Vascular complications in prediabetes and type 2 diabetes: a continuous process arising from a common pathology

Abstract

The term, “prediabetes”, describes a state of hyperglycaemia that is intermediate between true normoglycaemia and the diagnostic cut-offs for indices of glycaemia that are used to diagnose type 2 diabetes. The presence of prediabetes markedly increases the risk of developing type 2 diabetes. Numerous randomized, controlled evaluations of various agents have demonstrated significant prevention or delay of the onset of type 2 diabetes in subjects with prediabetes. Intensive lifestyle interventions and metformin have been studied most widely, with the lifestyle intervention being more effective in the majority of subjects. The application of therapeutic interventions at the time of prediabetes to preserve long-term outcomes has been controversial, however, due to a lack of evidence relating to the pathogenic effects of prediabetes and the effectiveness of interventions to produce a long-term clinical benefit. Recent studies have confirmed that prediabetes, however defined, is associated with a significantly increased risk of macrovascular and microvascular complications essentially identical to those of diabetes, and also with subclinical derangements of the function of microvasculature and neurons that likely signify increased risk of compilations in future. Normoglycaemia, prediabetes and type 2 diabetes appear to be part of a continuum of increased risk of adverse outcomes. Long-term (25–30 years) post-trial follow up of two major diabetes prevention trials have shown that short-term interventions to prevent diabetes lead to long-term reductions in the risk of complications. These findings support the concept of therapeutic intervention to preserve long-term health in people with prediabetes before type 2 diabetes becomes established.

Keywords:

• Prediabetes
• hyperglycaemia
• type 2 diabetes
• diabetic complications
• antidiabetic therapy

Introduction

Hyperglycaemia is associated with many acute and chronic complications. Chronic hyperglycaemia leads to micro- and macrovascular complications, and, inevitably, a reduced health- and lifespan in people living with diabetes^1,2. In an attempt to mitigate the increased mortality and morbidity in type 2 diabetes (T2D), current guidelines are aimed at the optimal management of T2D and include complex algorithms as the clinical course of the disease progresses. Considering the high burden of cardiovascular disease (CVD) and the risk of adverse outcome in people with T2D who develop CVD, the T2D guidelines prioritize cardiovascular outcomes^3,4. Many clinicians, however, are advocating for earlier intervention to mitigate CV risk more effectively, even before the onset of T2D. Both T2D and prediabetes are both characterized by abnormal glucose homeostasis, albeit individuals with prediabetes have intermediate indices of hyperglycaemia. Prediabetes therefore positions between normal glucose tolerance and overt diabetes⁵.

Timely intervention in individuals with prediabetes, using intensive lifestyle intervention and/or with pharmacologic antidiabetic agents, have the potential to delay or prevent the progression from prediabetes to T2D and its complications^5,6. The benefit of early intervention in the prediabetic population has however not received the same attention as secondary and tertiary prevention of CVD in T2D. Here, we review the current best evidence on the impact of prediabetes on long-term clinical outcomes and consider the potential benefits and harms that may arise from therapeutic intervention in this high-risk population.

What is prediabetes?

Diagnosis of prediabetes and type 2 diabetes

Prediabetes refers to a state of intermediate hyperglycaemia with glucose levels higher than “normal” (fasting and/or postprandial plasma glucose (PPG) values) but not exceeding diagnostic thresholds for T2D (fasting plasma glucose (FPG) values ≥7 mmol/L and/or 2-hour post load glucose (2hPG) ≥11.1 mmol/L and/or HbA1c ≥6.5%)⁴. Impairments of FPG (impaired fasting glucose [IFG]) and/or 2hPG post-load glucose (impaired glucose tolerance [IGT]) may occur simultaneously. An individual with FPG ≥5.6–6.9 mmol/L has IFG, and if the 2-hour post load glucose (2hPG) is ≥7.8–11 mmol/L during a standard 75 g oral glucose tolerance test (OGTT), he/she has IGT. Note that the World Health Organization (WHO), and American Diabetes Association (ADA), use different FPG values to diagnose IFG (WHO: FPG 6.1–6.9 mmol/L and ADA: 5.6–6.9 mmol/L), and the WHO does not currently endorse the use of HbA1c for the diagnosis of prediabetes⁷. More recently, glycated haemoglobin (HbA1c) ≥5.7–6.4% has been incorporated as a convenient alternative test. Considering the ethnic disparities in HbA1c, it is however important that the ideal HbA1c cut off is further corroborated by performing an OGTT in certain populations.

The diagnoses of prediabetes or T2D are categorical in nature. Thus, a subject with an FPG of 6.9 mmol/L and HbA1c of 6.4% has prediabetes, provided the 2hPG is not diagnostic of diabetes. Given the limitations in terms of reproducibility of the OGTT, any equivocal results should be scrutinized, and testing repeated. The FPG cut-off for T2D, interestingly, was proposed by the ADA in 1997, based on the association between FPG and the incidence of certain microvascular complications, particularly retinopathy¹. The HbA1c cut off of 6.5% was proposed by the ADA in 2010, following an expert review published in the previous year⁸. It is important to note that the robustness of the diagnostic cut-offs has been questioned and the relationship of other complications to the HbA1c level may differ from that of retinopathy^9–11. The relationships between hyperglycaemia (HbA1c or blood glucose level) and the risk of macrovascular complications is continuous in T2D, with no clear inception point where risk increases in parallel with glycaemia^12,13. There is no doubt that the cut-offs used for diagnosing diabetes have been a great advance in facilitating provision of care for people at risk of the devastating adverse long-term consequences of T2D. Nevertheless, these cut-offs do not define the presence vs. absence of risk of complications nor adverse outcome, and there appears to be a continuum of (especially microvascular) risk as the prevailing level of glycaemia increases¹⁰.

Prediabetes and clinical outcomes

Incident type 2 diabetes

The impact of prediabetes on the subsequent risk of developing T2D has been well described and will be addressed here only briefly, as the main focus of our review is on the relationships between prediabetes and clinical vascular outcomes. The landmark early diabetes prevention trials, the Diabetes Prevention Program (DPP; USA)¹⁴ and the Diabetes prevention Study (DPS; Finland)¹⁵ were conducted in populations with IGT, following observations in the “Diabetes epidemiology: collaborative analysis of diagnostic criteria in Europe” (DECODE) study and elsewhere that post-load hyperglycaemia was a much stronger predictor of adverse long-term clinical outcomes than FPG¹⁶. The rates of incident diabetes in the control groups of these trials were high (110/1,000 person-years in the DPP¹⁴ and 78 cases/1000 person years (DPS)¹⁵. An early meta-analysis of observational studies published in 1997 that contributed to the design of the DPP indicated lower incidence rates of T2D with IGT (57 cases/1,000-patient-years) compared with the control groups in the DPP and FDPS¹⁷. This highlights the fact that the randomised intervention trials described above did not in fact include a truly normoglycaemic control group.

These studies included subjects with FPG 5.3–6.9 mmol/L (DPP) or <7.8 mmol/L (DPS), with mean baseline FPG in control groups of 5.9 mmol/L and 6.1 mmol/L, respectively. Thus, most participants in these trials had plasma glucose levels that were broadly consistent with co-occurring IFG and IGT, and it would be interesting to see an analysis where the populations were stratified more rigorously for commonly-used definitions of prediabetes/intermediate hyperglycaemia.

A meta-analysis of prospective cohort studies demonstrated a 5–8-fold increase in the relative risk of developing T2D in people with IGT or IFG vs. NGT. When IGT and IFG co-occurred, a 12-fold increase was noted vs. NGT¹⁸. A Cochrane systematic review and meta-analysis presented pooled hazard ratios (HR, vs. NGT) for developing T2D using ADA IFG criteria (FPG ≥5.6 mmol/L) with reported HRs of 4.3 (95% CI 2.6, 7.1)) for IFG only, 3.6 (2.31, 5.64) for IGT only and 6.9 (4.2, 11.5) for combined IFG and IGT¹⁹. This analysis also confirmed the increased risk to develop T2D with HbA1c in the ADA prediabetes range (HR of 5.6 (2.8, 11.1) vs. NGT). A subsequent systematic review reported a steep increase in the incidence of T2D in people with HbA1c 5.5% to 6.5% (which corresponds roughly to the current ADA criteria for diagnosing type 2 diabetes using HbA1c [Table 1]), compared with lower HbA1c values²⁰. A recent cohort study that included 3,492 subjects with ADA-defined IFG demonstrated a lifetime risk of developing diabetes of 46% (men) and 58% (women), at age 45 years²¹.

Table 1. Summary of diagnostic criteria for different severities of dysglycaemia according to the European Association for the Study of Diabetes/European Society of Cardiology and the American Diabetes Association^3,4.

	Fasting plasma glucose [mmol/L (mg/dL)]	2-h OGTT plasma glucose [mmol/L (mg/dL)]	HbA1c [% (mmol/mol)]
Normal glucose tolerance (NGT)	<5.6 (<100)	<7.8 (<140)	<5.6 (39–47)
Impaired fasting glucose (IFG)	5.6–6.9 (100–125)	<11.1 (<200)	–
Isolated IFG	5.6–6.9 (100–125)	<7.8 (<140)	–
Impaired glucose tolerance (IGT)	<7.0 (<126)	7.8–11.0 (140–199)	–
Isolated IGT	<5.6 (<100)	7.8–11.0 (140–199)	–
Combined IFG/IGT	5.6–6.9 (100–125)	7.8–11.0 (140–199)	–
Type 2 diabetes	≥7.0 (≥126)	≥11.1 (≥200)	≥6.5 (≥48)

The lower cut-off for diagnosing IFG is 6.1 mmol/L (110 mg/dL) according to the World Health Organization (WHO)⁷. FPG, fasting plasma glucose; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; NGT, normal glucose tolerance; OGTT, oral glucose tolerance test.

Macrovascular complications

Prediabetes is associated with a significantly increased risk of CV events and mortality^1,2. Table 2 summarizes the main findings of studies of the effects of prediabetes or glycaemia in the prediabetic range on macrovascular outcomes. Mendelian randomization studies report a strong association between hyperglycaemia in the prediabetes range and genetic variants that predict an increased risk of CVD linearly, over and above the severity of prediabetes²². The AusDiab study reported that IFG and a prior diagnosis of T2D were significant independent predictors of all-cause and CV death after multivariable adjustment for other risk factors (IGT significantly predicted all-cause death, but not CV death)²³. Similar findings were presented from a cohort in Taiwan with IFG (WHO criteria)²⁴, from the Baltimore Health in Aging study, where IFG (WHO) and IGT predicted adverse CV outcomes (especially when IFG and IGT cooccurred)²⁵, and from the National Health and Nutrition Examination Survey (NHANES) cohort²⁶. A further large cohort study found no significant association between prediabetes and adverse cardiovascular outcomes; a low incidence of T2D and short follow-up after the development of T2D may partly explain this finding²⁷. Finally, the presence of prediabetes and diabetes increased the risk of a composite cardiovascular endpoint in a population of patients with CKD (renal outcomes from this study are summarized in a separate section, below)²⁸.

Table 2. Overview of key studies on the relationship between prediabetes or glycaemia in the prediabetic range and macrovascular complications.

Ref	Design	N	Follow up	Summary of key findings
22	Mendelian randomisation study of 40 variants associated with both diabetes and HbA1c	324,830^a	Not given	Each additional 1 mmol/mol of HbA1c predicted an increase in CHD risk of 11% (5–18)
23	Population-based, prospective, observational cohort study	10,428^b	5.2 y	Increased risk of death associated with known T2D (HR 2.3 [1.6–3.2]), IFG (HR 1.6 [1.0–2.4]) and IGT (HR 1.5 [1.1–2.0).
24	Prospective, observational cohort study of government employees and schoolteachers	36,386	11 y	1.5-fold increase in the risk of cardiovascular disease for FPG 110–126 mg/dL (6.1–7 mmol/L) vs. 90–99 mg/dL (5–5.5 mmol/L); no significant increase for 100–109 mg/dL (5.6–6.1 mmol/L).
25	Population-based, prospective, observational cohort study^c	1,236	13.4 y	RR for mortality 1.41 (1.01 to 1.97) for FPG 6.1–6.9 mmol/L vs. 4.1–5.2 mmol/L; corresponding RR for diabetes (glucose 7.0–7.9 mmol/L) was 2.02 (1.09–3.73).
26	Population-based, prospective, observational, nationally representative (USA) cohort study^d	3,174	42,130 person-years	Increased RR of all-cause death (1.42 [1.08–1.87]) for IGT vs. normoglycaemia; corresponding RR for known diabetes was 2.11 (1.56–2.84).
27	Population-based, prospective, observational cohort study^e	4,602	13 y	No significant association for prediabetes vs. no prediabetes for incident HF (HR 0.98 [0.85–1.14), MI (HR 1.02 [0.81–1.28]), angina (HR 0.93 [0.77–1.12), stroke (HR 0.86 [0.70–1.06]) or all-cause mortality (HR 0.99[0.88–1.11).
28	Population-based, prospective, observational cohort study^f	3,701	7.5 y	ADA-defined prediabetes was associated with increased risk of composite cardiovascular outcome vs. NGT (HR 1.38 [1.05–1.82]); HR for diabetes vs. NGT was 1.63[1.27–2.11])

^aSubjects without known diabetes from the UK Biobank; ^bAusDiab Study; ^cBaltimore Longitudinal Study of Aging; ^dSecond National Health and Nutrition Examination Survey Mortality Study; ^eCardiovascular Health Study; ^fChronic Renal Insufficiency Cohort; ^gHF, MI, stroke or all-cause mortality. Figures in parentheses are 95% confidence intervals. HF, heart failure; HR, hazard ratio; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; MI, myocardial infarction; RR, relative risk; T2D, type 2 diabetes.

Prediabetes, and T2D are increasingly noted to be heterogenous states of abnormal glucose homeostasis, and it therefore does not come as a surprise that the various measures of hyperglycaemia in prediabetes leads to a range of clinical outcomes, not disregarding the other CV risk factors that is often also present. For example, in one study, prediabetes diagnosed using HbA1c (5.7–6.4% [ADA criteria] or 6.0–6.4% [International Expert Committee criteria]) predicted higher rates of adverse clinical outcomes (CVD or death) vs. prediabetes diagnosed using measures of blood glucose²⁹. However, clustering of additional CV risk factors appeared to account for most of the difference. A report from the United Kingdom (UK) Biobank supported the importance of comorbid classical CV risk factors in driving adverse CV outcomes in the setting of prediabetes³⁰. Interestingly, higher vs. lower blood glucose 30 min into an OGTT predicted mortality more accurately than the FPG or 2-hPG in another study, highlighting the shortcomings of the OGTT timepoints with regards to predicting adverse outcomes³¹.

Increased carotid intima-media thickness (cIMT) is a marker of increased overall burden of atherosclerosis and is a strong independent predictor of future major adverse CV outcomes³². Average measures of cIMT were higher in people with prediabetes vs. NGT and were higher still in people with diabetes (60 subjects in each group)³³. Data from the US Dallas heart Study also demonstrated increased cIMT and higher coronary calcium scores (a marker of coronary atherosclerosis) in people with vs. without prediabetes³⁴. Accordingly, cIMT may be useful for further CV risk stratification in people with prediabetes.

Data on prediabetes and CV risk are conflicting. A systematic review and meta-analysis by Mutie et al. (average follow-up 10 years) concluded that prediabetes is causally related to the onset of coronary artery disease, but not to other diabetes complications³⁵. Other systematic reviews report that both IFG and IGT are associated with increased risk of CV and all-cause death, CV events and adverse cardiac events^36,37. Figure 1 summarizes the results of the more recent of these analyses (2021)³⁶.

Figure 1. An example of associations between prediabetic states and cardiovascular disease events, coronary heart disease events, or mortality from a recent systematic review. ADA, American Diabetes Association; IFG, impaired fasting glucose; IFG, impaired glucose tolerance; WHO, World Health organization. See Table 1 for details of diagnostic criteria. Reference values were glycaemic categories below the lower cut-off value for each prediabetes condition. Drawn from data presented in reference 36.

Figure 2. Association between prediabetes and heart failure from a systematic review of 15 studies. ADA, American Diabetes Association; IFG, impaired fasting glucose; IFG, impaired glucose tolerance; WHO, World Health organization. See Table 1 for details of diagnostic criteria. The reference value was normoglycaemia in each case. Drawn from data presented in reference 40.

Heart failure (HF) is recognized increasingly as a common complication of T2D with a high burden of morbidity and premature mortality³⁸. Individual studies of the association of prediabetes with HF have also produced conflicting results^27,36,39. A recent (2021) meta-analysis of 15 studies that included 9,827,430 individuals found significant associations between prediabetes vs. normoglycaemia and incident HF (Figure 2)⁴⁰. Several reports have associated prediabetes (sometimes previously undiagnosed) with worsened outcomes in patients with pre-existing HF^41–43.

Coronary flow reserve (CFR), a measure of cardiac microvascular function, was lower in people with prediabetes vs. NGT in a cross-sectional study, with a larger defect in people with T2D⁴⁴. It has been reported however, that changes in CFR only become clinically significant when T2D is established⁴⁵. Impaired endothelial function may be an early event in the pathogenesis of microvascular dysfunction in people with prediabetes or T2D⁴⁶. In addition, cardiac troponin-T, a sensitive marker of myocardial damage, has been shown to be elevated in people with prediabetes as well as T2D⁴⁷. Together, these reports suggest that the presence of subclinical cardiovascular impairment in people with prediabetes may position these subjects at increased risk of established diabetes-like complications in future, and that overall glycaemic exposure over time contributes.

Microvascular complications

The eye

Participants in the DPP who had elevated blood glucose, but no history of diabetes, demonstrated a prevalence of retinopathy at baseline of 7.9%⁴⁸. Further data from the same study found that the prevalence of retinopathy was higher among people who developed T2D during the randomised phase of the trial (12.6%), compared with those who did not (7.9%, p = .03 for the difference)⁴⁸. Retinal arteriolar dilatation in response to metabolic stress (induced by flickering light), was used to measure retinal vascular function in the community-based Maastricht Study cohort⁴⁹. Mean arteriolar dilatation was 3.4% in 1,269 people with NGT, 3.0% in 335 people with prediabetes, and 2.3% in 609 people with diagnosed T2D⁴⁹. A more recent study from this cohort found no significant correlation between prediabetes and arteriolar risk, but demonstrated a linear trend between the blood glucose level and retinal microvascular width across the range of prediabetic and diabetic dysglycaemia⁵⁰. The results of a study using similar methodology in a different cohort supported the possibility of an association between prediabetes and reduced retinal arteriolar diameter⁵¹. Macular thinning has also been observed as subclinical manifestations of retinal damage in the Maastricht cohort, again with a severity intermediate between NGT and T2D⁵².

Other studies have demonstrated reduced contrast-sensitivity function and photo stress-recovery time in people with prediabetes vs. NGT⁵³ or ultrastructural derangements of retinal micro vessels in the absence of loss of retinal structure (normal cone cell density)⁵⁴. These studies represent further demonstrations of a subclinical deterioration in retinal vascular function in the setting of prediabetes. Subjects in a primary care database in the UK with newly-diagnosed T2D had a higher incidence of retinopathy if they had vs. had not received a diagnosis of prediabetes beforehand (adjusted odds ratio 1.76 [1.69–1.85]), consistent with increased risk of complications from a higher prior exposure to hyperglycaemia⁵⁵. Impaired retinal function has not been observed in the setting of prediabetes in all studies, however⁵⁶.

Nerves and brain

In the NHANES cohort (1999–2004), the relative risk of peripheral neuropathy was 1.1 (0.9, 1.3) for prediabetes and 1.7 (1.5, 2.0) for diabetes, each vs. NGT, although the 95%CI for the RR associated with prediabetes crossed zero in this case⁵⁷. A study from the Cooperative Health Research in the Region of Augsburg (KORA) cohort in Germany found a similar prevalence of distal sensorimotor polyneuropathy (DSPN) in people with combined IGT + IFG (24%) and T2D (22%); associations between isolated IFG or IGT and DSPN did not reach statistical significance, however⁵⁸. A systematic review suggested an increased incidence of neuropathy associated with prediabetes, but wide variation between studies precluded a meta-analytic approach⁵⁹. In contrast, other studies have associated type 2 diabetes – but not prediabetes – with forms of neuropathy, but these are highly dependent on the sensitivity of the diagnostic tools used to diagnose DSPN^60,61.

The Maastricht cohort included a measure of peripheral neural function (heat-induced hyperaemia in skin); with results that paralleled those in the retina, indicating a significant defect in skin microvascular function in prediabetes, and a larger defect in T2D, vs. NGT⁴⁹. The authors concluded that prediabetes was associated with “generalized microvascular dysfunction”.

Reduced length and density of corneal nerves provided another manifestation of early neuropathy in prediabetes^62,63. Cardiac autonomic neuropathy also appears to be already present in people with prediabetes^64,65. Sensory nerve superexcitability, a marker of subclinical impairment of nerve conduction, was demonstrated to occur to a greater extent in 40 prediabetic subjects vs. 20 normoglycaemic subjects; again the defect was more pronounced in people with T2D⁶⁶.

Data from Sweden associated prediabetes significantly with loss of brain volume (especially white matter) and progression of cognitive impairment over the span of 9 years⁶⁷. Similarly, the Maastricht study participants with prediabetes were found to have more cerebral lacunar infarcts, more white matter lesions and more loss of brain volume (driven by loss of white matter) compared with NGT subjects; these deficits were more severe in participants with T2D⁶⁸. Increasing levels of hyperglycaemia drove deficits in executive function in the NHANES 2011–2014 cohort⁶⁹.

The kidney

Data from the NHANES 1999–2006 cohort indicated the prevalence of chronic kidney disease (CKD) was intermediate in subjects with prediabetes (17%; ADA criteria), compared with diagnosed diabetes (33%) or NGT (12%), after adjustment for race/ethnicity, age and gender⁷⁰. Nine years of prospective follow-up in a population in Korea identified prediabetes (IGT or based on HbA1c) as an independent predictor of incident CKD⁷¹. Cross sectional data from the KORA registry, adjusted for fasting glucose, 2hPG, HbA1c, fasting insulin and insulin resistance, associated isolated IFG with increased risk of microalbuminuria or CKD (∼50% increase in risk), with 2.6-fold increase in the risk of reduced kidney function with IFG + IGT⁷².

The Chronic Renal Insufficiency Cohort followed a population of 3,701 subjects with chronic kidney disease (CKD) for an average of 7.5 y²⁸. The main composite renal outcome included end-stage renal disease (renal transplantation or dialysis), a 50% decline in eGFR from baseline to ≤5 mL/min/1.73 m², or doubling of urine protein:creatinine ratio to ≥0.22 g/g creatinine. Subjects with prediabetes (based on ADA criteria for HbA1c or FPG) were not at greater risk of the composite renal outcome vs. NGT (adjusted HR 1.13 [95%CI, 0.96–1.32]), although diabetes increased the risk of this outcome significantly (adjusted HR 1.47 [95% CI, 1.27–1.70]). However, there was a 23% increase in the adjusted risk of progression of proteinuria (p = .002) associated with prediabetes vs. no prediabetes, vs. a risk increase of 50% for T2D (p < .001). Thus, the presence of prediabetes did not affect a broad renal composite outcome significantly, although there was a sign of early renal injury signified by exacerbation of proteinuria.

A prospective, 2-year cohort study from Japan demonstrated that prediabetes was associated with an increased risk of proteinuria, but not with decline of eGFR⁷³. Data from the study in UK primary care patients with new type 2 diabetes (3 years of retrospective follow-up) demonstrated a significant association of a prior diagnosis of prediabetes with nephropathy (OR 1.14 [1.10–1.19])⁵⁴. Increased renal cortical stiffness has been identified in people with both prediabetes and T2D, compared with NGT, in a cross-sectional study⁷⁴. Increased glomerular hyperfiltration may be an early (and reversible) step in the pathogenesis of CKD associated with prediabetes or diabetes^75–78. Prediabetes and CKD interacted to increase night time systolic blood pressure in a community-dwelling elderly population⁷⁹.

Meta-analyses support these observations: one associated the presence vs. absence of prediabetes with a “modest” increase in the risk of CKD (RR 1.12 [1.02, 1.21])⁸⁰. Another meta-analysis significantly associated increased risk of CKD with IFG and elevated, non-diabetic HbA1c levels (both ADA criteria), but not with IGT³⁵.

Pathogenetic mechanisms

Cardiovascular risk factors and CKD are prevalent among people with diabetes, and in the USA interventions to mitigate CV risk has been applied less often in the prediabetes population than for people with T2D⁸¹. In one observational study, findings of increased markers of atherosclerosis (and CKD) in people with vs. without prediabetes were attenuated by adjustment for other CV risk factors, suggesting that these were responsible for much or all of the increased atherosclerotic risk, providing a rationale for intensive comprehensive multifactorial intervention to mitigate them³³. However, IFG and IGT were independent predictors of all-cause and/or CV death in the AusDiab cohort²³, as was IFG in the cohort from Taiwan²⁴. The early (1999) study from the DECODE group suggested that only elevated post-load (not fasting) glucose predicted adverse outcomes; however, both IFG and IGT (and especially both together) predicted averse outcomes in the more recent studies described above⁸².

Hyperglycaemia and prediabetes, in particular, share common pathogenetic mechanisms. People with prediabetes are at least twice as likely to be obese, compared with people with normal glucose regulation⁸³, and the prevalence of prediabetes increases in line with the severity of obesity⁸⁴. More than 8 individuals in every 10 with prediabetes in the USA is overweight or obese⁸⁵. Accordingly, obesity is recognized as a major risk factor for developing diabetes⁸⁶. Visceral adipocytes in the setting of abdominal obesity secrete a range of inflammatory cytokines that impair insulin sensitivity and pancreatic β-cell function⁸⁷. The accumulation of ectopic fat in key organs responsible for metabolic regulation is a well recognized consequence of obesity⁸⁸. The deposition of ectopic fat promotes hyperglycaemia and insulin resistance in liver and skeletal muscle, oxidative stress, endothelial dysfunction and accelerated atherosclerosis in the vasculature and remodelling of the heart that increases the risk of heart failure and atrial fibrillation⁸⁹. These mechanisms are likely to accelerate the progression of the dysglycaemic continuum, and to contribute to the pathogenetic mechanisms linking prediabetes and diabetes with vascular disease. Other mechanisms linked to obesity, such as increased sympathetic drive, may also promote adverse outcomes such as neuropathy independently of dysglycaemia, to some degree⁹⁰. However, it is difficult to separate such mechanisms, as sympathetic overdrive is also associated with insulin resistance⁹¹. It is important to note that many of the studies of associations between prediabetes and vascular outcomes summarized above adjusted for the effects of individual cardiometabolic risk factors. Accordingly, prediabetes per se appears to be associated with an increased risk of adverse vascular outcomes.

People with prediabetes are at higher long-term exposure to elevated blood glucose than people with NGT, which likely accounts for much of the observed risk of microvascular complications. Consistent with this view, the microvascular dysfunction in the Maastricht cohort (described above) was driven mainly by hyperglycaemia⁴⁵. Increased levels of methyglyoxal, a toxic, inflammatory intermediate in the generation of advanced glycation end products in the setting of hyperglycaemia, were associated with microvascular, but not macrovascular complications in this cohort⁹². The generation of extracellular vesicles containing specific microRNAs has been proposed as a novel pathogenetic mechanism for the development of vascular complications in prediabetes and diabetes⁹³. Increased mitochondrial production of superoxide free radicals has been proposed as a common early step in pathways that cause damage to cells in the setting of hyperglycaemia⁹⁴. Obesity, insulin resistance and hypertension drive cerebral microvascular dysfunction in prediabetes, providing a pathogenetic link to the genesis of macrovascular complications⁹⁵.

Other medical issues may impact further on long-term outcomes in people with prediabetes in some areas. For example, in South Africa an epidemiological transition is underway from infectious diseases (e.g. tuberculosis and human immunodeficiency virus [HIV]) to non-communicable diseases including prediabetes and T2D^96,97. While improved healthcare has reduced the impact of infections diseases across the population, these continue coexist with non-communicable diseases at a high prevalence⁹⁸. Specific interactions between infectious and non-communicable disease, e.g. the exacerbation of CV risk factors by HIV, and other potential consequences of this high combined disease burden, e.g. pressure on healthcare availability and delivery, may impact on the evolution of dysglycaemia in countries such as South Africa in the future.

Does intervention in prediabetes reduce the long-term risk of complications?

Intensive lifestyle intervention and metformin are the most widely studied interventions for the prevention or delay of T2D in subjects with prediabetes. The DPP showed that lifestyle intervention was more effective than metformin for diabetes prevention in certain subgroups, although other smaller randomized diabetes prevention trials produced variable results^6,99. National and international guidelines recommend lifestyle intervention for all who are able to undertake it, with advice on improved health behaviours, such as diet, physical activity, and smoking cessation; addressing all of these factors will improve clinical outcomes. Some guidelines provide recommendations on the use of metformin for specific subpopulations, particularly women with prior gestational diabetes, based largely on the results of the DPP (reviewed elsewhere)^6,99.

The (cluster) randomized phase of the DaQing diabetes prevention study compared an intensive lifestyle intervention with a control group who received standard care; the randomized phase was followed by a long-term epidemiological study, with no attempt to maintain the prior randomized treatments¹⁰⁰. Subjects who were previously randomized to the lifestyle intervention benefitted from reductions in the risk of cardiovascular events (HR 0.74 [0.59, 0.92], p = .006), microvascular complications (HR 0.65 [0.45, 0.95]; p = .025), cardiovascular deaths (HR 0.67 [0.48, 0.94], p = .022), and death from any cause (HR 0.74 [0.61, 0.89], p = .0015) during a total of 30 years of follow-up¹⁰⁰.

An epidemiologic follow-up study was also conducted after the randomized phase of the DPP (the DPP Outcomes study [DPPOS])¹⁰¹. Again, no attempt was made to enforce prior randomized treatment, although all who wished to continue on (unmasked) metformin could do so; placebo was discontinued, former recipients of the intensive lifestyle intervention received additional lifestyle support, and all groups received group-based lifestyle support¹⁰¹. A significant degree of diabetes prevention persisted throughout 25 years of follow-up, with the difference in effectiveness of the lifestyle intervention and metformin narrowing over time¹⁰². Prior randomisation to neither the intensive lifestyle intervention nor metformin was associated with a significant reduction in the risk of a composite microvascular outcome in the total cohort after 15 years¹⁰³ or 25 years^102,104 of follow-up in the DPPOS, although a recent report from this study has not demonstrated a significant effect on the incidence of retinopathy¹⁰⁵. However, prevention of diabetes per se was associated with a significant microvascular benefit compared with those who did develop diabetes. Regression from prediabetes to IGT was also associated with a lower risk of microvascular disease in the DPPOS, which was attributed to a lower level of long-term exposure to hyperglycaemia¹⁰⁶. Follow-up in the DPPOS continues, and future analyses containing more cardiovascular events may shed more light on the effectiveness of the initial interventions for improving long-term clinical outcomes.

Other antidiabetes drugs have been shown to prevent or delay the onset of type 2 diabetes in at-risk subjects, including acarbose, and thiazolidinediones, although tolerability issues have limited their use on practice (reviewed elsewhere¹⁰⁷). A prespecified analysis of the STOP-NIDDM trial with acarbose reported a reduction in the risk of cardiovascular events¹⁰⁸, although the number of events was small and methodological flaws with the analysis have been reported¹⁰⁹; this finding therefore requires confirmation.

For the future, members of the newer classes of SGLT2 inhibitors¹¹⁰ and GLP-1 receptor agonists¹¹¹ have been shown to reduce the risk of new-onset type 2 diabetes in subjects at risk. Long-term follow-up in these populations will be needed to establish whether these agents reduce the risk of adverse cardiovascular outcomes in people with prediabetes.

Summary and conclusions

Professor John S Yudkin of University College London, UK, wrote in 2014 that “Pre-diabetes is an artificial category with virtually zero clinical relevance”¹¹². This was a reasonable standpoint based on the data available at that time: the evidence base associating the prediabetic state with an increased risk of diabetes-like vascular complications was sparse and conflicting, with no data on the impact on outcomes of intervening therapeutically at this time, other than a reduced risk of developing clinical T2D. Moreover, successive revisions to the glycaemic cut-offs for diagnosing T2D led to a substantial increase the proportion of the population diagnosed with prediabetes. For example, the US Centers for Disease Control estimated in 2020 that 88 million adult Americans – more than one third of the US population (34.5%) have prediabetes according to current diagnostic criteria.¹¹³ This represents a potentially huge burden on the public health, which should clearly not be accepted lightly.

The (mostly recent) data summarized above demonstrate significant associations of prediabetes with macrovascular and, increasingly, microvascular complications that are essentially identical to those seen in T2D. A number of studies demonstrated clinically significant increases in the prevalence of complications in people with prediabetes that were intermediate in magnitude between the prevalence values seen in people with NGT and those in people with established clinical type 2 diabetes. Moreover, several studies not only cited evidence for subclinical disturbances of the function of microvascular vessels and nerves that likely put the subject at risk of developing established diabetes-like complications in the future. Overall, these findings support the concept of prediabetes lying within a continuum of risk that increases steadily as patients’ glycaemia climbs above the normoglycaemic range, through the prediabetic range and into levels that are diagnostic for T2D, that will become increasingly relevant as life expectancy increases.

These findings identify prediabetes as a pathologic state associated with increased risk of adverse long-term vascular and renal outcomes. We know that preventing diabetes via intensive lifestyle intervention (DaQing study¹⁰⁰), or by lifestyle intervention or metformin (DPP/DPPOS^99,102,103) has been associated with improved clinical outcomes during post-trial follow-up periods that long outlasted the duration of the original randomized interventions. We await further data from the DPPOS on the effects of the individual interventions, supported by more clinical events.

For the future, we will need better ways to predict who will and will not progress from prediabetes to clinical diabetes. The OGTT has the advantage of being able to diagnose both IFG and IGT, but it is relatively little used in many countries, and its reproducibility is poor¹¹⁴. In addition, a substantial proportion of people with prediabetes will never develop diabetes, especially older persons¹¹⁵. Standardization of methodology to diagnose prediabetes would be helpful, as this would facilitate quantitative evaluation of the risk of adverse clinical outcomes associated with different manifestations of prediabetes. It is worth noting, however, that all forms of prediabetes have been associated with increased vascular risk, as described above. This is perhaps unsurprising, as prediabetes is a state of intermediate hyperglycaemia, which would be detected by any of the criteria in use.

Early, intensive intervention in the dysglycaemic continuum has been shown to provide long-term “legacy benefits” of reduced risk of adverse cardiovascular outcomes and mortality long after the randomized intervention had ended¹¹⁶. This was seen following the DaQing study (with lifestyle intervention in people with IGT⁹¹) and the UK Prospective Diabetes Study¹¹⁷ and the Diabetes Control and Complications Trial¹¹⁸ (in people with newly-diagnosed type 2 and type 1 diabetes, respectively). By contrast, such legacy benefits were not seen in large, randomized evaluations of more vs. less intensive glycaemic control in populations with advanced, long-standing diabetes¹¹⁹. Given the ever-strengthening relationship between prediabetes and adverse vascular outcomes, therapeutic intervention at the stage of prediabetes, however diagnosed, is a rational outcome to support long-term cardiovascular, microvascular and renal health.

Supporting information

About this review

Search strategy

This is a narrative review, based on the results of structured PubMed searches: (prediabetes [ti] OR nondiabetic hyperglycemia [ti] OR intermediate hyperglycemia [ti]) AND (microvascular OR nephropathy OR retinopathy OR CKD), and (prediabetes [ti] OR nondiabetic hyperglycemia [ti] OR intermediate hyperglycemia [ti]) AND (macrovascular [ti] OR mortality [ti] OR cardiovascular [ti]). Reference lists were another source of information.

Prediabetes, non-diabetic hyperglycaemia, or intermediate hyperglycaemia?

The term, “prediabetes” is controversial: by no means all people with prediabetic hyperglycaemia will ever develop type 2 diabetes, and applying the label of “prediabetes” to this population risks stigma, anxiety, and economic consequences, e.g. related to health insurance¹²⁰. Some expert societies including the World Health Organization and the International Diabetes Federation have moved away from this term, in favour of some of the alternative terms described above; others, including the European Association for the Study of Diabetes, the American Diabetes Association and the European Society of Cardiology retain the term in their guidelines. In the absence of a consensus on terminology, we have used the term, “prediabetes” here for its conciseness and wide use in the literature.

Transparency

Declaration of funding

Merck KGaA, Darmstadt, Germany funded Fast Track review, colour printed figures, and open access publication for this article. No payments were made for authorship of this article and no other funding applied.

Declaration of financial/other relationships

UH is an employee of Merck Healthcare KGaA, Darmstadt, Germany. MG has provided paid editorial consultancy services to Merck Healthcare KGaA, Darmstadt, Germany. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgements

None.