.63

DDI, U/kg

0.48 (0.17)

0.47 (0.15)

0.49 (0.18)

0.82

HDL cholesterol, mmol/l

1.7 (1.4–2.3)

2.0 (1.7–2.5)

1.7 (1.4–2.2)

0.25a

LDL cholesterol, mmol/l

2.6 (0.8)

2.3 (0.9)

2.7 (0.8)

0.14

Triglycerides, mmol/l

1.0 (0.7–1.3)

0.9 (0.6–1.3)

1.1 (0.8–1.3)

0.13a

Hs-CRP, ug/ml

1.5 (0.6–3.3)

3.3 (0.4–4.8)

1.4 (0.7–3.0)

0.70a

C-peptide, ng/ml

0.03 (0.01–0.05)

0.04 (0.02–0.06)

0.03 (0.01–0.04)

0.17a

eGFR, ml/min/1.73m2

82.2 (12.1)

81.0 (12.2)

82.8 (2.2)

0.67

Albuminuria, n (%)

Yes

7 (15.5)

0

7 (15.5)

0.09

No

38 (84.4)

13 (28.9)

25 (55.5)

Peripheral polyneuropathy, n (%)

Yes

24 (53.3)

5 (11.1)

19 (42.2)

0.32

No

21 (46.7)

8 (17.8)

13 (28.9)

Smoking (current or past), n (%)

Yes

6 (13.3)

1 (2.2)

5 (11.1)

0.66

No

39 (86.7)

12 (26.7)

27 (60.0)

Hypertension, n (%)

Yes

31 (68.9)

8 (17.8)

23 (51.1)

0.50

No

14 (31.1)

5 (11.1)

9 (20.0)

SBP, mmHg

125 (120–136)

130 (123–138)

125 (119–135)

0.26a

DBP, mmHg

70 (66–80)

78 (70–80)

70 (65–75)

0.049a

CVD, n (%)

Yes

20 (44.4)

4 (8.9)

16 (35.6)

0.33

No

25 (55.6)

9 (20.0)

16 (35.6)

Stroke, n (%)

Yes

3 (6.7)

1 (2.2)

2 (4.4)

>0.99

No

42 (93.3)

12 (26.7)

30 (66.7)

CAD, n (%)

Yes

17 (37.8)

3 (6.7)

14 (31.1)

0.31

No

28 (67.2)

10 (22.2)

18 (40.)

MI, n (%)

Yes

7 (15.6)

1 (2.2)

6 (13.3)

0.65

No

38 (84.4)

12 (26.7)

26 (57.8)

PAD, n (%)

Yes

4 (9.1)

1 (2.3)

3 (6.8)

>0.99

No

40 (90.9)

11 (25.0)

29 (65.9)

ACEIs/ARBs, n (%)

Yes

34 (77.3)

8 (18.2)

26 (59.1)

0.13

No

10 (22.7)

5 (11.4)

5 (11.4)

ASA, n (%)

Yes

26 (59.1)

7 (15.9)

19 (43.2)

0.74

No

18 (40.9)

6 (13.6)

12 (27.3)

Statins, n (%)

Yes

35 (79.5)

11 (25.0)

24 (54.5)

0.70

No

9 (20.5)

2 (4.5)

7 (15.9)

a P values were derived from the U test. In the remaining cases, P values were derived from the Fisher exact test or the t test.

Data are presented as mean (SD) or median (interquartile range) unless indicated otherwise.

Abbreviations: ACEI, angiotensin-converting-enzyme inhibitor; ARB, angiotensin II receptor blocker; ASA, acetylsalicylic acid; BMI, body mass index; CAD, coronary artery disease; hs-CRP, high-sensitivity C-reactive protein; CVD, cardiovascular disease; DBP, diastolic blood pressure; DDI, daily dose of insulin; eGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MI, myocardial infarction; PAD, peripheral artery disease; SBP, systolic blood pressure

The next-generation sequencing analysis identified 9 patients as carriers of 10 variants in the 7 analyzed genes; 1 patient was a carrier of 2 variants. The identified variants are summarized in Table 2, while the detailed clinical characteristics of mutation carriers are presented in Table 3.

Table 2. Summary of the identified variants in monogenic diabetes genes

Patient ID

Type

Gene

Exon

Codon_change

aa_change

VarSome prediction

VarSome predicted pathogenicity

6KL

snp

ABCC8

35

Gcc/Acc

p.Ala1410Thr/c.4228G>A

PM1, PM2, PP2, PP3

Likely pathogenic

21WM

snp

ABCC8

38

Cgc/Tgc

p.Arg1530Cys/c.4588C>T

PM1, PM2, PP2, PP3

Likely pathogenic

11ZT

indel

ABCC8

25

tgc/tgcCT

p.Ser1051fs/c.3150_3151insCT

PVS1, PM1, PM2, PPS3

Pathogenic

20FJ

snp

ABCC8

15

N/A

c.2117–1G>C

PVS1, PM2, PP3, PP5

Pathogenic

19MZ

snp

ABCC8

21

Gtt/Att

p.Val849Ile/c.2545G>A

PM1, PM2, PP2, BP4

Uncertain significance

37JA

snp

GCK

9

caC/caA

p.His380Gln/c.1140C>A

PVS1, PM2, BP4

Likely pathogenic

23SM

snp

GCK

2

gaG/gaT

p.Glu22Asp/c.66G>T

PM1, PM2, PP2, PP3

Likely pathogenic

23SM

snp

HNF1B

4

Cac/Gac

p.His336Asp/c.1006C>G

PM1, PM5, PP2, PP3, BS1. BS2

Uncertain significance

30CH

snp

HNF1B

4

Cac/Gac

p.His336Asp/c.1006C>G

PM1, PM5, PP2, PP3, BS1. BS2

Uncertain significance

46SM

snp

HNF1A

9

cGg/cAg

p.Arg583Gln/c.1748G>A

PM5, PP2, PP3, BS1, BS2, BS3

Benign

Table 3. Clinical and biochemical characteristics of the carriers of variants in monogenic diabetes genes

Patient ID

Gene

Sex

Age at diagnosis, y

Age at examination, y

BMI, kg/m2

HbA1c, %

C-peptide, ng/ml

DDI, U/kg

eGFR, ml/min/1.73 m2

Chronic complications

Family history of diabetes

6KL

ABCC8

F

8

53

20.0

7.1

0.029

0.58

80

None

Father, mother

21WM

ABCC8

F

16

60

34.0

7.4

0.039

0.36

86

PDR, peripheral polyneuropathy

Mother

11ZT

ABCC8

F

18

63

26.0

4.9

0.01

0.14

79

PDR, peripheral polyneuropathy

No

20FJ

ABCC8

F

14

60

20.4

7.0

0.038

0.42

62

NPDR, peripheral polyneuropathy

No

19MZ

ABCC8

M

19

64

21.3

6.1

0.035

0.55

83

PDR, peripheral polyneuropathy

Sibling

37JA

GCK

F

20

60

27.8

5.8

0.042

0.53

95

PDR

No

23SM

GCK

F

23

70

22.5

7.4

0.01

0.58

73

PDR

Sibling

23SM

HNF1B

F

23

70

22.5

7.4

0.01

0.58

73

PDR

Sibling

30CH

HNF1B

F

12

64

29.4

8.7

0.01

0.71

92

PDR, peripheral polyneuropathy

Sibling

46SM

HNF1A

F

19

59

24.7

9.5

0.1

0.67

99

NPDR, peripheral polyneuropathy

No

Abbreviations: F, female; M, male; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy; others, see Table 1

Five variants were found in the ABCC8 gene, including 2 missense mutations classified as likely pathogenic. The first mutation was a new Ala1410Thr variant found in a woman aged 8 at diagnosis and 53 at the time of the examination. She was on a rather low dose of insulin (26 U/d), and her glycemic control was good with an HbA1c level of 7.1% (54.1 mmol/l). Her BMI was 20.0 kg/m2. She was free from diabetic complications. Both her parents (aged 82 at the time of the study) were diagnosed with T2DM; however, they refused genetic testing. The other missense mutation, Arg1530Cys, was found in a female patient diagnosed with T1DM at the age of 16. At the time of examination, she was 60 years old, and she received intensive insulin therapy (multiple daily injections), with a daily insulin requirement of 31 units combined with metformin due to obesity. She also developed proliferative retinopathy. The Arg1530Cys missense mutation is also present in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/) with uncertain significance annotation. However, our patient and her family were unavailable for further evaluation.

There were 2 new null variants in the ABCC8 gene in our cohort, one frameshift (Ser1051fs/c.3150_3151insCT) and one splicing (c.2117-1G>C). Finally, the ABCC8 Val849Ile variant detected in a single patient was classified as a sequence difference of uncertain significance according to the ACMG criteria.

Next, we found 2 missense variants in the GCK gene classified as likely pathogenic. One of them, a newly identified missense mutation, Glu22Asp, was observed in a female patient who was also a carrier of the HNF1B His336Asp variant. The other one, the His380Gln mutation, was previously reported but without data on the frequency and clinical significance.25 Both female carriers were characterized by an insulin requirement typical for T1DM (daily dose of insulin, 35 U/d and 40 U/d, respectively) and good glycemic control (HbA1c, 7.4% [57.4 mmol/mol] and 5.8% [39.9 mmol/mol], respectively). They also developed proliferative retinopathy requiring laser therapy, and their C-peptide levels were barely detectable (<⁠0.1 ng/ml) at the time of examination.

We also detected an rs138986885 sequence difference corresponding to the His336Asp missense mutation in the HNF1B gene in 2 unrelated participants. This was classified as a variant of uncertain significance.

Finally, an rs137853242 variant was found in the HNF1A gene corresponding to the missense mutation Arg583Gln in exon 9. The female carrier of the rare Gln variant was diagnosed with T1DM at the age of 19, and her age at examination was 59. Her current C-peptide levels were almost undetectable (0.1 ng/ml), and her HbA1c level was 9.5% (80.3 mmol/mol).

Discussion

This study reports the clinical, biochemical, and genetic characteristics of a highly selected group of Polish patients with long-term T1DM. Because our population included only patients without advanced microvascular complications, it is not representative for individuals with T1DM in general. In particular, all participants had an eGFR higher than 60 ml/min/1.73 m2, and albuminuria (but not overt proteinuria) was present in only 7 of the 45 patients. Previous studies reported a potential association between genetic factors and the risk of all microvascular complications in patients with T1DM,26,27 while the genetic background of diabetic nephropathy is well determined.28-30 It seems that some patients with T1DM do not develop diabetic nephropathy despite long-term glycemic exposure. The German Diabetes Documentation System reported any retinopathy in more than 80% of individuals with diabetes duration longer than 40 years.31 Although most of our study participants were diagnosed with either nonproliferative or proliferative retinopathy, no case of blindness was observed, which is consistent with the study entry criteria. In the Medalist Study, the diagnosis of retinopathy was reported in 53.4% of patients.13 Peripheral polyneuropathy was also common, as it was diagnosed in more than half of participants. This is in line with findings from a population-based cohort study by Dyck et al.32 Unlike diabetic nephropathy, both retinopathy and neuropathy seem to depend more on environmental factors, such as glycemic exposure, than on hereditary factors.

Interestingly, in our highly selected population of patients with T1DM and well-preserved kidney function, there was no association between the HbA1c level and proliferative retinopathy or albuminuria. This may be explained by the fact that HbA1c levels were measured at a single time point, and no long-term data on glycemic control were available. Moreover, the mean HbA1c level was relatively close to the recommended target. Of note, the Medalist Study did not report such an association either.13 Our findings of high HDL cholesterol levels, low insulin requirement, and near-normal BMI as potential factors protecting against advanced complications and premature death are also in line with the results of the Medalist Study and the Golden Years Cohort (Table 4).13-15 In those studies, HDL cholesterol levels were higher by about 0.3 mmol/l than those reported in a population-based study by Eeg-Olofsson et al.33 High HDL cholesterol levels are known to have a strong genetic background and to be associated with lower cardiovascular risk.34-36 Of note, while most of our patients were on statins, this class of lipid-lowering drugs seems to have limited impact on the HDL cholesterol level in autoimmune diabetes.37,38

Table 4. Clinical and biochemical characteristics of patients in the current study vs the Joslin 50-Year Medalist Study and the Golden Years Cohort

Parameter

Polish cohort

Joslin 50-Year Medalist Study13

Golden Years Cohort14

No. of participants

45

326

400

Male sex, %

55

45.3

54

Age at diagnosis, y

14.6

12.6

13.7

Age at examination, y

59.2

69.5

68.9

Diabetes duration, y

44.6

57.1

55.8

BMI, kg/m2

26.4

24.5

25.0

HbA1c, %

7.6

7.0a

7.6

DDI, U/kg

0.48

0.50

0.52

Triglycerides, mmol/l

1.1

–

1.49

HDL cholesterol, mmol/l

1.9

1.75

1.84

LDL cholesterol, mmol/l

2.6

–

–

Hypertension, %

69

51

–

Proliferative retinopathy, %

71.1

48.1

–

Creatinine, µmol/l

77

–

125

eGFR, ml/min/1.73 m2

82.2

–

–

Albuminuria, %

16

–

35

Neuropathy, %

62.2

53.1

–

Smoking (current or past), %

13

–

64

CAD, %

38

–

34

MI / stroke, %

22.2

–

–

Data presented as means unless indicated otherwise.

a Median

Abbreviations: see Table 1

It was reported that patients with an established diagnosis of T1DM and a positive family history of diabetes are frequently misdiagnosed and the actual disease is MODY. For example, a study assessing participants in the Czech T1DM Prediction Programme revealed a significant proportion of MODY in families where at least 2 family members were affected by diabetes and the proband had an initial clinical diagnosis of T1DM. The authors reported MODY in 45% of families with multiple occurrences of diabetes.39 Genetic testing performed within the expanded Joslin Medalist Study in a group of patients with long-duration T1DM showed that almost 8% of the population were carriers of a likely pathogenic variant in monogenic diabetes genes.16 In our study, a positive family history of diabetes was reported by 16 of the 45 individuals. Overall, we identified 10 variants, but we were not able to confirm that any of them had a causative role in the disease. The Arg1530Cys variant was previously reported in the Norwegian cohort of children with a clinical diagnosis of T1DM but absence of T1DM-related autoantibodies.40 Functional analyses performed in that study suggested that the variant was involved in the pathogenesis of diabetes in the carrier. However, sulfonylurea treatment was unsuccessful, most probably due to the fact that the patient developed autoimmune diabetes.40

Next, the Arg1530Cys missense mutation was present in the ClinVar database with an annotation of uncertain significance. The 2 null variants in the ABCC8 gene found in our patients were not the cause of diabetes, because the biallelic null variants in this gene cause hyperinsulinism, and not diabetes. Therefore, diabetic individuals heterozygous for the ABCC8 null variant detected by next-generation sequencing were incidental carriers of hyperinsulinism.41 Additionally, the presence of the Val849Ile variant in 4 heterozygotes in the gnomAD database (rs770722134) suggests that it is likely a benign or a recessive hyperinsulinism variant. Overall, it was an unlikely cause of diabetes in our patient. It is also unlikely that GCK-MODY was the only etiology of the disease in our carriers of the GCK variants. However, we cannot exclude that T1DM was superimposed on monogenic GCK-related diabetes. The rare His336Asp variant in the HNF1B gene was reported in a Spanish pediatric cohort with diabetes and negative autoimmunity.42 This variant was also reported in patients with kidney disorders but not diabetes.43 The Arg583Gln variant was initially described as a likely causative mutation in T2DM or MODY cohorts.44-46 However, more recent studies did not confirm its pathogenicity—it was also present in a nondiabetic population at a frequency of 2% to 3%.47,48

Of note, half of the variants detected in our study occurred in the ABCC8 gene. This large gene containing 39 exons was previously described as highly polymorphic, which makes it difficult to interpret the identified variants in the context of diabetes.41 Unless there is clear evidence for the presence of neonatal diabetes, MODY-like diabetes with sensitivity to sulfonylurea treatment, or sufficient cosegregation with diabetes in the patient or family members, the novel ABCC8 missense variants should not be reported as causative ones. Therefore, in the absence of additional supporting clinical information, the 3 missense ABCC8 variants identified in our study should be considered as of uncertain significance and should not be reported as the cause of diabetes. The confirmation of monogenic diabetes in such patients might be important for possible modification of treatment and the introduction of sulfonylurea therapy.49 Still, it is unlikely that switching from insulin therapy to oral hypoglycemic agents would be successful in patients with a diabetes duration of more than 40 years, even if monogenic diabetes was confirmed.50 Detecting monogenic diabetes is also important for predicting the course of diabetes in subsequent generations, including the early institution of targeted therapy.

Our study has several limitations. First, the sample size was small in comparison with the 50-Years Medalist Study and the Golden Years Study. Second, diabetes duration in our cohort was shorter by 10 years compared with the 2 other studies. Third, our patients were not assessed for the presence of human leukocyte antigen genotypes for the risk of T1DM or the presence of T1DM autoantibodies. Fourth, the different types of cardiovascular disease were diagnosed on the basis of medical records and questionnaires. Finally, as there was no control group, the presence of variants in monogenic diabetes genes was not tested in the general T1DM population.

In conclusion, patients with long-term T1DM and well-preserved kidney function were characterized by good glycemic control, high HDL cholesterol levels, low insulin requirement, and near-normal BMI. Factors associated with proliferative retinopathy and / or albuminuria included diabetes duration and LDL cholesterol levels. The only factor associated with a composite cardiovascular end point (myocardial infarction and / or stroke) was smoking. Mutations in monogenic diabetes genes were rare in our population.