ABSTRACT
Objectives
We aimed to identify microribonucleic acids (miRNAs) that could serve as biomarkers for slowly progressive type 1 diabetes mellitus (SPIDDM).
Materials and Methods
We conducted a retrospective study of adult patients with SPIDDM who fulfilled the criteria for SPIDDM (definite) (n = 60), type 2 diabetes mellitus (T2DM) (n = 59), and healthy controls (n = 50) to analyze the serum expression levels of 40 miRNAs reported to be altered in latent autoimmune diabetes in adults and type 1 diabetes using quantitative real‐time polymerase chain reaction, specifically analyzing serum collected at the time of diagnosis as SPIDDM (probable) in the SPIDDM group. For miRNAs showing significant differences between the groups, we performed logistic regression analysis. In the SPIDDM group, we performed linear regression analysis on miRNAs correlated with the C‐peptide index (CPI) at the last outpatient visit as identified by Spearman's rank correlation.
Results
The serum expression levels of miR‐143‐3p, miR‐21‐5p, miR‐223‐3p, and miR‐517b‐3p differed significantly in the SPIDDM group compared with the type 2 diabetes mellitus and control groups. Multivariate logistic regression analysis distinguishing SPIDDM from type 2 diabetes mellitus revealed miR‐21‐5p, miR‐223‐3p, and miR‐517b‐3p as independent discriminators. Correlation and multivariate linear regression analyses revealed that miR‐150‐5p was selected as an independent predictor of CPI at the last outpatient visit.
Conclusion
This study identified serum miRNAs as potential biomarkers in adult patients with SPIDDM, particularly miR‐150‐5p as a candidate marker for predicting subsequent decline in insulin secretion.
Keywords: Microribonucleic acids, Slowly progressive type 1 diabetes mellitus
This study evaluated serum miRNAs as biomarkers in adult patients with SPIDDM. Serum miRNAs were measured at the time of SPIDDM (probable) diagnosis, and miR‐150‐5p was inversely associated with CPI at the last outpatient visit, suggesting its potential utility for predicting subsequent insulin secretory decline.
INTRODUCTION
Type 1 diabetes mellitus is characterized by autoimmune destruction of pancreatic β‐cells, leading to insulin dependence. 1 Slowly progressive T1DM (SPIDDM) is characterized by islet‐associated autoantibodies, such as glutamic acid decarboxylase autoantibody (GADA), and gradual β‐cell functional decline without ketosis or ketoacidosis at diabetes diagnosis. 2 , 3 Because patients with SPIDDM often initially resemble those with type 2 diabetes mellitus, they may be followed as having type 2 diabetes mellitus before SPIDDM is established. Once SPIDDM is suspected or diagnosed, predicting subsequent decline in endogenous insulin secretion becomes clinically important. Insulin therapy is useful for treating SPIDDM. 4 However, according to the latest clinical practice guidelines of the Japan Diabetes Society, DPP4 inhibitors and metformin can also be considered when insulin secretory capacity is maintained. 5 Thus, therapeutic decisions depend partly on residual insulin secretion, and biomarkers predicting its decline may help optimize follow‐up and treatment.
Diagnostic biomarkers for SPIDDM include islet‐associated autoantibodies, such as GADA and insulinoma‐associated antigen‐2 autoantibody (IA‐2A), which are included in the diagnostic criteria. 6 , 7 However, these markers do not fully predict subsequent β‐cell functional decline. Because some patients with non‐fulminant ttype 1 diabetes mellitus test negative for GADA, 8 whereas GADA can be detected in nondiabetic autoimmune diseases, 9 , 10 additional biomarkers are needed to predict insulin secretory decline after the stage of SPIDDM (probable).
Microribonucleic acids (miRNAs) are small single‐stranded noncoding RNAs 19–22 nucleotides in length that regulate gene expression by degrading or suppressing translation of target mRNAs. 11 , 12 MiRNAs are stably present in body fluids, including blood, urine, and saliva, 13 , 14 , 15 and can be detected in stored serum samples, making them useful circulating biomarker candidates. 16 In diabetes, circulating miRNAs have been associated with β‐cell dysfunction, immune‐mediated β‐cell injury, and metabolic abnormalities, 17 , 18 , 19 , 20 suggesting their potential to predict deterioration of endogenous insulin secretion. Latent autoimmune diabetes in adults (LADA) shares features with SPIDDM, 21 , 22 , 23 and altered circulating miRNA profiles have been reported in LADA or type 1 diabetes mellitus. 24 , 25 , 26 , 27 , 28 However, no study has examined circulating miRNAs in SPIDDM or their association with subsequent insulin secretory decline.
Therefore, in this study, we aimed to identify serum miRNAs that could serve as biomarkers for adult patients initially diagnosed with SPIDDM (probable), with particular emphasis on their potential to predict subsequent decline in insulin secretion.
MATERIALS AND METHODS
Study design and population
In this retrospective observational study, we enrolled participants using the procedure shown in Figure 1. Specifically, we excluded 49 patients from the 119 patients diagnosed with SPIDDM (probable) at University of Yamanashi Hospital between April 2006 and December 2022 according to the exclusion criteria. Of the remaining 70 patients, 60 patients who fulfilled the diagnostic criteria for SPIDDM (definite) at the last outpatient observation in December 2022 were included in the analysis of circulating miRNAs. The diagnosis of SPIDDM (probable) and SPIDDM (definite) was based on the latest diagnostic criteria published by the Japan Diabetes Society in 2023. 6 , 7 For the comparative analysis of miRNA expression levels in patients with SPIDDM, 59 patients with type 2 diabetes mellitus admitted to the University Hospital of Yamanashi and 50 healthy adult volunteers who underwent a medical check‐up at Isawa Hot Spring Hospital were included. The study was approved by the certified review boards of the University of Yamanashi and Isawa Hot Spring Hospital. Informed consent was waived owing to the retrospective nature of the study, and information regarding the study was disclosed through an online opt‐out process. Participants were given the opportunity to decline participation.
Figure 1.
Flowchart showing the participant selection process. The process of selecting study participants is shown. SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Clinical data collection
Figure 2 shows the timeline of this study, which focused specifically on the SPIDDM group. In the SPIDDM group, data were collected at the time of diagnosis as SPIDDM (probable), and only data for the C‐peptide index (CPI) were also collected at the last outpatient clinic visit when all included patients had fulfilled the criteria for SPIDDM (definite). Specifically, data on demographics, physiological measurements, complications of DM, laboratory test outcomes, and prescribed medications were collected. Blood samples were collected in the morning after fasting for at least 12 h. Hemoglobin A1c (HbA1c) levels were measured using latex agglutination, and fasting plasma glucose (FPG) levels were measured using a glucose oxidase method. GADA titers were measured by enzyme immunoassay (EIA) (Cosmic Co., Tokyo, Japan), whereas IA‐2A titers were measured by radioimmunoassay kits (RIA) (Cosmic Co.). Endogenous insulin secretion capacity was evaluated using the CPI, calculated as fasting C‐peptide (ng/mL) divided by FPG (mg/dL) × 100.
Figure 2.
Timeline of this study: CPI, C‐peptide index; DM, diabetes mellitus; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
HLA typing
In the SPIDDM group only, human leukocyte antigen (HLA) genotypes were determined using residual blood samples obtained at the time of diagnosis as SPIDDM (probable) by the polymerase chain reaction (PCR)‐reverse sequence‐specific oligonucleotide method. 29 Based on a previous report, 30 HLA genotypes were classified as DR2 positivity, DR4/DR4, DR4/DR8, DR4/DR9, DR8/DR9, DR9/DR9, or others. DR2 was defined as DRB1*15:01‐DQB1*06:02 or DRB1*15:02‐DQB1*06:01; DR4/DR4 as DRB1*04:05‐DQB1*04:01/DRB1*04:05‐DQB1*04:01; DR4/DR8 as DRB1*04:05‐DQB1*04:01/DRB1*08:02‐DQB1*03:02; DR4/DR9 as DRB1*04:05‐DQB1*04:01/DRB1*09:01‐DQB1*03:03; DR8/DR9 as DRB1*08:02‐DQB1*03:02/DRB1*09:01‐DQB1*03:03; and DR9/DR9 as DRB1*09:01‐DQB1*03:03/DRB1*09:01‐DQB1*03:03.
Serum miRNA extraction and quantitative real‐time polymerase chain reaction
Serum miRNA levels were analyzed in all participants, with serum collected at diagnosis as SPIDDM (probable) used for the SPIDDM group. Fasting blood samples were centrifuged within 1 h, and serum was stored at −80°C until analysis. Total RNA was extracted from 200 μL of serum using a miRNeasy Serum/Plasma Kit (Qiagen, Valencia, CA, USA), and reverse transcription was performed using a MIR‐x miRNA First‐Strand Synthesis Kit (Takara Bio USA, San Jose, CA, USA), according to the manufacturers' instructions. Quantitative real‐time PCR (qRT‐PCR) was performed using an ABI PRISM 7900 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) with a miRCURY LNA miRNA PCR System (Qiagen). Relative miRNA expression was calculated using the 2−ΔΔCT method, with hsa‐miR‐16‐5p as the internal control. 28 This internal control showed minimal variation across the three groups, confirming its stability (Figure S1). All miRNA measurements were performed twice; when the difference exceeded three standard deviations, the assay was repeated, and the reanalyzed values were used. Expression levels were compared after adjusting the mean value in the SPIDDM group to 1. Using these protocols, 40 serum miRNAs previously reported to be altered in patients with LADA (Table S1) or type 1 diabetes mellitus (Table S2) were measured. Type 1 diabetes mellitus‐associated miRNAs were selected if they had been reported in at least two studies, with reference to a previous review. 31 The target miRNAs are listed in Table S3. Custom‐designed primers were used for miR‐517b‐3p and miR‐720, and predesigned primers were used for the other miRNAs. Although miR‐720 has been withdrawn from miRBase and is currently regarded as a transfer RNA‐derived small RNA, it was included based on previous reports demonstrating its clinical relevance.
Statistical analysis
Statistical analyses were performed using Prism 9 (GraphPad Software Inc., San Diego, CA, USA). Normally distributed continuous variables were compared using the t‐test or analysis of variance, whereas non‐normally distributed variables were compared using the Mann–Whitney U test or Kruskal–Wallis test. Nominal variables were analyzed using Fisher's exact test, and correlations were assessed using Spearman's rank correlation analysis. In multivariate logistic and linear regression analyses, Model 1 used a stepwise procedure including age, sex, body mass index (BMI), current smoking status, duration of diabetes, duration of SPIDDM, physiological measurements, diabetic complications, laboratory parameters, and medications. In logistic regression, Models 2, 3, and 4 additionally included duration of diabetes, duration of diabetes plus HbA1c, and duration of diabetes plus CPI as forced‐entry variables, respectively. In linear regression, Models 2, 3, and 4 additionally included age plus female sex, duration of diabetes plus HbA1c, and IA‐2A positivity plus DR2 positivity as forced‐entry variables, respectively. The receiver operating characteristic (ROC) curve analysis was performed to evaluate diagnostic performance. Non‐normally distributed continuous variables, including duration of diabetes, were natural log‐transformed before inclusion in multivariable regression analyses. For additional analyses of candidate miRNAs, based on previous reports, 30 , 32 , 33 , 34 patients were stratified by GADA titer using a cutoff value of 180 U/mL, islet‐associated autoantibody status as single GADA positivity or GADA plus IA‐2A positivity, age at SPIDDM (probable) diagnosis using a cutoff value of 45 years, and HLA genotype as DR2 positivity, DR4/DR4, or other genotypes. A P‐value <0.05 was considered statistically significant.
RESULTS
Comparison of the clinical characteristics of all participants
The clinical characteristics of participants subjected to serum miRNA analysis are given in Table 1. HLA genotype is also presented for the SPIDDM group. All 60 patients in the SPIDDM group were GADA‐positive (titer: ≥5.0 U/mL). Age at serum collection was significantly younger in the SPIDDM group than in the other groups, while BMI was lower than in the type 2 diabetes mellitus group and higher than in healthy controls. No significant differences were observed between the two diabetes groups in sex, smoking status, blood pressure, diabetic complications, FPG, HbA1c, or CPI.
Table 1.
Baseline characteristics of the study participants
| Overall (n = 169) | ||||
|---|---|---|---|---|
| SPIDDM n = 60 | T2DM n = 59 | Healthy controls n = 50 | P‐value | |
| Clinical characteristics | ||||
| Age at the time of serum collection for miRNA analysis (years) | 48.0 ± 15.4 | 62.4 ± 12.0*** | 61.8 ± 8.5*** | <0.001 |
| Age at the last outpatient clinic visit (years) | 57.9 ± 15.4 | N/A | N/A | |
| Female sex, n (%) | 34 (57) | 25 (42) | 22 (44) | 0.255 |
| BMI (kg/m2) | 24.5 ± 3.7 | 26.3 ± 5.5* | 22.5 ± 2.9** | <0.001 |
| Current Smoker, n (%) | 20 (33) | 20 (34) | 17 (34) | 1.000 |
| Duration of DM at the time of serum collection for miRNA analysis (years) | 5 (3–7) | 8 (4–18)*** | N/A | |
| Duration of SPIDDM at the last outpatient clinic visit (years) | 6 (4–8) | N/A | N/A | |
| Physiological testing | ||||
| SBP (mmHg) | 132 ± 19 | 133 ± 22 | 117 ± 10*** | <0.001 |
| DBP (mmHg) | 74 ± 13 | 73 ± 10 | 73 ± 8 | 0.772 |
| Diabetic complications, n (%) | ||||
| Neuropathy | 32 (53) | 34 (58) | 0 (0) | |
| Retinopathy | 19 (32) | 19 (32) | 0 (0) | |
| Nephropathy | 27 (45) | 33 (56) | 0 (0) | |
| Laboratory parameters | ||||
| FPG (mg/dL) | 186 (127–208) | 174 (137–212) | 96 (90–103)*** | <0.001 |
| HbA1c (%) | 8.6 (6.6–10.4) | 8.7 (6.8–11.1) | 5.6 (5.4–5.7)*** | <0.001 |
| GADA positivity, n (%) | 60 (100) | 0 (0) | N/A | |
| GADA titer (U/mL) | 30 (13–132) | <5 | N/A | |
| IA‐2A positivity, n (%) | 17 (28) | 0 (0) | N/A | |
| CPI at the time of serum collection for miRNA analysis | 1.08 (0.91–1.43) | 1.11 (0.79–1.85) | N/A | |
| CPI at the last outpatient clinic visit | 0.43 ± 0.21 | N/A | N/A | |
| Total cholesterol (mg/dL) | 198 (177–236) | 203 (171–246) | 201 (181–213) | 0.829 |
| Triglycerides (mg/dL) | 119 (93–158) | 129 (108–203) | 75 (55–106)*** | <0.001 |
| AST (IU/L) | 24 (19–29) | 25 (20–31) | 17 (13–20) | <0.001 |
| ALT (IU/L) | 31 (22–38) | 31 (23–38) | 21 (16–27) | <0.001 |
| eGFR (ml/min/1.73 m2) | 78.0 ± 12.4 | 75.9 ± 18.0 | 71.4 ± 9.7** | 0.047 |
| HLA genotypes | ||||
| DR2 positivity † | 16 (27) | N/A | N/A | |
| DR4/DR4 ‡ | 5 (8) | N/A | N/A | |
| DR4/DR8 § | 0 (0) | N/A | N/A | |
| DR4/DR9 ¶ | 3 (5) | N/A | N/A | |
| DR8/DR9 # | 0 (0) | N/A | N/A | |
| DR9/DR9 †† | 2 (3) | N/A | N/A | |
| Others | 34 (57) | N/A | N/A | |
| Prescribed medication, n (%) | ||||
| Insulin | 16 (27) | 14 (24) | 0 (0) | |
| Sulfonylureas | 4 (7) | 5 (8) | 0 (0) | |
| Glinides | 5 (8) | 6 (10) | 0 (0) | |
| GLP‐1 RAs | 14 (23) | 16 (27) | 0 (0) | |
| DPP4 inhibitors | 20 (33) | 24 (41) | 0 (0) | |
| Metformins | 35 (58) | 33 (56) | 0 (0) | |
| TZDs | 5 (8) | 3 (5) | 0 (0) | |
| Alpha‐GIs | 2 (3) | 4 (7) | 0 (0) | |
| SGLT2 inhibitors | 34 (57) | 31 (53) | 0 (0) | |
| ARBs | 28 (47) | 31 (53) | 0 (0) | |
| CCBs | 18 (30) | 14 (24) | 0 (0) | |
| Statins | 30 (50) | 32 (54) | 0 (0) | |
| Fibrates | 18 (30) | 21 (24) | 0 (0) | |
Data are expressed as the mean ± standard deviation, the median (interquartile range), or number (%), and are compared among the three groups using one‐way analysis of variance, the Kruskal–Wallis test, or Fisher's exact test. Data are compared between each pair of two groups using Student's t‐test, the Mann–Whitney U test, or Fisher's exact test. *P < 0.05, **P < 0.01, ***P < 0.001 vs. SPIDDM. Items without a specified time statement were assessed at the time of serum collection for miRNA analysis.
ALT, alanine transaminase; ARBs, angiotensin receptor blockers; AST, aspartate transaminase; BMI, body mass index; CCB, calcium channel blocker; CPI, C‐peptide index; DBP, diastolic blood pressure; DKD, diabetic kidney disease; DM, diabetes mellitus; DPP4, dipeptidyl peptidase‐4; eGFR, estimated glomerular filtration rate; FPG, fasting blood glucose; GADA, glutamic acid decarboxylase autoantibody; GIs, glycosidase inhibitors; GLP‐1 RA, glucagon‐like peptide‐1 receptors agonist; HbA1c, hemoglobin A1c; HLA, human leukocyte antigen; IA‐2A, insulinoma‐associated antigen‐2 autoantibody; N/A, not available; SBP, systolic blood pressure; SGLT2, sodium–glucose cotransporter 2; SPIDDM, slow progressive type 1 diabetes mellitus; TZDs, thiazolidinediones; T2DM, type 2 diabetes mellitus.
DRB1*15:01‐DQB1*06:02 or DRB1*15:02‐DQB1*06:01.
DRB1*04:05‐DQB1*04:01/DRB1*04:05‐DQB1*04:01.
DRB1*04:05‐DQB1*04:01/DRB1*08:02‐DQB1*03:02.
DRB1*04:05‐DQB1*04:01/DRB1*09:01‐DQB1*03:03.
DRB1*08:02‐DQB1*03:02/DRB1*09:01‐DQB1*03:03.
DRB1*09:01‐DQB1*03:03/DRB1*09:01‐DQB1*03:03.
Serum miRNA expression levels among the three groups
Serum expression levels of 40 miRNAs reported to be altered in LADA or type 1 diabetes mellitus were analyzed by qRT‐PCR in the SPIDDM, type 2 diabetes mellitus, and healthy control groups. Four miRNAs were significantly altered in the SPIDDM group compared with the other groups (Figure 3). Specifically, miR‐143‐3p, miR‐21‐5p, and miR‐223‐3p levels were significantly higher, whereas miR‐517b‐3p levels were significantly lower in the SPIDDM group. No significant differences were observed in the other 36 miRNAs (Figure S2).
Figure 3.
Results of the qRT‐PCR analysis of four differentially expressed miRNAs among the three groups: patients with SPIDDM (n = 60), T2DM (n = 59), and healthy controls (n = 50). Data were compared using the Mann–Whitney U test, with two groups compared at a time. *P < 0.05, **P < 0.01, ***P < 0.001 vs. SPIDDM. Controls, healthy controls; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Logistic regression analysis to identify the miRNAs for differentiating SPIDDM from T2DM
To examine whether these four miRNAs could discriminate patients with SPIDDM from those with type 2 diabetes mellitus, univariate and multivariate logistic regression analyses were performed. In univariate analysis, all four miRNAs were significant factors, along with age, BMI, and duration of diabetes (Table S4). After adjustment for age and sex, miR‐21‐5p, miR‐223‐3p, and miR‐517b‐3p remained significant (Table 2). In multivariate model 1, adjusted for age, sex, and BMI, these three miRNAs were selected as significant factors (Table 2). They also remained significant in models 2, 3, and 4, suggesting that these miRNAs may contribute to the disease characterization of SPIDDM.
Table 2.
Logistic regression models examining the association between each miRNA and its ability to discriminate patients with SPIDDM (probable) from those with T2DM
| Dependent variables: patients with SPIDDM (probable) (control: patients with T2DM) | |||
|---|---|---|---|
| Independent variables | OR | 95% CI | P‐value |
| Univariate model | |||
| miR‐143‐3p † | 1.720 | 1.090–2.720 | 0.020 |
| miR‐21‐5p † | 5.330 | 3.100–9.160 | <0.001 |
| miR‐223‐3p † | 2.240 | 1.750–2.870 | <0.001 |
| miR‐517b‐3p † | 0.666 | 0.514–0.864 | 0.002 |
| Age and gender adjusted model | |||
| miR‐143‐3p † | 1.340 | 0.792–2.260 | 0.276 |
| miR‐21‐5p † | 4.990 | 2.800–8.900 | <0.001 |
| miR‐223‐3p † | 2.250 | 1.700–3.000 | <0.001 |
| miR‐517b‐3p † | 0.687 | 0.516–0.914 | 0.010 |
| Multivariate model 1 ‡ | |||
| miR‐143‐3p † | 1.380 | 0.798–2.370 | 0.251 |
| miR‐21‐5p † | 5.540 | 2.890–10.600 | <0.001 |
| miR‐223‐3p † | 2.320 | 1.700–3.180 | <0.001 |
| miR‐517b‐3p † | 0.650 | 0.480–0.879 | 0.005 |
| Multivariate model 2 § | |||
| miR‐143‐3p † | 1.350 | 0.784–2.320 | 0.281 |
| miR‐21‐5p † | 5.550 | 2.860–10.800 | <0.001 |
| miR‐223‐3p † | 2.480 | 1.730–3.560 | <0.001 |
| miR‐517b‐3p † | 0.630 | 0.460–0.865 | 0.004 |
| Multivariate model 3 ¶ | |||
| miR‐143‐3p † | 1.180 | 0.673–2.090 | 0.557 |
| miR‐21‐5p † | 5.490 | 2.810–10.700 | <0.001 |
| miR‐223‐3p † | 2.520 | 1.770–3.590 | <0.001 |
| miR‐517b‐3p † | 0.628 | 0.454–0.870 | 0.005 |
| Multivariate model 4 # | |||
| miR‐143‐3p † | 1.310 | 0.758–2.270 | 0.331 |
| miR‐21‐5p † | 5.580 | 2.820–11.000 | <0.001 |
| miR‐223‐3p † | 2.470 | 1.740–3.510 | <0.001 |
| miR‐517b‐3p † | 0.643 | 0.469–0.881 | 0.006 |
All miRNAs and adjustment variables that showed non‐normal distributions, including duration of DM, HbA1c, and CPI, were log‐transformed using the natural logarithm. CI, confidence interval; OR, odds ratio; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Log‐transformed using the natural logarithm.
Each miRNA was entered into the multivariate model along with age, sex, and BMI at the time of collecting serum for analyzing miRNAs.
Each miRNA was entered into the multivariate model along with age, sex, BMI, and duration of DM at the time of collecting serum for analyzing miRNAs.
Each miRNA was entered into the multivariate model along with age, sex, BMI, duration of DM, and HbA1c at the time of collecting serum for analyzing miRNAs.
Each miRNA was entered into the multivariate model along with age, sex, BMI, duration of DM, and CPI at the time of collecting serum for analyzing miRNAs.
ROC curve analysis of miRNAs selected as independent discriminators
ROC curve analysis was performed to evaluate the ability of selected miRNAs to discriminate the SPIDDM group from the type 2 diabetes mellitus group (Table 3). The combination of miR‐21‐5p, miR‐223‐3p, and miR‐517b‐3p yielded an AUC of 0.971, indicating better diagnostic performance than individual miRNAs or any two‐miRNA combination. In addition, the combination of all four miRNAs identified in Figure 3 yielded an AUC of 0.981. These findings suggest that individual miRNAs showed diagnostic performance, while combining multiple miRNAs further improved discrimination between SPIDDM and type 2 diabetes mellitus.
Table 3.
Receiver operating characteristic curve analysis of each miRNA comparing its diagnostic performance in discriminating patients with SPIDDM (probable) from those with T2DM
| Dependent variables: patients with SPIDDM (probable) (control: patients with T2DM) | |||
|---|---|---|---|
| Independent variables | Sensitivity | Specificity | AUC |
| miR‐21‐5p † | 1.000 | 0.831 | 0.925 |
| miR‐223‐3p † | 1.000 | 0.831 | 0.924 |
| miR‐517b‐3p † | 0.283 | 0.983 | 0.638 |
| miR‐21‐5p † + miR‐223‐3p † | 1.000 | 0.932 | 0.958 |
| miR‐21‐5p † + miR‐517b‐3p † | 0.983 | 0.847 | 0.945 |
| miR‐223‐3p † + miR‐517b‐3p † | 0.967 | 0.847 | 0.953 |
| miR‐21‐5p † + miR‐223‐3p † + miR‐517b‐3p † | 1.000 | 0.932 | 0.971 |
| miR‐143‐3p † + miR‐21‐5p † + miR‐223‐3p † + miR‐517b‐3p † | 0.983 | 0.932 | 0.981 |
AUC, area under the curve; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Log‐transformed using the natural logarithm.
Analysis of miRNAs predicting decreased insulin secretion in the SPIDDM group
Because insulin secretion gradually decreases after GADA detection in SPIDDM, 6 , 7 we examined whether serum miRNA expression at diagnosis as SPIDDM (probable) predicted subsequent insulin secretory decline. Correlation analysis was performed between CPI at the last outpatient visit and serum expression levels of the 40 miRNAs. The level of miR‐150‐5p was significantly inversely correlated with CPI (Figure 4), whereas the other 39 miRNAs showed no correlation (data not shown).
Figure 4.
Spearman's rank correlation analysis between the CPI at the last outpatient clinic visit and the individual expression levels of the 40 miRNAs. The results of the correlation analysis that were determined to be significant are presented. CPI, C‐peptide index; ρ, Spearman's correlation coefficient.
We then performed univariate and multivariate linear regression analyses using CPI at the last outpatient visit as the dependent variable. In univariate analysis, miR‐150‐5p was a significant factor, along with duration of SPIDDM, nephropathy, GADA titer, CPI at SPIDDM (probable) diagnosis, and DR2 positivity (Table S5). After adjustment for age and sex, miR‐150‐5p remained significant (Table 4). In multivariate model 1, adjusted for duration of SPIDDM, GADA titer, and CPI at SPIDDM (probable) diagnosis, miR‐150‐5p was selected as a significant factor (Table 4). It also remained significant in models 2, 3, and 4, suggesting its potential as a biomarker for predicting decreased insulin secretion in patients with SPIDDM (probable).
Table 4.
Multivariate linear regression models examining the association between miR‐150‐5p and the CPI at the last outpatient clinic visit in the SPIDDM group
| Dependent variables: CPI at the last outpatient clinic visit | |||
|---|---|---|---|
| Independent variables | Standardized β | 95% CI | P‐value |
| Univariate model | (Adjusted R 2 = 0.137) | ||
| miR‐150‐5p † | −0.049 | −0.080 to −0.019 | 0.002 |
| Age and gender adjusted model | (Adjusted R 2 = 0.155) | ||
| miR‐150‐5p † | −0.052 | −0.082 to −0.021 | 0.001 |
| Age (years) | 0.001 | −0.002 to 0.005 | 0.400 |
| Female sex | 0.075 | −0.029 to 0.179 | 0.156 |
| Multivariate model 1 | (Adjusted R 2 = 0.379) | ||
| miR‐150‐5p † | −0.034 | −0.062 to −0.006 | 0.018 |
| Duration of SPIDDM (years) † ‡ | −0.081 | −0.153 to −0.010 | 0.026 |
| GADA titer (U/mL) † | −0.040 | −0.067 to −0.012 | 0.006 |
| CPI at SPIDDM (probable) diagnosis † | 0.162 | 0.046–0.279 | 0.007 |
| Multivariate model 2 | (Adjusted R 2 = 0.381) | ||
| miR‐150‐5p † | −0.035 | −0.063 to −0.007 | 0.014 |
| Age (years) | 0.001 | −0.002 to 0.004 | 0.557 |
| Female sex | 0.059 | −0.033 to 0.150 | 0.205 |
| Duration of SPIDDM (years) † ‡ | −0.083 | −0.154 to −0.011 | 0.024 |
| GADA titer (U/mL) † | −0.039 | −0.067 to −0.012 | 0.006 |
| CPI at SPIDDM (probable) diagnosis † | 0.143 | 0.024–0.262 | 0.020 |
| Multivariate model 3 | (Adjusted R 2 = 0.417) | ||
| miR‐150‐5p † | −0.035 | −0.062 to −0.008 | 0.012 |
| Duration of DM (years) † § | −0.037 | −0.093 to 0.019 | 0.194 |
| Duration of SPIDDM (years) † ‡ | −0.076 | −0.145 to −0.007 | 0.032 |
| HbA1c (%) † | 0.148 | −0.026 to 0.322 | 0.093 |
| GADA titer (U/mL) † | −0.041 | −0.068 to −0.014 | 0.003 |
| CPI at SPIDDM (probable) diagnosis † | 0.136 | 0.017–0.254 | 0.026 |
| Multivariate model 4 | (Adjusted R 2 = 0.494) | ||
| miR‐150‐5p † | −0.026 | −0.052 to −0.001 | 0.045 |
| IA‐2A positivity | −0.116 | −0.213 to −0.019 | 0.020 |
| DR2 positivity ¶ | 0.186 | 0.084–0.288 | <0.001 |
| Duration of SPIDDM (years) † ‡ | −0.050 | −0.117 to 0.016 | 0.136 |
| GADA titer (U/mL) † | −0.041 | −0.066 to −0.016 | 0.002 |
| CPI at SPIDDM (probable) diagnosis † | 0.099 | −0.011 to 0.209 | 0.076 |
Items without a specified time statement were assessed at the time of SPIDDM (probable) diagnosis. CI, confidence interval; CPI, C‐peptide index; DM, diabetes mellitus; GADA, glutamic acid decarboxylase autoantibody; HbA1c, hemoglobin A1c; IA–2A, insulinoma‐associated antigen‐2 autoantibody; β, partial regression coefficient.
Log‐transformed using the natural logarithm.
Duration of SPIDDM at the last outpatient clinic visit.
Duration of DM at the time of SPIDDM (probable) diagnosis.
DRB1*15:01‐DQB1*06:02 or DRB1*15:02‐DQB1*06:01.
Correlation between GADA titers and the expression levels of miRNAs in the SPIDDM group
We examined the correlation between GADA titers and the expression levels of the four candidate miRNAs that were found to be useful in this study for differentiating SPIDDM from type 2 diabetes mellitus or for predicting impaired insulin secretion in the SPIDDM group. The results showed that GADA titers were significantly negatively correlated with the expression levels of miR‐517b‐3p (Figure S3).
Additional stratified analyses of candidate miRNAs according to known progression‐related factors in the SPIDDM group
In autoimmune diabetes, high‐GADA titers, 32 multiple islet‐associated autoantibody positivity, 33 and younger age 34 have been associated with progression of insulin deficiency, whereas HLA genotypes including DR2 positivity and DR4/DR4 have been associated with preserved insulin secretion. 30 Thus, we compared candidate miRNA levels according to these factors. Consistent with these previous reports, CPI at the last outpatient visit was significantly lower in the high‐GADA titer (Table S6) and GADA plus IA‐2A‐positive groups (Table S7). CPI at the last outpatient visit did not differ according to age at SPIDDM (probable) diagnosis (Table S8) and was significantly higher in the DR2‐positive group (Table S9). However, candidate miRNA levels did not significantly differ according to GADA titer, GADA plus IA‐2A positivity, age < 45 vs ≥45 years at SPIDDM (probable) diagnosis, or HLA genotype classification (Tables S6–S9).
DISCUSSION
In this study, using the residual serum samples of patients with SPIDDM, type 2 diabetes mellitus, and healthy controls, we analyzed the expression levels of 40 circulating miRNAs. As a result, we identified several miRNAs as potential biomarkers for SPIDDM.
Because immediate insulin dependency is absent in the early disease stage and islet‐associated autoantibodies are not perfect biomarkers, SPIDDM may be misdiagnosed as type 2 diabetes mellitus. Previous studies have reported that 5–10% of Japanese patients treated as type 2 diabetes mellitus are GADA‐positive, 23 , 35 , 36 and that obesity may delay the diagnosis of SPIDDM by leading to treatment as type 2 diabetes mellitus. 37 Although combined measurement of multiple islet‐associated autoantibodies is useful for diagnosing SPIDDM, 10 , 38 the timing of autoantibody appearance varies widely, from several months to years before type 1 diabetes mellitus diagnosis. 39 , 40 , 41 Thus, additional biomarkers are needed for disease characterization and for predicting subsequent β‐cell functional decline in SPIDDM.
Studies of miRNAs in diabetes have suggested their potential as biomarkers for disease onset and progression. 17 , 18 , 19 , 20 , 24 , 25 , 26 , 27 , 28 However, no study has examined circulating miRNAs in SPIDDM. Therefore, this study aimed to identify serum miRNAs associated with SPIDDM, particularly those capable of predicting subsequent decline in insulin secretion at initial diagnosis of SPIDDM (probable).
Our study showed that three serum miRNAs were upregulated and one was downregulated in the SPIDDM group. In addition, miR‐517b‐3p was significantly correlated with GADA titers, whereas miR‐517b‐3p levels did not differ significantly between the high‐ and low‐GADA titer groups. This may indicate that the relationship between miR‐517b‐3p and GADA titers is modest or continuous and may not be fully captured by dichotomization using a cutoff value of 180 U/mL. Therefore, the association between miR‐517b‐3p and islet‐associated autoimmunity should be interpreted cautiously.
Logistic regression analysis identified three miRNAs as independent factors associated with SPIDDM, and ROC analysis showed that their combination achieved an AUC of ≥0.9 for discriminating SPIDDM from type 2 diabetes mellitus. Combining multiple miRNAs to improve diagnostic performance has been explored in cancer research. 42 , 43 , 44 Because multiple miRNAs can be measured from a single minimally invasive blood sample, this approach may also be useful for characterizing autoimmune diabetes.
The prediction of progressive insulin deficiency is important in the management of SPIDDM. Serum miRNA levels were measured at the time of diagnosis of SPIDDM (probable), before progression to SPIDDM (definite), and the miR‐150‐5p level was associated with CPI at the last outpatient visit. Notably, miR‐150‐5p was not identified as a marker that discriminated SPIDDM from type 2 diabetes mellitus, suggesting that it may not primarily reflect the onset or initial identification of SPIDDM (probable), but rather subsequent disease progression and β‐cell functional decline. In additional stratified analyses according to known factors related to the decline or preservation of insulin secretion, the levels of candidate miRNAs identified in the present study did not significantly differ among the subgroups. Thus, although miR‐150‐5p is not a substitute for established clinical and genetic factors associated with disease progression, it may provide complementary information as a marker of pathophysiological aspects of SPIDDM that are not directly determined by known progression‐related factors. In this context, miR‐150‐5p may help predict subsequent insulin secretory decline during the transition from SPIDDM (probable) to SPIDDM (definite). Furthermore, from the perspective of clinically applying the present findings to the management of SPIDDM, miRNA measurement may be most useful at the SPIDDM (probable) stage, when endogenous insulin secretion is still preserved, and may help identify patients at high risk of subsequent decline in insulin secretion.
The functions of each miRNA identified from our results have been examined in previous studies. miR‐143‐3p may be involved in the progression of autoimmune DM by controlling inflammatory responses through the suppression of Fos‐related antigen 2 expression, an important factor that determines T‐cell maturation and function. 27 It has been reported that β‐cell extracellular vesicle miR‐21‐5p cargo increases in response to inflammatory cytokines and may serve as a biomarker of type 1 diabetes mellitus. Thus, increased serum miR‐21‐5p in SPIDDM may reflect autoimmune inflammation or β‐cell stress. 45 In addition, transforming growth factor‐β has been hypothesized to link miR‐223‐3p and autoimmune DM. 24 Although miR‐517b‐3p has not been extensively studied, given its correlation with GADA titers, future studies should explore its involvement in autoimmune responses. Previous research has reported that serum miR‐150‐5p levels and insulin secretory capacity are inversely correlated in GADA‐positive patients. 41 This miRNA may be associated with the mechanism of pancreatic β‐cell dysfunction in type 1 diabetes mellitus. These pathways may serve as novel targets for elucidating the pathophysiology of SPIDDM onset and progression.
The present study has several limitations. First, only 40 miRNAs were analyzed. Although we considered performing comprehensive miRNA profiling using microarrays based on the present findings, this was not possible because of insufficient residual total RNA samples containing miRNAs. Therefore, additional miRNAs that were not evaluated in this study may also be associated with SPIDDM and require further investigation. Second, this was a retrospective study with a relatively small sample size. The number of participants was insufficient to simultaneously include all factors that may influence miRNA expression or SPIDDM pathogenesis in multivariate models. In addition, although we performed subgroup analyses according to known progression‐related factors, including GADA titer, islet‐associated autoantibody status, age, and HLA genotype, the number of patients in each subgroup was limited. Therefore, the relationships between established progression‐related factors and candidate miRNAs should be interpreted cautiously and validated in larger cohorts. Third, there were limitations regarding immunological and genetic characterization. GADA was measured by EIA rather than RIA, and other islet‐associated autoantibodies, such as zinc transporter 8 autoantibody and islet cell antibody, were not measured. Thus, direct comparison with RIA‐based GADA stratification or broader autoantibody‐defined risk classification was not possible. Furthermore, although HLA genotypes were analyzed in patients with SPIDDM, these could not be performed in the type 2 diabetes mellitus or healthy control groups due to a shortage of remaining blood samples. Therefore, the influence of HLA type on the discrimination of SPIDDM from other groups could not be evaluated. Patients with acute‐onset type 1 diabetes mellitus and childhood‐onset type 1 diabetes mellitus, which are more common in Caucasian populations, 46 were also not included. Future studies including these patient groups are needed to clarify the disease‐subtype and age‐specific significance of circulating miRNAs in autoimmune diabetes. Fourth, serum miRNA levels were measured only at diagnosis as SPIDDM (probable), and longitudinal serum samples were unavailable. Therefore, we could not determine whether the identified miRNAs change during the disease course or whether such changes are associated with progression to insulin deficiency. Although internal quality‐control procedures were applied to ensure measurement reproducibility, clinical implementation of serum miRNA testing will require validation of inter‐assay and inter‐laboratory reproducibility and standardized protocols for serum collection, storage, RNA extraction, normalization, and qRT‐PCR analysis. Finally, the functions, cellular origins, and cost‐effectiveness of the identified miRNAs were not examined. In particular, because a recent anti‐islet autoantibody assay can simultaneously measure multiple autoantibodies in a single well (3 Screen ICA EIA), 47 whether serum miRNA measurement offers additional cost‐effectiveness compared with such assays remains to be determined.
In summary, this study identified serum miRNAs as potential biomarkers in adult patients with SPIDDM. Among the identified miRNAs, miR‐150‐5p may be particularly useful for predicting subsequent decline in insulin secretion in patients initially diagnosed with SPIDDM (probable). Further prospective studies are warranted to validate the utility of serum miRNAs for risk stratification and therapeutic decision‐making in patients with SPIDDM.
DISCLOSURE
The authors declare no conflicts of interest.
Approval of the research protocol: The study protocol for this research project has been approved by the certified review board of the University of Yamanashi and Isawa Hot Spring Hospital.
Informed consent: Informed consent was waived owing to the retrospective nature of the study, and information regarding the study was disclosed through an online opt‐out process. Participants were given the opportunity to decline participation.
Registry and the registration no. of the study/trial: Approval date of registry by the certified review board of the University of Yamanashi was February 16, 2022, and approval number 2272. Approval date of registry by the certified review board of Isawa Hot Spring Hospital was February 7, 2018, and approval number 2018–003.
Animal studies: N/A.
Supporting information
Figure S1 Result of the qRT‐PCR analysis of miR‐16‐5p among the three groups: SPIDDM (n = 60), T2DM (n = 59), and healthy controls (n = 50). Data were compared using the Mann–Whitney U test, with two groups compared at a time [SPIDDM vs T2DM, P = 0.805; SPIDDM vs Controls, P = 0.546; T2DM vs Controls, P = 0.731]. Controls, healthy controls; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Figure S2 Results of the qRT‐PCR analysis of 36 miRNAs whose expression levels did not differ significantly among the three groups: SPIDDM (n = 60), T2DM (n = 59), and healthy controls (n = 50). Data were compared using the Mann–Whitney U test, with two groups compared at a time. SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; controls, healthy controls.
Figure S3 Spearman's rank correlation analysis between the expression level of the four miRNAs and GADA titers. GADA, glutamic acid decarboxylase autoantibody; ρ, Spearman's correlation coefficient.
Table S1 The miRNAs reported to be significantly altered in expression in patients with LADA.
Table S2 The miRNAs whose expression has been reported to be significantly altered in patients with type 1 diabetes mellitus in at least two studies.
Table S3 Description of target miRNAs.
Table S4 Univariate logistic regression analysis examining the association between characteristics and their ability to discriminate patients with SPIDDM (probable) from those with T2DM.
Table S5 Univariate linear regression analysis examining the association between participant characteristics and CPI at the last outpatient clinic visit in the SPIDDM group.
Table S6 Clinical characteristics and candidate serum miRNA levels according to GADA titer in the SPIDDM group.
Table S7 Clinical characteristics and candidate serum miRNA levels according to GADA and IA‐2A positivity in the SPIDDM group.
Table S8 Clinical characteristics and candidate serum miRNA levels according to age at SPIDDM (probable) diagnosis.
Table S9 Clinical characteristics and candidate serum miRNA levels according to HLA genotype in the SPIDDM group.
ACKNOWLEDGMENTS
This work was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI (Grant Number JP25K21120) and Young Researcher Grants (2021) from The Japan Diabetes Society. We appreciate Dr. Asako Miyazaki for her contribution to the collection of samples from healthy controls.
Contributor Information
Hideyuki Okuma, Email: hokuma@yamanashi.ac.jp.
Kyoichiro Tsuchiya, Email: tsuchiyak@yamanashi.ac.jp.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Pugliese A. The multiple origins of type 1 diabetes. Diabet Med 2013; 30: 135–146. [DOI] [PubMed] [Google Scholar]
- 2. Kawasaki E, Matsuura N, Eguchi K. Type 1 diabetes in Japan. Diabetologia 2006; 49: 828–836. [DOI] [PubMed] [Google Scholar]
- 3. Kobayashi T, Tamemoto K, Nakanishi K, et al. Immunogenetic and clinical characterization of slowly progressive IDDM. Diabetes Care 1993; 16: 780–788. [DOI] [PubMed] [Google Scholar]
- 4. Maruyama T, Tanaka S, Shimada A, et al. Insulin intervention in slowly progressive insulin‐dependent (type 1) diabetes mellitus. J Clin Endocrinol Metab 2008; 93: 2115–2121. [DOI] [PubMed] [Google Scholar]
- 5. Shimada A, Kawasaki E, Abiru N, et al. Statement concerning Management of Slowly Progressive Insulin‐dependent Diabetes Mellitus (probable). J Nihon Rinsho Denshi Kenbikyo Gakkai‐shi 2023; 66: 807–814 (In Japanese). [Google Scholar]
- 6. Shimada A, Kawasaki E, Abiru N, et al. New diagnostic criteria (2023) for slowly progressive type 1 diabetes (SPIDDM): Report from committee on type 1 diabetes of the Japan diabetes society (English version). J Diabetes Investig 2024; 15: 254–257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Shimada A, Kawasaki E, Abiru N, et al. New diagnostic criteria (2023) for slowly progressive type 1 diabetes (SPIDDM): Report from committee on type 1 diabetes in Japan diabetes society (English version). Diabetol Int 2024; 15: 1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kawasaki E, Nakamura K, Kuriya G, et al. Differences in the humoral autoreactivity to zinc transporter 8 between childhood‐ and adult‐onset type 1 diabetes in Japanese patients. Clin Immunol 2011; 138: 146–153. [DOI] [PubMed] [Google Scholar]
- 9. Kawasaki E, Yasui J, Tsurumaru M, et al. Sequential elevation of autoantibodies to thyroglobulin and glutamic acid decarboxylase in type 1 diabetes. World J Diabetes 2013; 4: 227–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Kawasaki E. Anti‐islet autoantibodies in type 1 diabetes. Int J Mol Sci 2023; 24: 10012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Yates LA, Norbury CJ, Gilbert RJ. The long and short of microRNA. Cell 2013; 153: 516–519. [DOI] [PubMed] [Google Scholar]
- 12. Radenkovic M, Silver C, Arvastsson J, et al. Altered regulatory T cell phenotype in latent autoimmune diabetes of the adults (LADA). Clin Exp Immunol 2016; 186: 46–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Flynt AS, Lai EC. Biological principles of microRNA‐mediated regulation: Shared themes amid diversity. Nat Rev Genet 2008; 9: 831–842. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Kroh EM, Parkin RK, Mitchell PS, et al. Analysis of circulating microRNA biomarkers in plasma and serum using quantitative reverse transcription‐PCR (qRT‐PCR). Methods 2010; 50: 298–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Mraz M, Malinova K, Mayer J, et al. MicroRNA isolation and stability in stored RNA samples. Biochem Biophys Res Commun 2009; 390: 1–4. [DOI] [PubMed] [Google Scholar]
- 16. Keller A, Rounge T, Backes C, et al. Sources to variability in circulating human miRNA signatures. RNA Biol 2017; 14: 1791–1798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Zampetaki A, Kiechl S, Drozdov I, et al. Plasma microRNA profiling reveals loss of endothelial miR‐126 and other microRNAs in type 2 diabetes. Circ Res 2010; 107: 810–817. [DOI] [PubMed] [Google Scholar]
- 18. Osmai M, Osmai Y, Bang‐Berthelsen CH, et al. MicroRNAs as regulators of beta‐cell function and dysfunction. Diabetes Metab Res Rev 2016; 32: 334–349. [DOI] [PubMed] [Google Scholar]
- 19. Seyhan AA, Nunez Lopez YO, Xie H, et al. Pancreas‐enriched miRNAs are altered in the circulation of subjects with diabetes: A pilot cross‐sectional study. Sci Rep 2016; 6: 31479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Nielsen LB, Wang C, Sørensen K, et al. Circulating levels of microRNA from children with newly diagnosed type 1 diabetes and healthy controls: Evidence that miR‐25 associates to residual beta‐cell function and glycaemic control during disease progression. Exp Diabetes Res 2012; 2012: 896362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Buzzetti R, Maddaloni E, Gaglia J, et al. Adult‐onset autoimmune diabetes. Nat Rev Dis Primers 2022; 8: 63. [DOI] [PubMed] [Google Scholar]
- 22. Buzzetti R, Di Pietro S, Giaccari A, et al. High titer of autoantibodies to GAD identifies a specific phenotype of adult‐onset autoimmune diabetes. Diabetes Care 2007; 30: 932–938. [DOI] [PubMed] [Google Scholar]
- 23. Tuomi T, Carlsson A, Li H, et al. Clinical and genetic characteristics of type 2 diabetes with and without GAD antibodies. Diabetes 1999; 48: 150–157. [DOI] [PubMed] [Google Scholar]
- 24. Fan W, Pang H, Li X, et al. Plasma‐derived exosomal miRNAs as potentially novel biomarkers for latent autoimmune diabetes in adults. Diabetes Res Clin Pract 2023; 197: 110570. [DOI] [PubMed] [Google Scholar]
- 25. Sørgjerd EP, Mjelle R, Beisvåg V, et al. Small RNAs are differentially expressed in autoimmune and non‐autoimmune diabetes and controls. Eur J Endocrinol 2022; 187: 231–240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Yu K, Huang Z, Zhou J, et al. Transcriptome profiling of microRNAs associated with latent autoimmune diabetes in adults (LADA). Sci Rep 2019; 9: 11347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Pan S, Li M, Yu H, et al. microRNA‐143‐3p contributes to inflammatory reactions by targeting FOSL2 in PBMCs from patients with autoimmune diabetes mellitus. Acta Diabetol 2021; 58: 63–72. [DOI] [PubMed] [Google Scholar]
- 28. Liu Y, Ma M, Yu J, et al. Decreased serum microRNA‐21, microRNA‐25, microRNA‐146a, and microRNA‐181a in autoimmune diabetes: Potential biomarkers for diagnosis and possible involvement in pathogenesis. Int J Endocrinol 2019; 2019: 8406438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kawabata Y, Ikegami H, Awata T, et al. Differential association of HLA with three subtypes of type 1 diabetes: Fulminant, slowly progressive and acute‐onset. Diabetologia 2009; 52: 2513–2521. [DOI] [PubMed] [Google Scholar]
- 30. Noso S, Chujo D, Imagawa A, et al. Rapid and slow progressors toward β‐cell depletion and their predictors in type 1 diabetes: Prospective longitudinal study in Japanese type 1 diabetes (TIDE‐J). Diabetes Care 2025; 48: 1438–1445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Assmann TS, Recamonde‐Mendoza M, De Souza BM, et al. MicroRNA expression profiles and type 1 diabetes mellitus: Systematic review and bioinformatic analysis. Endocr Connect 2017; 6: 773–790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Haisa A, Oikawa Y, Satomura A, et al. High glutamic acid decarboxylase antibody titers may be associated with a decline in β‐cell function over time and future insulin deficiency in latent autoimmune diabetes in adults. Diabetes Res Clin Pract 2024; 215: 111799. [DOI] [PubMed] [Google Scholar]
- 33. Kawasaki E, Nakamura K, Kuriya G, et al. Autoantibodies to insulin, insulinoma‐associated antigen‐2, and zinc transporter 8 improve the prediction of early insulin requirement in adult‐onset autoimmune diabetes. J Clin Endocrinol Metab 2010; 95: 707–713. [DOI] [PubMed] [Google Scholar]
- 34. Tanaka S, Okubo M, Nagasawa K, et al. Predictive value of titer of GAD antibodies for further progression of beta cell dysfunction in slowly progressive insulin‐dependent (type 1) diabetes (SPIDDM). Diabetol Int 2016; 7: 42–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Yamada K, Yuan X, Inada C, et al. Combined measurements of GAD65 and ICA512 antibodies in acute onset and slowly progressive IDDM. Diabetes Res Clin Pract 1997; 35: 91–98. [DOI] [PubMed] [Google Scholar]
- 36. Takino H, Yamasaki H, Abiru N, et al. Antibodies to GAD in Japanese patients classified as type 2 diabetes at diagnosis. High titre of GAD ab is a predictive marker for early insulin treatment‐‐report of west Japan (Kyushu, Yamaguchi, Osaka) study for GAD ab(+) diabetes. Diabet Med 2002; 19: 730–734. [DOI] [PubMed] [Google Scholar]
- 37. Hoshina S, Miura J, Sugizawa E, et al. Clinical features of slowly progressive type 1 (insulin‐dependent) diabetes mellitus: A comparative study based on degree of obesity at diagnosis of diabetes. Diabetol Int 2015; 6: 91–97. [Google Scholar]
- 38. Verge CF, Gianani R, Kawasaki E, et al. Prediction of type I diabetes in first‐degree relatives using a combination of insulin, GAD, and ICA512bdc/IA‐2 autoantibodies. Diabetes 1996; 45: 926–933. [DOI] [PubMed] [Google Scholar]
- 39. Sosenko JM, Skyler JS, Palmer JP, et al. A longitudinal study of GAD65 and ICA512 autoantibodies during the progression to type 1 diabetes in diabetes prevention trial‐type 1 (DPT‐1) participants. Diabetes Care 2011; 34: 2435–2437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Ilonen J, Hammais A, Laine AP, et al. Patterns of β‐cell autoantibody appearance and genetic associations during the first years of life. Diabetes 2013; 62: 3636–3640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Snowhite IV, Allende G, Sosenko J, et al. Association of serum microRNAs with islet autoimmunity, disease progression and metabolic impairment in relatives at risk of type 1 diabetes. Diabetologia 2017; 60: 1409–1422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Liu C, Yu Z, Huang S, et al. Combined identification of three miRNAs in serum as effective diagnostic biomarkers for HNSCC. EBioMedicine 2019; 50: 135–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Tang S, Cheng J, Yao Y, et al. Combination of four serum Exosomal MiRNAs as novel diagnostic biomarkers for early‐stage gastric cancer. Front Genet 2020; 11: 237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Shimomura A, Shiino S, Kawauchi J, et al. Novel combination of serum microRNA for detecting breast cancer in the early stage. Cancer Sci 2016; 107: 326–334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Lakhter AJ, Pratt RE, Moore RE, et al. Beta cell extracellular vesicle miR‐21‐5p cargo is increased in response to inflammatory cytokines and serves as a biomarker of type 1 diabetes. Diabetologia 2018; 61: 1124–1134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Kawasaki E, Eguchi K. Is type 1 diabetes in the Japanese population the same as among Caucasians? Ann N Y Acad Sci 2004; 1037: 96–103. [DOI] [PubMed] [Google Scholar]
- 47. Raab J, Haupt F, Scholz M, et al. Capillary blood islet autoantibody screening for identifying pre‐type 1 diabetes in the general population: Design and initial results of the Fr1da study. BMJ Open 2016; 6: e011144. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1 Result of the qRT‐PCR analysis of miR‐16‐5p among the three groups: SPIDDM (n = 60), T2DM (n = 59), and healthy controls (n = 50). Data were compared using the Mann–Whitney U test, with two groups compared at a time [SPIDDM vs T2DM, P = 0.805; SPIDDM vs Controls, P = 0.546; T2DM vs Controls, P = 0.731]. Controls, healthy controls; SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus.
Figure S2 Results of the qRT‐PCR analysis of 36 miRNAs whose expression levels did not differ significantly among the three groups: SPIDDM (n = 60), T2DM (n = 59), and healthy controls (n = 50). Data were compared using the Mann–Whitney U test, with two groups compared at a time. SPIDDM, slowly progressive type 1 diabetes mellitus; T2DM, type 2 diabetes mellitus; controls, healthy controls.
Figure S3 Spearman's rank correlation analysis between the expression level of the four miRNAs and GADA titers. GADA, glutamic acid decarboxylase autoantibody; ρ, Spearman's correlation coefficient.
Table S1 The miRNAs reported to be significantly altered in expression in patients with LADA.
Table S2 The miRNAs whose expression has been reported to be significantly altered in patients with type 1 diabetes mellitus in at least two studies.
Table S3 Description of target miRNAs.
Table S4 Univariate logistic regression analysis examining the association between characteristics and their ability to discriminate patients with SPIDDM (probable) from those with T2DM.
Table S5 Univariate linear regression analysis examining the association between participant characteristics and CPI at the last outpatient clinic visit in the SPIDDM group.
Table S6 Clinical characteristics and candidate serum miRNA levels according to GADA titer in the SPIDDM group.
Table S7 Clinical characteristics and candidate serum miRNA levels according to GADA and IA‐2A positivity in the SPIDDM group.
Table S8 Clinical characteristics and candidate serum miRNA levels according to age at SPIDDM (probable) diagnosis.
Table S9 Clinical characteristics and candidate serum miRNA levels according to HLA genotype in the SPIDDM group.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.