European Journal of Endocrinology, 2023, 188, 1–9

https://doi.org/10.1093/ejendo/lvac001
Advance access publication 10 January 2023
Original Research

Persistence or regression of prediabetes and coronary

artery calcification among adults without diabetes
Yoosun Cho,1 Yoosoo Chang,2,3,4,* Seungho Ryu,2,3,4,* Yejin Kim,2 Hyun-Suk Jung,1
Jeonggyu Kang,1,2 In Young Choi,1 Chan-won Kim,1 Hyungseok Oh,1 Sarah H. Wild,5,*
and Christopher D Byrne6,7
1
Total Healthcare Center, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul 04514, Republic of Korea
2
Center for Cohort Studies, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul 04514, Republic of Korea

Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024

3
Department of Occupational and Environmental Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine, Seoul
04514, Republic of Korea
4
Department of Clinical Research Design & Evaluation, Samsung Advanced Institute for Health Sciences & Technology, Sungkyunkwan
University, Seoul 06355, Republic of Korea
5
Centre for Population Health Sciences, Usher Institute, University of Edinburgh, Edinburgh EH8 9AG, United Kingdom
6
Nutrition and Metabolism, Faculty of Medicine, University of Southampton, Southampton SO16 6YD, United Kingdom
7
National Institute for Health Research Southampton Biomedical Research Centre, University Hospital Southampton, Southampton SO16 6YD,
United Kingdom
*Corresponding authors: Seungho Ryu, MD, PhD, Department of Occupational and Environmental Medicine, Kangbuk Samsung Hospital, Sungkyunkwan
University School of Medicine, Samsung Main Building B2, 250, Taepyung-ro 2ga, Jung-gu, Seoul 04514, Republic of Korea. Email: sh703.yoo@gmail.com.
Yoosoo Chang, MD, PhD, Department of Occupational and Environmental Medicine, Kangbuk Samsung Hospital, Sungkyunkwan University School of Medicine,
Samsung Main Building B2, 250, Taepyung-ro 2ga, Jung-gu, Seoul 04514, Republic of Korea. Email: yoosoo.chang@gmail.com. Sarah H. Wild, MB, BChir, PhD,
Usher Institute, University of Edinburgh, Edinburgh, EH8 9AG, UK. Email: Sarah.Wild@ed.ac.uk

Abstract
Objective: The effect of changes in glycemic status on subclinical atherosclerosis is uncertain. We assessed the association of persistence,
regression, or progression of prediabetes with coronary artery calcium score (CACS) as a measure of subclinical atherosclerosis.
Design: A cross-sectional study, comprising 126 765 adults, and longitudinal sub-study, comprising 40 622 adults (with baseline and at least 1
follow-up computed tomography scan to assess changes in CACS), were undertaken.
Methods: Changes in glycemic status over 1.5 years (interquartile range, 1.0-2.0) before the first CACS assessment were categorized according
to 6 groups: persistent normoglycemia (reference), normoglycemia to prediabetes, normoglycemia to diabetes, prediabetes to normoglycemia,
persistent prediabetes, and prediabetes to diabetes. Logistic regression was used to calculate the odds ratios (ORs) and 95% CIs for prevalent
coronary artery calcification (CAC). Mixed models with random intercepts and random slopes were used to estimate 5-year CAC progression
rates.
Results: Mean (SD) age was 41.3 (7.0) years (74.7% male) (n = 126 765). Multivariable-adjusted OR for prevalent CAC was 1.13 (95% CI, 1.08-
1.18) for persistent prediabetes, 1.05 (0.98-1.12) for regression to normoglycemia, and 1.46 (95% CI, 1.27-1.67) for progression from prediabetes
to diabetes, compared with persistent normoglycemia. Coronary artery calcification progression increased significantly in all prediabetes groups.
Multivariable-adjusted ratio of 5-year CAC progression rates was 1.19 (95% CI, 1.16-1.22) (persistent prediabetes), 1.11 (1.07-1.14) (regression to
normoglycemia), and 1.63 (95% CI, 1.26-2.10) (progression from prediabetes to diabetes).
Conclusions: Unfavorable changes in glycemic status, including persistence of prediabetes or progression to diabetes from prediabetes, were
associated with increased risk of CAC.
Keywords: cardiovascular disease, coronary artery calcification, prediabetes, cohort study

Significance

We present the data from a very large cohort study showing for the first time that persistent prediabetes and progression
from prediabetes to diabetes were significantly associated with a higher prevalence of coronary artery calcification (CAC)
at baseline and higher 5-year CAC progression rates over time. Compared with the groups with persistent prediabetes, those
who reverted from prediabetes to normoglycemia had a significantly reduced 5-year progression rate of CAC. We suggest
that reversion from prediabetes to normoglycemic could prevent the progression of atherosclerosis.

Received: June 22, 2022. Revised: October 31, 2022. Editorial Decision: November 7, 2022. Accepted: November 10, 2022
© The Author(s) 2023. Published by Oxford University Press on behalf of (ESE) European Society of Endocrinology.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
2 European Journal of Endocrinology, 2023, Vol. 188, No. 1

Introduction and missing covariates including values of glucose, hemoglo­

bin A1, or body mass index (BMI). Some participants met
The prevalence of prediabetes or hyperglycemia, defined as
more than 1 of the exclusion criteria, resulting in a total of
glycemic levels above normal but below the diabetes thresh­
126 765 eligible participants (Figure 1). A timeline of the study
old, is increasing worldwide. It is estimated that prediabetes
design is shown in Figure 2. We also evaluated the prospective
will affect more than 470 million people by 2030,1 70% of
association between changes in glycemic status and CAC pro­
whom will eventually develop diabetes.2 Prediabetes increases
gression. This analysis included all study participants who had
not only the risk of progression to diabetes but also the risk of
at least 1 follow-up cardiac CT to measure the CACS until
various adverse outcomes, including cardiovascular diseases
December 31, 2019 (n = 40 622). Study participants have
(CVD) and all-cause mortality.3-6 Prediabetes was suggested
been recruited continuously since 2010, and half of the partic­
to cause CVD in a recent Mendelian randomization analysis
ipants recruited in more recent years did not have a second
study.7 Considering the high prevalence of prediabetes, its po­
CACS measurement included in the dataset used in the current
tential to progress to diabetes, and its complications; addition­
study. Consequently, half of the participants (n = 40 622) had
al attention and appropriate management of prediabetes is
a follow-up CACS and were included in the analysis of the
needed, to potentially reduce cardiovascular risk and other
prospective association between changes in glycemic status

Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024

complications.
and CAC progression.
Pathophysiological evidence suggests that atherosclerotic
This study was approved by the Institutional Review Board
changes occur before the clinical manifestation of diabetes.8
of Kangbuk Samsung Hospital, which waived the requirement
A higher atherosclerotic burden and lipid-rich coronary pla­
for informed consent because we used only de-identified data
ques have been found in individuals with prediabetes.8,9
routinely collected as part of health screening examinations
Additionally, prediabetic status has been associated with in­
(IRB No. 2022-02-001). All procedures used in this study ad­
flammation and vasoconstriction, which may promote athero­
hered to the ethical principles of the Declaration of Helsinki
sclerosis in the coronary arteries.10,11 Nevertheless,
for Medical Research Involving Human Subjects outlined in
observational studies have reported inconsistent findings re­
2013.
garding the relationship between prediabetes and subclinical
atherosclerosis, a potential precursor of subsequent CVD
events.12-14 While previous studies have also suggested that Definition of diabetes, prediabetes, and
an unfavorable change in glycemic status or persistent predia­ normoglycemia
betes is significantly associated with an increased risk of CVD,
Diabetes mellitus was defined as fasting serum glucose levels
such as risks of myocardial infarction,15,16 stroke, and all-
≥126 mg/dL, hemoglobin A1C (HbA1c) level ≥6.5%
cause mortality in patients,16 it is currently uncertain whether
(48 mmol/mol), a history of diabetes, or the use of blood
changes in glycemic status are associated with changes in the
glucose-lowering agents. Impaired fasting glucose (IFG) was
progression of subclinical atherosclerosis among patients
defined as fasting glucose levels ranging from 100 to
without diabetes.
125 mg/dL. Participants with IFG or HbA1c levels of
Coronary artery calcification (CAC), which can be easily as­
5.7%-6.4% (39-47 mmol/mol) were classified as predia­
sessed by high-resolution computed tomography (CT), is an
betic.21 Participants meeting all criteria of fasting glucose
independent predictor of cardiovascular events and a useful
<100 mg/dL, HbA1c < 5.7% (<39 mmol/mol), no history of
indicator of subclinical coronary atherosclerosis.17,18 In add­
diabetes, and no use of blood glucose-lowering agents were
ition, CAC progression over time is a significant predictor of
classified as having normoglycemia.
mortality.19 Therefore, in patients who did not have diabetes
Participants were classified into 6 groups based on their gly­
at baseline, we aimed to evaluate (a) the association between
cemic status at 2 time points prior to the first (baseline) CT
glycemic status change and baseline subclinical atherosclerosis
scan to assess CACS: (1) persistent normoglycemia (refer­
determined by coronary artery calcium score (CACS) in a
ence), (2) normoglycemia to prediabetes, (3) normoglycemia
cross-sectional study and (b) the association between glycemic
to diabetes, (4) prediabetes to normoglycemia, (5) persistent
status change and CACS progression over time (between base­
prediabetes, and (6) prediabetes to diabetes (Figure 2).
line and follow-up CT scans) in a retrospective cohort study.

Measurement of CAC by multidetector CT

Methods
Study population
The present large-scale study was conducted in a subsample of
participants of the Kangbuk Samsung Health Study, for adults
aged 18 years and older who underwent annual or biennial
health screenings at the Kangbuk Samsung Hospital Total
Healthcare Centers in Seoul and Suwon, South Korea20
(Supplemental Data).
This study included 129 350 men and women who under­
went at least 2 visits for health examination between 2010
and 2019, including measurement of CACS via CT scans at
the visit preceding the baseline (and first) CAC scan
(CACS-1) and at the first CAC scan (CACS-1st). Among the
individuals, we excluded 2585 participants for the following
reasons: history of type 2 diabetes; family history of CVD;
CT scans were performed in both the Seoul and Suwon centers
with a Lightspeed VCT XTe-64 slice MDCT scanner (GE
Healthcare, Tokyo, Japan) using the standardized scanning
protocol: 2.5-mm thickness, 400 ms rotation time, 120 kV
tube voltage, and 124 mAS (310 mA × 0.4 s) tube current
under ECG-gated dose modulation. Coronary artery calcifica­
tion Agatston scores were calculated by summing the CACS of
all foci in the epicardial coronary system.22 The prevalence of
CAC was defined as an Agatston score of >0 at the time of the
first CT scan. The CAC progression was defined as any in­
crease in CAC that consists of the following: (a) conversion
from CAC of 0 to detectable calcification and (b) a CACS in­
dicating progression among participants with a baseline
CACS > 0. The CACS had good inter- and intra-observer reli­
abilities (intraclass correlation coefficient of 0.99).20
Yoosun Cho et al.
Figure 1. Flow chart of study participants.
Data collection
See Supplemental Data for data collection.
Statistical analysis
To assess the relationship between glycemic status category
and CAC prevalence, a logistic regression model was used to
estimate the odds ratios (ORs) with 95% CIs for the presence
of CACS >0. Multivariable models were adjusted for age, sex,
center, year of screening examination, smoking status, alcohol
intake, physical activity, education level, medication for
hypertension, lipid-lowering medication, BMI, LDL-C, and
systolic blood pressure (SBP) (Supplemental Data).
We used a Tobit regression model for natural log (CACS + 1)
with a Huber–White estimation of SEs in a sensitivity analysis
with the CACS as a continuous variable.23 By comparing the
categories of change in glycemic status to the reference (persist­
ent normoglycemia), Tobit models were used to estimate the
CACS ratio and 95% CI across glycemic change categories.
The estimated value of the Tobit model is expressed as expo­
nentiation. Tobit regression coefficient (CACS ratio) approxi­
mates the relative CACS increment comparing glycemic
change categories to the reference category (persistent normo­
glycemia). For instance, a CACS of 1.50 is interpreted as a
50% increase in the CACS of a particular category compared
with the reference category.
We further evaluated the prospective association between
changes in glycemic status and CAC progression, with the se­
cond screening visit as the start of the follow-up. We used lin­
ear mixed models with random intercepts and random
slopes24 to estimate CAC progression. The multivariable
model included glycemic status change as well as smoking
status, alcohol intake, physical activity, lipid-lowering medi­
cation, medication for hypertension, BMI, LDL-C, and SBP
as time-varying variables, and age at baseline, sex, center,
year of screening examination, education level, and time-
3
Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024
fixed variables (See Supplemental Data for selection of
confounders).
We then estimated the 5-year change in the adjusted CACS
for each category of glycemic change and compared these
estimates to the 5-year change in the reference category
(persistent normoglycemia). To account for potential differen­
ces between participants with a single CAC measurement and
those with ≥2 CAC measurements, we performed an analysis
using inverse probability weights for selection. We estimated
the probability of having ≥2 CAC measurements using base­
line characteristics and weighted each individual using the in­
verse of the predicted probability of having ≥2 CAC
measurements in the analyzes. Moreover, because the CACS
were right-skewed, we performed the analysis after transform­
ing the CACS into loge(CACS + 1). The estimates from this
model were then exponentiated to obtain the geometric means
of CACS. Based on the glycemic change categories, the 5-year
progression rate with 95% CI was estimated.
All statistical analyzes were performed using Stata version
16.0 (StataCorp LP; College Station, TX, USA). All reported
P-values were 2-sided, and comparisons were considered stat­
istically significant at P < 0.05.
Results
The mean age of the study participants (n = 126 765) at the
first CAC visit was 41.3 years (SD 7.0), and 74.7% were
men (Table 1). The 2 largest subgroups of short-term glycemic
patterns were persistent normoglycemia (42.4%) and persist­
ent prediabetes (29.9%). The interval between 2 visits for gly­
cemic measurements was 1.5 years (interquartile range,
1.0-2.0). The overall prevalence of a CACS >0 was 11.6%.
Individuals with persistent prediabetes or progression to
worse glycemic status were more likely to have unfavorable
cardiometabolic profiles and prevalent CAC than those with
persistent normoglycemia.
4 European Journal of Endocrinology, 2023, Vol. 188, No. 1

Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024

Figure 2. Timeline of the study design. The glycemic status such as normoglycemia, prediabetes, or diabetes was defined (based on the fasting blood
glucose or HbA1c level) at the visit preceding the baseline (and first) coronary artery calcium (CAC) CT scan (CACS-1) and at the first CAC CT scan
(CACS-1). Changes in glycemic status were then ascertained for each patient; the participants were divided into 6 groups: (1) persistent normoglycemia
(reference), (2) normoglycemia to prediabetes, (3) normoglycemia to diabetes, (4) prediabetes to normoglycemia, (5) persistent prediabetes, and (6)
prediabetes to diabetes. Accordingly, the associations were determined between the 6 groups for short-term change in glycemic status and (A) the
baseline CAC scores (CACS-1) (cross-sectional study) and (B) CAC progression (the difference between CACS-1 and subsequent follow-up CT scan [if the
latter were available; prospective study]). The median follow-up duration between the CAC CT scan at the first visit (CAC-1) and the last visit was 4.2 years
(interquartile range, 2.8-6.0).

Prevalent CAC by short-term change in glycemic
status: a cross-sectional analysis
We evaluated the prevalence of CAC, defined as an Agatston
score > 0 at the time of the first CT scan according to the
changes in glycemic status between the first CAC CT scan
and the prior visit. Specifically, the changes in glycemic status
indicate glycemic status at the visit preceding the baseline [and
first] CAC CT scan [CACS-1] and the glycemic status at the
first CAC CT scan [CACS-1]) (Table 2). The prevalence rates
of CAC were 12% and 43% higher for individuals with per­
sistent prediabetes and for those who had progressed from
prediabetes to diabetes, respectively, than that for those with
persistent normoglycemia (reference group), after adjustment
for cardiovascular risk factors and other confounders. There
was a trend toward a higher prevalence of CAC in the group
that changed its glycemic status from prediabetes to normo­
glycemia, but this association was attenuated and no longer
statistically significant (OR, 1.05; 95% CI, 0.98-1.12) after
adjusting for confounders.
The group that progressed from normoglycemia to predia­
betes or diabetes tended to have a higher prevalence of CAC;
however, these associations did not reach statistical signifi­
cance. In a sensitivity analysis using the Tobit regression mod­
el (Table S1)25 with log (CACS + 1) as the dependent variable,
the relationship between a change in glycemic status and the
CACS was similarly observed.
CAC progression by short-term change in glycemic
status: a prospective analysis
In the prospective analysis (Table 3), we further evaluated the
CAC progression rates between the first CT scan and the
follow-up CT scan according to the changes in the glycemic
status (as defined above) during a median follow-up of 4.2
years. The risks of 5-year CAC progression, estimated as the
ratios of the 5-year CAC progression rates, were 19%, 63%,
and 11% higher for individuals with persistent prediabetes,
those who progressed from prediabetes to diabetes, and those
who regressed from prediabetes to normoglycemia, compared
with that for patients who had persistent normoglycemia (ref­
erence group), even after adjusting for changes in cardiovascu­
lar risk factors over time as time-varying covariates. However,
compared with that of the persistent prediabetes group, the re­
gression from prediabetes to normoglycemia group showed a
7% lower ratio of the 5-year CAC progression rate (the ratio
of the 5-year progression rates, 0.93; 95% CI, 0.90-0.96)
(Table S2).25
Only 2 patients progressed from normoglycemia to overt
diabetes, and therefore insufficient to estimate significance.
Discussion
In this large-scale study of over 126 000 young and middle-
aged Korean adults without diabetes at the first visit, persistent
prediabetes and progression from prediabetes to diabetes were
both significantly associated with a higher prevalence of sub­
clinical atherosclerosis measured using CACS (than that ob­
served in the persistent normoglycemia group) in the
cross-sectional analysis. In the subsample of participants
with repeated CT measurements, the ratios of estimated
5-year CAC progression rates were significantly higher among
people with persisting prediabetes than among those with per­
sistent normoglycemia in the prospective analysis, even after
adjusting for time-varying cardiovascular risk factors and oth­
er confounders. This key analysis allowed us to adjust for
change over time in potential confounding cardiovascular
risk factors between the baseline and follow-up.
Table 1. General characteristics of study participants at visit 2 by glycemic status category at visits 1 and 2a (n = 126 765).

Characteristics Overall Normal at visit 1 Prediabetes glucose at visit 1

Normal at visit 2 Prediabetes at visit 2 Diabetes at visit 2 Normal at visit 2 Prediabetes at visit 2 Diabetes at visit 2
Number of participants 126 765 53 807 18 952 51 14 627 37 857 1471
Age (years)b 41.3 (7.0) 39.9 (6.3) 41.1 (6.9) 43 (8.8) 41.1 (6.7) 43.2 (7.5) 44.6 (8.4)
Yoosun Cho et al.

Male (%) 74.7 70.4 77.7 64.7 72.3 79.8 83.5

Seoul center(%) 47.0 48.2 48.0 41.2 44.2 46.1 46.2
Current smoker (%) 21.8 19.2 23.1 26.0 21.0 24.8 28.1
Alcohol intake (%)c 43.2 38.6 45.8 39.1 42.3 48.4 53.4
HEPA (%)d 15.8 15.0 16.0 8.9 16.9 16.4 15.5
Education level (%)e 85.1 87.1 85.0 76.5 84.1 82.7 80.3
Medication for hypertension 5.8 3.3 6.0 17.7 5.1 9.1 16.4
Family history of CVD(%) 12.6 11.8 12.7 11.8 12.2 13.8 13.0
Lipid-lowering medication 3.31 1.86 3.36 1.96 2.68 5.43 7.12
Obesity (%) 38.3 28.9 41.8 72.6 34.9 49.7 73.2
Fatty liver(%) 39.08 28.05 42.47 74.51 34.97 52.95 81.8
BMI (kg/m2)b 24.3 (3.3) 23.5 (3.0) 24.6 (3.3) 28.3 (5.7) 24 (3.1) 25.2 (3.4) 27.5 (4.0)
Systolic BP (mmHg)b 112.2 (12.4) 109.9 (11.8) 113.7 (12.4) 121.9 (15.7) 111.2 (12.2) 115 (12.6) 121.1 (12.8)
Diastolic BP (mmHg)b 72.7 (9.8) 70.9 (9.3) 73.7 (9.7) 79.5 (11.4) 71.9 (9.6) 74.9 (9.9) 79.3 (10.3)
Glucose (mg/dL)b 95.9 (9.7) 90.7 (5.4) 98.7 (7.4) 130.1 (21.7) 92.7 (4.9) 101.6 (8.5) 132.2 (28.2)
eGFR(mL/min) 90.9 (15.5) 92.4 (15.7) 90.2 (15.3) 90.3 (17.2) 91.2 (15.6) 89.1 (15.2) 89.8 (16.5)
HbA1c (mg/dL)b 5.5 (0.3) 5.3 (0.2) 5.6 (0.2) 6.6 (1.1) 5.5 (0.2) 5.8 (0.2) 6.7 (1.0)
Total cholesterol (mg/dL)b 198.5 (34.0) 193.2 (32.3) 200.5 (33.7) 205.6 (44.8) 197.7 (33.3) 204.7 (35.0) 211.5 (40.2)
LDL-C (mg/dl)b 129.4 (31.7) 125.1 (30.7) 131.1 (31.5) 134.1 (40.6) 128.5 (31.3) 134.7 (32.3) 140 (36.0)
HDL-C (mg/dL)b 55.9 (14.7) 58.1 (15.2) 55 (14.2) 49.1 (14.0) 56.6 (14.8) 53.1 (13.6) 48.2 (12)
Triglycerides (mg/dL)f 110 (77-160) 96 (69-139) 117 (83-167) 150 (109-222) 103 (73-148) 128 (90-185) 173 (120-245)
ALT (U/L)f 21 (15-31) 19 (14-27) 22 (16-33) 35 (23-87) 20 (14-29) 24 (17-36) 38 (24-64)
AST (U/L)f 20 (17-25) 19 (16-24) 21 (17-26) 26 (21-48) 20 (17-24) 21 (18-27) 27 (21-40)
GTP (U/L)f 25 (16-42) 21 (15-34) 28 (18-46) 42 (24-73) 23 (15-38) 32 (20-53) 54 (33-88)
hs-CRP (mg/L)f 0.5 (0.3-1.0) 0.4 (0.3-0.8) 0.5 (0.3-1.0) 1.4 (0.7-3.2) 0.5 (0.3-0.9) 0.6 (0.3-1.2) 1.1 (0.6-2.4)
HOMA-IRf 1.44 (0.96-2.14) 1.23 (0.83-1.76) 1.65 (1.11-2.42) 5.11 (2.86-7.58) 1.29 (0.89-1.83) 1.78 (1.19-2.61) 3.42 (2.23-5.38)
CAC >0 (%) 11.6 8.1 11.3 15.7 10.5 16.6 26.9
CACS if CAC >0 17 (5-54) 14 (4-43) 17 (5-56) 38 (30-129) 16 (4-48) 20 (5-63) 18 (5-70)

Abbreviations: ALT, alanine aminotransferase; BP, blood pressure; CAC, coronary artery calcification; CACS, coronary artery calcium score; CVD, cardiovascular disease; GGT, γ-glutamyltransferase; HEPA, health-
enhancing physical activity; HDL-C, high-density lipoprotein-cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; hs-CRP, high-sensitivity C-reactive protein; LDL-C, low-density
lipoprotein-cholesterol.
a
Visit 1; the visit preceding the baseline (and first) CAC CT scan (CACS-1), visit 2; the visit at the first CAC CT scan (CACS-1).
b
Data are means (SD).
c
≥20 g of ethanol per day.
d
HEPA as meeting either of two criteria: (1) vigorous intensity activity on ≥3 days/week accumulating ≥1500 metabolic equivalent (MET) min/week or (2) 7 days with any combination of walking, moderate intensity, or
vigorous intensity activities, reaching at least 3000 MET min/week.
e
≥College graduate.
f
Median (interquartile range).
5

Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024

6 European Journal of Endocrinology, 2023, Vol. 188, No. 1

Table 2. Cross-sectional analysis; the absolute and relative prevalence of coronary artery calcificationa at the time of the first coronary artery calcium CT
scan according to the changes in glycemic status between the 2 visits: preceding the baseline (and first) CAC CT scan (CACS-1) and at the first CAC CT scan
(CACS-1) (n = 126 765).

Glycemic status Total number CAC >0 (%) Age- and sex-adjusted OR (95% CI) Multivariable-adjusted OR
Visit 1 Visit 2
Normal Normal 53 807 4331 8.1 1.00 (reference) 1.00 (reference)
Normal Prediabetes 18 952 2139 11.3 1.16 (1.09-1.23) 1.01 (0.95-1.07)
Normal Diabetes 51 8 15.7 1.42 (0.62-3.25) 0.98 (0.42-2.25)
Prediabetes Normal 14 627 1534 10.5 1.13 (1.05-1.20) 1.05 (0.98-1.12)
Prediabetes Prediabetes 37 857 6264 16.6 1.40 (1.33-1.46) 1.13 (1.08-1.18)
Prediabetes Diabetes 1471 396 26.9 2.22 (1.95-2.54) 1.46 (1.27-1.67)

Abbreviations: CAC, coronary artery calcification; CACS, coronary artery calcium score; CI, confidence interval; OR, odds ratio.
a
Estimated from binomial logistic regression models. The multivariable model was age, sex, center, year of screening examination, smoking status, alcohol
intake, physical activity, education level, medication for hypertension, lipid-lowering medication, BMI, LDL-C, and SBP at baseline.

Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024

Importantly, these data showed that compared with the group
with persistent prediabetes, those who reverted from predia­
betes to normoglycemia had a significantly reduced 5-year
progression rate of CAC, supporting the potential benefit of
reversion from prediabetes to normoglycemia, in preventing
the progression of atherosclerosis.
In our study, persistent prediabetes, in the period before the
first CACS, was positively associated with both the presence of
CAC on the first CT scan and CAC progression between base­
line and follow-up CT scans compared with the group with
persistent normoglycemia. Prediabetes or mildly increased
glucose concentration is an important determinant of the
future risk of subclinical atherosclerosis, including CAC
progression.26,27 Epidemiological evidence supports a con­
tinuous relationship between glycemic parameters and CVD,
similar to the relationship between cholesterol and blood pres­
sure.28-30 A longer duration of hyperglycemia has been linked
to a higher risk of subclinical atherosclerosis.31,32 The cutoff
for prediabetes can be arbitrarily drawn but is widely used
in clinical practice. We classified the participants according
to the definition of prediabetes widely used in clinical prac­
tice.21 It is important to understand the clinical and prognostic
implications for patients classified as having prediabetes, as
many clinicians can make this diagnosis using the HbA1c
measurement. Furthermore, the impact of persistent prediabe­
tes on CAC over a period of time has not been clearly under­
stood, with only a few studies examining this association.
According to the CARDIA study, a long-term cohort study
of approximately 3600 black and white young adults without
prediabetes or diabetes at baseline, the duration of prediabetes
or diabetes estimated during a 25-year period was associated
with the presence of CAC, highlighting that cumulative expos­
ure to chronic hyperglycemia is associated with an increased
risk of subclinical atherosclerosis.32
In our study, individuals with progression from normogly­
cemia to prediabetes showed a higher risk of CAC progression
than those with persistent normoglycemia during our relative­
ly short-term follow-up period in which we assessed changes
in glycemic status. Importantly, these data suggest that even
over this relatively short period in which we assessed changes
in glycemic status, there is evidence that exposure to prediabe­
tes has an adverse effect on the progression of subclinical ath­
erosclerosis. Our data therefore add to previous findings by
revealing that prediabetes, even when persistent for a relative­
ly short period, may lead to an increased risk of CAC
progression.
Several previous studies have reported the impact of
changes in glycemic status on CVD and all-cause mortality.
However, few studies have specifically focused on the associ­
ation between changes in glycemic status and the presence or
progression of subclinical atherosclerosis. The Whitehall II co­
hort study found that reversion from 2-hour glucose-defined
prediabetes to normoglycemia was associated with a de­
creased risk of CVD and mortality, although individuals re­
verting from fasting glucose- or HbA1c-defined prediabetes
to normoglycemia were not at reduced risk of future CVD
or death, compared with those with prediabetes, or those
who progressed to diabetes.33 These study results also

Table 3. Prospective analysis; the association between changes in glycemic status and 5-year progression rates of coronary artery calcium scorea between
the first CAC CT scan and the subsequent CAC CT scan (n = 40 622).

Glycemic status Total number 5-year progression rates of CAC Ratio of 5-year progression rate
Visit 1 Visit 2 Age- and sex-adjusted model Multivariable-adjusted model
Normal Normal 14 802 1.31 (1.29-1.33) 1.00 (reference) 1.00 (reference)
Normal Prediabetes 4679 1.43 (1.40-1.46) 1.09 (1.06-1.12) 1.09 (1.06-1.12)
Normal Diabetes 2 - - -
Prediabetes Normal 4926 1.45 (1.42-1.49) 1.11 (1.07-1.14) 1.11 (1.07-1.14)
Prediabetes Prediabetes 14 420 1.57 (1.54-1.59) 1.19 (1.16-1.22) 1.19 (1.16-1.22)
Prediabetes Diabetes 58 2.14 (1.66-2.77) 1.63 (1.26-2.12) 1.63 (1.26-2.10)

Abbreviation: CAC, coronary artery calcification.

a
Five-year rates of CAC progression and the ratio of 5-year progression rates were estimated using mixed models with random intercepts and random slopes
with natural log(CAC + 1) as the outcome and inverse probability weighting. The multivariable model included glycemic status change as well as smoking status,
alcohol intake, physical activity, lipid-lowering medication, medication for hypertension, BMI, LDL-C, and SBP as time-varying variables, and age at baseline,
sex, center, year of screening examination, education level, and time-fixed variables.
Yoosun Cho et al.
supported the notion, raised over 20 years ago, that 2-hour
glucose levels in the non-diabetic range are a stronger deter­
minant of all-cause and cardiovascular mortality than fasting
glucose or HbA1c levels.34 However, in contrast to this no­
tion, a recent prospective cohort study from China reported
that reversion from fasting glucose-defined prediabetes to
normoglycemia over 2 years was associated with a reduction
in the risk of CVD and all-cause mortality, compared with a
reference group who progressed to diabetes.35 Taken together,
these findings appear to indicate a benefit of reversion from
prediabetes to normoglycemia (irrespective of the definition
used) in CVD risk reduction, although prior to our study, no
previous study has evaluated CAC progression as an outcome.
We found that individuals who reverted from prediabetes (de­
fined by fasting glucose or HbA1c concentrations) to normo­
glycemia had a lower risk of CAC progression than those
with persistent prediabetes. Our study is therefore the first to
document that normalization of the glycemic state from pre­
diabetes may potentially help reduce the risk of subclinical
atherosclerosis. While there is ongoing debate regarding the
role of prediabetes on the risk of CVD,36 our findings lend fur­
ther support to the importance of prediabetes management as
a strategy for minimizing cardiovascular complications.
Therefore, we recommend that additional longitudinal cohort
studies with longer follow-up duration are needed to confirm
our findings.
There are some inherent limitations in our study. First, we
could not use 2-hour glucose levels to define the glycemic sta­
tus. Since our data set consists of the participants’ laboratory
data obtained in the fasting state only, and 2-hour glucose lev­
els were not available. Second, there was a lack of information
on CAC density or volume in our database. A more refined
measurement of CAC will provide a better understanding of
the association between changes in the glycemic state and
CAC progression. Third, even though a wide range of covari­
ates at baseline and follow-up was adjusted in the models, there
is an inherent possibility of residual confounding from the
measured CVD risk factors and confounding from the unmeas­
ured CVD risk factors. Fourth, the 1.5-year time interval (inter­
quartile range, 1.0-2.0) between the 2 glycemic status
measurements to define the change in glycemic status may be
relatively short to determine these changes. However, accord­
ing to a randomized controlled trial that assessed whether a
lifestyle-intervention program or the use of metformin would
prevent or delay the development of diabetes, a reduction
was observed in the mean fasting plasma glucose and HbA1c
levels in the metformin and lifestyle-intervention groups in
the first year of the trial.37 Therefore, it is plausible that a
1.5-year interval between the 2 glycemic status measurements
in our study may be sufficient to observe a change in the risk
of CAC progression. Finally, our findings may not be general­
izable to populations of different ethnicities; and previous re­
ports have noted ethnic differences in HbA1c trajectories.38
Moreover, only one-third of the participants were included in
our prospective analysis because the remainder did not under­
go a second CAC CT scan. This could limit the generalizability
of our findings (see Supplemental Data for more details).
However, the current study has important strengths. This
study represents a large sample of Korean participants of a sin­
gle ethnicity who have undergone detailed phenotyping with
baseline and follow-up validated measurements of the CACS
by high-resolution CT scanning. We also considered the
change in CVD risk factors between the baseline and follow-
7
up and took account of this in our analyzes by adjusting for
time-varying covariates in the analyzes.
In conclusion, we have demonstrated that individuals with
adverse changes in glycemic status, including persistent pre­
diabetes or development of new prediabetes, over a 1- to
2-year period preceding a measurement of CACS, have an in­
creased risk of subclinical atherosclerosis. Moreover, the risk
of CAC progression for individuals in whom prediabetes re­
gressed to normoglycemia was intermediate between that of
the persistent normoglycemia and persistent prediabetes
groups. Our findings consistently support prediabetes per se
as an independent predictor for subclinical atherosclerosis
and its progression.
Prediabetes is associated with CAC and the presence of
CAC identifies subjects at increased risk of CVD.8,9
Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024
Assessment of progression of subclinical atherosclerosis using
CACS is useful in clinical practice because measurement of
CAC improves CVD risk prediction, over and above tradition­
al cardiovascular risk factors.19 However, prior to our study,
it has been uncertain whether persistence or regression of pre­
diabetes influences CAC progression. Herein, we provide evi­
dence that persistent prediabetes represents a high-risk state
for atherosclerotic progression, whereas in contrast an im­
provement in glycemia from prediabetes to normoglycemia,
somewhat mitigates the risk of progression. Our findings em­
phasize that identifying and managing conventional cardio­
vascular risk factors in asymptomatic people with
prediabetes could reduce progression to overt CVD.
Acknowledgments
We thank our staff members at the Kangbuk Samsung Health
Study for their hard work, dedication, and continuing support.
Author contributions
Y.C., Y.C., S.R., and C.D.B. planned, designed, and imple­
mented the study, including quality assurance and control.
S.R. analyzed the data and designed the study’s analytical
strategy. Y.C. and S.R. supervised field activities. Y.C., S.R.,
Y.C., Y.K., I.C., C.W.K., H.J., and H.O. conducted the litera­
ture review and prepared the Methods and Discussion sections
of the text. Y.C. and Y.C. drafted the manuscript. All authors
interpreted the results, and S.R., S.H.W., and C.D.B. contrib­
uted to the critical revisions of the manuscript. All authors ap­
proved the final version of this manuscript.
Supplementary material
Supplementary material is available at European Journal of
Endocrinology online.
Funding
This study was supported by the SKKU Excellence in Research
Award Research Fund, Sungkyunkwan University, 2020,
and by the National Research Foundation of Korea, funded
by the Ministry of Science, ICT, and Future Planning
(NRF-2021R1A2C1012626). C.D.B. was supported in part
by the Southampton National Institute for Health Research
Biomedical Research Centre (IS-BRC-20004), UK.
Conflicts of interest: All authors declare that they have no con­
flict of interest.
8 European Journal of Endocrinology, 2023, Vol. 188, No. 1
Data availability
The data are not publicly available outside of the hospital be­
cause of Institutional Review Board restrictions (the data were
not collected in a way that could be distributed widely).
However, the analytical methods are available from the corre­
sponding author upon request.
References
1. Makaroff LE. The need for international consensus on prediabetes.
Lancet Diabetes Endocrinol. 2017;5(1):5-7. https://doi.org/10.
1016/S2213-8587(16)30328-X
2. American Diabetes Association. 3. Prevention or delay of type 2 dia­
betes: standards of medical care in diabetes-2020. Diabetes Care.
2020;43(Suppl_1):S32-S36. https://doi.org/10.2337/dc20-S003
3. Tabak AG, Herder C, Rathmann W, Brunner EJ, Kivimaki M.
Prediabetes: a high-risk state for diabetes development. Lancet.
2012;379(9833):2279-2290. https://doi.org/10.1016/S0140-6736(12)
60283-9
4. Echouffo-Tcheugui JB, Narayan KM, Weisman D, Golden SH, Jaar
BG. Association between prediabetes and risk of chronic kidney dis­
ease: a systematic review and meta-analysis. Diabet Med. 2016;33­
(12):1615-1624. https://doi.org/10.1111/dme.13113
5. Huang Y, Cai X, Mai W, Li M, Hu Y. Association between predia­
betes and risk of cardiovascular disease and all cause mortality: sys­
tematic review and meta-analysis. BMJ. 2016;355 (8083):i5953.
https://doi.org/10.1136/bmj.i5953
6. Schlesinger S, Neuenschwander M, Barbaresko J, et al. Prediabetes
and risk of mortality, diabetes-related complications and comorbid­
ities: umbrella review of meta-analyses of prospective studies.
Diabetologia. 2022;65(2):275-285. https://doi.org/10.1007/
s00125-021-05592-3
7. Mutie PM, Pomares-Millan H, Atabaki-Pasdar N, et al. An investi­
gation of causal relationships between prediabetes and vascular
complications. Nat. Commun. 2020;11(1):4592. https://doi.org/
10.1038/s41467-020-18386-9
8. Acar B, Ozeke O, Karakurt M, et al. Association of prediabetes with
higher coronary atherosclerotic burden among patients with first di­
agnosed acute coronary syndrome. Angiology. 2019;70(2):
174-180. https://doi.org/10.1177/0003319718772420
9. Amano T, Matsubara T, Uetani T, et al. Abnormal glucose regula­
tion is associated with lipid-rich coronary plaque: relationship to in­
sulin resistance. JACC Cardiovasc Imaging. 2008;1(1):39-45.
https://doi.org/10.1016/j.jcmg.2007.09.003
10. Bergman M. Pathophysiology of prediabetes and treatment implica­
tions for the prevention of type 2 diabetes mellitus. Endocrine.
2013;43(3):504-513. https://doi.org/10.1007/s12020-012-9830-9
11. Ferrannini E, Gastaldelli A, Iozzo P. Pathophysiology of prediabe­
tes. Med Clin North Am. 2011;95(2):327-339, vii-viii. https://doi.
org/10.1016/j.mcna.2010.11.005
12. Xing FYF, Neeland IJ, Gore MO, et al. Association of prediabetes
by fasting glucose and/or haemoglobin A1c levels with subclinical
atherosclerosis and impaired renal function: observations from
the Dallas Heart Study. Diab Vasc Dis Res. 2014;11(1):11-18.
https://doi.org/10.1177/1479164113514239
13. Won K-B, Han D, Lee JH, et al. Evaluation of the impact of glycemic
status on the progression of coronary artery calcification in asymp­
tomatic individuals. Cardiovasc Diabetol. 2018;17(1):4. https://doi.
org/10.1186/s12933-017-0653-0
14. Park GM, Cho YR, Lee SW, et al. Prediabetes is not a risk factor for
subclinical coronary atherosclerosis. Int J Cardiol. 2017;243:
479-484. https://doi.org/10.1016/j.ijcard.2017.05.073
15. Jin C, Chen S, Vaidya A, et al. Longitudinal change in fasting blood
glucose and myocardial infarction risk in a population without dia­
betes. Diabetes Care. 2017;40(11):1565-1572. https://doi.org/10.
2337/dc17-0610
16. Lee G, Kim SM, Choi S, et al. The effect of change in fasting glucose
on the risk of myocardial infarction, stroke, and all-cause mortality:
a nationwide cohort study. Cardiovasc Diabetol. 2018;17(1):51.
https://doi.org/10.1186/s12933-018-0694-z
17. Yeboah J, McClelland RL, Polonsky TS, et al. Comparison of novel
risk markers for improvement in cardiovascular risk assessment in
intermediate-risk individuals. JAMA. 2012;308(8):788-795.
https://doi.org/10.1001/jama.2012.9624
18. Shaw LJ, Raggi P, Schisterman E, Berman DS, Callister TQ.
Prognostic value of cardiac risk factors and coronary artery calcium
screening for all-cause mortality. Radiology. 2003;228(3):826-833.
https://doi.org/10.1148/radiol.2283021006
19. Budoff MJ, Hokanson JE, Nasir K, et al. Progression of coronary
artery calcium predicts all-cause mortality. JACC Cardiovasc
Imaging. 2010;3(12):1229-1236. https://doi.org/10.1016/j.jcmg.
2010.08.018
20. Chang Y, Ryu S, Sung KC, et al. Alcoholic and non-alcoholic fatty
liver disease and associations with coronary artery calcification: evi­
Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024
dence from the Kangbuk Samsung Health Study. Gut. 2019;68(9):
1667-1675. https://doi.org/10.1136/gutjnl-2018-317666
21. American Diabetes Association. Diagnosis and classification of dia­
betes mellitus. Diabetes Care. 2011;34(Suppl_1):S62-S69. https://
doi.org/10.2337/dc11-S062
22. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M
Jr, Detrano R. Quantification of coronary artery calcium using ul­
trafast computed tomography. J Am Coll Cardiol. 1990;15(4):
827-832. https://doi.org/10.1016/0735-1097(90)90282-T
23. Reilly MP, Wolfe ML, Localio AR, Rader DJ. Coronary artery cal­
cification and cardiovascular risk factors: impact of the analytic ap­
proach. Atherosclerosis. 2004;173(1):69-78. https://doi.org/10.
1016/j.atherosclerosis.2003.10.010
24. Gassett AJ, Sheppard L, McClelland RL, et al. Risk factors for long-
term coronary artery calcium progression in the multi-ethnic study
of atherosclerosis. J Am Heart Assoc. 2015;4(8):e001726. https://
doi.org/10.1161/JAHA.114.001726
25. Cho Y, Ryu S, Kim Y, et al. Supplementary data for “glycemic status
changes and the progression of coronary artery calcification among
adults without diabetes”. figshare, 2022.
26. McNeely MJ, McClelland RL, Bild DE, et al. The association be­
tween A1C and subclinical cardiovascular disease: the multi-ethnic
study of atherosclerosis. Diabetes Care. 2009;32(9):1727-1733.
https://doi.org/10.2337/dc09-0074
27. Moebus S, Stang A, Mohlenkamp S, et al. Association of impaired
fasting glucose and coronary artery calcification as a marker of sub­
clinical atherosclerosis in a population-based cohort–results of the
Heinz Nixdorf Recall Study. Diabetologia. 2009;52(1):81-89.
https://doi.org/10.1007/s00125-008-1173-y
28. Gerstein HC. Glucose: a continuous risk factor for cardiovascular
disease. Diabet Med. 1997;14(S3):S25-S31. https://doi.org/10.
1002/(SICI)1096-9136(199708)14:3+<S25::AID-DIA441>3.0.
CO;2-1
29. Khaw KT, Wareham N, Bingham S, Luben R, Welch A, Day N.
Association of hemoglobin A1c with cardiovascular disease and
mortality in adults: the European prospective investigation into can­
cer in Norfolk. Ann Intern Med. 2004;141(6):413-420. https://doi.
org/10.7326/0003-4819-141-6-200409210-00006
30. Park C, Guallar E, Linton JA, et al. Fasting glucose level and the risk
of incident atherosclerotic cardiovascular diseases. Diabetes Care.
2013;36(7):1988-1993. https://doi.org/10.2337/dc12-1577
31. Kim J-J, Hwang B-H, Choi IJ, et al. Impact of diabetes duration on
the extent and severity of coronary atheroma burden and long-term
clinical outcome in asymptomatic type 2 diabetic patients: evalu­
ation by coronary CT angiography. Eur Heart J Cardiovasc
Imaging. 2015;16(10):1065-1073. https://doi.org/10.1093/ehjci/
jev106
32. Reis JP, Allen NB, Bancks MP, et al. Duration of diabetes and pre­
diabetes during adulthood and subclinical atherosclerosis and car­
diac dysfunction in middle age: the CARDIA study. Diabetes
Care. 2018;41(4):731-738. https://doi.org/10.2337/dc17-2233
33. Vistisen D, Kivimaki M, Perreault L, et al. Reversion from predia­
betes to normoglycaemia and risk of cardiovascular disease and
Yoosun Cho et al.
mortality: the Whitehall II cohort study. Diabetologia. 2019;62(8):
1385-1390. https://doi.org/10.1007/s00125-019-4895-0
34. de Vegt F, Dekker JM, Ruhé HG, et al. Hyperglycaemia is associ­
ated with all-cause and cardiovascular mortality in the Hoorn
population: the Hoorn Study. Diabetologia. 1999;42(8):926-931.
https://doi.org/10.1007/s001250051249
35. Liu X, Wu S, Song Q, Wang X. Reversion from pre-diabetes
mellitus to normoglycemia and risk of cardiovascular disease and
all-cause mortality in a Chinese population: a prospective cohort
study. J Am Heart Assoc. 2021;10(3):e019045. https://doi.org/10.
1161/JAHA.120.019045
9
36. Yudkin JS, Montori VM. The epidemic of pre-diabetes: the medi­
cine and the politics. BMJ. 2014;349 (7967):g4485. https://doi.
org/10.1136/bmj.g4485
37. 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(6):393-403. https://doi.org/10.1056/
NEJMoa012512
38. Miller KM. Racial differences in trajectories of hemoglobin A1c:
further evidence of gaps in care. JAMA Network Open.
2018;1(5):e181882. https://doi.org/10.1001/jamanetworkopen.
2018.1882
Downloaded from https://academic.oup.com/ejendo/article/188/1/1/6979715 by guest on 12 April 2024