Review of insulin-associated hypoglycemia and its impact on the management of diabetes in Southeast Asian countries

Generalization of a Deep Learning Model for Continuous Glucose Monitoring-Based Hypoglycemia Prediction: Algorithm Development and Validation Study

Background

Predicting hypoglycemia while maintaining a low false alarm rate is a challenge for the wide adoption of continuous glucose monitoring (CGM) devices in diabetes management. One small study suggested that a deep learning model based on the long short-term memory (LSTM) network had better performance in hypoglycemia prediction than traditional machine learning algorithms in European patients with type 1 diabetes. However, given that many well-recognized deep learning models perform poorly outside the training setting, it remains unclear whether the LSTM model could be generalized to different populations or patients with other diabetes subtypes.

Objective

The aim of this study was to validate LSTM hypoglycemia prediction models in more diverse populations and across a wide spectrum of patients with different subtypes of diabetes.

Methods

We assembled two large data sets of patients with type 1 and type 2 diabetes. The primary data set including CGM data from 192 Chinese patients with diabetes was used to develop the LSTM, support vector machine (SVM), and random forest (RF) models for hypoglycemia prediction with a prediction horizon of 30 minutes. Hypoglycemia was categorized into mild (glucose=54-70 mg/dL) and severe (glucose<54 mg/dL) levels. The validation data set of 427 patients of European-American ancestry in the United States was used to validate the models and examine their generalizations. The predictive performance of the models was evaluated according to the sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).

Results

For the difficult-to-predict mild hypoglycemia events, the LSTM model consistently achieved AUC values greater than 97% in the primary data set, with a less than 3% AUC reduction in the validation data set, indicating that the model was robust and generalizable across populations. AUC values above 93% were also achieved when the LSTM model was applied to both type 1 and type 2 diabetes in the validation data set, further strengthening the generalizability of the model. Under different satisfactory levels of sensitivity for mild and severe hypoglycemia prediction, the LSTM model achieved higher specificity than the SVM and RF models, thereby reducing false alarms.

Conclusions

Our results demonstrate that the LSTM model is robust for hypoglycemia prediction and is generalizable across populations or diabetes subtypes. Given its additional advantage of false-alarm reduction, the LSTM model is a strong candidate to be widely implemented in future CGM devices for hypoglycemia prediction.

A Cross-Sectional study on risk factors for severe hypoglycemia among Insulin-Treated elderly type 2 diabetes Mellitus (T2DM) patients in Singapore

OBJECTIVE

This study investigates the risk factors for severe hypoglycemia among Southeast Asian T2DM patients.

METHODS

Insulin-treated T2DM patients greater than 65 years old with HbA1c < 8% were recruited. They completed questionnaires detailing their experience of hypoglycemia and presence of impaired hypoglycemia awareness (IAH). Data on insulin treatment regimens, glycated haemoglobin (Hba1c) and comorbidities were also collected.

RESULTS

Of the 92 participants, 15.2% had at least one episode of severe hypoglycemia over the past year. Comparison between both groups showed that patients with severe hypoglycemia had lower Hba1c, higher Gold score (3.9 ± 1.9 vs. 2.5 ± 1.4; p < .05) and higher Hypoglycemia Fear Survey (HFS) worry score (39.1 ± 14.3 vs. 31.8 ± 11.8; p < .05). There were no significant differences in duration of diabetes and insulin treatment, treatment regimens and diabetes associated comorbidities except peripheral vascular disease. Furthermore, no significant differences were noted in HFS behavior score, hypoglycemia risk modifying behavior and social economic status.

CONCLUSIONS

Patients with severe hypoglycemia had tighter glycemic control, greater IAH and higher worry scores regardless of treatment regimens. Clinicians may play a significant role in tightening glycemic control and influencing the risk of severe hypoglycemia. Standard structured diabetes education may help reduce the risk of severe hypoglycemia among this group of patients.

Ramadan-focused nutrition therapy for people with diabetes: A narrative review

AIMS

This narrative review aimed to synthesize the evidence on the effects of Ramadan-focused nutrition therapy for people with diabetes.

METHODS

We searched MEDLINE (via PubMed) and Science Direct databases for articles that included the component of nutrition for adult patients with type 2 diabetes (T2D), published in English between 2010 and 2020.

RESULTS

Fourteen studies met the criteria. Eight of 14 studies had an intervention with a control arm. In comparison to the control group, all studies (n = 8) showed a reduction in hypoglycemic events. However, only half of these studies (n = 4) had shown at least one positive clinical outcome. Features of nutrition therapy that appeared to have favorable clinical outcomes include individualized caloric prescription; distributing carbohydrates equally between Suhoor, Iftar and snacks; providing meal plans; adjusting food intake to suit Ramadan; and incorporating diabetes-specific formula as part of Suhoor or snack.

CONCLUSIONS

The review provides evidence for the effectiveness of Ramadan-focused nutrition therapy among people with T2D and identifies key features of nutrition therapy that may provide favourable clinical outcomes. Additional data on dietary quality and adequacy during Ramadan fasting warrants further studies.

A SERS quenching method for the sensitive determination of insulin

In this work, we utilise the disulphide bond structure of insulin and a new benzothiazole Raman probe for the detection of human insulin using surface-enhanced Raman spectroscopy (SERS). The disulphide bond structure of the insulin was reduced to generate free sulfhydryl terminal groups. When reacted with benzothiazole-functionalised gold nanoparticles, the reduced protein desorbs the Raman probe and causes its Raman signal intensity to quench. Using this approach, insulin was quantified in the concentration range of 1 × 10^-14 -1 × 10^-8 M by SERS quenching. The limit of quantification of insulin by the SERS quenching method was found to be 1 × 10^-14 M (0.01 pM or 58 pg/L), which satisfies the requirements for monitoring its blood concentration in patients. Because many proteins and peptides have disulphide bonds in their molecular structures, the new SERS quenching method has a strong potential for the rapid determination of ultralow concentrations of proteins in formulations and biological fluids.

Real-world data reveal unmet clinical needs in insulin treatment in Asian people with type 2 diabetes: the Joint Asia Diabetes Evaluation (JADE) Register

AIMS

To explore the pattern of insulin use and glycaemic control in Asian people with type 2 diabetes, stratified by gender, young-onset diabetes (YOD; diagnosed before age 40 years), and diabetic kidney disease (DKD; estimated glomerular filtration rate [eGFR] < 60 mL/min/1.73m² ).

MATERIALS AND METHODS

We conducted a cross-sectional analysis of 97 852 patients from 11 Asian countries/regions (2007-2017) included in the prospective Joint Asia Diabetes Evaluation (JADE) Register.

RESULTS

Among 18 998 insulin users (47% women, mean ± SD age 59.2 ± 11.7 years, diabetes duration 13.2 ± 8.3 years, glycated haemoglobin [HbA1c] 72 ± 21.4 mmol/mol [8.74 ± 1.95%], median total daily insulin dose [TDD] 0.27-0.82 units/kg), 25% and 29.5% had YOD and DKD, respectively. Premixed (44%) and basal-only (42%) insulin were the most common regimens. Despite being more commonly treated with these two regimens with higher insulin dosages, patients with YOD had worse HbA1c levels than their late-onset peers (73 ± 20.5 vs. 71 ± 21.2 mmol/mol [8.82 ± 1.87% vs. 8.66 ± 1.94%]; P < 0.001). Fewer women than men attained an HbA1c level < 53 mmol/mol (7%; 15.7% vs 17.1%; P = 0.018). Adjusting for age, diabetes duration, TDD, HbA1c, eGFR, and use of oral glucose-lowering drugs at baseline, the odds of self-reported hypoglycaemia were higher in women (vs. men: adjusted odds ratio [aOR] 1.16, 95% confidence interval [CI] 1.05-1.28) and in patients with DKD treated with a premixed regimen (1.81 [95% CI 1.54-2.13] vs. 1.34 [95% CI 1.16-1.54] in non-DKD; P_interaction  < 0.001). Compared to basal-only regimens, premixed and basal-bolus regimens had similar HbA1c reductions but were independently associated with increased odds of hypoglycaemia (1.65 [95% CI 1.45-1.88] and 1.88 [95% CI 1.58-2.23], respectively).

CONCLUSIONS

In this Asian population, there were varying patterns of insulin regimens with suboptimal glycaemic control, despite relatively high TDDs, which were influenced by gender, DKD, and YOD status.

Self-reported hypoglycaemia in insulin-treated patients with diabetes mellitus: results from the Singapore cohort of the International Operations Hypoglycaemia Assessment Tool study

INTRODUCTION

Hypoglycaemia constitutes a significant barrier to achieving glycaemic control with insulin in both Type 1 (T1DM) and Type 2 diabetes mellitus (T2DM). The International Operations Hypoglycaemia Assessment Tool (IO HAT) study was designed to determine the incidence of hypoglycaemia in insulin-treated patients with T1DM and T2DM.

METHODS

The IO HAT study retrospectively and prospectively assessed the incidence of hypoglycaemia in patients with insulin-treated diabetes mellitus in nine countries. This sub-analysis included patients from Singapore with T1DM or T2DM who were aged ≥ 21 years and had completed two self-assessment questionnaires (SAQ1 and SAQ2).

RESULTS

Of the 50 T1DM and 320 T2DM patients who completed the SAQ1, 39 T1DM and 265 T2DM patients completed SAQ2; 100% and 90.9%, respectively, experienced at least one hypoglycaemic event prospectively. The incidence rates of any hypoglycaemia were 49.5 events per patient-year (EPPY) and 16.1 EPPY for T1DM and T2DM patients, respectively, in the four-week prospective period. Hypoglycaemia rate did not differ in terms of glycated haemoglobin level. The vast majority of T1DM or T2DM patients (92.0% and 90.7%, respectively) knew the overall definition of hypoglycaemia before study participation, although over half of the patients (T1DM 54.0%, T2DM 51.9%) defined hypoglycaemia based only on symptoms.

CONCLUSION

High proportions of insulin-treated patients with diabetes mellitus in Singapore reported hypoglycaemic events prospectively, showing that they had underreported hypoglycaemic episodes retrospectively. Patient education can help in improving hypoglycaemia awareness and its management in the region.

Toward Label-Free SERS Detection of Proteins through Their Disulfide Bond Structure

The molecular structure of many proteins contains disulfide bonds between their cysteine residues. In this work we demonstrate the utilization of the disulfide bond structure of proteins for their label-free determination by surface-enhanced Raman spectroscopy (SERS). The new approach for label-free SERS detection of proteins is demonstrated for human insulin. The protein was selectively extracted from spiked plasma samples using target-specific functionalized nanomaterial. Enzyme-linked immune assay (ELISA) was used to detect insulin in the blood plasma and cross-validate the SERS method. The disulfide bonds in the molecular structure of the protein were chemically reduced and used for their chemisorption onto the gold-coated copper oxide substrate in a unified orientation at a very short distance from the hotspots. The oriented chemisorption of the protein caused significant enhancement to the signal intensity of its Raman vibration modes. This is attributed to the strong short-range electromagnetic and chemical enhancement effects that are experienced by the immobilized protein. Using this approach, label-free and reproducible SERS detection of insulin, down to 10 zM (relative standard deviation [RSD] = 5.52%), was achieved. Sixty-five percent of proteins contain disulfide bonds in their molecular structure. Therefore, the new label-free SERS detection method has strong potential for the determination of ultralow concentrations of proteins at pathology labs and in biology research.

Basal insulin therapy: Unmet medical needs in Asia and the new insulin glargine in diabetes treatment

Diabetes remains a global epidemic and a tremendous health challenge, especially in the Asian population. Dramatic increases in the prevalence of diabetes across different countries or areas in Asia have been reported in recent epidemiological studies. Although clinical guidelines have strengthened appropriate antihyperglycemic medications and lifestyle modifications for optimal diabetes management, inadequate glycemic control still occurs in many patients with an increased risk of developing microvascular and macrovascular complications. Insulin administration is the main therapy for diabetes in response to the inability to secrete insulin, and is recommended in current guidelines to treat patients with type 2 diabetes after failure of oral antidiabetic drugs. Clinical studies have shown that long-acting insulin analogs improve basal glycemic control with reduced risk of hypoglycemia. In the present review, we discuss previous challenges with basal insulin therapy in Asia, the pharmacological development of insulin analogs to overcome the unmet medical needs and recent clinical studies of the new ultra-long-acting insulin analog, insulin glargine U300. Furthermore, relevant findings of current real-world evidence are also included for the comparison of the efficacy and safety of different insulin formulations. Based on the accumulating evidence showing a low incidence of hypoglycemia and technical benefits of dose titration, treatment with glargine U300 can be a promising strategy for Asian diabetes patients to achieve glycemic targets with favorable safety.

Dipeptidyl peptidase-4 inhibitors as add-on therapy to insulin in patients with type 2 diabetes mellitus: a meta-analysis of randomized controlled trials

PURPOSE

Addition of the dipeptidyl peptidase-4 (DPP4) inhibitors to insulin in patients with type 2 diabetes mellitus (T2DM) may achieve better glycemic control. However, results of pilot randomized controlled trials (RCTs) are inconsistent. We aimed to perform a meta-analysis of RCTs to evaluate efficacy and safety of DPP4 inhibitors compared with placebo/no treatment as add-on therapy to insulin in T2DM patients.

MATERIALS AND METHODS

Relevant studies were identified via a search of PubMed, Cochrane Library, and Embase databases. A fixed or random effect model was applied according to the heterogeneity.

RESULTS

Overall, 22 RCTs with 6,957 T2DM patients were included. Addition of DPP4 inhibitors to insulin was associated with significantly reduced HbA1c as compared with controls (weighed mean difference [WMD]: -0.54%, p<0.001). The benefits of DPP4 inhibitors as add-on therapy on HbA1c were independent of study design, follow-up duration, categories of DPP4 inhibitors used, and using of fixed/adjustable insulin doses as indicated by predefined subgroup analyses. Moreover, addition of DPP4 inhibitors to insulin was associated with significantly reduced fasting blood glucose (WMD: -0.47mmol/L, p<0.001), postprandial glucose at 2 hrs (WMD: -2.03 mmol/L, p<0.001), and daily dose of insulin (WMD: -2.73U/d, p<0.001), while body weight (WMD: 0.02 g, p=0.81) or risk of symptomatic hypoglycemia (risk ratio: 0.92, p=0.37) were not affected.

CONCLUSIONS

Addition of DPP4 inhibitors to insulin significantly improved the glycemic control in T2DM patients without further increasing the risk of weight gain and hypoglycemia.

Beyond HbA1c: Comparing Glycemic Variability and Glycemic Indices in Predicting Hypoglycemia in Type 1 and Type 2 Diabetes

BACKGROUND

Hypoglycemia is the major impediment to therapy intensification in diabetes. Although higher individualized HbA1c targets are perceived to reduce the risk of hypoglycemia in those at risk of hypoglycemia, HbA1c itself is a poor predictor of hypoglycemia. We assessed the use of glycemic variability (GV) and glycemic indices as independent predictors of hypoglycemia.

METHODS

A retrospective observational study of 60 type 1 and 100 type 2 diabetes subjects. All underwent professional continuous glucose monitoring (CGM) for 3-6 days and recorded self-monitored blood glucose (SMBG). Indices were calculated from both CGM and SMBG. Statistical analyses included regression and area under receiver operator curve (AUC) analyses.

RESULTS

Hypoglycemia frequency (53.3% vs. 24%, P < 0.05) and %CV (40.1% ± 10% vs. 29.4% ± 7.8%, P < 0.001) were significantly higher in type 1 diabetes compared with type 2 diabetes. HbA1c was, at best, a weak predictor of hypoglycemia. %CV_CGM, Low Blood Glucose Index (LBGI)_CGM, Glycemic Risk Assessment Diabetes Equation (GRADE)Hypoglycemia_CGM, and Hypoglycemia Index_CGM predicted hypoglycemia well. %CV_CGM and %CV_SMBG consistently remained a robust discriminator of hypoglycemia in type 1 diabetes (AUC 0.88). In type 2 diabetes, a combination of HbA1c and %CV_SMBG or LBGI_SMBG could help discriminate hypoglycemia.

CONCLUSION

Assessment of glycemia should go beyond HbA1c and incorporate measures of GV and glycemic indices. %CV_SMBG in type 1 diabetes and LBGI_SMBG or a combination of HbA1c and %CV_SMBG in type 2 diabetes discriminated hypoglycemia well. In defining hypoglycemia risk using GV and glycemic indices, diabetes subtypes and data source (CGM vs. SMBG) must be considered.

Hypoglycaemia among Insulin-Treated Patients with Diabetes: Southeast Asia Cohort of IO HAT Study

Objective

To provide real-world data on hypoglycaemia incidence in patients with type 1 (T1D) or type 2 diabetes (T2D) from the Southeast Asian cohort of the International Operations Hypoglycaemia Assessment Tool (IO HAT) study.

Methodology

IO HAT was a non-interventional, multicentre, 6-month retrospective and 4-week prospective study of hypoglycaemic events among insulin-treated adults with T1D or T2D, including four countries in Southeast Asia (Singapore, Philippines, Indonesia, and Bangladesh). Data were collected using a two-part self-assessment questionnaire (SAQ1 for retrospective and SAQ2 for prospective). The primary endpoint was the percentage of patients experiencing at least one hypoglycaemic event during the 4-week prospective observational period (ClinicalTrials.gov Identifier: NCT02306681).

Results

A total of 2594 patients completed SAQ1. Nearly all patients reported experiencing any hypoglycaemic event in the 4-week prospective period (T1D, 100%; T2D, 97.3%), with all patients reporting higher rates in the prospective versus retrospective period. Severe hypoglycaemia was also reported higher prospectively (57.2% and 76.9%) than retrospectively (33.9% and 12.2%) in both T1D and T2D, respectively. Nocturnal hypoglycaemia was reported higher retrospectively than prospectively.

Conclusion

Incidence of any and severe hypoglycaemia in the Southeast Asian cohort of IO HAT was higher prospectively versus retrospectively, suggesting hypoglycaemia has previously been under-reported in this region.

Use of 50/50 Premixed Insulin Analogs in Type 2 Diabetes: Systematic Review and Clinical Recommendations

INTRODUCTION

Premixed insulin analogs represent an alternative to basal or basal-bolus insulin regimens for the treatment of type 2 diabetes (T2D). "Low-mix" formulations with a low rapid-acting to long-acting analog ratio (e.g., 25/75) are commonly used, but 50/50 formulations (Mix50) may be more appropriate for some patients. We conducted a systematic literature review to assess the efficacy and safety of Mix50, compared with low-mix, basal, or basal-bolus therapy, for insulin initiation and intensification.

METHODS

MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, ClinicalTrials.gov, LillyTrials.com, and NovoNordisk-trials.com were searched (11 or 13 Dec 2016) using terms for T2D, premixed insulin analogs, and/or Mix50. Studies (randomized, nonrandomized, or observational; English only) comparing Mix50 with other insulins (except human) and reporting key efficacy [glycated hemoglobin (HbA1c), fasting and postprandial glucose] and/or safety (hypoglycemia, weight gain) outcomes were eligible for inclusion. Narrative reviews, letters, editorials, and conference abstracts were excluded. Risk of bias in randomized trials was assessed using the Cochrane tool.

RESULTS

MEDLINE and EMBASE searches identified 716 unique studies, of which 32 met inclusion criteria. An additional three studies were identified in the other databases. All 19 randomized trials except one were open label; risk of other biases was generally low. Although not conclusive, the evidence suggests that Mix50 may provide better glycemic control (HbA1c reduction) and, particularly, postprandial glucose reduction in certain patients, such as those with high carbohydrate diets and Asian patients, than low-mix and basal therapy. Based on this evidence and our experience, we provide clinical guidance on factors to consider when deciding whether Mix50 is appropriate for individual patients.

CONCLUSIONS

Mix50 may be more suitable than low-mix therapy for certain patients. Clinicians should consider not only efficacy and safety but also patient characteristics and preferences when tailoring insulin treatment to individuals with T2D.

FUNDING

Eli Lilly.

Review of insulin-associated hypoglycemia and its impact on the management of diabetes in Southeast Asian countries

Although the incidence of diabetes is rising in Southeast Asia, there is limited information regarding the incidence and manifestation of insulin-associated hypoglycemia. The aim of the present review was to discuss what is currently known regarding insulin-associated hypoglycemia in Southeast Asia, including its known incidence and impact in the region, and how the Southeast Asian population with diabetes differs from other populations. We found a paucity of data regarding the incidence of hypoglycemia in Southeast Asia, which has contributed to the adoption of Western guidelines. This might not be appropriate, as Southeast Asians have a range of etiological, educational and cultural differences from Western populations with diabetes that might place them at greater risk of hypoglycemia if not managed optimally. For example, Southeast Asians with type 2 diabetes tend to be younger, with lower body mass indexes than their Western counterparts, and the management of type 2 diabetes with premixed insulin preparations is more common in Southeast Asia. Both of these factors might result in higher rates of hypoglycemia. In addition, Southeast Asians are often poorly educated about hypoglycemia and its management, including during Ramadan fasting. We conclude there is a need for more information about Southeast Asian populations with diabetes to assist with the construction of more appropriate national and regional guidelines for the management of hypoglycemia, more closely aligned to patient demographics, behaviors and treatment practices. Such bespoke guidelines might result in a greater degree of implementation and adherence within clinical practice in Southeast Asian nations.