The HKDR was the first to report a linear relationship between HbA1c and cancer risk, with every 1% increase in HbA1c associated with 18% increased risk of all-site cancer [10]. Patients treated with insulin and various glucose-lowering drugs including metformin, sulphonylurea and thiazolidinediones had 40% to 60% lower cancer risk compared to non-users [11]. Apart from hyperglycaemia, the CUHK-PWH team also identified non-linear risk associations of cancer with body mass index (BMI), LDL-C, high density lipoprotein cholesterol (HDL-C), triglycerides (TG), and white blood cell (WBC) count in V-shaped or A-shaped manners [4]. Supported by experimental studies in animals, we first reported that the use of statins and renin-angiotensin system (RAS) blockers was associated with reduced risk for cancer in patients with T2DM in the HKDR. Subsequent detailed analysis revealed various drug-subphenotype interactions associated with cancer risk [12].

These subphenotypes suggested complex pathways linking lipid and glucose metabolism with cancer biology (Fig. 2). Due to the close links between diabetes, obesity and insulin resistance and given the co-sharing of insulin and insulin-like growth factor 1 (IGF-1) receptors, hyperinsulinaemia was often proposed as a major mechanism for promoting tumorigenesis through activation of insulin receptors and IGF-1 receptors expressed by tumor cells. However, insulin played a key role in de novo lipogenesis through activation of multiple sterol regulatory element-binding proteins (SREBPs). Thus, insulin deficiency might lead to reduced SREBP-1 and low TG. This might upregulate the IGF-1 pathway to increase other SREBPs which regulated the cholesterol pathway through activation of the hydroxymethylglutaryl-CoA reductase (HMGCR). The latter could lead to increased synthesis of mevalonate and farnesylated proteins, the latter being upstream signaling pathways of cell mitogenesis. Besides, hyperglycaemia resulting from insulin deficiency could activate the RAS to promote cellular growth [1516]. In a series of experimental studies, we reported development of renal cell carcinoma in uninephrectomized rat which exhibited a combined phenotype of hyperglycaemia, albuminuria and high LDL-C levels with upregulated expression of HMGCR and components of the RAS and IGF-1 pathways. Angiotensin converting enzyme inhibitors (ACEIs) treatment normalized expression of these proteins and reduced renal cell carcinoma formation. These findings suggested crosstalk between the RAS, HMGCR, and IGF-1 pathways in carcinogenesis which might explain the clinical observations of reduced cancer risk in patients treated with RAS blockers and statins [12].

Hepatocellular carcinoma (HCC) was a leading cause of cancer deaths in Hong Kong due to the high local prevalence of chronic hepatitis B viral (HBV) infection. The CUHK-PWH diabetes team was among the first to report the unfavorable modifying effect of chronic HBV infection on cardiovascular-renal outcomes in patients with diabetes [47]. Using the data from the HKDR, we first reported that HbA1c greater than 7% and the lipid phenotype (LDL-C less than 2.8 mmol/L plus TG less than 1.7 mmol/L) markedly increased the hazard ratios for incident HCC amongst HBV carriers, which were attenuated by the use of insulin and statins [17]. These results also corroborated the findings from an experimental study where insulin treatment reversed the increased liver tumors in insulin-deficient cancer-prone mice [18].

Both metformin and apolipoprotein A-I (APOA-I), the latter being the main fraction of HDL-C, were activators of the adenosine 5′-monophosphate-activated protein kinase (AMPK) pathway, a master regulator of energy metabolism [15]. In animal studies, inhibition of this pathway led to tumor formation whereas use of metformin activated the pathway and inhibited cancer cell growth. Insulin played a key role in regulating lipid metabolism and patients with insulin resistance were prone to have low HDL-C and APOA-I levels which might explain the interaction between metformin and low HDL-C phenotype in reducing all-cancer risk as reported in the HKDR [15].

In summary, detailed analysis of the HKDR have provided new insights into the potential cross-talks amongst these four biological pathways (HMGCR, RAS, IGF-1, and AMPK), explaining the diabetes-cancer link (Table 2). Dysregulation of these pathways have been reported in experimental studies of cancer which supported these clinical observations. Thus, hyperglycaemia, either due to insulin resistance and/or insulin deficiency could trigger these pathways to increase oxidative stress, inflammation and RAS activity to promote atherogenesis and oncogenesis. Similar to the multi-pronged approach needed to reduce cardiovascular-renal complications, the CUHK-PWH diabetes team hypothesized that an integrated approach was needed to optimize the metabolic milieu for reducing cancer risk in T2DM, although randomized clinical trials would be needed to confirm this hypothesis [11].

Asians had higher prevalence of diabetic kidney disease than Caucasians, accounting for 40% to 55% of those on dialysis, compared with less than 30% of those in Western countries [19]. They also had a greater propensity for earlier development of micro- or macroalbuminuria, affecting 60% of diabetic patients compared with 30% to 40% in Caucasians, given similar duration of diabetes or risk factor control [20]. The JADE risk equation for ESRD comprised of only three independent predictors: urinary albumin-creatinine ratio (ACR), estimated glomerular filtration rate (eGFR), and haematocrit. Of these predictors, urinary ACR was the most modifiable risk factor for controlling multiple cardiometabolic risk factors [21]. In a cohort study from the HKDR, for every 10 mm Hg increase in systolic BP above 130 mm Hg, CHD risk was increased by 1.13-fold amongst Hong Kong women with T2DM [22]. However the significance was lost after adjustment of urinary ACR and/or eGFR, suggesting that the effects of elevated BP on increased CHD risk was mediated by urinary ACR, which might be a surrogate marker of endothelial dysfunction [2324]. In another analysis, we found a linear association between albuminuria and ischemic stroke most notably amongst those with HbA1c greater than 6.2% [25]. Based on these findings, it may be inferred that in patients whose HbA1c could not be reduced safely to less than 6.2%, stroke risk may still be reduced by aggressive control of albuminuria using RAS blockers [26]. Both the CHD and stroke risk equations (which included urinary ACR as a predictor) derived from the HKDR performed better than the respective United Kingdom Prospective Diabetes Study (UKPDS) risk engines (which did not include urinary ACR) in our local population [2728]. This series of analyses highlighted the critical importance of screening for and reducing albuminuria in Asian patients with diabetes. Apart from albuminuria, intrarenal arterial resistance (resistance index greater than 0.8) as measured by Doppler was associated with deterioration in renal function as well as other microvascular complications [29]. These findings suggested that abnormal microvascular flow in the kidney might reflect generalized vasculopathy in diabetes. Taken together, all these findings supported the utility of using albuminuria as a prognosticator for cardiovascular-renal complications.

In Chinese patients with T2DM, components of metabolic syndrome (central obesity, hypertriglyceridaemia, and hypertension) predicted new onset of chronic kidney disease (CKD) (eGFR less than 60 mL/min/1.73 m²) independent of disease duration, glycaemic control, and albuminuria [30]. As a surrogate marker of visceral adiposity, waist circumference but not BMI was associated with increased intrarenal atherosclerosis and glomerulosclerosis, suggesting adverse effects of adipokines on renal haemodynamics and glomerular inflammation. In T2DM, the frequent coexistence of cardiometabolic risk factors and insulin resistance caused earlier onset of nephropathy than type 1 diabetes, making multifactorial management of T2DM of paramount importance in preventing ESRD [31].

In the HKDR, use of statins was associated with over 60% risk reduction in the development of CKD and 40% risk reduction in new onset CHD. These benefits were independent of LDL-C levels, use of RAS inhibitors, glucose lowering drugs or insulin, and other baseline risk factors [532]. These findings echoed recent recommendations from international guidelines, which advocated the use of statins based on a patient's absolute cardiovascular risk rather than a specific LDL-C level. In a regional expert statement for lipid management in Asians, due to its safer pharmacokinetics and more potent effects on reducing proteinuria, atorvastatin was preferred over rosuvastatin [33]. Should rosuvastatin be chosen, it should be prescribed at a lower starting dose (5 mg daily) in Asians, as the plasma concentrations in Japanese and Chinese were approximately double that in Caucasians.

In the landmark trials Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan (RENAAL) and Irbesartan Diabetic Nephropathy Trial (IDNT), angiotensin receptor blockers (ARBs) delayed progression to ESRD in T2DM patients independent of BP reduction [3435]. In a subgroup analysis from the RENAAL study, the renoprotective effect of ARB was more pronounced in Asians compared to Caucasians [36]. Although Chinese patients were more prone to develop cough from ACEI use (up to 50% chance) [37], the use of ARB as an alternative RAS blocker was far from optimal, ranking second to calcium channel blockers in Hong Kong (Table 3). That said, apart from usage of RAS blockers, optimal control of BP was of paramount importance for organ protection. In a post hoc analysis of the Olmesartan Reducing Incidence of Endstage Renal Disease in Diabetic Nephropathy Trial (ORIENT-BP), reducing systolic BP to less than 130 mm Hg improved renal outcomes in Asians with T2DM, especially in those with heavy proteinuria (greater than 1 g/g Cr). However, similar to blood glucose control, BP targets should be individualized since patients with atherosclerotic diseases might not benefit from tight control of BP [38].

Compared with Caucasians in whom the rising prevalence of diabetes was driven by aging, we are now seeing a rapid increase in the incidence of young onset diabetes (YOD; defined as diagnosis of diabetes before 40 years old) in Asian countries. In a recent national survey in China, the prevalence of YOD and pre-diabetes was approximately 45%, with the majority being undiagnosed [39]. In the HKDR, one in five adult patients had YOD [40]. In a 7-year prospective analysis of 9,509 patients with T2DM, those with YOD had 30% to 50% higher age-adjusted incidence of cardiovascular and renal events compared with patients with late onset diabetes (LOD) for the same attained age. The increased risk was mainly driven by the prolonged disease duration of the YOD group. Despite being younger by 20 years on average, those with YOD had poorer metabolic profiles than those with LOD. Another prospective analysis of patients with YOD from the HKDR showed that overweight patients with T2DM had 15- and 5-fold greater hazard of progression to cardiovascular disease (CVD) and ESRD respectively compared with patients with type 1 diabetes, even when adjusted for age, sex, time from diagnosis, and HbA1c [41]. Adjustment of other metabolic risk factors (BMI, BP, lipids) rendered the difference non-significant. Moreover, there was no difference in hazards for either CVD or ESRD between patients with type 1 diabetes and those with normal weight T2DM. These results suggested that apart from glycaemic control, early implementation of a multifaceted approach to control blood glucose and optimize other aspects of the metabolic syndrome was of particular importance in YOD. This was in line with suggestions from the landmark UKPDS [42] and Steno-2 studies [43], both demonstrating reduction in long-term risk of complications with early control of cardiometabolic risk factors.

In Hong Kong Chinese, a family history of YOD was associated with 6- to 7-fold odds of developing incident diabetes during a follow-up period of 12 years [44]. Although these participants had poorer metabolic profiles than the control subjects, the risk association of T2DM with family history of YOD remained robust after controlling for baseline glycaemic indices and metabolic risk factors. From genome-wide association studies, the majority of identified genetic loci for T2DM were implicated in β-cell biology compatible with the importance of reduced insulin secretory capacity as a main culprit in T2DM [45]. In support of this, those with family history of YOD had lower β-cell function (disposition index) and insulin sensitivity than those with family history of LOD or those without a family history. This suggested that in the younger population, heritability or shared environment reflected by positive family history, took on greater significance in the prediction of progression to diabetes, over and above the presence of metabolic syndrome. These findings called for earlier detection of diabetes in young individuals who had first degree relatives with YOD [44].

Recent epidemiological analyses from the HKDR revealed high rates of complications and low rates of achievement of treatment goals especially in patients with YOD. In the HKDR, 5% of YOD already had CVD or CKD at enrollment [40]. As many as 37% had micro- or macroalbuminuria, and yet only 15% of them were treated with RAS blockers. More than half had hypertension and three-quarters had dyslipidaemia, yet they were often less intensively treated with BP or lipid lowering therapies than their older counterparts, especially during the early stages of their disease. An analysis of the JADE Register which enrolled more than 40,000 patients from nine Asian countries revealed similar treatment gaps amongst the YOD group, despite diversity in health-care policies, clinical practices, and insurance coverages [46]. On the one hand, there is the lack of management guidelines for YOD and clinical inertia by the treating physician, and on the other hand, there is the poor treatment adherence. In these young patients, competing priorities in other aspects of life and difficult behavioral characteristics contributed to their poor control of cardiometabolic risk factors. Although the American Diabetes Association (ADA) guideline suggested statins for patients with diabetes aged less than 40 years with any cardiovascular risk factors [47], in real world practice, statins were seldom prescribed to these young patients. Further research and interventional studies are required to target patients with YOD for aggressive metabolic control and empowerment in order to reduce the burden of chronic diabetic complications.

Asians are more likely to have a higher percentage of body fat at a lower BMI and waist circumference compared with Europeans. Thus, for similar BMI cut-off points, they were more likely to harbor cardiovascular risk factors and obesity-related comorbidities [59]. In the United Kingdom Biobank Study, the prevalence of diabetes at a BMI of 22 kg/m² among South Asian participants was comparable to that at a BMI of 30 kg/m² amongst participants of white European descent [60]. In a cross-sectional study on Asian Americans, lowering the cut-off value of BMI from 25 to 23 kg/m² increased the sensitivity of detecting pre-diabetes and diabetes from 50.2% to 74.1% [59]. Accordingly, the latest consensus suggested that the BMI thresholds for bariatric surgery should be 2.5 kg/m² lower for Asians across all risk groups [61]. Surgical outcomes in Asians showed favorable results similar to that seen in the Western population, and these favorable outcomes are likely due to the rapid reduction of insulin resistance and recovery of early insulin secretion postoperatively. In a prospective cohort of patients who underwent laparoscopic sleeve gastrectomy at the CUHK, half of them achieved remission of T2DM starting from the second year after surgery, and this figure was maintained at 5 years. Amongst insulin-treated patients, two-thirds were able to discontinue insulin therapy upon follow-up [62]. In addition, 64% of patients had resolution of metabolic syndrome with sustained weight loss (defined as 50% or more reduction of excessive weight) at long term follow-up. In long-term observational studies, bariatric surgery reduced risk of microvascular complications compared with usual care, regardless of baseline glycaemic status [63]. Using the ABCD Diabetes Surgery Score system, young age, high BMI, high C-peptide level, and short duration of diabetes predicted remission of T2DM after metabolic surgery [61]. Taken together, patients with YOD who often had higher metabolic burden might benefit more from metabolic surgery than patients with LOD. However, bariatric surgery remained an invasive treatment modality which might not be well accepted by patients. In Asia, there remained uncertainty over long-term adverse health consequences, especially if supporting service for follow-up were suboptimal. Therefore, medical options for management of diabesity remained the cornerstone option for most patients.

The ADA recommended combination therapy with insulin and/or multiple oral glucose lowering drugs if the target HbA1c of less than 7% was not achieved after 3 months of lifestyle modifications plus metformin alone [47]. In Asians with predominant insulin insufficiency, as suggested by lean body mass and especially in those with YOD, early insulinization might preserve β-cell dysfunction by alleviating glucotoxicity [64]. Thus, in contrast to guidelines from the West, the Standards of Care in China 2016 recommended short term intensive insulin therapy in patients with HbA1c higher than 9% or symptomatic hyperglycaemia at presentation (fasting plasma glucose ≥11 mmol/L) [65]. In the International Diabetes Management Practice Study (IDMPS), short disease duration, use of few oral glucose lowering drugs, and absence of microvascular complications were predictors for achieving glycaemic control [66]. These findings supported avoidance of delayed use of insulin in patients treated with multiple oral glucose lowering drugs with inadequate glycaemic control. However, in real world practice, these algorithms were seldom adhered to and insulin was often delayed despite failure on dual or triple oral glucose lowering drugs. Even for those who were initiated on insulin, without adequate patient education and empowerment, titration to goal was often delayed [6768]. In the IDMPS, self-adjustment of insulin dosages was identified as a predictor of glycaemic control in Asians [66].

A pooled analysis of 16 randomized controlled trials had shown that Asians required higher basal insulin doses to achieve adequate glycaemic control compared with Caucasians (0.18 units/kg vs. 0.16 units/kg, P<0.001), despite the fact that they were often younger with lower BMI, and had similar disease duration. In this cohort, the C-peptide level was 20% lower in Asian than in non-Asian patients. Even when similar fasting plasma glucose targets were achieved, Asian patients were less likely to achieve HbA1c target less than 7% compared with Caucasians, suggesting a need to control postprandial hyperglycaemia [69]. In another observational survey of Chinese patients with T2DM, insulin-treated patients with low fasting C-peptide levels had better cardiovascular outcomes than those with high fasting C-peptide levels and treated with insulin, supporting the potential role of using C-peptide levels to guide therapy [70].

Large scale epidemiological studies indicated that postprandial blood glucose was a better independent predictor of cardiovascular risk than fasting blood glucose or HbA1c [71]. In a cross-sectional study in Asian Americans, screening for diabetes with HbA1c and fasting plasma glucose identified only half of those with diabetes compared to the use of 75 g oral glucose tolerance test which only missed 15% of people with diabetes [59]. Asians had higher propensity for postprandial glucose excursions due to their preference for high-carbohydrate diets, reduced early-phase insulin secretion and blunted incretin effect. Compared with the use of short-acting insulin, short-acting glucagon-like peptide-1 receptor agonists (GLP-1RAs) offered additional advantages of weight reduction and reduced risk of hypoglycaemia. In Japan, patients had lower endogenous GLP-1 levels and meal-induced GLP-1 secretion than Caucasians, rendering them more responsive to lixisenatide, a GLP-1RA in postprandial glucose reduction [72]. Asians with lower BMI often required a lower dose of GLP-1RAs compared to their Caucasian counterparts due to higher rates of gastrointestinal side effects and greater drug exposure. The combination of insulin and GLP-1RA theoretically worked best for those with poor β-cell reserve, impaired incretin effect and obesity.

For those with needle phobia or difficulties in handling needle injections, oral dipeptidyl peptidase-4 (DPP-4) inhibitors is an alternative to GLP1-RAs as an incretin-mimetic drug in augmenting postprandial insulin secretion and suppressing glucagon secretion, especially when used in the early stage of the disease. In a meta-analysis, Asians exhibited better response to DPP-4 inhibitors in HbA1c reduction than Caucasian patients, along with improvement in β-cell function [73]. In patients with newly diagnosed T2DM and HbA1c greater than 9%, the ADA guideline suggested initial combination therapy to achieve a target HbA1c of less than 7% earlier [47]. Here, metformin reduced fasting plasma glucose by inhibiting hepatic gluconeogenesis while DPP-4 inhibitor reduced postprandial glucose. In a post hoc subgroup analysis of pooled data from randomized clinical trials, more than half of the Asian patients with initial HbA1c 8.5% to 12% started on metformin and linagliptin achieved glycaemic control within 24 weeks, without weight gain or hypoglycaemia [74].

A number of randomized controlled studies have demonstrated similar blood glucose lowering efficacy of sodium-glucose cotransporter-2 (SGLT-2) inhibitors in Asians compared with Caucasians [75]. By decreasing glucotoxicity through insulin-independent mechanisms, this class of drugs improved both β-cell function and peripheral insulin sensitivity. Moreover, the favorable effects on BP, body weight, visceral fat, and waist circumference made SGLT-2 inhibitors an attractive treatment choice for patients with T2DM and metabolic syndrome. In the EMPA-REG (Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients–Removing Excess Glucose) trial, empagliflozin reduced the rate of decline of renal function as well as cardiovascular-renal events in high risk patients with T2DM, when added to usual care [76]. Since Asians were more prone to developing renal complications, further real-world data and longer-term clinical trials in this population would be useful to support the preferred use of this class of drugs in Asian patients with T2DM. Here, a combination of SGLT-2 inhibitors and GLP-1RAs with different mechanisms of action and fewer drug interactions showed potential promise given their efficacy in reducing body weight and cardiometabolic risk factors.