Fasting plasma glucose outperformed 1-hour plasma glucose in predicting diabetes incidence in individuals with family history of young-onset type 2 diabetes

Abstract

Introduction

Family history (FmH) of young-onset type 2 diabetes (YOD) and 1-hour plasma glucose (PG) during the 75-g oral glucose tolerance test predicts incident diabetes, although their interactions remain unknown.

Research design and methods

In a workforce cohort established in 1998–2003, we ascertained their glycemic status in 2012–2014. We examined the interaction between FmH-YOD and 1-hour PG in predicting diabetes and used receiver operating characteristics (ROC) analysis to compare the performance of 1-hour PG in participants with or without FmH-YOD.

Results

Among 583 participants (median age (IQR)=41 (36–47) years, 43.7% men, body mass index=23.3 (21.2–26) kg/m², 40.3% (n=235) had FmH-YOD, 1-hour PG=8.1 (6.4–10.1) mmol/L), 99 (17%) had developed diabetes at a follow-up of 12.1 (11.3–13.1) years. In the FmH-YOD group, 45% in the high 1-hour PG group and 17% in the normal 1-hour PG group developed diabetes. The respective figures were 16% and 1.8% in the FmH-NONE group. Both FmH-YOD and 1-hour PG predicted diabetes with a negative interaction between FmH-YOD and 1-hour PG (OR 0.72, 95% CI 0.55 to 0.93, p=0.013). Compared with (FmH-NONE/normal 1-hour PG) group, the ORs of incident diabetes in (FmH-NONE/high 1-hour PG), (FmH-YOD/normal 1-hour PG), (FmH-YOD/high 1-hour PG) groups were 7.4 (95% CI 1.6 to 35.1, p=0.011), 18 (95% CI 3.3 to 98.1, p=0.001) and 28.2 (95% CI 5.5 to 145.9, p<0.001), respectively. In ROC analysis, the C-statistics of 1-hour PG dropped from 0.83 (95% CI 0.76 to 0.90, p<0.001) in the FmH-NONE group to 0.69 (95% CI 0.62 to 0.76, p<0.001) in the FmH-YOD group (difference=0.14 (95% CI 0.04–0.24), p=0.006) where fasting PG (FPG) was the best predictor (0.792 (95% CI 0.730–0.853), p<0.001).

Conclusions

FPG outperformed 1-hour PG in predicting incident diabetes in people with FmH-YOD, calling for precise classification and preventive strategies.

Correspondence to Professor Juliana C N Chan; [email protected]

WHAT IS ALREADY KNOWN ON THIS TOPIC

• In a recent position statement, the International Diabetes Federation concluded the superiority of 1-hour plasma glucose (PG), over fasting PG (FPG) and 2-hour PG, in predicting incident diabetes and associated complications across age, sex, and ethnicities based on 40 years of epidemiological evidence, and advocated the use of 1-hour PG ≥8.6 mmol/L to define intermediate hyperglycaemia.

• Given the strong association between family history of young-onset diabetes and incident diabetes, whether the good predictive power of 1-hour PG still holds in individuals with a family history of young-onset diabetes, in itself, and in comparison to other glycemic parameters (FPG, and 2-hour PG), is unknown.

WHAT THIS STUDY ADDS

• The good predictive power of 1-hour PG for incident diabetes was greatly attenuated in individuals with a family history of young-onset diabetes, and was outperformed by FPG.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Compared to the general population, individuals with a family history of young-onset diabetes may have different pathophysiologies, such as hepatic insulin resistance and/or glucose sensing dysfunction, with a predominant phenotype of high FPG.

• Documentation of family history of young-onset diabetes and genetic variants related to hepatic insulin resistance and/or glucose sensing dysfunction may improve the precision of risk stratification and prevention of diabetes.

• Accurate sub-phenotyping of prediabetes is important due to different underlying pathophysiologies with likely different responses to lifestyle and/or pharmacological interventions.

Introduction

There are multiple pathways underlying glucose metabolism encompassing glucose sensing and insulin synthesis, storage, and secretion at the pancreatic level, as well as insulin actions at hepatic and muscle levels, in addition to pathways implicated in gut hormones, central regulation, and postreceptor insulin signaling. Thus, the heterogeneity of phenotypes of diabetes and prediabetes is now increasingly recognized.^{1 2} There are also human physiological studies indicating that fasting plasma glucose (FPG) and postglucose loading PG have different mechanisms related to insulin secretion and its peripheral actions.³ While 1-hour PG shows robust performance in predicting diabetes after adjusting for confounders in the general population,^{4 5} the rising epidemic of young-onset type 2 diabetes (YOD) with marked phenotypic and genotypic heterogeneity, as well as the poor response of people with isolated impaired fasting glucose (i-IFG) to lifestyle intervention and metformin raises questions about whether IFG may have different implications in people at risk of YOD.^6–10

Although there are many well-characterized prospective cohorts that have informed the significance of PG during oral glucose tolerance tests (OGTTs) on the progression of hyperglycemia,⁵ few cohorts predominantly recruited young-to-middle-aged people to examine the natural progression of glycemic status, especially those with concomitant strong family history (FmH) of YOD. The Hong Kong Family Diabetes Study (HKFDS) cohort and Better Health for Better Hong Kong (BHBHK) cohort recruited people of working age who first reported the high risk of diabetes and prediabetes in family members of people with YOD.¹¹

Against this background, we performed various analyses in these young-to-middle-aged individuals to test the following hypotheses: (1) Does 1-hour PG have a different performance in predicting incident diabetes in people with FmH of YOD? (2) What are the relative performances of FPG and 1-hour PG in predicting diabetes in people with FmH of YOD?

Methods

Study participants

The participants of the analysis came from two established community-based cohorts in Hong Kong, namely, the HKFDS and the BHBHK.^11–15 The HKFDS cohort was established in 1998–2003 by the Chinese University of Hong Kong (CUHK) Diabetes Care and Research Team, where 192 index cases with diabetes (149 had YOD) and their family members were recruited, giving a total of 1076 participants for studying genetic and environmental causes of diabetes in the Chinese population.^{11 12} The index cases were identified in the diabetes complication assessment program at the Prince of Wales Hospital, followed by the invitation of their relatives. In 2001–2003, the BHBHK cohort was established as part of a community-based health promotion campaign to screen for cardiovascular risk factors, including obesity and diabetes in the workforce.^13–15 The HKFDS cohort and a random BHBHK subcohort (n=863) underwent structured assessment, including personal and FmH, anthropometric measurements, and collection of blood and urine samples for metabolic profiling. They underwent a 2-hour 75-g oral glucose tolerance test (OGTT) with five timepoint measurements of PG accompanied by a DNA/serum biobank. In 2012–2014, the diabetes status of both cohorts was ascertained using medical records, a 75-g OGTT, and glycated hemoglobin (HbA1c).¹¹ The study of both cohorts has been approved by the Joint CUHK-New Territories East Cluster Clinical Research Ethics Committee (Ref No. 2019.080-T, 2012.171), and written consent has been obtained from all participants.

Inclusion and exclusion criteria

In this analysis, we included any individuals in the BHBHK cohort and siblings of the probands in HKFDS without diabetes at baseline. Individuals with (1) missing baseline data of 75-g OGTT (fasting, 1-hour PG, and 2-hour PG) or no information on FmH of diabetes (father, mother, and/or siblings) with age of diagnosis, (2) missing data of ascertained diabetes status at 2012–2014 follow-up, and (3) known FmH of type 1 diabetes in first-degree relatives were excluded.

Definitions

At baseline, diabetes status was defined by the known history of diabetes, use of glucose-lowering medications, and 75-g OGTT (FPG ≥7 mmol/L or 2-hour PG ≥11.1 mmol/L according to the latest criteria of the American Diabetes Association (ADA)).¹⁶ Other definitions included YOD with an age of diagnosis before 40 years; IFG as an FPG of 5.6–6.9 mmol/L and impaired glucose tolerance (IGT) as a 2-hour PG of 7.8–11.0 mmol/L¹. We adopted the International Diabetes Federation (IDF) criteria and defined high 1-hour PG as ≥8.6 mmol/L during the 75-g OGTT.⁵ A positive FmH-YOD was defined by a known history of diabetes in the father, mother, or any siblings of the individual at baseline and the earliest age of diagnosis before 40 years. A negative FmH of diabetes (FmH-NONE) was defined by no known history of diabetes in the father, mother, and all siblings of the individual at baseline. At the 2012–2014 follow-up, diabetes was defined by clinical diagnosis of diabetes from electronic medical records, use of glucose-lowering medications, FPG ≥7 mmol/L, 2-hour PG ≥11.1 mmol/L, and/or HbA1c ≥6.5%.¹⁶

Statistical analysis

All data are expressed as median (IQR) if continuous and count (percentage) if categorical. Mann-Whitney U test, χ² test, and Fisher’s exact test were used for between-group comparison of clinical characteristics at baseline. Generalized estimating equations binary logistic regression with an exchangeable matrix to account for intrafamily correlations was used to examine the association of FmH-YOD and 1-hour PG at baseline with incident diabetes at follow-up expressed as OR with 95% CI. Receiver operating characteristics (ROC) analysis with area under the curve (AUC) was used to compare the performance of different glycemic parameters at baseline in predicting incident diabetes at follow-up, stratified by FmH status.

Results

From the original cohort of 477 siblings in the HKFDS and 863 presumably unrelated individuals in the BHBHK, 583 individuals were included in the present analysis, of whom 235 individuals from 104 families had a FmH-YOD and 348 unrelated individuals reported no FmH of diabetes (FmH-NONE) (figure 1 and online supplemental figure 1). In this prospective cohort, 43.7% were men. The median age was 41 (36–47) years with a body mass index (BMI) of 23.3 (21.2–26) kg/m², an FPG of 4.9 (4.6–5.2) mmol/L, 1-hour PG of 8.1 (6.4–10.1) mmol/L and 2-hour PG of 6.0 (4.9–7.3) mmol/L (table 1). Compared with the FmH-NONE group, the FmH-YOD group was younger (38 vs 44 years, p<0.001) and had higher BMI (23.8 vs 22.9 kg/m², p=0.006), systolic blood pressure (120 vs 110 mm Hg, p<0.001), plasma triglycerides (1.1 vs 1 mmol/L, p=0.043), and a lower high-density lipoprotein-cholesterol (1.3 vs 1.6 mmol/L, p<0.001). The FmH-YOD group was more likely to smoke (15.5% vs 8.7%, p=0.011), have IFG (12.8% vs 6.6%, p=0.011), and IGT (27.2% vs 12.9%, p<0.001) than the FmH-NONE group (table 1).

Figure 1. Overview of study participants stratified by FmH of diabetes and 1-hour plasma glucose during the 75-g oral glucose tolerance test at baseline. FmH, family history; YOD, young-onset type 2 diabetes; DM, diabetes mellitus.

Table 1

Baseline characteristics of study participants by FmH of diabetes

Clinical characteristics	All participants	FmH-young-onset diabetes	FmH-NONE	P value
Age (years)	41 (36–47)	38 (33–43)	44(38–48)	<0.001
Men	255/583 (43.7%)	95/235 (40.4%)	160/348 (46%)	0.185
Active smoker	66/578 (11.4%)	36/232 (15.5%)	30/346 (8.7%)	0.011
Active alcohol drinker	83/579 (14.3%)	36/232 (15.5%)	47/347 (13.5%)	0.507
Body mass index (kg/m2)	23.3 (21.2–26)	23.8 (21.4–26.8)	22.9 (21–25.3)	0.006
Systolic BP (mm Hg)	116 (107–129)	120 (109–130)	110 (102–123)	<0.001
Diastolic BP (mm Hg)	73 (69–80)	73 (66–80)	72 (70–80)	0.662
Total cholesterol (mmol/L)	5.1 (4.5–5.7)	5.1 (4.4–5.7)	5.1 (4.5–5.7)	0.497
High-density lipoprotein-cholesterol (mmol/L)	1.5 (1.2–1.8)	1.3 (1.1–1.6)	1.6 (1.3–1.9)	<0.001
Low-density lipoprotein-cholesterol (mmol/L)	3.0 (2.5–3.6)	3 (2.5–3.7)	2.9 (2.4–3.5)	0.075
Triglycerides (mmol/L)	1.1 (0.7–1.5)	1.1 (0.7–1.6)	1.0 (0.7–1.5)	0.043
Estimated glomerular filtration rate (mL/min/1.73m2)	102 (89–113)	112 (102–118)	95 (85–106)	<0.001
Urine albumin-to-creatinine ratio (mg/mmol)	0.68 (0.42–1.35)	0.76 (0.51–1.74)	0.64 (0.39–1.25)	<0.001
Use of antihypertensives	13/582 (2.2%)	9/235 (3.8%)	4/347 (1.2%)	0.032
Fasting plasma glucose (mmol/L)	4.9 (4.6–5.2)	4.9 (4.6–5.3)	4.8 (4.6–5.1)	0.003
1-hour plasma glucose (mmol/L)	8.1 (6.4–10.1)	9.0 (6.8–11.1)	7.6 (6.3–9.2)	<0.001
2-hour plasma glucose (mmol/L)	6.0 (4.9–7.3)	6.3 (4.9–8)	5.8 (4.9–6.8)	0.005
Impaired fasting glucose	53/583 (9.1%)	30/235 (12.8%)	23/348 (6.6%)	0.011
High 1-hour plasma glucose ≥8.6 mmol/L	253/583 (40.3%)	127/235 (54.0%)	126/348 (36.2%)	<0.001
Impaired glucose tolerance	109/583 (18.7%)	64/235 (27.2%)	45/348 (12.9%)	<0.001

BP, blood pressure; FmH, family history.

At baseline, 54% in the FmH-YOD group had high 1-hour PG versus 36.1% in the FmH-NONE group (p<0.001). In the whole group, 40.3% and 19.8% of the high 1-hour PG group had concurrent IGT and IFG, respectively, with high 1-hour PG capturing most cases of IGT (93.6%) and IFG (94.3%) (online supplemental table 1). In the FmH-YOD group, 31.9% (75/235) developed diabetes compared with 6.9% (24/348) in the FmH-NONE group after a median follow-up of 12.1 (11.3–13.1) years. In the high 1-hour PG group, 45% (57/127) developed diabetes in FmH-YOD compared with 16% (20/126) in the FmH-NONE group. In the unadjusted model, FmH-YOD, 1-hour PG (centered at 4 mmol/L), and the interaction term (FmH-YOD×1-hour PG) had ORs of 20.2 (95% CI 5.3 to 77.6, p<0.001), 1.68 (95% CI 1.41 to 2.01, p<0.001), and 0.78 (95% CI 0.63 to 0.96, p=0.022), respectively, for incident diabetes (online supplemental table 3a). After adjusting for age, sex, BMI, and other cardiometabolic risk factors, including IFG and IGT, the ORs of FmH-YOD, 1-hour PG and (FmH-YOD×1-hour PG) for incident diabetes remained significant with respective values of 36.3 (95% CI 6.1 to 216.7, p<0.001), 1.48 (95% CI 1.16 to 1.88, p=0.002), and 0.72 (95% CI 0.55 to 0.93, p=0.013), respectively (online supplemental table 3b). The results remained similar when IFG and IGT were replaced with FPG and 2-hour PG as continuous covariates. We repeated the analysis by dichotomizing 1-hour PG with the cut-off at 8.6 mmol/L. After adjusting for the aforementioned confounders, including IFG and IGT, the negative association of (FmH-YOD×1-hour PG) with incident diabetes showed marginal significance (OR 0.211 (95% CI 0.037 to 1.20), p=0.079). Replacing IFG and IGT with FPG and 2-hour PG as covariates rendered the interactive term of (FmH-YOD×1-hour PG) nearly significant with an OR of 0.179 (95% CI 0.032 to 1.014, p=0.052) (online supplemental table 4).

To further examine this negative interaction between 1-hour PG and FmH-YOD, we classified participants into four groups: (1) FmH-NONE/normal 1-hour PG, (2) FmH-NONE/high 1-hour PG, (3) FmH-YOD/normal 1-hour PG, and (4) FmH-YOD/high 1-hour PG. After adjusting for confounders including IFG and IGT, (FmH-NONE/high 1-hour PG), (FmH-YOD/normal 1-hour PG), and (FmH-YOD and high 1-hour PG) groups had ORs of 7.4 (95% CI 1.6 to 35.1, p=0.011), 18 (95% CI 3.3 to 98.1, p=0.001), and 28.2 (95% CI 5.5 to 145.9, p<0.001), respectively, when compared with the (FmH-NONE/normal 1-hour PG) group (table 2).

Table 2

Risk of incident diabetes at 2012–2014 follow-up, stratified by family history and 1-hour plasma glucose during 75-g oral glucose tolerance test at baseline

Models	No FmH-YOD and normal 1-hour PG	No FmH-YOD and high 1-hour PG	FmH-YOD and normal 1-hour PG	FmH-YOD and high 1-hour PG
Unadjusted model	Reference	OR 20.2 (95% CI 4.6 to 88.4), p<0.001	OR 20.5 (95% CI 4.6 to 91.8), p<0.001	OR 86.9 (95% CI 20.5 to 368.9), p<0.001
Adjusted model 1	Reference	OR 14.7 (95% CI 3.3 to 66.0), p<0.001	OR 23.7 (95% CI 4.8 to 116.6), p<0.001	OR 79.2 (95% CI 17.2 to 364.3), p<0.001
Adjusted model 2	Reference	OR 12.9 (95% CI 2.9 to 58.1), p=0.001	OR 17.9 (95% CI 3.4 to 96.1), p=0.001	OR 57.4 (95% CI 11.8 to 278.4), p<0.001
Adjusted model 3	Reference	OR 7.4 (95% CI 1.6 to 35.1), p=0.011	OR 18.0 (95% CI 3.3 to 98.1), p=0.001	OR 28.2 (95% CI 5.5 to 145.9), p<0.001

Model 1: adjusted for age, sex, body mass index, use of tobacco or alcohol.

Model 2: Model 1+adjusted for systolic blood pressure, Ln (triglyceride), high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, estimated glomerular filtration rate, Ln (urine albumin-to-creatinine ratio) and blood pressure lowering drugs.

Model 3: Model 2+adjusted for impaired fasting glucose and impaired glucose tolerance.

FmH, family history; PG, plasma glucose; YOD, young-onset type 2 diabetes.

In ROC analysis, 1-hour PG predicted incident diabetes in both the FmH-YOD (area under ROC curve (AUROC) 0.688 (95% CI 0.616 to 0.760, p<0.001)) and FmH-NONE groups (0.829 (95% CI 0.758 to 0.899), p<0.001) with considerably lower performance in the FmH-YOD group (difference in AUROC (FmH-NONE–FmH-YOD)=0.141 (95% CI 0.040 to 0.241), p=0.006) (figure 2). In the FmH-NONE group, 1-hour PG had the highest AUROC (0.829 (95% CI 0.758 to 0.899), p<0.001) compared with FPG (AUROC 0.743 (95% CI 0.647 to 0.839), p<0.001) and 2-hour PG (0.783 (95% CI 0.688 to 0.877), p<0.001), although the differences were short of significance. In the FmH-YOD group, FPG had the highest AUROC (0.792 (95% CI 0.730 to 0.853), p<0.001) compared with 1-hour PG (AUROC 0.688 (95% CI 0.616 to 0.760), p<0.001) and 2-hour PG (AUROC 0.706 (95% CI 0.636 to 0.776), p<0.001). These differences in AUROC were significant (Difference in AUROC (FPG–1-hour PG) 0.104 (95% CI 0.033 to 0.174), p=0.004; Difference in AUROC (FPG–2-hour PG) 0.086 (95% CI 0.006 to 0.165), p=0.035). Among people with i-IFG (n=27), the conversion rate to diabetes was 66.7% in the FmH-YOD and 6.7% in the FmH-NONE group. We repeated the ROC analysis by randomly selecting one person from each family in the cohort, and the results remained similar and significant.

Out of the 583 participants included in the main analysis, 149 who were not defined to have diabetes at the 12-year follow-up based on electronic medical records did not have a complete set of OGTT (FPG, 2-hour PG) and HbA1c data for definite ascertainment of non-diabetes status at follow-up. As a sensitivity analysis, we excluded these 149 individuals and repeated the main analysis (n=434) including regression and ROC analysis, and the results remained similar and significant (all p<0.05 as in the main analysis [results not shown].

Figure 2. Receiver operator characteristics analysis of plasma glucose parameters during the 75-g oral glucose tolerance test for incident diabetes at the 2012-2014 follow-up. AUROC, area under receiver operator characteristics curve; FmH-NONE, no family history of diabetes; FmH-YOD, family history of young-onset type 2 diabetes; FPG, fasting plasma glucose.

Discussion

In this 12-year follow-up study of 583 young to middle-aged Hong Kong Chinese without diabetes at baseline, we had three main findings: (1) in the FmH-NONE group, 1-hour PG was a strong predictor for incident diabetes, (2) in the FmH-YOD group, the predictive power of 1-hour PG was largely attenuated, and (3) in the FmH-YOD group, FPG was more robust than 1-hour PG in predicting incident diabetes.

Importance of YOD

In Asia, one in five adult patients with type 2 diabetes (T2D) under clinical care was diagnosed under the age of 40 years with a high risk of morbidity and premature mortality.^{17 18} Using register data, our group modeled that a person diagnosed with diabetes at the age of 33 years accrued 97 hospitalization bed-days by the age of 75.¹ By delaying the age of onset from 30 to 50 years, the hospitalization bed-days could be reduced by more than one-third.¹⁹ Given the legacy effect of tight glycemic control at an early stage of T2D as reported in the 44-year follow-up analysis of the United Kingdom Prospective Diabetes Study (UKPDS) and the long duration of exposure to hyperglycemia in young people with prediabetes or diabetes, there is an urgent need to develop an effective strategy to reduce the societal and personal burden of YOD.²⁰

Importance of family history of YOD

In our previous analysis of this combined cohort of HKFDS and BHBHK, we reported a dose-dependent association between age of diagnosis of affected family members and incident T2D at 12-year follow-up.¹¹ The relative risk of incident diabetes dropped from sevenfold in those with FmH of diabetes diagnosed at <30 years to twofold in those with FmH diagnosed at ≥50 years¹. In the Joint Asia Diabetes Evaluation Register with pooled data from >100 000 patients with T2D from 11 Asian countries, a history of T2D in more family members was also associated with an earlier age of diagnosis.²¹ Interestingly, among individuals with FmH of T2D, self-reporting of adherence to a healthy lifestyle was associated with a later age of diabetes,²¹ highlighting the importance of nature versus nurture in the development of T2D.

Confirmation of utility of 1-hour PG in FmH-NONE

Prediabetes is a state of intermediate hyperglycemia between normal glucose intolerance (NGT) and T2D, conventionally defined by IFG, IGT, and/or intermediately HbA1c, although the cut-off values are not uniform across different definitions.²² Individuals with such defined prediabetes had a fourfold to eightfold higher 5-year risk of incident diabetes than those with NGT.²² In a recent position statement, the IDF advocated the use of 1-hour PG ≥8.6 mmol/L during OGTT to define intermediate hyperglycemia.⁵ Supported by 40 years of epidemiological evidence, high 1-hour PG predicted incident diabetes and its complications, including death, across age, sex, and ethnicities, with superiority over FPG, 2-hour PG, and HbA1c.⁵ In line with this, our results in the FmH-NONE group showed the strong 12-year risk of incident diabetes imparted by high 1-hour PG at baseline, with odds ranging from 7 to 20 in unadjusted and adjusted models. The AUROC (0.83) of 1-hour PG in predicting diabetes also tended to be higher than other glycemic measurements. Besides, the majority (94%) of people with IGT in this cohort had concomitant high 1-hour PG. These findings echoed other reports on the precedence of high 1-hour PG before the development of IGT in the majority of cases.^{5 23} These data solidly supported the use of 1-hour PG as a risk stratifier to identify high-risk people even before the development of IGT and would reduce the time and cost associated with the 2-hour OGTT.

The United States Diabetes Prevention Program recruited people with IGT and confirmed that metformin and lifestyle intervention effectively prevented diabetes and improved outcomes.^{24 25} In the post-trial Diabetes Prevention Program Outcome Study (DPP/DPPOS), these benefits were maintained, although with some attenuation.^{24 25} Economic analysis with simulated lifetime modeling supported the cost-effectiveness of DPP and was cost-saving in young people.²⁶ While a high 1-hour PG usually precedes IGT, whether such clinical and economic benefits could be applied with a high 1-hour PG in lieu of IGT as the selection criteria for intervention would require further confirmatory evidence.

Pathophysiology of 1-hour PG hints predominant mechanism of diabetes in FmH-NONE

The underlying pathophysiology of high 1-hour PG might be the predominant mechanism of diabetes development in people without FmH of diabetes. In this regard, several recent Asian studies from Japan, China, Korea, and Singapore have revealed that elevated 1-hour PG was associated with substantial beta-cell dysfunction based on calculated indices, intravenous glucose tolerance tests, and euglycemic hyperinsulinemic clamp studies.^27–31 These researchers also demonstrated that 1-hour PG exhibited stronger associations with higher sensitivity and accuracy than 2-hour PG. The progression from NGT with high 1-hour PG to combined high 1-hour PG and IGT was accompanied by a further decline in beta-cell function.^{27 28} Although some researchers reported an association of 1-hour PG with insulin resistance, the relative importance was less prominent than beta-cell dysfunction, while some studies did not support such the association between 1-hour PG and insulin resistance.^27–31 In people without FmH, lifestyle factors may play a key role in the development of prediabetes and diabetes, as suggested by the associations of low dietary quality and physical inactivity with IGT.³ The efficacy of increased physical activity in people with IGT was in part attributed to increased glucose disposal at the muscle level, which might be better reflected by the 1-hour PG.³

Attenuation of the predictive value of 1-hour PG in FmH-YOD

While 1-hour PG performed satisfactorily in multiple subpopulations, its predictive power in people with an FmH of YOD had not been examined in Chinese with a high prevalence of YOD. In the current cohort, FmH-YOD conferred 4.6-fold higher crude risk than FmH-NONE with significant negative interaction between 1-hour PG and FmH-YOD. The FmH-YOD/normal 1-hour PG group had a similar or even higher risk than the FmH-NONE/high 1-hour PG group. In the FmH-YOD group, the fully adjusted odds for incident diabetes were less than twofold between the normal 1-hour PG and high 1-hour PG groups. The AUROC also significantly dropped from 0.83 to 0.69 when moving from FmH-NONE to FmH-YOD. Given that 1 hour was closely associated with beta-cell dysfunction and lifestyle factors, these findings suggested that FPG might also be important in the development of diabetes in the FmH-YOD group.

Importance of FPG in the context of FmH-YOD

In contrast to 1-hour PG, the AUROC of FPG in predicting diabetes was similar between the FmH-NONE and FmH-YOD groups, although with a higher tendency in the FmH-YOD group. Thus, FPG also predicted diabetes in the general population, although familial factors in the FmH-YOD group might amplify its effect size. In the FmH-YOD group, the higher ROC of FPG than that of 1-hour PG for incident diabetes warrants further discussion. In this cohort of the workforce, 136 people had prediabetes defined by IFG and/or IGT, one in five of whom had i-IFG. Among these people with i-IFG at baseline, the status of FmH-YOD was 89% among progressors to diabetes versus 29% among non-progressors at 12 years. These data highlighted the importance of FPG or i-IFG, especially in those with FmH of YOD.

Population-based studies suggested a lower progression rate of i-IFG compared with isolated IGT (i-IFG) to diabetes. There is also a lack of conclusive evidence regarding the efficacy of intensive lifestyle intervention or metformin in people with i-IFG. Thus, some experts argued that these prevention programs should be limited to people with IGT with or without IFG.^{7 8 10 32 33} However, our results suggested that in people with FmH-YOD, FPG was a more robust predictor of progression than 1-hour PG and 2-hour PG and that further research should be performed to elucidate the underlying causes to inform intervention. From an implementation perspective, whenever possible, documentation of FmH-YOD should be ascertained, and family members of people with YOD would benefit from regular screening to detect early prediabetes and diabetes.

In contrast to i-IGT, which was more related to lifestyle factors, researchers had reported a stronger genetic component associated with i-IFG.³ In a recent global analysis, there were more people with IGT than people with IFG (FPG: 6–6.9 mmol/L by WHO criteria) in all regions except Southeast Asia.³⁴ Using ADA criteria (FPG: 5.6–6.9 mmol/L), several Caucasian cohorts showed a higher prevalence of IFG than IGT in people with prediabetes.³⁵ In our Chinese cohort, we observed a higher prevalence of IGT (80.1%) than IFG (39%) among those with prediabetes. Different settings and population profiles might contribute to this heterogeneity in the distribution of IFG and IGT among people with prediabetes.

However, in the global analysis, there was a clear association between age and i-IGT, while the prevalence of i-IFG was similar across all age groups.³⁴ Pathophysiologically, we postulated that impaired glucose sensing due to abnormal function of glucokinase (GK), GK regulatory protein, and glucose-6-phosphatase might contribute to high FPG as exemplified by GK-maturity-onset diabetes of the young and raise the question of whether GK activator (eg, dorzagliatin) might benefit people with i-IFG.^{3 36–39} Likewise, in people with dysregulation of fat metabolism, such as metabolic dysfunction-associated fatty liver disease with excess hepatic glucose production leading to high FPG, drugs such as peroxisome proliferator-activated receptor (PPAR)-gamma activator (eg, pioglitazone) might improve fat metabolism and prevent diabetes. The Actos Now for the prevention of diabetes study recruited people with combined IGT and FPG 5.3–6.9 mmol/L. In this study, pioglitazone reduced the risk of conversion to diabetes by 72% after a median follow-up of 2.4 years⁴⁰. In subgroup analysis, the group with combined IFG and IGT had a numerically lower HR than i-IGT but without statistical significance. Likewise, the Insulin Resistance Intervention after Stroke trial recruited people without diabetes who had insulin resistance (defined by Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) >3) and a recent history of cerebrovascular disease. In this trial, pioglitazone reduced FPG and HOMA-IR with a 52% lower risk of conversion to diabetes after a median follow-up of 4.8 years. Importantly, this effect was predominately driven by those with IFG at baseline with an HR of 0.41 compared with placebo.⁴¹

Lifestyle-wise, some researchers have reported associations of IFG with smoking.³ In our analysis, people with FmH-YOD were more likely to be smokers. Other researchers have reported an association of sugar-sweetened beverages (SSB) with FPG. This association was positively modulated by a polygenic risk score including variants implicated in beta-cell function, circadian rhythm, and hepatic resistance, including PPAR-gamma.⁴² This knowledge inferred that reduced intake of SSB might confer greater risk reduction in people with a genetic predisposition. These aforementioned hypotheses of intervention targeting prediabetes subphenotypes will need further testing, especially in people with FmH of YOD and i-IFG.

Limitations

This study has several limitations. First, baseline diabetes status was defined by known history, use of glucose-lowering medications, and OGTT, but not HbA1c. The latter was not used to diagnose diabetes 20 years ago. However, at the 2012–2014 follow-up, we included HbA1c as a diagnostic criterion. Thus, some individuals with NGT might have undiagnosed diabetes based on HbA1c ≥6.5% if this had been measured. Second, the age of diagnosis of incident diabetes was not available except for the newly diagnosed cases. Thus, we could not use a time-to-event models, such as the Cox proportional hazards regression model to estimate the HR or progression rates of diabetes. Third, for FmH, the age of diagnosis of diabetes was available for parents but not siblings, except those siblings in the HKFDS cohort. Since defining FmH of late-onset diabetes (FmH-LOD) would require the age of diagnosis of diabetes in all first-degree relatives, we did not include FmH-LOD in the analysis. Fourth, the sample size was relatively small, and despite the significance and consistency of our observations, which lent support to our hypotheses, the CIs of the ORs were wide in regression models. A larger sample size or longer follow-up duration will be needed to improve the precision of estimation. Fifth, the 1-hour PG cut-off of 8.6 mmol/L for predicting incident diabetes was mainly based on European data.⁵ A few prospective studies in Asia suggested a higher cut-off values, which might be due to more rapid gastric emptying with greater postprandial glucose excursion rather than abnormal glucose metabolism.³¹ Sixth, all participants were Chinese with a baseline age of 41 (36–47). This limited generalization of our results to non-Chinese, youth, and elderly populations.

There are reports from other populations indicating that high FPG had more genetic components than 2-hour PG and that high 1-hour PG and IGT were more related to lifestyle factors.³ Thus, we speculated that the trend of increased predictive power of FPG and decreased predictive power of 1-hour PG, when moving from individuals with FmH-NONE to FmH-YOD, could be replicated in other ethnic groups. Large-scale East Asian or multiethnic genome-wide association studies supported the association of multiple variants with FPG.^{39 43} However, whether FPG will outperform 1-hour PG would depend on the distribution and relative importance of diabetes of different underlying predominant mechanisms as well as other factors such as genetics, ethnicity, environment, and lifestyles.

Conclusions

Based on these findings, we concluded that while 1-hour PG is a robust marker for progression to diabetes in the general population, its differential performance in people with different status of FmH (attenuated in people with FmH of YOD and being outperformed by FPG) calls for better understanding of the pathophysiology and genetics affecting FPG. This knowledge might inform targeted prevention with improved efficacy in people with FmH-YOD and/or i-IFG. Our data call for better documentation of FmH and recognition of YOD in clinical practice for surveillance, more granular analysis in ethnic-specific cohorts to unravel the heterogeneity of prediabetes, and the development of new strategies to prevent the progression from i-IFG to diabetes, especially in people with familial and genetic predisposition.

Data availability statement

Due to local law and regulation, no data can be shared with external parties. Summary statistics may be shared upon reasonable request to corresponding author.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee (CREC Ref. No.: 2019.080-T, 2012.171). Participants gave informed consent to participate in the study before taking part.

Footnote

Contributors JCNC and AL conceptualized the research question. CKO and ESHL performed the data analysis. CKO, BF, JPYH, ESHL, GTCK, JNML, EC, APSK, RCWM, AL, and JCNC contributed to the interpretation of the data. CKO wrote the first draft. All authors critically reviewed the manuscript and approved the final version. JCNC is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding Dr C-K O is supported by the Hong Kong Genome Institute, the Government of the Hong Kong Special Administrative Region (HKSAR Government). The study was partially supported by Commissioned Grant by the Hong Kong Health and Medical Research Fund (CFS-CUHK2), Health Bureau, HKSAR Government. The establishment of the Better Health for Better Hong Kong Cohort was supported by the Li Ka Shing Charity Foundation. Some data in this manuscript were presented in the 60th Annual Meeting of the European Association for the Study of Diabetes in the form of a published abstract and a short oral discussion in 2024.

Competing interests JCNC has received research grants (through institutions) and/or honoraria for consultancy and/or giving lectures from Applied Therapeutics, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Hua Medicine, Powder Pharmaceuticals, Merck Serono, MSD, Pfizer, Sanofi, Viatris and Zuellig Pharma. AL has served as an advisory committee member for AstraZeneca, Boehringer Ingelheim, Sanofi, and Amgen and has received research grants and travel grants from AstraZeneca, Boehringer Ingelheim, MSD, Novartis, Novo Nordisk, Sanofi, and Amgen. RCWM has received research grants for clinical trials from AstraZeneca, Bayer, MSD, Novo Nordisk, Sanofi, Roche, and Tricida and honoraria for consultancy or lectures from AstraZeneca and Boehringer Ingelheim. APSK has received research grants and/or speaker honoraria from Abbott, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Merck Serono, Nestle, Novo Nordisk, and Sanofi. JCNC and RCWM are inventors of patents for using biogenetic markers to predict risk of diabetes and its complications, owned by the Chinese University of Hong Kong. JCNC and RCWM are co-founders of GemVCare, a biotech start-up supported by the Technology Start-up Support Scheme for Universities of the Hong Kong Government Innovation and Technology Commission.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Ahlqvist E, Storm P, Käräjämäki A, et al. Novel subgroups of adult-onset diabetes and their association with outcomes: a data-driven cluster analysis of six variables. Lancet Diabetes Endocrinol 2018; 6: 361–9. doi:10.1016/S2213-8587(18)30051-2

² Zheng R, Xu Y, Li M, et al. Data-driven subgroups of prediabetes and the associations with outcomes in Chinese adults. Cell Rep Med 2023; 4: 100958. doi:10.1016/j.xcrm.2023.100958

³ Faerch K, Borch-Johnsen K, Holst JJ, et al. Pathophysiology and aetiology of impaired fasting glycaemia and impaired glucose tolerance: does it matter for prevention and treatment of type 2 diabetes? Diabetologia 2009; 52: 1714–23. doi:10.1007/s00125-009-1443-3

⁴ Ho J-Y, Lau E-H, Kwan O C, et al. One-hour post-load plasma glucose level predicts future type 2 diabetes in a community-based study of Hong Kong Chinese workforce. Diabetes Res Clin Pract 2024; 212: 111718. doi:10.1016/j.diabres.2024.111718

⁵ Bergman M, Manco M, Satman I, et al. International Diabetes Federation Position Statement on the 1-hour post-load plasma glucose for the diagnosis of intermediate hyperglycaemia and type 2 diabetes. Diabetes Res Clin Pract 2024; 209: 111589. doi:10.1016/j.diabres.2024.111589

⁶ Misra S, Ke C, Srinivasan S, et al. Current insights and emerging trends in early-onset type 2 diabetes. Lancet Diabetes Endocrinol 2023; 11: 768–82. doi:10.1016/S2213-8587(23)00225-5

⁷ Sheng Z, Cao J-Y, Pang Y-C, et al. Effects of Lifestyle Modification and Anti-diabetic Medicine on Prediabetes Progress: A Systematic Review and Meta-Analysis. Front Endocrinol (Lausanne) 2019; 10: 455. doi:10.3389/fendo.2019.00455

⁸ Sathish T, Khunti K, Narayan KMV, et al. Effect of Conventional Lifestyle Interventions on Type 2 Diabetes Incidence by Glucose-Defined Prediabetes Phenotype: An Individual Participant Data Meta-analysis of Randomized Controlled Trials. Diabetes Care 2023; 46: 1903–7. doi:10.2337/dc23-0696

⁹ Herman WH, Ye W. Precision Prevention of Diabetes. Diabetes Care 2023; 46: 1894–6. doi:10.2337/dci23-0052

¹⁰ Zhang L, Zhang Y, Shen S, et al. Safety and effectiveness of metformin plus lifestyle intervention compared with lifestyle intervention alone in preventing progression to diabetes in a Chinese population with impaired glucose regulation: a multicentre, open-label, randomised controlled trial. Lancet Diabetes Endocrinol 2023; 11: 567–77. doi:10.1016/S2213-8587(23)00132-8

¹¹ Zhang Y, Luk AOY, Chow E, et al. High risk of conversion to diabetes in first-degree relatives of individuals with young-onset type 2 diabetes: a 12-year follow-up analysis. Diabet Med 2017; 34: 1701–9. doi:10.1111/dme.13516

¹² Li JKY, Ng MCY, So WY, et al. Phenotypic and genetic clustering of diabetes and metabolic syndrome in Chinese families with type 2 diabetes mellitus. Diabetes Metab Res Rev 2006; 22: 46–52. doi:10.1002/dmrr.577

¹³ Ko GTC, Chan JCN, Chan AWY, et al. Low levels of awareness of suboptimal health conditions in a high-risk working population: the “better health for better Hong Kong” health promotion campaign. Int J Behav Med 2007; 14: 63–9. doi:10.1007/BF03004170

¹⁴ Ko GTC, Chan JCN, Chan AWY, et al. Association between sleeping hours, working hours and obesity in Hong Kong Chinese: the “better health for better Hong Kong” health promotion campaign. Int J Obes (Lond) 2007; 31: 254–60. doi:10.1038/sj.ijo.0803389

¹⁵ Ko GT, So W, Chow C, et al. Risk associations of obesity with sugar-sweetened beverages and lifestyle factors in Chinese: the “Better Health for Better Hong Kong” health promotion campaign. Eur J Clin Nutr 2010; 64: 1386–92. doi:10.1038/ejcn.2010.181

¹⁶ American Diabetes Association Professional Practice Committee. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes-2024. Diabetes Care 2024; 47: S20–42. doi:10.2337/dc24-S002

¹⁷ Yeung RO, Zhang Y, Luk A, et al. Metabolic profiles and treatment gaps in young-onset type 2 diabetes in Asia (the JADE programme): a cross-sectional study of a prospective cohort. Lancet Diabetes Endocrinol 2014; 2: 935–43. doi:10.1016/S2213-8587(14)70137-8

¹⁸ Nanayakkara N, Curtis AJ, Heritier S, et al. Impact of age at type 2 diabetes mellitus diagnosis on mortality and vascular complications: systematic review and meta-analyses. Diabetologia 2021; 64: 275–87. doi:10.1007/s00125-020-05319-w

¹⁹ Ke C, Lau E, Shah BR, et al. Excess Burden of Mental Illness and Hospitalization in Young-Onset Type 2 Diabetes: A Population-Based Cohort Study. Ann Intern Med 2019; 170: 145–54. doi:10.7326/M18-1900

²⁰ Adler AI, Coleman RL, Leal J, et al. Post-trial monitoring of a randomised controlled trial of intensive glycaemic control in type 2 diabetes extended from 10 years to 24 years (UKPDS 91). Lancet 2024; 404: 145–55. doi:10.1016/S0140-6736(24)00537-3

²¹ Cheung JTK, Lau E, Tsui CCT, et al. Combined associations of family history and self-management with age at diagnosis and cardiometabolic risk in 86,931 patients with type 2 diabetes: Joint Asia Diabetes Evaluation (JADE) Register from 11 countries. BMC Med 2022; 20: 249. doi:10.1186/s12916-022-02424-y

²² Lee CMY, Colagiuri S, Woodward M, et al. Comparing different definitions of prediabetes with subsequent risk of diabetes: an individual participant data meta-analysis involving 76 513 individuals and 8208 cases of incident diabetes. BMJ Open Diabetes Res Care 2019; 7: e000794. doi:10.1136/bmjdrc-2019-000794

²³ Bergman M, Abdul-Ghani M, Chan J, et al. Staging schema for early diagnosis of prediabetes. Lancet Diabetes Endocrinol 2024; 12: 873–6. doi:10.1016/S2213-8587(24)00320-6

²⁴ Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393–403. doi:10.1056/NEJMoa012512

²⁵ Diabetes Prevention Program Research Group. Long-term effects of lifestyle intervention or metformin on diabetes development and microvascular complications over 15-year follow-up: the Diabetes Prevention Program Outcomes Study. Lancet Diabetes Endocrinol 2015; 3: 866–75. doi:10.1016/S2213-8587(15)00291-0

²⁶ Herman WH. The cost-effectiveness of diabetes prevention: results from the Diabetes Prevention Program and the Diabetes Prevention Program Outcomes Study. Clin Diabetes Endocrinol 2015; 1: 9. doi:10.1186/s40842-015-0009-1

²⁷ Lu J, Ni J, Su H, et al. One-Hour Postload Glucose Is a More Sensitive Marker of Impaired β-Cell Function Than Two-Hour Postload Glucose. Diabetes 2025; 74: 36–42. doi:10.2337/db24-0652

²⁸ Nie Q, Jin X, Mu Y, et al. Insulin resistance and β-cell dysfunction in individuals with normal glucose tolerance but elevated 1-h post-load plasma glucose. Front Endocrinol (Lausanne) 2025; 16: 1507107. doi:10.3389/fendo.2025.1507107

²⁹ Sato M, Tamura Y, Someya Y, et al. Characteristics associated with elevated 1-h plasma glucose levels during a 75-g oral glucose tolerance test in non-obese Japanese men. J Diabetes Investig 2020; 11: 1520–3. doi:10.1111/jdi.13245

³⁰ Oh TJ, Min SH, Ahn CH, et al. Normal Glucose Tolerance with a High 1-Hour Postload Plasma Glucose Level Exhibits Decreased β-Cell Function Similar to Impaired Glucose Tolerance. Diabetes Metab J 2015; 39: 147–53. doi:10.4093/dmj.2015.39.2.147

³¹ Lee MH, Febriana E, Lim M, et al. Performance of the 1 h oral glucose tolerance test in predicting type 2 diabetes and association with impaired β-cell function in Asians: a national prospective cohort study. Lancet Reg Health West Pac 2025; 54: 101278. doi:10.1016/j.lanwpc.2024.101278

³² Narayan KMV, Kondal D, Chang HH, et al. Natural History of Type 2 Diabetes in Indians: Time to Progression. Diabetes Care 2024; 47: 858–63. doi:10.2337/dc23-1514

³³ Campbell MD, Sathish T, Zimmet PZ, et al. Benefit of lifestyle-based T2DM prevention is influenced by prediabetes phenotype. Nat Rev Endocrinol 2020; 16: 395–400. doi:10.1038/s41574-019-0316-1

³⁴ Rooney MR, Fang M, Ogurtsova K, et al. Global Prevalence of Prediabetes. Diabetes Care 2023; 46: 1388–94. doi:10.2337/dc22-2376

³⁵ Yip WCY, Sequeira IR, Plank LD, et al. Prevalence of Pre-Diabetes across Ethnicities: A Review of Impaired Fasting Glucose (IFG) and Impaired Glucose Tolerance (IGT) for Classification of Dysglycaemia. Nutrients 2017; 9: 1273. doi:10.3390/nu9111273

³⁶ Chow E, Wang K, Lim CKP, et al. Dorzagliatin, a Dual-Acting Glucokinase Activator, Increases Insulin Secretion and Glucose Sensitivity in Glucokinase Maturity-Onset Diabetes of the Young and Recent-Onset Type 2 Diabetes. Diabetes 2023; 72: 299–308. doi:10.2337/db22-0708

³⁷ Tam CHT, Ho JSK, Wang Y, et al. Common polymorphisms in MTNR1B, G6PC2 and GCK are associated with increased fasting plasma glucose and impaired beta-cell function in Chinese subjects. PLoS One 2010; 5: e11428. doi:10.1371/journal.pone.0011428

³⁸ Tam CHT, Ma RCW, So WY, et al. Interaction effect of genetic polymorphisms in glucokinase (GCK) and glucokinase regulatory protein (GCKR) on metabolic traits in healthy Chinese adults and adolescents. Diabetes 2009; 58: 765–9. doi:10.2337/db08-1277

³⁹ Hwang J-Y, Sim X, Wu Y, et al. Genome-wide association meta-analysis identifies novel variants associated with fasting plasma glucose in East Asians. Diabetes 2015; 64: 291–8. doi:10.2337/db14-0563

⁴⁰ DeFronzo RA, Tripathy D, Schwenke DC, et al. Pioglitazone for diabetes prevention in impaired glucose tolerance. N Engl J Med 2011; 364: 1104–15. doi:10.1056/NEJMoa1010949

⁴¹ Inzucchi SE, Viscoli CM, Young LH, et al. Pioglitazone Prevents Diabetes in Patients With Insulin Resistance and Cerebrovascular Disease. Diabetes Care 2016; 39: 1684–92. doi:10.2337/dc16-0798

⁴² López-Portillo ML, Huidobro A, Tobar-Calfucoy E, et al. The Association between Fasting Glucose and Sugar Sweetened Beverages Intake Is Greater in Latin Americans with a High Polygenic Risk Score for Type 2 Diabetes Mellitus. Nutrients 2021; 14. doi:10.3390/nu14010069

⁴³ Downie CG, Dimos SF, Bien SA, et al. Multi-ethnic GWAS and fine-mapping of glycaemic traits identify novel loci in the PAGE Study. Diabetologia 2022; 65: 477–89. doi:10.1007/s00125-021-05635-9