Comparative predictive ability of visit-to-visit HbA1c variability measures for microvascular disease risk in type 2 diabetes

Type 2 diabetes mellitus: A cross-sectional analysis of glycemic controls and brain health outcomes

In this cross-sectional analysis, we explored how fluctuations in glycemic levels impact executive functions and psychosocial outcomes in patients with type 2 diabetes mellitus (T2DM). The goal was to understand the relationship between glycemic control and both neuropsychological and psychosocial health. We stratified participants into well-controlled and poorly controlled groups based on glycated hemoglobin (HbA1c) levels and variability, including a healthy control group for comparison. The study consisted of neuropsychological tests and psychosocial assessments. Results indicated that the poorly controlled T2DM group experienced significant executive dysfunction and scored lower on the Tower of London, Wisconsin Card Sorting, and Digit Span Tests, reflecting a broader impact on quality of life and resilience. These findings support the importance of maintaining stable glycemic levels for better executive and psychosocial outcomes and highlight the need for regular neuropsychological and psychosocial assessments in diabetes care.

Effect of Preoperative Level of Glycemic Control with Pulsed Radiofrequency on the Incidence of Postherpetic Neuralgia in Patients with Herpes Zoster Combined with Type 2 Diabetes Mellitus: A Cohort Study

Purpose

To investigate the correlation between the level of glycosylated hemoglobin (HbA1c) and postherpetic neuralgia (PHN).

Patients and Methods

This cohort study included 100 patients with herpes zoster (HZ) undergoing treatment with pulsed radiofrequency (PRF). Patients with comorbid type 2 diabetes mellitus (T2DM) were divided into three groups based on their glycemic control levels: good [HbA1c < 7% (53.01 mmol/mol), group D1], fair [7% (53.01 mmol/mol) ≤ HbA1c < 9% (74.86 mmol/mol), group D2], and poor [9% (74.86 mmol/mol) ≤ HbA1c, group D3]. The control group (group N) consisted of patients without T2DM. The main outcome measured was the occurrence of PHN in the four groups.

Results

A total of 90 patients were included in the cohort. The occurrence of PHN was found to be higher in groups D2 and D3 when compared to group N (N vs D2, P = 0.007; N vs D3, P < 0.001). Furthermore, the occurrence of PHN was higher in groups D2 and D3 in comparison to group D1 (D1 vs D2, P = 0.022; D1 vs D3, P < 0.001), with the incidence of PHN in group D3 being greater than in group D2 (P < 0.001).

Conclusion

Preoperative HbA1c predicts the incidence of PHN after PRF in T2DM patients.

Relationship Between Glycosylated Hemoglobin Variability and the Severity of Coronary Artery Disease in Patients With Type 2 Diabetes Mellitus

Background: Glycosylated hemoglobin (HbA1c) variability is a risk factor for cardiovascular complications in patients with Type 2 diabetes mellitus (T2DM), but its relationship with the severity of coronary artery disease (CAD) is unclear. Methods: Patients with T2DM who underwent coronary angiography due to angina were enrolled. HbA1c variability was expressed as coefficient of variation (CV), standard deviation (SD), variability independent of mean (VIM), and time in range (TIR). The severity of CAD was expressed by the number of involved vessels and Gensini score. Multivariate regression models were constructed to test the relationship between HbA1c variability, number of involved vessels, and the Gensini score, followed by linear regression analysis. Results: A total of 147 patients were included. In multivariate analysis, VIM-HbA1c (OR = 2.604; IQR: 1.15, 5.90; r = 0.026) and HbA1cTIR (OR = 0.13; IQR: 0.04, 0.41; r < 0.001) were independent risk factors for the number of involved vessels. After adjustment, HbA1cTIR (OR = 0.01; IQR: 0.002, 0.04; r < 0.001), SD-HbA1c (OR = 4.12, IQR: 1.64, 10.35; r = 0.001), CV-HbA1c (OR = 1.41, IQR: 1.04, 1.92; r = 0.007), and VIM-HbA1c (OR = 3.26; IQR: 1.43, 7.47; r = 0.003) were independent risk factors for the Gensini score. In the linear analysis, the Gensini score was negatively correlated with HbA1cTIR (β = -0.629; r < 0.001) and positively correlated with SD-HbA1c (β = 0.271; r = 0.001) and CV-HbA1c (β = 0.176; r = 0.033). After subgroup analysis, HbA1cTIR was a risk factor for the number of involved vessels. The Gensini score was negatively correlated with HbA1cTIR and positively correlated with SD-HbA1c at subgroups of subjects with a mean HbA1c ≤ 7%. Conclusions: Our analysis indicates that HbA1c variability, especially HbA1cTIR, plays a role for the severity of CAD in patients with T2DM. HbA1c variability may provide additional information and require management even at the glycemic target. Translational Aspects: Studies have shown that HbA1c variability is related to cardiovascular complications. Further, we explore the correlation between HbA1c variability and the severity of CAD. HbA1c variability is a risk factor for coronary stenosis in T2DM. It may be a potential indicator reflecting glycemic control for the prevention and treatment of cardiovascular complications.

Combination of retagliptin and henagliflozin as add-on therapy to metformin in patients with type 2 diabetes inadequately controlled with metformin: A multicentre, randomized, double-blind, active-controlled, phase 3 trial

AIM

This study assessed the efficacy and safety of co-administering retagliptin and henagliflozin versus individual agents at corresponding doses in patients with type 2 diabetes mellitus who were inadequately controlled with metformin.

METHODS

This multicentre, phase 3 trial consisted of a 24-week, randomized, double-blind, active-controlled period. Patients with glycated haemoglobin (HbA1c) levels between 7.5% and 10.5% were randomized to receive once-daily retagliptin 100 mg (R100; n = 155), henagliflozin 5 mg (H5; n = 156), henagliflozin 10 mg (H10; n = 156), co-administered R100/H5 (n = 155), or R100/H10 (n = 156). The primary endpoint was the change in HbA1c from baseline to week 24.

RESULTS

Based on the primary estimand, the least squares mean reductions in HbA1c at week 24 were significantly greater in the R100/H5 (-1.51%) and R100/H10 (-1.54%) groups compared with those receiving the corresponding doses of individual agents (-0.98% for R100, -0.86% for H5 and -0.95% for H10, respectively; p < .0001 for all pairwise comparisons). Achievement of HbA1c <7.0% at week 24 was observed in 27.1% of patients in the R100 group, 21.2% in the H5 group, 24.4% in the H10 group, 57.4% in the R100/H5 group and 56.4% in the R100/H10 group. Reductions in fasting plasma glucose and 2-h postprandial glucose were also more pronounced in the co-administration groups compared with the individual agents at corresponding doses. Decreases in body weight and systolic blood pressure were greater in the groups containing henagliflozin than in the R100 group. The incidence rates of adverse events were similar across all treatment groups, with no reported episodes of severe hypoglycaemia.

CONCLUSIONS

For patients with type 2 diabetes mellitus inadequately controlled by metformin monotherapy, the co-administration of retagliptin and henagliflozin yielded more effective glycaemic control through 24 weeks compared with the individual agents at their corresponding doses.

An interpretable predictive deep learning platform for pediatric metabolic diseases

OBJECTIVES

Metabolic disease in children is increasing worldwide and predisposes a wide array of chronic comorbid conditions with severe impacts on quality of life. Tools for early detection are needed to promptly intervene to prevent or slow the development of these long-term complications.

MATERIALS AND METHODS

No clinically available tools are currently in widespread use that can predict the onset of metabolic diseases in pediatric patients. Here, we use interpretable deep learning, leveraging longitudinal clinical measurements, demographical data, and diagnosis codes from electronic health record data from a large integrated health system to predict the onset of prediabetes, type 2 diabetes (T2D), and metabolic syndrome in pediatric cohorts.

RESULTS

The cohort included 49 517 children with overweight or obesity aged 2-18 (54.9% male, 73% Caucasian), with a median follow-up time of 7.5 years and mean body mass index (BMI) percentile of 88.6%. Our model demonstrated area under receiver operating characteristic curve (AUC) accuracies up to 0.87, 0.79, and 0.79 for predicting T2D, metabolic syndrome, and prediabetes, respectively. Whereas most risk calculators use only recently available data, incorporating longitudinal data improved AUCs by 13.04%, 11.48%, and 11.67% for T2D, syndrome, and prediabetes, respectively, versus models using the most recent BMI (P < 2.2 × 10-16).

DISCUSSION

Despite most risk calculators using only the most recent data, incorporating longitudinal data improved the model accuracies because utilizing trajectories provides a more comprehensive characterization of the patient's health history. Our interpretable model indicated that BMI trajectories were consistently identified as one of the most influential features for prediction, highlighting the advantages of incorporating longitudinal data when available.

Prognostic value of longitudinal HbA1c variability in predicting the development of diabetic sensorimotor polyneuropathy among patients with type 2 diabetes mellitus: A prospective cohort observational study

AIMS/INTRODUCTION

This prospective cohort study aims to identify the optimal measure of glycated hemoglobin (HbA1c) variability and to explore its relationship with the development of new diabetic sensorimotor polyneuropathy (DSPN) in individuals with type 2 diabetes mellitus, building upon previous cross-sectional studies that highlighted a significant association between HbA1c visit-to-visit variability and DSPN.

MATERIALS AND METHODS

In a prospective study, 321 participants diagnosed with type 2 diabetes mellitus underwent comprehensive clinical assessments, neurophysiologic studies, and laboratory evaluations at enrollment and follow-up. Various indices, including HbA1c standard deviation (HbA1c SD), coefficient of variation (HbA1c CV), HbA1c change score (HbA1c HVS), and average real variability (HbA1c ARV), were employed to calculate the visit-to-visit variability HbA1c based on 3 month intervals. The investigation focused on examining the associations between these indices and the development of new DSPN.

RESULTS

The average follow-up duration was 16.9 ± 6.9 months. The Cox proportional hazards model identified age (P = 0.001), diabetes duration (P = 0.024), and HbA1C ARV (P = 0.031) as the sole factors associated with the development of new DSPN. Furthermore, the cumulative risk of developing DSPN over 1 year demonstrated a significant association with HbA1C ARV (P = 0.03, log-rank test).

CONCLUSIONS

Apart from age and diabetes duration, HbA1c variability emerged as a robust predictor for the occurrence of new DSPN. Among the various measures of HbA1c variability evaluated, HbA1c ARV demonstrated the highest potential as a reliable indicator for anticipating the onset of new DSPN.

Effects of intensive glycemic control on microvascular outcomes in type 2 diabetes mellitus are modified by long-term HbA1c variability: A post hoc analysis of the ACCORD trial

AIMS

To assess the impact of long-term visit-to-visit variability in HbA1c on microvascular outcomes in type 2 diabetes mellitus (T2DM), and its influence on the effects of intensive glycemic control.

METHODS

Included were participants with T2DM enrolled in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) who had at least three measurements of HbA1c prior to new-onset microvascular outcomes, namely nephropathy, retinopathy and neuropathy. Variability in HbA1c was defined as the coefficient of variation (CV) across HbA1c measurements obtained from enrollment to the transition from intensive to standard glycemic therapy.

RESULTS

During a median of 22,005, 23,121, and 13,080 person-years of follow-up, 2,905 nephropathy, 2,655 retinopathy, and 1,974 neuropathy cases were recorded, respectively. Median CV (IQR) was 7.91 % (5.66 %-10.76 %) in the standard treatment group and 9.79 % (7.32 %-13.35 %) in the intensive treatment group. In the standard treatment group, lower HbA1c-CV (the first versus the second quartile) was associated with a higher risk of all microvascular outcomes, while higher HbA1c-CV (the fourth quartile) was associated with a higher risk of nephropathy only. In the intensive treatment group, only higher HbA1c-CV was associated with a higher risk of developing the microvascular outcomes. Intensive therapy reduced all microvascular outcomes among individuals with lower HbA1c-CV, but increased the risk among those with the highest HbA1c-CV (all P values for interaction < 0.0001). For example, hazard ratios (95 % CI) of retinopathy comparing intensive with standard treatments were 0.65 (0.56-0.75), 0.84 (0.71-0.98), 0.97 (0.82-1.14) and 1.28 (1.08-1.53) across the lowest to the highest quartiles of HbA1c variability.

CONCLUSIONS

The effects of intensive glycemic control on microvascular outcomes in T2DM appear to be modified by the variability of HbA1c during the treatment process, suggesting the significance of dynamic monitoring of HbA1c levels and timely adjustments to the therapeutic strategy among individuals with a high HbA1c variability.

Glycated hemoglobin variability and the risk of cardiovascular events in patients with prediabetes and type 2 diabetes mellitus: A post-hoc analysis of a prospective and multicenter study

AIMS/INTRODUCTION

High glycated hemoglobin (HbA1c) variability has been reported to be linked with cardiovascular events in type 2 diabetes patients. Only a few studies have been carried out on Asian patients. This study aimed to investigate the association of prediabetes and type 2 diabetes in Asian patients by performing a post-hoc analysis of a multicenter, prospective, observational study.

MATERIALS AND METHODS

Data for prediabetes and type 2 diabetes patients were retrieved from a multicenter national registry entitled "CORE-Thailand study." The primary outcome was 4P-MACE (major adverse cardiovascular events, including non-fatal myocardial infarction, heart failure hospitalization, non-fatal stroke and all-cause death). Patients were stratified according to quartiles of HbA1c standard deviation. The Cox proportional hazards regression model was used to estimate the association of HbA1c variability with incident cardiovascular disease.

RESULTS

A total of 3,811 patients with prediabetes and type 2 diabetes were included. The median follow-up duration was 54 months. In the fully adjusted model, the highest quartile of HbA1c variability showed a statistically significant association with 4P-MACE (hazard ratio [HR] 2.77, 95% confidence interval [CI] 1.77-4.35), fatal and non-fatal myocardial infarction (HR 6.91, 95% CI 1.90-25.12), hospitalization for heart failure (HR 3.34, 95% CI 1.20-9.26) and all-cause death (HR 3.10, 95% CI 1.72-5.57). All these outcomes were statistically significantly different among four quartiles of HbA1c (log-rank P-value <0.05). Fatal and non-fatal stroke showed no statistically significant association with high HbA1c variability.

CONCLUSION

High HbA1c variability in the highest quartile showed a statistically significant association with multiple adverse cardiovascular events in an Asian population. Minimizing HbA1c fluctuation during long-term follow up should be another important objective for type 2 diabetes patients.

Key indices of glycaemic variability for application in diabetes clinical practice

Near normal glycaemic control in diabetes consists to target daily glucose fluctuations and quarterly HbA1c oscillations in addition to overall glucose exposure. Consequently, the prerequisite is to define simple, and mathematically undisputable key metrics for the short- and long-term variability in glucose homeostasis. As the standard deviations (SD) of either glucose or HbA1c are dependent on their means, the coefficient of variation (CV = SD/mean) should be applied instead as it that avoids the correlation between the SD and mean values. A CV glucose of 36% is the most appropriate threshold between those with stable versus labile glucose homeostasis. However, when near normal mean glucose concentrations are achieved a lower CV threshold of <27 % is necessary for reducing the risk for hypoglycaemia to a minimal rate. For the long-term variability in glucose homeostasis, a CVHbA1c < 5 % seems to be a relevant recommendation for preventing adverse clinical outcomes.

Correlates of disordered eating and insulin restriction behavior and its association with psychological health in Taiwanese youths with diabetes mellitus

BACKGROUND

Adolescents and young adults (AYAs) with diabetes mellitus (DM) are prone to eating disorders that may worsen metabolic control. This study investigated the clinical and behavioral correlates of disordered eating and insulin restriction (DE/IR) behavior and its association with psychological health among AYAs with DM.

METHODS

We enrolled patients with DM aged 10-30 years receiving insulin treatment in a tertiary medical center from 2019 to 2021. After obtaining informed consent, we assessed various visit-to-visit HbA1c measures indicating glycemic control, DE/IR behavior using the modified SCOFF questionnaire, weight-control practices (e.g., self-medication, induced vomiting, and over-exercising), and anxious and depressive symptoms using the Hospital Anxiety and Depression Scale. Correlation and hierarchical regression analyses were applied to understand the clinical and behavioral correlates of DE/IR behavior and its association with anxiety and depression.

RESULTS

Among the 110 patients with type 1 and type 2 DM recruited, we found 17.6% restricting insulin use and 6.3% self-medicating for weight control (higher in type 2 DM than type 1 DM). Hierarchical regression analyses showed HbA1c standard deviation (odds ratio = 2.18, [95% confidence interval 1.07-4.42]), body image (1.83, [1.05-3.20]), and dieting (4.74, [1.70-13.23]) associated with DE/IR behavior. Moreover, DE/IR behavior was further associated with anxiety (1.17 [1.08-1.27]) and depression (1.12 [1.03-1.22]).

CONCLUSION

DE/IR behavior is not uncommon among AYAs with DM, particularly those with type 2 DM, and may be associated with anxiety and depressive symptoms. In addition, HbA1c variability is correlated with DE/IR behavior, and the clinical implications need further exploration.

A study of factors influencing long-term glycemic variability in patients with type 2 diabetes: a structural equation modeling approach

Aim

The present study aims to utilize structural equation modeling (SEM) to investigate the factors impacting long-term glycemic variability among patients afflicted with type 2 diabetes.

Method

The present investigation is a retrospective cohort study that involved the collection of data on patients with type 2 diabetes mellitus who received care at a hospital located in Chengdu, Sichuan Province, over a period spanning from January 1, 2013, to October 30, 2022. Inclusion criteria required patients to have had at least three laboratory test results available. Pertinent patient-related information encompassing general demographic characteristics and biochemical indicators was gathered. Variability in the dataset was defined by standard deviation (SD) and coefficient of variation (CV), with glycosylated hemoglobin variation also considering variability score (HVS). Linear regression analysis was employed to establish the structural equation models for statistically significant influences on long-term glycemic variability. Structural equation modeling was employed to analyze effects and pathways.

Results

Diabetes outpatient special disease management, uric acid variability, mean triglyceride levels, mean total cholesterol levels, total cholesterol variability, LDL variability, baseline glycated hemoglobin, and recent glycated hemoglobin were identified as significant factors influencing long-term glycemic variability. The overall fit of the structural equation model was found to be satisfactory and it was able to capture the relationship between outpatient special disease management, biochemical indicators, and glycated hemoglobin variability. According to the total effect statistics, baseline glycated hemoglobin and total cholesterol levels exhibited the strongest impact on glycated hemoglobin variability.

Conclusion

The factors that have a significant impact on the variation of glycosylated hemoglobin include glycosylated hemoglobin itself, lipids, uric acid, and outpatient special disease management for diabetes. The identification and management of these associated factors can potentially mitigate long-term glycemic variability, thereby delaying the onset of complications and enhancing patients' quality of life.

Correlation Between Glycemic Variability and Diabetic Complications: A Narrative Review

Diabetes mellitus is a metabolic disorder with a complex etiology in which glycemic dynamics are disturbed and the body is unable to maintain the process of glucose homeostasis through the pancreas. Persistent symptoms of high blood glucose or low blood glucose may lead to diabetic complications, such as neuropathy, nephropathy, retinopathy, and cardiovascular diseases. Glycemic variability which can represent the presence of excessive glycemic excursions is an indicator for evaluating glucose homoeostasis. Limiting glycemic variability has gradually become an emerging therapeutic target in improve diabetes metabolism and prevent associated complications. This article reviews the progress of research on the various quantifiable parameters of glycemic variability and their relationships with vascular lesions and mechanisms.

Higher glucose fluctuation is associated with a higher risk of cardiovascular disease: Insights from pooled results among patients with diabetes

BACKGROUND

The relationship between glucose fluctuation and the risk of cardiovascular disease (CVD) in patients with diabetes remains elusive. Glycated hemoglobin (HbA1c) variability is a key parameter of glucose fluctuation.

METHODS

PubMed, Cochrane Library, Web of Science, and Embase were searched up to 1 July 2022. Studies reporting associations of HbA1c variability (HbA1c-SD), coefficient of variation of HbA1c (HbA1c-CV), and HbA1c variability score [HVS] with the risk of CVD among patients with diabetes were included. We used three different insights (a high-low value meta-analysis, a study-specific meta-analysis, and a non-linear dose-response meta-analysis) to explore the relationship between HbA1c variability and CVD risk. A subgroup analysis was also performed to screen the potential confounding factors.

RESULTS

A total of 14 studies with 254 017 patients with diabetes were eligible. The highest HbA1c variability was significantly associated with increased risks of CVD (HbA1c-SD, risk ratio [RR] 1.45; HbA1c-CV, RR 1.74; HVS, RR 2.46; all p < .001) compared to the lowest HbA1c variability. The RRs of CVD for per HbA1c variability were significantly >1 (all p < .001). The subgroup analysis for per HbA1c-SD found a significant exposure-covariate interaction in the types of diabetes (p = .003 for interaction). The dose-response analysis showed a positive association between HbA1c-CV and CVD risk (P for nonlinearity <.001).

CONCLUSIONS

Our study suggests that the higher glucose fluctuation is significantly associated with the higher CVD risk in diabetes patients based on HbA1c variability. The CVD risk associated with per HbA1c-SD might be higher among patients type 1 diabetes than patients with type 2 diabetes.

Long-Term Risk of Cardiovascular Disease Among Type 2 Diabetes Patients According to Average and Visit-to-Visit Variations of HbA1c Levels During the First 3 Years of Diabetes Diagnosis

BACKGROUND

It remains unclear whether a combination of glycemic variability and glycated hemoglobin (HbA1c) status leads to a higher incidence of cardiovascular disease (CVD). Therefore, to investigate CVD risk according to the glucose control status during early diabetes, we examined visit-to-visit HbA1c variability among patients with type 2 diabetes (T2DM).

METHODS

In this 9-year retrospective study, we measured HbA1c levels at each visit and tracked the change in HbA1c levels for 3 years after the first presentation (observation window) in newly diagnosed T2DM patients. We later assessed the occurrence of CVD in the last 3 years (target outcome window) of the study period after allowing a 3-year buffering window. The HbA1c variability score (HVS; divided into quartiles, HVS_Q1-4) was used to determine visit-to-visit HbA1c variability.

RESULTS

Among 4,817 enrolled T2DM patients, the mean HbA1c level was < 7% for the first 3 years. The group with the lowest HVS had the lowest rate of CVD (9.4%; 104/1,109 patients). The highest incidence of CVD of 26.7% (8/30 patients) was found in HVS [≥ 9.0%]_Q3, which was significantly higher than that in HVS [6.0-6.9%]_Q1 (P = 0.006), HVS [6.0-6.9%]_Q2 (P = 0.013), HVS [6.0-6.9%]_Q3 (P = 0.018), and HVS [7.0-7.9%]_Q3 (P = 0.040).

CONCLUSION

To our knowledge, this is the first long-term study to analyze the importance of both HbA1c change and visit-to-visit HbA1c variability during outpatient visits within the first 3 years. Lowering glucose levels during early diabetes may be more critical than reducing visit-to-visit HbA1c variability.

Associations Between Mean HbA1c, HbA1c Variability, and Both Mortality and Macrovascular Complications in Patients with Diabetes Mellitus: A Registry-Based Cohort Study

Background

We investigate the association between mean HbA1c, HbA1c variability, and all-cause mortality and diabetes-related macrovascular complications in patients with diabetes.

Methods

We performed a retrospective cohort study using patients present in the Singapore Health Services diabetes registry (SDR) during 2013 to 2014. We assessed mean HbA1c using three models: a baseline mean HbA1c for 2013-14, the mean across the whole follow-up period, and a time-varying yearly updated mean. We assessed HbA1c variability at baseline using the patient's HbA1c variability score (HVS) for 2013-14. The association between mean HbA1c, HVS, and 6 outcomes were assessed using Cox proportional hazard models.

Results

We included 43,837-53,934 individuals in the analysis; 99.3% had type 2 diabetes mellitus. The data showed a J-shaped distribution in adjusted hazard ratios (HRs) for all-cause mortality, ischemic heart disease, acute myocardial infarction, peripheral arterial disease, and ischemic stroke, with an increased risk of developing these outcomes at HbA1c <6% (42 mmol/mol) and ≥8% (64 mmol/mol). With the addition of HVS, the J-shaped distribution was maintained for the above outcomes, but HRs were greater at HbA1c <6.0% (42 mmol/mol) and reduced at HbA1c ≥8.0% (64 mmol/mol) when compared to models without HVS. The risk for all outcomes increased substantially with increasing glycaemic variability.

Conclusion

Both low (<6.0% [42 mmol/mol]) and high (≥8.0% [64 mmol/mol]) levels of glycaemic control are associated with increased all-cause mortality and diabetes-related macrovascular complications. Glycaemic variability is independently associated with increased risk for these outcomes. Therefore, patients with stable glycaemic level of 6-8% (42-64mmol/mol) are at lowest risk of all-cause mortality and diabetes-related macrovascular complications.

Development and validation of a novel nomogram to predict diabetic kidney disease in patients with type 2 diabetic mellitus and proteinuric kidney disease

PURPOSE

Differentiating between diabetic kidney disease (DKD) and non-diabetic kidney disease (NDKD) in patients with Type 2 diabetes mellitus (T2DM) is important due to implications on treatment and prognosis. Clinical methods to accurately distinguish DKD from NDKD are lacking. We aimed to develop and validate a novel nomogram to predict DKD in patients with T2DM and proteinuric kidney disease to guide decision for kidney biopsy.

METHODS

A hundred and two patients with Type 2 Diabetes Mellitus (T2DM) who underwent kidney biopsy from 1st January 2007 to 31st December 2016 were analysed. Univariate and multivariate analyses were performed to identify predictive variables and construct a nomogram. The discriminative ability of the nomogram was assessed by calculating the area under the receiver operating characteristic curve (AUROC), while calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test and calibration plot. Internal validation of the nomogram was assessed using bootstrap resampling.

RESULTS

Duration of T2DM, HbA1c, absence of hematuria, presence of diabetic retinopathy and absence of positive systemic biomarkers were found to be independent predictors of DKD in multivariate analysis and were represented as a nomogram. The nomogram showed excellent discrimination, with a bootstrap-corrected C statistic of 0.886 (95% CI 0.815-0.956). Both the calibration curve and the Hosmer-Lemeshow goodness-of-fit test (p = 0.242) showed high degree of agreement between the prediction and actual outcome, with the bootstrap bias-corrected curve similarly indicating excellent calibration.

CONCLUSIONS

A novel nomogram incorporating 5 clinical parameters is useful in predicting DKD in type 2 diabetes mellitus patients with proteinuric kidney disease.

HbA1c Variability and Cardiovascular Events in Patients with Prostate Cancer Receiving Androgen Deprivation Therapy

Background

Androgen deprivation therapy (ADT) worsens glycaemic control and cardiovascular outcomes. The prognostic value of visit-to-visit HbA1c variability (VVHV) has been unexplored in prostate cancer (PCa) patients receiving ADT.

Objective

To explore the effect of ADT on VVHV and the cardiovascular prognostic value of VVHV.

Design setting and participants

PCa patients receiving ADT in Hong Kong between January 1, 1993 and March 31, 2021 were included in this retrospective cohort study. Those with fewer than three HbA1c results available within 3 yr after ADT initiation, <6 mo of ADT, missing baseline HbA1c, prior diagnosis of any component of major adverse cardiovascular events (MACEs), and MACEs occurring within 3 yr were excluded. Patients were followed up until September 31, 2021.

Outcome measurements and statistical analysis

The outcome was MACEs (composite of heart failure, myocardial infarction, stroke, and cardiovascular mortality). VVHV was calculated from HbA1c levels within 3 yr after and, separately where available, before ADT initiation using coefficient of variation (CV; standard deviation [SD] divided by mean) and average real variability (ARV; average difference between consecutive measurements).

Results and limitations

Altogether, 1065 patients were analysed (median age 74.4 yr old [interquartile range 68.3-79.5 yr]). In 709 patients with VVHV available before and after ADT initiation, VVHV increased after ADT initiation (p < 0.001), with 473 (66.2%) and 474 (66.9%) having increased CV and ARV, respectively. Over a median follow-up of 4.3 yr (2.8-6.7 yr), higher VVHV was associated with a higher risk of MACEs (adjusted hazard ratio [per SD] for CV 1.21 [95% confidence interval: 1.02, 1.43], p = 0.029; ARV 1.25 [1.06, 1.48], p = 0.008). Limitations included residual confounding and selection bias.

Conclusions

In PCa patients receiving ADT, VVHV increased after ADT initiation. Higher VVHV was associated with an increased risk of MACEs.

Patient summary

In prostate cancer patients receiving androgen deprivation therapy (ADT), glycaemic control is less stable after initiating ADT, which was associated with an increased cardiovascular risk.

Association of visit-to-visit HbA1c variability with cardiovascular diseases in type 2 diabetes within or outside the target range of HbA1c

Background

Although a growing attention has been recently paid to the role of HbA1c variability in the risk of diabetic complications, the impact of HbA1c variability on cardiovascular diseases (CVD) in type 2 diabetes is still debated. The aim of the study is to investigate the association of HbA1c variability with CVD in individuals within or outside the target range of HbA1c.

Methods

Using data from Action in Diabetes and Vascular disease: preterAx and diamicroN-MR Controlled Evaluation (ADVANCE), we enrolled 855 patients with type 2 diabetes in China. The primary outcomes included major macrovascular events and major microvascular events. Visit-to-visit HbA1c variability was expressed as the coefficient of variation (CV) of five measurements of HbA1c taken 3-24 months after treatment. Cox proportional hazard models were used to estimate adjusted hazard ratios (aHR).

Results

Among 855 patients in the intensive glucose treatment group, 563 and 292 patients were assigned to the group of "within the target range of HbA1c" (WTH) (updated mean HbA1c ≤ 7.0%) and "outside the target range of HbA1c" (OTH) (updated mean HbA1c > 7.0%), respectively. HbA1c variability was positively associated with the risk of major microvascular events in all patients and both the subgroups during a median follow-up period of 4.8 years. Particularly, the risk related to HbA1c variability was higher in patients in WTH group for the new or worsening nephropathy [aHR: 3.35; 95% confidence interval (CI): 1.05-10.74; P = 0.042].

Conclusions

This retrospective cohort study confirmed the positive correlation between HbA1c variability and major microvascular events, especially in subjects in WTH or OTH.

Identification of robust deep neural network models of longitudinal clinical measurements

Deep learning (DL) from electronic health records holds promise for disease prediction, but systematic methods for learning from simulated longitudinal clinical measurements have yet to be reported. We compared nine DL frameworks using simulated body mass index (BMI), glucose, and systolic blood pressure trajectories, independently isolated shape and magnitude changes, and evaluated model performance across various parameters (e.g., irregularity, missingness). Overall, discrimination based on variation in shape was more challenging than magnitude. Time-series forest-convolutional neural networks (TSF-CNN) and Gramian angular field(GAF)-CNN outperformed other approaches (P < 0.05) with overall area-under-the-curve (AUCs) of 0.93 for both models, and 0.92 and 0.89 for variation in magnitude and shape with up to 50% missing data. Furthermore, in a real-world assessment, the TSF-CNN model predicted T2D with AUCs reaching 0.72 using only BMI trajectories. In conclusion, we performed an extensive evaluation of DL approaches and identified robust modeling frameworks for disease prediction based on longitudinal clinical measurements.

Association of hemoglobin A1c time in range with risk for diabetes complications

INTRODUCTION

We assessed the association between hemoglobin A1c time in range (A1c TIR), based on unique patient-level A1c target ranges, with risks of developing microvascular and macrovascular complications in older adults with diabetes.

RESEARCH DESIGN AND METHODS

We used a retrospective observational study design and identified patients with diabetes from the Department of Veterans Affairs (n=397 634). Patients were 65 years and older and enrolled in Medicare during the period 2004-2016. Patients were assigned to individualized A1c target ranges based on estimated life expectancy and the presence or absence of diabetes complications. We computed A1c TIR for patients with at least four A1c tests during a 3-year baseline period. The association between A1c TIR and time to incident microvascular and macrovascular complications was studied in models that included A1c mean and A1c SD.

RESULTS

We identified 74 016 patients to assess for incident microvascular complications and 89 625 patients to assess for macrovascular complications during an average follow-up of 5.5 years. Cox proportional hazards models showed lower A1c TIR was associated with higher risk of microvascular (A1c TIR 0% to <20%; HR=1.04; 95%) and macrovascular complications (A1c TIR 0% to <20%; HR=1.07; 95%). A1c mean was associated with increased risk of microvascular and macrovascular complications but A1c SD was not. The association of A1c TIR with incidence and progression of individual diabetes complications within the microvascular and macrovascular composites showed similar trends.

CONCLUSIONS

Maintaining stability of A1c levels in unique target ranges was associated with lower likelihood of developing microvascular and macrovascular complications in older adults with diabetes.

A hybrid approach for risk stratification and predictive modelling of 30-days unplanned readmission of comorbid patients with diabetes

OBJECTIVES

When comorbid patients with diabetes have 30-days Unplanned Readmission (URA), they attract more burdens to the healthcare system due to increased cost of treatment, insurance penalties to hospitals, and unavailable bed spaces for new patients. This paper, therefore, aims to develop a risk stratification and a predictive model for identifying patients at various risk severities of 30-days URA.

METHODS

Patients records of comorbid patients with diabetes treated with different medications were collected from different hospitals and analysed with Principal Component Analysis (PCA) and Multivariate Logistic Regression (MLR) to determine the probability of 30-days URA, which is classified into very low, low, moderate, high, and very high. The risk classes are later modelled using ANOVA feature selection to identify the optimal predictors and the best random forest (RF) hyperparameters for 30-days URA risk stratification. Synthetic Minority Oversampling Technique (SMOTE) was used to balance the risk classes while employing a10-fold cross-validation.

RESULTS

After analysing 17,933 episodes of comorbid diabetes patients' treatment, 10.71% are identified to have 30-days URA with 61.95% of patients at moderate risk, 35.5% at low risk, 2.25% at very low risk, 0.37% at high risk, and 0.08% at very high risk. The predictive accuracy of RF is: - recall: 0.947 ± 0.035, precision: 0.951 ± 0.033, F1-score: 0.947 ± 0.035, AUC: 0.994 ± 0.007 and Average Precision (AP) of 0.99. The predictive accuracies of the risk classes measured with F1-score are: - very low: 0.985 ± 0.019, low risk: 0.871 ± 0.079, moderate: 0.881 ± 0.093, high: 0.999 ± 0.003, and very high: 1.000 ± 0.00.

CONCLUSION

This study identified the risk severity of comorbid patients with diabetes treated with different medications, making it easier to identify those that will be prioritized on hospitalization to minimize 30-days URA. By relying on the technique developed, vulnerable patients to 30-days URA can be given better post-discharge monitoring to build critical self-management skills that will minimize the cost of diabetes care and improve the quality of life.

Prevention of Peripheral Distal Polyneuropathy in Patients with Diabetes: A Systematic Review

Background: Diabetic peripheral neuropathy (DPN) is the most frequent chronic complication and is that which generates the highest disability and mortality in diabetes mellitus (DM). As it is currently the only microvascular complication of DM without a specific treatment, prevention is essential. The aim of this study was to determine the most effective preventive strategy to avoid or delay the appearance and/or development of DPN in patients with DM. Methods: A systematic search was carried out in the main health science databases (PubMed, Scopus, CINAHL, PEDro and The Cochrane Library) from 1 January 2010 to 31 August 2020. The study selection was conducted by two independent reviewers and data extraction was performed by the author. The eligibility criteria included randomized clinical trials (RCTs) and cohort studies from RCTs. Results: Eleven studies were selected that included 23,595 participants with DM. The interventions evaluated were intensive or standard glycemic control, the use of drugs to achieve glycemic control, and the promotion of a healthy lifestyle and exercise. Intensive glucose control achieved a significant reduction in the development of DPN in TIDM patients, and lifestyle modifications and exercise achieved it moderately in TIIDM patients. Conclusions: The main preventive strategy for DPN is intensive glycemic control with a target HbA1c < 6% in patients with TIDM and standard control of 7.0−7.9 in patients with TIIDM, incorporating lifestyle modifications.

HbA1c variability predicts cardiovascular complications in type 2 diabetes regardless of being at glycemic target

Background

HbA1c variability has emerged as risk factor for cardiovascular diseases in diabetes. However, the impact of HbA1c variability on cardiovascular diseases in subjects within the recommended HbA1c target has been relatively unexplored.

Methods

Using data from a large database, we studied 101,533 people with type 2 diabetes without cardiovascular diseases. HbA1c variability was expressed as quartiles of the standard deviation of HbA1c during three years (exposure phase). The primary composite outcome included non-fatal myocardial infarction, non-fatal stroke, all-cause mortality and was assessed during five years following the first three years of exposure to HbA1c variability (longitudinal phase). An expanded composite outcome including non-fatal myocardial infarction, non-fatal stroke, coronary revascularization/reperfusion procedures, peripheral revascularization procedures, and all-cause mortality was also considered, as well as a series of specific cardiovascular complications. Cox models were adjusted for a large range of risk factors and results were expressed as adjusted hazard ratios.

Results

An association between HbA1c variability and all the outcomes considered was found. The correlation between HbA1c variability and cardiovascular complications development was confirmed in both the subgroups of subjects with a mean HbA1c ≤ 53 mmol/mol (recommended HbA1c target) or > 53 mmol/mol during the exposure phase. The risk related to HbA1c variability was higher in people with mean HbA1c ≤ 53 mmol/mol for the primary outcome (p for interaction 0.004), for the expanded secondary outcome (p for interaction 0.001) and for the stroke (p for interaction 0.001), even though HbA1c remained at the target during the follow-up.

Conclusions

These findings suggest that HbA1c variability may provide additional information for an optimized management of diabetes, particularly in people within the target of HbA1c.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-022-01445-4.

Effect of Hemoglobin A1c Trajectories on Future Outcomes in a 10-Year Cohort With Type 2 Diabetes Mellitus

BACKGROUND

Hemoglobin A1c (HbA1c) variability may be a predictor of diabetic complications, but the predictive values of HbA1c trajectories remain unclear. We aimed to classify long-term HbA1c trajectories and to explore their effects on future clinical outcomes in a 10-year cohort with type 2 diabetes mellitus (T2DM).

METHODS

A total of 2,161 participants with T2DM from the Beijing Community Diabetes Study were included. The 10-year follow-up was divided into two stages for the present data analysis. Stage I (from 2008 to 2014) was used to identify the HbA1c trajectories and to calculate the adjusted SD of HbA1c (HbA1c-adjSD), or the coefficient of variation of HbA1c (HbA1c-CV). Stage II (from 2014 to 2018) was used to collect the records of the new occurrence of diabetes-related clinical outcomes. Latent growth mixture models were used to identify HbA1c trajectories. Cox proportional hazards models were used to explore the relationship between HbA1c trajectories, HbA1c-adjSD, or HbA1c-CV and the future outcomes.

RESULTS

Three HbA1c trajectories were identified, including low stable (88.34%), gradual decreasing (5.83%), and pre-stable and post-increase (5.83%). Either the risk of death or the chronic complications were significantly higher in the latter two groups compared to the low stable group after adjustment for average HbA1c and other traditional risk factors, the adjusted hazard ratios (HRs) for renal events, composite endpoint, and all-cause death for the pre-stable and post-increase group were 2.83 [95%CI: 1.25-6.41, p = 0.013], 1.85 (95%CI: 1.10-3.10, p = 0.020), and 3.01 (95%CI: 1.13-8.07, p = 0.028), respectively, and the adjusted HR for renal events for the gradual decreasing group was 2.37 (95%CI: 1.08-5.21, p = 0.032). In addition, both univariate and multivariate Cox HR models indicated that participants in the fourth and third quartiles of HbA1c-CV or HbA1c-adjSD were at higher risk of renal events compared to participants in the first quartile.

CONCLUSIONS

HbA1c trajectories, HbA1c-CV, and HbA1c-adjSD could all predict diabetes-related clinical outcomes. HbA1c trajectories could reflect long-term blood glucose fluctuation more intuitively, and non-stable HbA1c trajectories may predict increased risk of renal events, all-cause death, and composite endpoint events, independent of average HbA1c.

Predictive Ability of Visit-to-Visit Variability of HbA1c Measurements for the Development of Diabetic Kidney Disease: A Retrospective Longitudinal Observational Study

AIMS

This study is aimed at clarifying the relationship between visit-to-visit variability of glycated hemoglobin (HbA1c) and the risk of diabetic kidney disease (DKD) and to identifying the most useful index of visit-to-visit variability of HbA1c.

METHODS

This clinic-based retrospective longitudinal study included 699 Japanese type 2 diabetes mellitus patients. Visit-to-visit variability of HbA1c was calculated as the internal standard deviation of HbA1c (HbA1c-SD), the coefficient of variation of HbA1c (HbA1c-CV), the HbA1c change score (HbA1c-HVS), and the area under the HbA1c curve (HbA1c-AUC) with 3-year serial HbA1c measurement data, and the associations between these indices and the development/progression of DKD were examined.

RESULTS

Cox proportional hazards models showed that the HbA1c-SD and HbA1c-AUC were associated with the incidence of microalbuminuria, independently of the HbA1c level. These results were verified and replicated in propensity score (PS) matching and bootstrap analyses. Moreover, the HbA1c-SD and HbA1c-AUC were also associated with oxidized human serum albumin (HSA), an oxidative stress marker.

CONCLUSIONS

Visit-to-visit variability of HbA1c was an independent risk factor of microalbuminuria in association with oxidative stress among type 2 diabetes mellitus patients. HbA1c-AUC, a novel index of HbA1c variability, may be a potent prognostic indicator in predicting the risk of microalbuminuria.

Machine learning based study of longitudinal HbA1c trends and their association with all-cause mortality: Analyses from a National Diabetes Registry

OBJECTIVE

The association of long-term HbA1c variability with mortality has been previously suggested. However, the significance of HbA1c variability and trends in different age and HbA1c categories is unclear.

RESEARCH DESIGN AND METHODS

Data on patients with diabetes listed in the Israeli National Diabetes Registry during years 2012-2016 (observation period) were collected. Patients with >4 HbA1c measurements, type 1 diabetes, eGFR < 30mg/ml/min, persistent HbA1c < 6% or malignancy were excluded. Utilizing machine learning methods, patients were classified into clusters according to their HbA1c trend (increasing, stable, decreasing). Mortality risk during 2017-2019 was calculated in subgroups defined by age (35-54, 55-69, 70-89 years) and last HbA1c (≤7% and >7%) at end of observation period. Models were adjusted for demographic, clinical and laboratory measurements including HbA1c, standard deviation (SD) of HbA1c and HbA1c trend.

RESULTS

This historical cohort study included 293,314 patients. Increased HbA1c variability (high SD) during the observation period was an independent predictor of mortality in patients aged more than 55 years (p < 0.01). The HbA1c trend was another independent predictor of mortality. Patients with a decreasing versus stable HbA1c trend had a greater mortality risk; this association persisted in all age groups in patients with HbA1c > 7% at the end of the observation period (p = 0.02 in age 35-54; p < 0.01 in aged >55). Patients with an increasing versus stable HbA1c trend had a greater mortality risk only in the elderly group (>70), yet in both HbA1c categories (p < 0.01).

CONCLUSIONS

HbA1c variability and trend are important determinants of mortality risk and should be considered when adjusting glycaemic targets.

Fasting Glucose Variability as a Risk Indicator for End-Stage Kidney Disease in Patients with Diabetes: A Nationwide Population-Based Study

Given the fact that diabetes remains a leading cause of end-stage kidney disease (ESKD), multi-aspect approaches anticipating the risk for ESKD and timely correction are crucial. We investigated whether fasting glucose variability (FGV) could anticipate the development of ESKD and identify the population prone to the harmful effects of GV. We included 777,192 Koreans with diabetes who had undergone health examinations more than three times in 2005-2010. We evaluated the risk of the first diagnosis of ESKD until 2017, according to the quartile of variability independent of the mean (VIM) of FG using multivariate-adjusted Cox proportional hazards analyses. During the 8-year follow-up, a total of 7290 incidents of ESKD were found. Subjects in the FG VIM quartile 4 had a 27% higher risk for ESKD compared to quartile 1, with adjustment for cardiovascular risk factors and the characteristics of diabetes. This effect was more distinct in patients aged < 65 years; those with a long duration of diabetes; the presence of hypertension or dyslipidemia; and prescribed angiotensin-converting enzyme inhibitors, metformin, sulfonylurea, α-glucosidase inhibitors, and insulin. In contrast, the relationship between baseline FG status and ESKD risk showed a U-shaped association. FGV is an independent risk factor for kidney failure regardless of FG.

The Association Between Mean Glycated Haemoglobin or Glycaemic Variability and the Development of Retinopathy in People with Diabetes: A Retrospective Observational Cohort Study

INTRODUCTION

To determine the association between mean glycated haemoglobin (HbA_1c) or glycaemic variability and the development of diabetic retinopathy (DR) in people with diabetes.

METHODS

An observational cohort study with people registered with a DR eye screening service between October 2012 and October 2017. Those who had no DR at the start of the study were followed for a maximum of 5 years. HbA_1c measures were used to calculate HbA_1c variability and mean HbA_1c to assess any relationship between these and the risk of developing new onset DR.

RESULTS

A total of 2511 individuals were followed up for up to 5 years. Of these, 542 (21.6%) developed DR. After adjustment, HbA_1c variability was not significantly associated with the development of DR (p = 0.3435). However, the mean HbA_1c was (p < 0.0001). Those with type 1 diabetes had an odds of 1.63 (95% CI 1.11-2.40) of a retinopathy diagnosis compared to those with type 2 diabetes.

CONCLUSIONS

We have shown that mean HbA_1c is associated with an increased risk of developing diabetic retinopathy. However, after adjustment for sex, age, diabetes type and the mean, the HbA_1c variability no longer remained significant. Our data suggest that optimizing long-term glycaemic control remains paramount.

Glycemic Gap as a Useful Surrogate Marker for Glucose Variability and Progression of Diabetic Retinopathy

(1) Background: Recent studies have reported that the glucose variability (GV), irrespective of glycosylated hemoglobin (HbA1c), could be an additional risk factor for the development of diabetic retinopathy (DR). However, measurements for GV, such as continuous glucose monitoring (CGM) and fasting plasma glucose (FPG) variability, are expensive and time consuming. (2) Methods: This present study aims to explore the correlation between the glycemic gap as a measurement of GV, and DR. In total, 2565 patients were included in this study. We evaluated the effect of the different types of glycemic gaps on DR progression. (3) Results: We found that the area under the curve (AUC) values of both the glycemic gap and negative glycemic gap showed an association with DR progression. (4) Conclusions: On eliminating the possible influences of chronic blood glucose controls, the results show that GV has deleterious effects that are associated with the progression of DR. The glycemic gap is a simple measurement of GV, and the predictive value of the negative glycemic gap in DR progression shows that GV and treatment-related hypoglycemia may cause the development of DR. Individual treatment goals with a reasonable HbA1c and minimal glucose fluctuations may help in preventing DR.

Higher long-term visit-to-visit glycemic variability predicts new-onset atrial fibrillation in patients with diabetes mellitus

Background

Atrial fibrillation (AF) is prevalent in patients with type 2 diabetes mellitus (T2DM). Glycemic variability (GV) is associated with risk of micro- and macrovascular diseases. However, whether the GV can increase the risk of AF remains unknown.

Methods

The cohort study used a database from National Taiwan University Hospital, a tertiary medical center in Taiwan. Between 2014 and 2019, a total of 27,246 adult patients with T2DM were enrolled for analysis. Each individual was assessed to determine the coefficients of variability of fasting glucose (FGCV) and HbA1c variability score (HVS). The GV parameters were categorized into quartiles. Multivariate Cox regression models were employed to estimate the relationship between the GV parameters and the risk of AF, transient ischemic accident (TIA)/ischemic stroke and mortality in patients with T2DM.

Results

The incidence rates of AF and TIA/ischemic stroke were 21.31 and 13.71 per 1000 person-year respectively. The medium follow-up period was 70.7 months. In Cox regression model with full adjustment, the highest quartile of FGCV was not associated with increased risk of AF [Hazard ratio (HR): 1.12, 95% confidence interval (CI) 0.96–1.29, p = 0.148] or TIA/ischemic stroke (HR: 1.04, 95% CI 0.83–1.31, p = 0.736), but was associated with increased risk of total mortality (HR: 1.33, 95% CI 1.12–1.58, p < 0.001) and non-cardiac mortality (HR: 1.41, 95% CI 1.15–1.71, p < 0.001). The highest HVS was significantly associated with increased risk of AF (HR: 1.29, 95% CI 1.12–1.50, p < 0.001), total mortality (HR: 2.43, 95% CI 2.03–2.90, p < 0.001), cardiac mortality (HR: 1.50, 95% CI 1.06–2.14, p = 0.024) and non-cardiac mortality (HR: 2.80, 95% CI 2.28–3.44, p < 0.001) but was not associated with TIA/ischemic stroke (HR: 0.98, 95% CI 0.78–1.23, p = 0.846). The Kaplan–Meier analysis showed significantly higher risk of AF, cardiac and non-cardiac mortality according to the magnitude of GV (log-rank test, p < 0.001).

Conclusions

Our data demonstrate that high GV is independently associated with the development of new-onset AF in patients with T2DM. The benefit of maintaining stable glycemic levels to improve clinical outcomes warrants further studies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-021-01341-3.