Clinical characterization of data-driven diabetes subgroups in Mexicans using a reproducible machine learning approach

Proteomics profiling and association with cardiorenal complications in type 2 diabetes subtypes in Asian population

AIM

Among multi-ethnic Asians, type 2 diabetes (T2D) clustered in three subtypes; mild obesity-related diabetes (MOD), mild age-related diabetes with insulin insufficiency (MARD-II) and severe insulin-resistant diabetes with relative insulin insufficiency (SIRD-RII) had differential cardio-renal complication risk. We assessed the proteomic profiles to identify subtype specific biomarkers and its association with diabetes complications.

METHODS

1448 plasma proteins at baseline were measured and compared across the T2D subtypes. Multivariable cox regression was used to assess associations between significant proteomics features and cardio-renal complications.

RESULTS

Among 645 T2D participants (SIRD-RII [19%], MOD [45%], MARD-II [36%]), 295 proteins expression differed significantly across the groups. These proteins were enriched in cell adhesion, neurogenesis and inflammatory response processes. In SIRD-RII group, ADH4, ACY1, THOP1, IGFBP2, NEFL, ENTPD2, CALB1, HAO1, CTSV, ITGAV, SCLY, EDA2R, ERBB2 proteins significantly associated with progressive CKD and LILRA5 protein with incident heart failure (HF). In MOD group, TAFA5, RSPO3, EDA2R proteins significantly associated with incident HF. In MARD-II group, FABP4 protein significantly associated with progressive CKD and PTPRN2 protein with major adverse cardiovascular events. Genetically determined NEFL and CALB1 were associated with kidney function decline.

CONCLUSIONS

Each T2D subtype has unique proteomics signature and association with clinical outcomes and underlying mechanisms.

Cluster analysis of adult individuals with type 1 diabetes: Treatment pathways and complications over a five-year follow-up period

AIMS

To identify subgroups of adults with type 1 diabetes and analyse their treatment pathways and risk of diabetes-related complications over a 5-year follow-up.

METHODS

We performed a k-means cluster analysis using the T1DExchange Registry (n = 6,302) to identify subgroups based on demographic and clinical characteristics. Annual reassessments linked treatment trajectories with these clusters, considering drug and technology use. Complication risks were analysed using Cox regression.

RESULTS

Five clusters were identified: 1) A favourable combination of all variables (31.67 %); 2) Longer diabetes duration (22.63 %); 3) Higher HbA1c levels (13.28 %); 4) Higher BMI (15.25 %); 5) Older age at diagnosis (17.17 %). Two-thirds of patients remained in their initial cluster annually. Technology adoption showed improved glycaemic control over time. Cox proportional hazards showed different risk patterns: Cluster 1 had low complication risk; Cluster 2 had the highest risk for retinopathy, coronary artery disease and autonomic neuropathy; Cluster 3 had the highest risk for albuminuria, depression and diabetic ketoacidosis; Cluster 4 had increased risk for multiple complications; Cluster 5 had the highest risk for hypertension and severe hypoglycaemia, with elevated coronary artery disease risk.

CONCLUSIONS

Clinical characteristics can identify subgroups of patients with T1DM showing differences in treatment and complications during follow-up.

Unsupervised clustering identified clinically relevant metabolic syndrome endotypes in UK and Taiwan Biobanks

Metabolic syndrome (MetS) is a collection of cardiovascular risk factors; however, the high prevalence and heterogeneity impede effective clinical management. We conducted unsupervised clustering on individuals from UK Biobank to reveal endotypes. Five MetS subgroups were identified: Cluster 1 (C1): non-descriptive, Cluster 2 (C2): hypertensive, Cluster 3 (C3): obese, Cluster 4 (C4): lipodystrophy-like, and Cluster 5 (C5): hyperglycemic. For all of the endotypes, we identified the corresponding cardiometabolic traits and their associations with clinical outcomes. Genome-wide association studies (GWASs) were conducted to identify associated genotypic traits. We then determined endotype-specific genotypic traits and constructed polygenic risk score (PRS) models specific to each endotype. GWAS of each MetS clusters revealed different genotypic traits. C1 GWAS revealed novel findings of TRIM63, MYBPC3, MYLPF, and RAPSN. Intriguingly, C1, C3, and C4 were associated with genes highly expressed in brain tissues. MetS clusters with comparable phenotypic and genotypic traits were identified in Taiwan Biobank.

Trajectories of clinical characteristics, complications and treatment choices in data-driven subgroups of type 2 diabetes

AIMS/HYPOTHESIS

This study aimed to explore the added value of subgroups that categorise individuals with type 2 diabetes by k-means clustering for two primary care registries (the Netherlands and Scotland), inspired by Ahlqvist's novel diabetes subgroups and previously analysed by Slieker et al. METHODS: We used two Dutch and Scottish diabetes cohorts (N=3054 and 6145; median follow-up=11.2 and 12.3 years, respectively) and defined five subgroups by k-means clustering with age at baseline, BMI, HbA1c, HDL-cholesterol and C-peptide. We investigated differences between subgroups by trajectories of risk factor values (random intercept models), time to diabetes-related complications (logrank tests and Cox models) and medication patterns (multinomial logistic models). We also compared directly using the clustering indicators as predictors of progression vs the k-means discrete subgroups. Cluster consistency over follow-up was assessed.

RESULTS

Subgroups' risk factors were significantly different, and these differences remained generally consistent over follow-up. Among all subgroups, individuals with severe insulin resistance faced a significantly higher risk of myocardial infarction both before (HR 1.65; 95% CI 1.40, 1.94) and after adjusting for age effect (HR 1.72; 95% CI 1.46, 2.02) compared with mild diabetes with high HDL-cholesterol. Individuals with severe insulin-deficient diabetes were most intensively treated, with more than 25% prescribed insulin at 10 years of diagnosis. For severe insulin-deficient diabetes relative to mild diabetes, the relative risks for using insulin relative to no common treatment would be expected to increase by a factor of 3.07 (95% CI 2.73, 3.44), holding other factors constant. Clustering indicators were better predictors of progression variation relative to subgroups, but prediction accuracy may improve after combining both. Clusters were consistent over 8 years with an accuracy ranging from 59% to 72%.

CONCLUSIONS/INTERPRETATION

Data-driven subgroup allocations were generally consistent over follow-up and captured significant differences in risk factor trajectories, medication patterns and complication risks. Subgroups serve better as a complement rather than as a basis for compressing clustering indicators.

Epistasis Between HLA-DRB1*16:02:01 and SLC16A11 T-C-G-T-T Reduces Odds for Type 2 Diabetes in Southwest American Indians

We sought to identify genetic/immunologic contributors of type 2 diabetes (T2D) in an indigenous American community by genotyping all study participants for both high-resolution HLA-DRB1 alleles and SLC16A11 to test their risk and/or protection for T2D. These genes were selected based on independent reports that HLA-DRB1*16:02:01 is protective for T2D and that SLC16A11 associates with T2D in individuals with BMI <35 kg/m2. Here, we test the interaction of the two loci with a more complete data set and perform a BMI sensitivity test. We defined the risk protection haplotype of SLC16A11, T-C-G-T-T, as allele 2 of a diallelic genetic model with three genotypes, SLC16A11*11, *12, and *22, where allele 1 is the wild type. Both earlier findings were confirmed. Together in the same logistic model with BMI ≥35 kg/m2, DRB1*16:02:01 remains protective (odds ratio [OR] 0.73), while SLC16A11 switches from risk to protection (OR 0.57 [*22] and 0.78 [*12]); an added interaction term was statistically significant (OR 0.49 [*12]). Bootstrapped (b = 10,000) statistical power of interaction, 0.4801, yielded a mean OR of 0.43. Sensitivity analysis demonstrated that the interaction is significant in the BMI range of 30-41 kg/m2. To investigate the epistasis, we used the primary function of the HLA-DRB1 molecule, peptide binding and presentation, to search the entire array of 15-mer peptides for both the wild-type and ancient human SLC16A11 molecules for a pattern of strong binding that was associated with risk and protection for T2D. Applying computer binding algorithms suggested that the core peptide at SLC16A11 D127G, FSAFASGLL, might be key for moderating risk for T2D with potential implications for type 1 diabetes.

ARTICLE HIGHLIGHTS

Harnessing the power of proteomics in precision diabetes medicine

Precision diabetes medicine (PDM) aims to reduce errors in prevention programmes, diagnosis thresholds, prognosis prediction and treatment strategies. However, its advancement and implementation are difficult due to the heterogeneity of complex molecular processes and environmental exposures that influence an individual's disease trajectory. To address this challenge, it is imperative to develop robust screening methods for all areas of PDM. Innovative proteomic technologies, alongside genomics, have proven effective in precision cancer medicine and are showing promise in diabetes research for potential translation. This narrative review highlights how proteomics is well-positioned to help improve PDM. Specifically, a critical assessment of widely adopted affinity-based proteomic technologies in large-scale clinical studies and evidence of the benefits and feasibility of using MS-based plasma proteomics is presented. We also present a case for the use of proteomics to identify predictive protein panels for type 2 diabetes subtyping and the development of clinical prediction models for prevention, diagnosis, prognosis and treatment strategies. Lastly, we discuss the importance of plasma and tissue proteomics and its integration with genomics (proteogenomics) for identifying unique type 2 diabetes intra- and inter-subtype aetiology. We conclude with a call for action formed on advancing proteomics technologies, benchmarking their performance and standardisation across sites, with an emphasis on data sharing and the inclusion of diverse ancestries in large cohort studies. These efforts should foster collaboration with key stakeholders and align with ongoing academic programmes such as the Precision Medicine in Diabetes Initiative consortium.

Diabetes subgroups and sociodemographic inequalities in Mexico: a cross-sectional analysis of nationally representative surveys from 2016 to 2022

Background

Differences in the prevalence of four diabetes subgroups have been reported in Mexico compared to other populations, but factors that may contribute to these differences are poorly understood. Here, we estimate the prevalence of diabetes subgroups in Mexico and evaluate their correlates with indicators of social disadvantage using data from national representative surveys.

Methods

We analyzed serial, cross-sectional Mexican National Health and Nutrition Surveys spanning 2016, 2018, 2020, 2021, and 2022, including 23,354 adults (>20 years). Diabetes subgroups (obesity-related [MOD], severe insulin-deficient [SIDD], severe insulin-resistant [SIRD], and age-related [MARD]) were classified using self-normalizing neural networks based on a previously validated algorithm. We used the density-independent social lag index (DISLI) as a proxy of state-level social disadvantage.

Findings

We identified 4204 adults (median age: 57, IQR: 47-66, women: 64%) living with diabetes, yielding a pooled prevalence of 16.04% [95% CI: 14.92-17.17]. When stratified by diabetes subgroup, prevalence was 6.62% (5.69-7.55) for SIDD, 5.25% (4.52-5.97) for MOD, 2.39% (1.95-2.83) for MARD, and 1.27% (1.00-1.54) for SIRD. SIDD and MOD clustered in Southern Mexico, whereas MARD and SIRD clustered in Northern Mexico and Mexico City. Each standard deviation increase in DISLI was associated with higher odds of SIDD (OR: 1.12, 95% CI: 1.06-1.12) and lower odds of MOD (OR: 0.93, 0.88-0.99). Speaking an indigenous language was associated with higher odds of SIDD (OR: 1.35, 1.16-1.57) and lower odds of MARD (OR 0.58, 0.45-0.74).

Interpretation

Diabetes prevalence in Mexico is rising in the context of regional and sociodemographic inequalities across distinct diabetes subgroups. SIDD is a subgroup of concern that may be associated with inadequate diabetes management, mainly in marginalized states.

Funding

This research was supported by Instituto Nacional de Geriatría in Mexico.

Prevalence of prediabetes in Mexico: a retrospective analysis of nationally representative surveys spanning 2016-2022

Background

Characterizing prediabetes phenotypes may be useful in guiding diabetes prevention efforts; however, heterogeneous criteria to define prediabetes have led to inconsistent prevalence estimates, particularly in low- and middle-income countries. Here, we estimated trends in prediabetes prevalence in Mexico across different prediabetes definitions and their association with prevalent cardiometabolic conditions.

Methods

We conducted a serial cross-sectional analysis of National Health and Nutrition Surveys in Mexico (2016-2022), totalling 22 081 Mexican adults. After excluding individuals with diagnosed or undiagnosed diabetes, we defined prediabetes using ADA (impaired fasting glucose [IFG] 100-125 mg/dL and/or HbA1c 5.7-6.4%), WHO (IFG 110-125 mg/dL), and IEC criteria (HbA1c 6.0-6.4%). Prevalence trends of prediabetes over time were evaluated using weighted Poisson regression and its association with prevalent cardiometabolic conditions with weighted logistic regression.

Findings

The prevalence of prediabetes (either IFG or high HbA1c [ADA]) in Mexico was 20.9% in 2022. Despite an overall downward trend in prediabetes (RR 0.973, 95% CI 0.957-0.988), this was primarily driven by decreases in prediabetes by ADA-IFG (RR 0.898, 95% CI 0.880-0.917) and WHO-IFG criteria (RR 0.919, 95% CI 0.886-0.953), while prediabetes by ADA-HbA1c (RR 1.055, 95% CI 1.033-1.077) and IEC-HbA1C criteria (RR 1.085, 95% CI 1.045-1.126) increased over time. Prediabetes prevalence increased over time in adults >40 years, with central obesity, self-identified as indigenous or living in urban areas. For all definitions, prediabetes was associated with an increased risk of cardiometabolic conditions.

Interpretation

Prediabetes rates in Mexico from 2016 to 2022 varied based on defining criteria but consistently increased for HbA1c-based definitions and high-risk subgroups.

Funding

This research was supported by Instituto Nacional de Geriatría in Mexico. JAS was supported by NIH/NIDDK Grant# K23DK135798.

Precision medicine for cardiometabolic disease: a framework for clinical translation

Cardiometabolic disease is a major threat to global health. Precision medicine has great potential to help to reduce the burden of this common and complex disease cluster, and to enhance contemporary evidence-based medicine. Its key pillars are diagnostics; prediction (of the primary disease); prevention (of the primary disease); prognosis (prediction of complications of the primary disease); treatment (of the primary disease or its complications); and monitoring (of risk exposure, treatment response, and disease progression or remission). To contextualise precision medicine in both research and clinical settings, and to encourage the successful translation of discovery science into clinical practice, in this Series paper we outline a model (the EPPOS model) that builds on contemporary evidence-based approaches; includes precision medicine that improves disease-related predictions by stratifying a cohort into subgroups of similar characteristics, or using participants' characteristics to model treatment outcomes directly; includes personalised medicine with the use of a person's data to objectively gauge the efficacy, safety, and tolerability of therapeutics; and subjectively tailors medical decisions to the individual's preferences, circumstances, and capabilities. Precision medicine requires a well functioning system comprised of multiple stakeholders, including health-care recipients, health-care providers, scientists, health economists, funders, innovators of medicines and technologies, regulators, and policy makers. Powerful computing infrastructures supporting appropriate analysis of large-scale, well curated, and accessible health databases that contain high-quality, multidimensional, time-series data will be required; so too will prospective cohort studies in diverse populations designed to generate novel hypotheses, and clinical trials designed to test them. Here, we carefully consider these topics and describe a framework for the integration of precision medicine in cardiometabolic disease.

Precision subclassification of type 2 diabetes: a systematic review

BACKGROUND

Heterogeneity in type 2 diabetes presentation and progression suggests that precision medicine interventions could improve clinical outcomes. We undertook a systematic review to determine whether strategies to subclassify type 2 diabetes were associated with high quality evidence, reproducible results and improved outcomes for patients.

METHODS

We searched PubMed and Embase for publications that used 'simple subclassification' approaches using simple categorisation of clinical characteristics, or 'complex subclassification' approaches which used machine learning or 'omics approaches in people with established type 2 diabetes. We excluded other diabetes subtypes and those predicting incident type 2 diabetes. We assessed quality, reproducibility and clinical relevance of extracted full-text articles and qualitatively synthesised a summary of subclassification approaches.

RESULTS

Here we show data from 51 studies that demonstrate many simple stratification approaches, but none have been replicated and many are not associated with meaningful clinical outcomes. Complex stratification was reviewed in 62 studies and produced reproducible subtypes of type 2 diabetes that are associated with outcomes. Both approaches require a higher grade of evidence but support the premise that type 2 diabetes can be subclassified into clinically meaningful subtypes.

CONCLUSION

Critical next steps toward clinical implementation are to test whether subtypes exist in more diverse ancestries and whether tailoring interventions to subtypes will improve outcomes.

Optimizing Clinical Diabetes Diagnosis through Generative Adversarial Networks: Evaluation and Validation

The escalating prevalence of Type 2 Diabetes (T2D) represents a substantial burden on global healthcare systems, especially in regions such as Mexico. Existing diagnostic techniques, although effective, often require invasive procedures and labor-intensive efforts. The promise of artificial intelligence and data science for streamlining and enhancing T2D diagnosis is well-recognized; however, these advancements are frequently constrained by the limited availability of comprehensive patient datasets. To mitigate this challenge, the present study investigated the efficacy of Generative Adversarial Networks (GANs) for augmenting existing T2D patient data, with a focus on a Mexican cohort. The researchers utilized a dataset of 1019 Mexican nationals, divided into 499 non-diabetic controls and 520 diabetic cases. GANs were applied to create synthetic patient profiles, which were subsequently used to train a Random Forest (RF) classification model. The study's findings revealed a notable improvement in the model's diagnostic accuracy, validating the utility of GAN-based data augmentation in a clinical context. The results bear significant implications for enhancing the robustness and reliability of Machine Learning tools in T2D diagnosis and management, offering a pathway toward more timely and effective patient care.

Deep learning on electronic medical records identifies distinct subphenotypes of diabetic kidney disease driven by genetic variations in the Rho pathway

Kidney disease affects 50% of all diabetic patients; however, prediction of disease progression has been challenging due to inherent disease heterogeneity. We use deep learning to identify novel genetic signatures prognostically associated with outcomes. Using autoencoders and unsupervised clustering of electronic health record data on 1,372 diabetic kidney disease patients, we establish two clusters with differential prevalence of end-stage kidney disease. Exome-wide associations identify a novel variant in ARHGEF18, a Rho guanine exchange factor specifically expressed in glomeruli. Overexpression of ARHGEF18 in human podocytes leads to impairments in focal adhesion architecture, cytoskeletal dynamics, cellular motility, and RhoA/Rac1 activation. Mutant GEF18 is resistant to ubiquitin mediated degradation leading to pathologically increased protein levels. Our findings uncover the first known disease-causing genetic variant that affects protein stability of a cytoskeletal regulator through impaired degradation, a potentially novel class of expression quantitative trait loci that can be therapeutically targeted.

Machine learning based study for the classification of Type 2 diabetes mellitus subtypes

PURPOSE

Data-driven diabetes research has increased its interest in exploring the heterogeneity of the disease, aiming to support in the development of more specific prognoses and treatments within the so-called precision medicine. Recently, one of these studies found five diabetes subgroups with varying risks of complications and treatment responses. Here, we tackle the development and assessment of different models for classifying Type 2 Diabetes (T2DM) subtypes through machine learning approaches, with the aim of providing a performance comparison and new insights on the matter.

METHODS

We developed a three-stage methodology starting with the preprocessing of public databases NHANES (USA) and ENSANUT (Mexico) to construct a dataset with N = 10,077 adult diabetes patient records. We used N = 2,768 records for training/validation of models and left the remaining (N = 7,309) for testing. In the second stage, groups of observations -each one representing a T2DM subtype- were identified. We tested different clustering techniques and strategies and validated them by using internal and external clustering indices; obtaining two annotated datasets Dset A and Dset B. In the third stage, we developed different classification models assaying four algorithms, seven input-data schemes, and two validation settings on each annotated dataset. We also tested the obtained models using a majority-vote approach for classifying unseen patient records in the hold-out dataset.

RESULTS

From the independently obtained bootstrap validation for Dset A and Dset B, mean accuracies across all seven data schemes were [Formula: see text] ([Formula: see text]) and [Formula: see text] ([Formula: see text]), respectively. Best accuracies were [Formula: see text] and [Formula: see text]. Both validation setting results were consistent. For the hold-out dataset, results were consonant with most of those obtained in the literature in terms of class proportions.

CONCLUSION

The development of machine learning systems for the classification of diabetes subtypes constitutes an important task to support physicians for fast and timely decision-making. We expect to deploy this methodology in a data analysis platform to conduct studies for identifying T2DM subtypes in patient records from hospitals.

A scoping review of the clinical application of machine learning in data-driven population segmentation analysis

OBJECTIVE

Data-driven population segmentation is commonly used in clinical settings to separate the heterogeneous population into multiple relatively homogenous groups with similar healthcare features. In recent years, machine learning (ML) based segmentation algorithms have garnered interest for their potential to speed up and improve algorithm development across many phenotypes and healthcare situations. This study evaluates ML-based segmentation with respect to (1) the populations applied, (2) the segmentation details, and (3) the outcome evaluations.

MATERIALS AND METHODS

MEDLINE, Embase, Web of Science, and Scopus were used following the PRISMA-ScR criteria. Peer-reviewed studies in the English language that used data-driven population segmentation analysis on structured data from January 2000 to October 2022 were included.

RESULTS

We identified 6077 articles and included 79 for the final analysis. Data-driven population segmentation analysis was employed in various clinical settings. K-means clustering is the most prevalent unsupervised ML paradigm. The most common settings were healthcare institutions. The most common targeted population was the general population.

DISCUSSION

Although all the studies did internal validation, only 11 papers (13.9%) did external validation, and 23 papers (29.1%) conducted methods comparison. The existing papers discussed little validating the robustness of ML modeling.

CONCLUSION

Existing ML applications on population segmentation need more evaluations regarding giving tailored, efficient integrated healthcare solutions compared to traditional segmentation analysis. Future ML applications in the field should emphasize methods' comparisons and external validation and investigate approaches to evaluate individual consistency using different methods.

Subtypes of Type 2 Diabetes and Incident Cardiovascular Disease Risk: UK Biobank and All of Us Cohorts

OBJECTIVE

To characterize and validate the subtypes of type 2 diabetes (T2D) using a novel clustering algorithm and to further assess their associations with the risk of incident cardiovascular disease (CVD) events.

METHODS

Unsupervised k-means clustering based on glycated hemoglobin level, age at onset of T2D, body mass index, and estimated glomerular filtration rate was conducted among participants with T2D from the UK Biobank (March 13, 2006, to October 1, 2010) and replicated in the All of Us cohort (May 30, 2017, to April 1, 2021).

RESULTS

Five distinct T2D clusters were identified in the UK Biobank and validated in the All of Us cohort, characterizing the phenotypically heterogeneous subtypes. With a median follow-up of 11.69 years for patients with T2D in the UK Biobank, risks of incident CVD events varied considerably between the clusters after adjustment for potential confounders and multiple testing (all P<.001). With cluster 1 characterized by early onset of T2D and mild abnormalities of other variables as the reference, patients in cluster 5 characterized by poor renal function had the highest risk of CVD events (hazard ratio [95% CI], 1.72 [1.45 to 2.03], 2.41 [1.93 to 3.02], and 1.62 [1.35 to 1.94] for composite CVD event, CVD mortality, and CVD incidence, respectively; all P<.001), followed by cluster 4 characterized by poor glycemic control and cluster 3 characterized by severe obesity. No consistently significant difference was found between cluster 2 characterized by late onset of T2D and cluster 1.

CONCLUSION

Our study, using a novel clustering algorithm to identify robust subtypes of T2D, found heterogeneous associations with incident CVD risk among patients with diabetes.

Systematic review of precision subclassification of type 2 diabetes

Heterogeneity in type 2 diabetes presentation, progression and treatment has the potential for precision medicine interventions that can enhance care and outcomes for affected individuals. We undertook a systematic review to ascertain whether strategies to subclassify type 2 diabetes are associated with improved clinical outcomes, show reproducibility and have high quality evidence. We reviewed publications that deployed 'simple subclassification' using clinical features, biomarkers, imaging or other routinely available parameters or 'complex subclassification' approaches that used machine learning and/or genomic data. We found that simple stratification approaches, for example, stratification based on age, body mass index or lipid profiles, had been widely used, but no strategy had been replicated and many lacked association with meaningful outcomes. Complex stratification using clustering of simple clinical data with and without genetic data did show reproducible subtypes of diabetes that had been associated with outcomes such as cardiovascular disease and/or mortality. Both approaches require a higher grade of evidence but support the premise that type 2 diabetes can be subclassified into meaningful groups. More studies are needed to test these subclassifications in more diverse ancestries and prove that they are amenable to interventions.

Two Distinct Groups Are Shown to Be at Risk of Diabetes by Means of a Cluster Analysis of Four Variables

Recent attempts to classify adult-onset diabetes using only six diabetes-related variables (GAD antibody, age at diagnosis, BMI, HbA1c, and homeostatic model assessment 2 estimates of b-cell function and insulin resistance (HOMA2-B and HOMA2-IR)) showed that diabetes can be classified into five clusters, of which four correspond to type 2 diabetes (T2DM). Here, we classified nondiabetic individuals to identify risk clusters for incident T2DM to facilitate the refinement of prevention strategies. Of the 1167 participants in the population-based Iwaki Health Promotion Project in 2014 (baseline), 868 nondiabetic individuals who attended at least once during 2015-2019 were included in a prospective study. A hierarchical cluster analysis was performed using four variables (BMI, HbA1c, and HOMA2 indices). Of the four clusters identified, cluster 1 (n = 103), labeled as "obese insulin resistant with sufficient compensatory insulin secretion", and cluster 2 (n = 136), labeled as "low insulin secretion", were found to be at risk of diabetes during the 5-year follow-up period: the multiple factor-adjusted HRs for clusters 1 and 2 were 14.7 and 53.1, respectively. Further, individuals in clusters 1and 2 could be accurately identified: the area under the ROC curves for clusters 1and 2 were 0.997 and 0.983, respectively. The risk of diabetes could be better assessed on the basis of the cluster that an individual belongs to.

Adherence to antidiabetic treatment in primary health care in individuals with type 2 diabetes. A survey including socio-demographic, patient related and clinical factors

BACKGROUND

Adherence to medications is a critical element in diabetes management.

OBJECTIVE

To investigate patient-related factors associated with adherence to antidiabetic treatment in patients with type 2 diabetes in primary health-care units.

METHODS

A cross-sectional study among patients receiving diabetes care in 18 primary care units in Mexico City was conducted. Data were collected from medical records and medical interviews. Adherence to antidiabetic therapy and patient-related factors associated with adherence were evaluated through a self-administered questionnaire. Socio-demographic, clinical, behavioral (self-care activities including a healthy diet, exercise or physical activity, self-monitoring of blood glucose testing, and foot care), and patients-adherence-related factors (health-care provider-patient communication; daily and social activities, and support network; alternative treatment beliefs; comorbidity, diabetes symptoms, and treatment; side effects and treatment access) were compared between the adherent and non-adherent group.

RESULTS

Of 319 outpatients, 48.3% were adherent to their antidiabetic therapy. In the adjusted analysis, patient-related factors associated to adherence were exercise and physical activity self-care behavior (OR=1.26; 95%CI 1.09-1.46), treatment interference with daily activities (OR=0.27; 95%CI 0.14-0.52), not satisfied with resolution of questions by the physician (OR=0.42; 95%CI 0.19-0.94), independently to increased cardiovascular risk and insulin treatment.

CONCLUSION

A low proportion of outpatients with T2D with regular primary care were adherent to diabetes treatment and this was associated with patient-related factors like answering patients' treatment-related questions by the physician. This information may be useful to identify patients at risk for low adherence, and to guide the design of quality-of-care strategies, like diabetes education programs to address suboptimal adherence in patients, and health-care professional communication skills training, particularly in primary care health systems.

Diabetes-Related Excess Mortality in Mexico: A Comparative Analysis of National Death Registries Between 2017-2019 and 2020

OBJECTIVE

To estimate diabetes-related mortality in Mexico in 2020 compared with 2017-2019 after the onset of the coronavirus disease 2019 (COVID-19) pandemic.

RESEARCH DESIGN AND METHODS

This retrospective, state-level study used national death registries of Mexican adults aged ≥20 years for the 2017-2020 period. Diabetes-related death was defined using ICD-10 codes listing diabetes as the primary cause of death, excluding certificates with COVID-19 as the primary cause of death. Spatial and negative binomial regression models were used to characterize the geographic distribution and sociodemographic and epidemiologic correlates of diabetes-related excess mortality, estimated as increases in diabetes-related mortality in 2020 compared with average 2017-2019 rates.

RESULTS

We identified 148,437 diabetes-related deaths in 2020 (177 per 100,000 inhabitants) vs. an average of 101,496 deaths in 2017-2019 (125 per 100,000 inhabitants). In-hospital diabetes-related deaths decreased by 17.8% in 2020 versus 2017-2019, whereas out-of-hospital deaths increased by 89.4%. Most deaths were attributable to type 2 diabetes (130 per 100,000 inhabitants). Compared with 2018-2019 data, hyperglycemic hyperosmolar state and diabetic ketoacidosis were the two contributing causes with the highest increase in mortality (128% and 116% increase, respectively). Diabetes-related excess mortality clustered in southern Mexico and was highest in states with higher social lag, rates of COVID-19 hospitalization, and prevalence of HbA1c ≥7.5%.

CONCLUSIONS

Diabetes-related deaths increased among Mexican adults by 41.6% in 2020 after the onset of the COVID-19 pandemic, occurred disproportionately outside the hospital, and were largely attributable to type 2 diabetes and hyperglycemic emergencies. Disruptions in diabetes care and strained hospital capacity may have contributed to diabetes-related excess mortality in Mexico during 2020.

Metabolic and proteomic signatures of type 2 diabetes subtypes in an Arab population

Type 2 diabetes (T2D) has a heterogeneous etiology influencing its progression, treatment, and complications. A data driven cluster analysis in European individuals with T2D previously identified four subtypes: severe insulin deficient (SIDD), severe insulin resistant (SIRD), mild obesity-related (MOD), and mild age-related (MARD) diabetes. Here, the clustering approach was applied to individuals with T2D from the Qatar Biobank and validated in an independent set. Cluster-specific signatures of circulating metabolites and proteins were established, revealing subtype-specific molecular mechanisms, including activation of the complement system with features of autoimmune diabetes and reduced 1,5-anhydroglucitol in SIDD, impaired insulin signaling in SIRD, and elevated leptin and fatty acid binding protein levels in MOD. The MARD cluster was the healthiest with metabolomic and proteomic profiles most similar to the controls. We have translated the T2D subtypes to an Arab population and identified distinct molecular signatures to further our understanding of the etiology of these subtypes.

Extending artificial intelligence research in the clinical domain: a theoretical perspective

Academic research to the utilization of artificial intelligence (AI) has been proliferated over the past few years. While AI and its subsets are continuously evolving in the fields of marketing, social media and finance, its application in the daily practice of clinical care is insufficiently explored. In this systematic review, we aim to landscape various application areas of clinical care in terms of the utilization of machine learning to improve patient care. Through designing a specific smart literature review approach, we give a new insight into existing literature identified with AI technologies in the clinical domain. Our review approach focuses on strategies, algorithms, applications, results, qualities, and implications using the Latent Dirichlet Allocation topic modeling. A total of 305 unique articles were reviewed, with 115 articles selected using Latent Dirichlet Allocation topic modeling, meeting our inclusion criteria. The primary result of this approach incorporates a proposition for future research direction, abilities, and influence of AI technologies and displays the areas of disease management in clinics. This research concludes with disease administrative ramifications, limitations, and directions for future research.

A novel diabetes typology: towards precision diabetology from pathogenesis to treatment

The current classification of diabetes, based on hyperglycaemia, islet-directed antibodies and some insufficiently defined clinical features, does not reflect differences in aetiological mechanisms and in the clinical course of people with diabetes. This review discusses evidence from recent studies addressing the complexity of diabetes by proposing novel subgroups (subtypes) of diabetes. The most widely replicated and validated approach identified, in addition to severe autoimmune diabetes, four subgroups designated severe insulin-deficient diabetes, severe insulin-resistant diabetes, mild obesity-related diabetes and mild age-related diabetes subgroups. These subgroups display distinct patterns of clinical features, disease progression and onset of comorbidities and complications, with severe insulin-resistant diabetes showing the highest risk for cardiovascular, kidney and fatty liver diseases. While it has been suggested that people in these subgroups would benefit from stratified treatments, RCTs are required to assess the clinical utility of any reclassification effort. Several methodological and practical issues also need further study: the statistical approach used to define subgroups and derive recommendations for diabetes care; the stability of subgroups over time; the optimal dataset (e.g. phenotypic vs genotypic) for reclassification; the transethnic generalisability of findings; and the applicability in clinical routine care. Despite these open questions, the concept of a new classification of diabetes has already allowed researchers to gain more insight into the colourful picture of diabetes and has stimulated progress in this field so that precision diabetology may become reality in the future.

Phenotypic and genetic classification of diabetes

The historical subclassification of diabetes into predominantly types 1 and 2 is well appreciated to inadequately capture the heterogeneity seen in patient presentations, disease course, response to therapy and disease complications. This review summarises proposed data-driven approaches to further refine diabetes subtypes using clinical phenotypes and/or genetic information. We highlight the benefits as well as the limitations of these subclassification schemas, including practical barriers to their implementation that would need to be overcome before incorporation into clinical practice.

Characterization of data-driven clusters in diabetes-free adults and their utility for risk stratification of type 2 diabetes

BACKGROUND

The prevention of type 2 diabetes is challenging due to the variable effects of risk factors at an individual level. Data-driven methods could be useful to detect more homogeneous groups based on risk factor variability. The aim of this study was to derive characteristic phenotypes using cluster analysis of common risk factors and to assess their utility to stratify the risk of type 2 diabetes.

METHODS

Data on 7317 diabetes-free adults from Sweden were used in the main analysis and on 2332 diabetes-free adults from Mexico for external validation. Clusters were based on sex, family history of diabetes, educational attainment, fasting blood glucose and insulin levels, estimated insulin resistance and β-cell function, systolic and diastolic blood pressure, and BMI. The risk of type 2 diabetes was assessed using Cox proportional hazards models. The predictive accuracy and long-term stability of the clusters were then compared to different definitions of prediabetes.

RESULTS

Six risk phenotypes were identified independently in both cohorts: very low-risk (VLR), low-risk low β-cell function (LRLB), low-risk high β-cell function (LRHB), high-risk high blood pressure (HRHBP), high-risk β-cell failure (HRBF), and high-risk insulin-resistant (HRIR). Compared to the LRHB cluster, the VLR and LRLB clusters showed a lower risk, while the HRHBP, HRBF, and HRIR clusters showed a higher risk of developing type 2 diabetes. The high-risk clusters, as a group, had a better predictive accuracy than prediabetes and adequate stability after 20 years.

CONCLUSIONS

Phenotypes derived using cluster analysis were useful in stratifying the risk of type 2 diabetes among diabetes-free adults in two independent cohorts. These results could be used to develop more precise public health interventions.

Detecting Sarcopenia Risk by Diabetes Clustering: A Japanese Prospective Cohort Study

CONTEXT

Previous studies have assessed the usefulness of data-driven clustering for predicting complications in patients with diabetes mellitus. However, whether the diabetes clustering is useful in predicting sarcopenia remains unclear.

OBJECTIVE

To evaluate the predictive power of diabetes clustering for the incidence of sarcopenia in a prospective Japanese cohort.

DESIGN

Three-year prospective cohort study.

SETTING AND PATIENTS

We recruited Japanese patients with type 1 or type 2 diabetes mellitus (n = 659) between January 2018 and February 2020 from the Fukushima Diabetes, Endocrinology, and Metabolism cohort.

INTERVENTIONS

Kaplan-Meier and Cox proportional hazards models were used to measure the predictive values of the conventional and clustering-based classification of diabetes mellitus for the onset of sarcopenia. Sarcopenia was diagnosed according to the Asian Working Group for Sarcopenia (AWGS) 2019 consensus update.

MAIN OUTCOME MEASURES

Onset of sarcopenia.

RESULTS

Cluster analysis of a Japanese population revealed 5 diabetes clusters: cluster 1 [severe autoimmune diabetes (SAID)], cluster 2 [severe insulin-deficient diabetes (SIDD)], cluster 3 (severe insulin-resistant diabetes, cluster 4 (mild obesity-related diabetes), and cluster 5 (mild age-related diabetes). At baseline, 38 (6.5%) patients met the AWGS sarcopenia criteria, and 55 had newly developed sarcopenia within 3 years. The SAID and SIDD clusters were at high risk of developing sarcopenia after correction for known risk factors.

CONCLUSIONS

This study reveals that among the 5 diabetes clusters, the SAID and SIDD clusters are at a high risk for developing sarcopenia. Clustering-based stratification may be beneficial for predicting and preventing sarcopenia in patients with diabetes.

Prescribing Patterns and Response to Antihyperglycemic Agents Among Novel Clusters of Type 2 Diabetes in Asian Indians

Aim: To assess the prescribing patterns and response to different classes of antihyperglycemic agents in novel clusters of type 2 diabetes (T2D) described in India. Materials and Methods: We attempted to replicate the earlier described clusters of T2D, in 32,867 individuals with new-onset T2D (within 2 years of diagnosis) registered between October 2013 and December 2020 at 15 diabetes clinics located across India, by means of k-means clustering utilizing 6 clinically relevant variables. Individuals who had follow-up glycated hemoglobin (HbA1c) up to 2 years were included for the drug response analysis (n = 13,247). Results: Among the 32,867 participants included in the study, 20,779 (63.2%) were males. The average age at diagnosis was 45 years and mean HbA1c at baseline was 8.9%. The same four clusters described in India earlier were replicated. Forty percent of the study participants belonged to the mild age-related diabetes cluster, followed by insulin-resistant obese diabetes (27%), severe insulin-deficient diabetes (21%), and combined insulin-resistant and insulin-deficient diabetes (12%) clusters. The most frequently used antihyperglycemic agents were sulfonylureas, metformin, and dipeptidyl peptidase-4 inhibitors apart from insulin. While there were significant differences in HbA1c reduction between drugs across clusters, these were largely driven by differences in the baseline (pretreatment) HbA1c. Conclusions: In this new cohort, we were able to reliably replicate the four subtypes of T2D earlier described in Asian Indians. Prescribing patterns show limited usage of newer antihyperglycemic agents across all clusters. Randomized clinical trials are required to establish differential drug responses between clusters.

Prevalence Trends of Diabetes Subgroups in the United States: A Data-driven Analysis Spanning Three Decades From NHANES (1988-2018)

CONTEXT

Data-driven diabetes subgroups were proposed as an alternative to address diabetes heterogeneity. However, changes in trends for these subgroups have not been reported.

OBJECTIVE

Here, we analyzed trends of diabetes subgroups, stratified by sex, race, education level, age categories, and time since diabetes diagnosis in the United States.

METHODS

We used data from consecutive NHANES cycles spanning the 1988-2018 period. Diabetes subgroups (mild obesity-related [MOD], severe insulin-deficient [SIDD], severe insulin-resistant [SIRD], and mild age-related diabetes [MARD]) were classified using validated self-normalizing neural networks. Severe autoimmune diabetes (SAID) was assessed for NHANES-III. Prevalence was estimated using examination sample weights considering bicyclic changes (BCs) to evaluate trends and changes over time.

RESULTS

Diabetes prevalence in the United States increased from 7.5% (95% CI 7.1-7.9) in 1988-1989 to 13.9% (95% CI 13.4-14.4) in 2016-2018 (BC 1.09%, 95% CI 0.98-1.31, P < .001). Non-Hispanic Black people had the highest prevalence. Overall, MOD, MARD, and SIDD had an increase during the studied period. Particularly, non-Hispanic Black people had sharp increases in MARD and SIDD, Mexican Americans in SIDD, and non-Hispanic White people in MARD. Males, subjects with secondary/high school, and adults aged 40-64 years had the highest increase in MOD prevalence. Trends in diabetes subgroups sustained after stratifying time since diabetes diagnosis.

CONCLUSION

Prevalence of diabetes and its subgroups in the United States has increased from 1988 to 2018. These trends were different across sex, ethnicities, education, and age categories, indicating significant heterogeneity in diabetes within the US obesity burden, population aging, socioeconomic disparities, and lifestyle aspects could be implicated in the increasing trends of diabetes in the United States.

Assessing reproducibility and utility of clustering of patients with type 2 diabetes and established CV disease (SAVOR -TIMI 53 trial)

OBJECTIVE

To assess the reproducibility and clinical utility of clustering-based subtyping of patients with type 2 diabetes (T2D) and established cardiovascular (CV) disease.

METHODS

The cardiovascular outcome trial SAVOR-TIMI 53 (n = 16,492) was used. Analyses focused on T2D patients with established CV disease. Unsupervised machine learning technique called "k-means clustering" was used to divide patients into subtypes. K-means clustering including HbA1c, age of diagnosis, BMI, HOMA2-IR and HOMA2-B was used to assign clusters to the following diabetes subtypes: severe insulin deficient diabetes (SIDD); severe insulin-resistant diabetes (SIRD); mild obesity-related diabetes (MOD); mild age-related diabetes (MARD). We refer these subtypes as "clustering-based diabetes subtypes". A simulation study using randomly generated data was conducted to understand how correlations between the above variables influence the formation of the cluster-based diabetes subtypes. The predictive utility of clustering-based diabetes subtypes for CV events (3-point MACE), renal function reduction (eGFR decrease >30%) and diabetic disease progression (introduction of additional anti-diabetic medication) were compared with conventional risk scores. Hazard ratios (HR) were estimated by Cox-proportional hazard models.

RESULTS

In the SAVOR-TIMI 53 trial based dataset, the percentage of the clustering-based T2D subtypes were; SIDD (18%), SIRD (17%), MOD (29%), MARD (37%). Using the simulated dataset, the diabetes subtypes could be largely reproduced from a log-normal distribution when including known correlations between variables. The predictive utility of clustering-based diabetic subtypes on CV events, renal function reduction, and diabetic disease progression did not show an advantage compared to conventional risk scores.

CONCLUSIONS

The consistent reproduction of four clustering-based T2D subtypes can be explained by the correlations between the variables used for clustering. Subtypes of T2D based on clustering had limited advantage compared to conventional risk scores to predict clinical outcome in patients with T2D and established CV disease.

Genome-wide association analyses highlight etiological differences underlying newly defined subtypes of diabetes

Type 2 diabetes has been reproducibly clustered into five subtypes with different disease progression and risk of complications; however, etiological differences are unknown. We used genome-wide association and genetic risk score (GRS) analysis to compare the underlying genetic drivers. Individuals from the Swedish ANDIS (All New Diabetics In Scania) study were compared to individuals without diabetes; the Finnish DIREVA (Diabetes register in Vasa) and Botnia studies were used for replication. We show that subtypes differ with regard to family history of diabetes and association with GRS for diabetes-related traits. The severe insulin-resistant subtype was uniquely associated with GRS for fasting insulin but not with variants in the TCF7L2 locus or GRS reflecting insulin secretion. Further, an SNP (rs10824307) near LRMDA was uniquely associated with mild obesity-related diabetes. Therefore, we conclude that the subtypes have partially distinct genetic backgrounds indicating etiological differences.

Novel strategies for glycaemic control and preventing diabetic complications applying the clustering-based classification of adult-onset diabetes mellitus: A perspective

Diabetes mellitus results from an interplay between insulin resistance and β-cell dysfunction. Since their relative contributions to its pathogenesis are difficult to quantify, therapeutic strategies for glycaemic control are determined primarily based on two limited metrics: plasma glucose and haemoglobin A1c. Recent attempts have been made to subclassify diabetes mellitus to better predict its associated pathology and plan appropriate therapeutic strategies. These classifications are based on data-driven cluster analysis using autoimmunity, age, obesity (metabolically unhealthy and healthy phenotypes), insulin secretory capacity and resistance, and ethnicity. This review addresses potential therapeutic strategies for the cluster-based classifications of adult-onset diabetes mellitus to achieve better glycaemic control and prevent or at least delay the concomitant complications.

Clinical Phenotypes in Patients With Type 2 Diabetes Mellitus: Characteristics, Cardiovascular Outcomes and Treatment Strategies

PURPOSE OF REVIEW

With recent advances in the pharmacological management of type 2 diabetes mellitus (T2DM), there is a growing need to understand which patients optimally benefit from these novel therapies. Various clinical clustering methodologies have emerged that utilise data-agnostic strategies to categorise patients that have similar clinical characteristics and outcomes; broadly, this characterisation is termed phenotyping. In patients with T2DM, we aimed to describe patient characteristics from phenotype studies, their cardiovascular risk profiles and the impact of antihyperglycemic treatment.

RECENT FINDINGS

Numerous phenotypic studies have been undertaken that have utilised a combination of clinical, biochemical, imaging and genetic variables. Each of these has produced phenotypes that display a spectrum of cardiovascular risk. Studies that aimed to describe pathophysiological phenotypes generally identified five phenotypes: autoimmune phenotype, insulin-related phenotypes (including permutations of insulin deficiency and resistance), obesity phenotype, ageing phenotype, and a sex-related phenotype. Studies examining risk profiles have demonstrated that across such phenotypes there is a spectrum of risk for diabetic complications. Few studies have examined treatment effects across these phenotypes, and thus provide little insights towards making phenotype-guided treatment decisions Clustering analyses in patients with T2DM have identified distinct phenotypes with unique risk profiles. Further studies are needed that harness the use of clinical, biochemical, imaging and genetic data to explore therapeutic heterogeneity and response to antihyperglycemic treatment across the spectrum of patient phenotypes.

Challenges Involved in Incorporating Personalised Treatment Plan as Routine Care of Patients with Diabetes

Diabetes is a heterogenous disorder, and patients with this disorder vary considerably in their clinical presentation, response to therapy and risk of complications. Expanding knowledge about the pathophysiology of various forms of diabetes has raised the possibility that diagnostic and therapeutic modalities can be tailored to the individual patient in a personalized manner. The recent publication of a Consensus Statement on precision diabetes care underlines the major strides made in this field in the recent past. However, while personalized diabetes care has the potential to significantly improve outcomes in patients with diabetes in a safe and cost-effective manner, its wider application presents several challenges, especially in resource-strained settings. These challenges pertain equally to precision diagnostics, precision therapeutics and precision monitoring. This article discusses some of the important challenges that care providers are likely to face in applying the personalized approach in caring for their patients with diabetes, in the context of diagnosis and management of type 1 diabetes, type 2 diabetes and monogenic forms of diabetes. Suggestions are also presented for overcoming some of these challenges.

Clusters of people with type 2 diabetes in the general population: unsupervised machine learning approach using national surveys in Latin America and the Caribbean

INTRODUCTION

We aimed to identify clusters of people with type 2 diabetes mellitus (T2DM) and to assess whether the frequency of these clusters was consistent across selected countries in Latin America and the Caribbean (LAC).

RESEARCH DESIGN AND METHODS

We analyzed 13 population-based national surveys in nine countries (n=8361). We used k-means to develop a clustering model; predictors were age, sex, body mass index (BMI), waist circumference (WC), systolic/diastolic blood pressure (SBP/DBP), and T2DM family history. The training data set included all surveys, and the clusters were then predicted in each country-year data set. We used Euclidean distance, elbow and silhouette plots to select the optimal number of clusters and described each cluster according to the underlying predictors (mean and proportions).

RESULTS

The optimal number of clusters was 4. Cluster 0 grouped more men and those with the highest mean SBP/DBP. Cluster 1 had the highest mean BMI and WC, as well as the largest proportion of T2DM family history. We observed the smallest values of all predictors in cluster 2. Cluster 3 had the highest mean age. When we reflected the four clusters in each country-year data set, a different distribution was observed. For example, cluster 3 was the most frequent in the training data set, and so it was in 7 out of 13 other country-year data sets.

CONCLUSIONS

Using unsupervised machine learning algorithms, it was possible to cluster people with T2DM from the general population in LAC; clusters showed unique profiles that could be used to identify the underlying characteristics of the T2DM population in LAC.

Novel Subgroups and Chronic Complications of Diabetes in Middle-Aged and Elderly Chinese:A Prospective Cohort Study

BACKGROUND

Diabetes mellitus, especially type 2 diabetes mellitus (T2DM), is regarded as highly heterogeneous. Novel diabetes phenotypes by cluster analysis have been proposed in Europeans but may show different cluster features in Asians. The applicability of cluster analysis in middle-aged and elderly Chinese community T2DM patients needs further investigation.

METHODS

Participants were recruited from Jiading community in Shanghai, China. We adopted k-means cluster analysis in 1130 patients (aged ≥ 40 years) with newly-diagnosed T2DM at baseline. Cluster analysis was performed based on seven variables, including fasting plasma glucose, 2 hours postprandial blood glucose, age at diagnosis, body mass index, hemoglobin A1c, homoeostatic model assessment estimates of β-cell function and insulin resistance. All subjects were re-examined at 4.4 years later. Metabolic associated fatty liver disease was diagnosed using B-ultrasound, hepatic fibrosis by non-invasive scores, renal and cardiovascular status by subclinical biomarkers. Multivariable logistic regression models were used to compare the risks of complications between clusters.

RESULTS

Patients were classified into 4 clusters. 381 (33.7%), 456 (40.4%), 87 (7.7%), and 206 (18.2%) patients were separately assigned to mild age-related diabetes (MARD), mild obesity-related diabetes (MOD), severe insulin-deficient and insulin-resistant diabetes (SIDRD), or severe obesity-related and insulin-resistant diabetes (SOIRD), respectively. Participants in MARD, SOIRD, and SIDRD clusters were associated with significantly increased risks of different complications. SOIRD and SIDRD showed novel features in Chinese T2DM patients that were different from those in Europeans.

CONCLUSIONS

The refined diabetes phenotypic approach was applicable to Chinese middle-aged and elderly T2DM patients. Patients in different clusters presented significantly different characteristics, progression of metabolic features, and risks of diabetic complications.