Вы находитесь на странице: 1из 12

Journal of Diabetes 10 (2018), 641–652

ORIGINAL ARTICLE

Utility of three novel insulin resistance-related lipid


indices for predicting type 2 diabetes mellitus among
people with normal fasting glucose in rural China

Highlights
• Data are limited regarding the prevention of type 2 diabetes mellitus (T2DM) in rural Chinese people.
• Elevated triglyceride glucose, visceral adiposity, and lipid accumulation product are significantly associated
with the risk of incident T2DM.
• Triglyceride glucose, visceral adiposity, and lipid accumulation product did not improve the T2DM prediction
compared with fasting plasma glucose or waist circumference.

Bingyuan WANG,1 Ming ZHANG,2 Yu LIU,3 Xizhuo SUN,3 Lu ZHANG,1 Chongjian WANG,1 Linlin LI,1
Yongcheng REN,1,2,3 Chengyi HAN,1,2,3 Yang ZHAO,1,2,3 Junmei ZHOU,2,3 Chao PANG,4 Lei YIN,4
Tianping FENG,4 Jingzhi ZHAO4 and Dongsheng HU1
1
Department of Epidemiology and Health Statistics, College of Public Health, Zhengzhou University, Zhengzhou, China, 2Department of
Preventive Medicine, Shenzhen University Health Science Center, Shenzhen, China, 3The Affiliated Luohu Hospital of Shenzhen University
Health Science Center, Shenzhen, China, and 4Department of Prevention and Health Care, Military Hospital of Henan Province, Zhengzhou,
China

Correspondence Abstract
Dongsheng Hu, Department of
Epidemiology and Health Statistics, Background: Inexpensive and easily measured indices are needed for the early
College of Public Health, Zhengzhou prediction of type 2 diabetes mellitus (T2DM) in rural areas of China. The aim
University, 100 Kexue Avenue,
of this study was to compare triglyceride glucose (TyG), visceral adiposity
Zhengzhou, Henan 450001, China.
Tel: +86 371 6778 1963
(VAI), and lipid accumulation product (LAP) with traditional individual mea-
Fax: +86 371 8667 1963 sures and their ratios for predicting T2DM.
Email: dongsheng_hu@zzu.edu.cn Methods: Data for 11 113 people with baseline normal fasting glucose in a
rural Chinese cohort were followed for a median of 6.0 years. Cox propor-
tional hazards regression was used to calculate covariate-adjusted hazard ratios
(aHRs) and 95% confidence intervals (95% CIs) and receiver operating charac-
Received 10 July 2017; revised 17
December 2017; accepted 7
teristic analysis was used to compare the ability of traditional measures and
January 2018. TyG, VAI, and LAP at baseline to predict T2DM at follow-up.
Results: Among individual measures, fasting plasma glucose (FPG) and waist
doi: 10.1111/1753-0407.12642 circumference (WC) were strongly associated with T2DM. Of all lipid ratios, an
elevated triglycerides (TG) to high-density lipoprotein cholesterol (HDL-C)
ratio was associated the most with T2DM. Compared with the first quartiles of
TyG, VAI, and LAP, their fourth quartiles were associated with T2DM for men
(aHR 3.54 [95% CI 2.08–6.03], 2.89 [1.72–4.87], and 5.02 [2.85–8.85], respec-
tively) and women (6.15 [3.48–10.85], 4.40 [2.61–7.42], and 6.49 [3.48–12.12],
respectively). For predicting T2DM risk, TyG, VAI, and LAP were mostly supe-
rior to the TG: HDL-C ratio, but did not differ from FPG and WC.
Conclusions: Prediction of T2DM was not improved by TyG, VAI, and LAP
versus FPG or WC alone. Therefore, TyG, VAI, and LAP may not be inex-
pensive tools for predicting T2DM in rural Chinese people.
Keywords: cohort study, diabetes, fat accumulation, insulin resistance, lipid
index.

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 641
Diabetes prevention in China B. WANG et al.

Introduction However, the ability of TyG and VAI to predict T2DM


incidence is contentious compared with that of other
China is home to the largest number of adults living
fatness indices.4,14–16,18 In addition, there are limited
with type 2 diabetes mellitus (T2DM) worldwide.1 In
data regarding the association between LAP and diabe-
the latest national survey of China in 2010, although
tes.17 These indices should be validated in different eth-
the prevalence of T2DM was lower for rural than urban
nic groups.
residents (10.3% vs 14.3%), that of prediabetes was
Therefore, in the present study we conducted a
greater for rural than urban residents (50.9% vs 48.4%), population-based prospective cohort study of rural Chi-
and the rates were lower for rural than urban residents nese participants with normal FPG at baseline to:
for awareness (24.6% vs 38.7%), treatment (21.5% vs (i) evaluate the relationship between the TyG, LAP,
32.7%), and control (38.6% vs 40.8%) of T2DM.2 Fur- and VAI indices and risk of incident T2DM; and
thermore, household income and health care expendi- (ii) compare the ability of TyG, LAP, VAI, and both
ture are lower in rural than urban areas of China.3 In traditional individual measures (FPG, WC, total choles-
terms of the epidemic state of T2DM for rural residents, terol [TC], TG, HDL-C, low-density lipoprotein choles-
the situation will be grim if we do not establish immedi- terol [LDL-C]) and lipid ratios (TG: HDL-C,
ate control measures. Therefore, we must explore inex- TC: HDL-C, TC: LDL-C, and LDL-C: HDL-C
pensive and easily measurable indices for the early ratios) to predict the risk of incident T2DM.
prediction of T2DM in rural areas.
Insulin resistance (IR) and/or visceral adiposity are
well-known risk factors for T2DM in the general popu- Methods
lation.4 People with T2DM often show prior impaired
fasting glucose and features of IR.5 In addition, ectopic Study population
fat deposition in the abdominal viscera and liver can Between July–August 2007 and July–August 2008,
result in inflammatory and adipokine dysregulation, as 20 194 participants age ≥ 18 years were recruited using
well as IR, which may accelerate the development of a cluster sampling method from a rural Chinese popula-
T2DM.6 For these reasons, identifying IR and visceral tion in Henan province and underwent baseline exami-
adiposity can be helpful in the primary and secondary nations; of these, 17 265 were re-examined in July–
prevention of T2DM. August 2013 and July–October 2014 (response rate
The hyperinsulinemic–euglycemic clamp is considered 85.5%). Details of the study design and characteristics
the gold standard to define IR,7 and imaging techniques of the study population have been described
such as magnetic resonance imaging and computed elsewhere.19–21 Participants were excluded from the
tomography can measure fat distribution and quantify study if they: (i) had T2DM (n = 1302) or type 1 diabe-
adiposity accurately.8 However, none of these tech- tes mellitus (n = 13) at baseline; (ii) had impaired fast-
niques is cheap or can be used in large epidemiological ing glucose (IFG; FPG 6.1–6.9 mmol/L;22 n = 1122) or
studies. Thus, we need indicators that are easily mea- unknown diabetic status (n = 10) at baseline; or
sured and widely applicable with a high ability to pre- (iii) had died (n = 1110) or did not have biochemical
dict to T2DM for the general population. data (n = 2583) during follow-up. Those who did not
Several studies have reported using the triglyceride have biochemical data primarily because they had
glucose (TyG) index, the product of triglycerides moved to another place were investigated by telephone
(TG) and fasting plasma glucose (FPG) levels, as a sur- follow-up. The final sample for the present analysis con-
rogate measure to identify IR.9 Compared with the sisted of 11 113 people with baseline normal fasting glu-
hyperinsulinemic–euglycemic clamp, TyG showed high cose (6844 women).
sensitivity (96.5%) and specificity (85%) for the diagno- This study was approved by the Shenzhen University
sis of IR in Mexican people.9 In addition, previous Medical Ethics Committee. All procedures were per-
studies proposed another two reliable lipid indices to formed in accordance with the ethical standards of the
estimate IR from visceral adiposity dysfunction, one responsible committee on human experimentation and
being the visceral adiposity index (VAI), based on body with the 1975 Declaration of Helsinki and its later
mass index (BMI), waist circumference (WC), TG and amendments.
high-density lipoprotein cholesterol (HDL-C) levels,10,11
and the other being the lipid accumulation product
(LAP), which is calculated from WC and fasting TG Data collection and measurements
levels.12 Several studies have reported significant associ- Data were collected in examination centers at local
ations between TyG, VAI, LAP, and T2DM.4,13–18 health stations or in community clinics in the

642 © 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
B. WANG et al. Diabetes prevention in China

participants’ residential areas. Participants who were VAI, and LAP were calculated using the following
unable to attend the examination center were examined formulas:
in their home. Trained research staff administered a
standard questionnaire to collect information on sex, TyG = ln ðTG × FPG=2Þ:
age, education level, marital status, smoking status,
alcohol intake, and physical activity. Participants who with TG and FPG both in mg/dL.9
had not smoked >100 cigarettes in their lifetime were    
considered never-smokers; the others were considered WC TG
VAI ðmenÞ = ×
smokers. Drinkers were defined as those who reported 39:68 + ð1:88 × BMIÞ 1:03
 
consuming alcohol >12 times in the previous 1:31
×
12 months. Physical activity level was classified using HDL − C
the short version of the International Physical Activity    
WC TG
Questionnaire23 into one of three levels: VAI ðwomenÞ = ×
36:58 + ð1:89 × BMIÞ 0:81
1. Low physical activity: individuals who did not meet  
1:52
the criteria for the two other categories. ×
2. Moderate physical activity: individuals meeting one HDL − C
of the following three criteria:
• ≥3 days/week of vigorous-intensity activity of at with WC in cm, BMI in kg/m2, and TG and HDL-C
least 20 min/day. both in mmol/L.10
• ≥5 days/week of moderate-intensity activity and/or
walking at least 30 min/day. LAP ðmenÞ = ðWC – 65Þ × TG
• ≥5 days/week of any combination of walking, LAP ðwomenÞ = ðWC – 58Þ × TG
moderate-intensity or vigorous-intensity activities
achieving a total score of at least 600 metabolic with WC in cm and TG in mmol/L.26
equivalent (MET)min/week. Data collection at follow-up was the same as at the
3. High physical activity: Individuals meeting one of baseline examination.
the following two criteria: Type 2 diabetes mellitus was defined as FPG ≥ 7.0
• Vigorous-intensity activity on at least 3 days/week mmol/L and/or the use of insulin or oral hypoglycemic
achieving a total score of at least 1500 agents, and/or a self-reported history of T2DM.22
METmin/week
• ≥7 Days/week of any combination of walking, Statistical analysis
moderate-intensity or vigorous-intensity activities
achieving a total score of at least 3000 MET All continuous data are described as the median (inter-
min/week. quartile range) because of skewed distribution and were
Participants were asked to wear light clothes and be analyzed using the Wilcoxon rank sum test. Categorical
barefoot when anthropometric indices were measured. data are presented as percentages and were analyzed
Height and weight were measured by trained investigators using the Chi-squared test. Fasting plasma glucose,
under standardized conditions following a standard pro- WC, TC, TG, HDL-C, LDL-C, TG: HDL-C, TC:
tocol. In standing participants, WC was measured mid- HDL-C, TC: LDL-C, LDL-C: HDL-C ratios, and
way between the lower edge of the costal arch and the TyG, VAI, and LAP indices were categorized into
upper edge of the iliac crest. Participants were measured quartiles. A Cox proportional hazard regression model
twice, and the average was used. Blood pressure was mea- was used to estimate hazard ratios (HRs) and 95% con-
sured using an electronic sphygmomanometer (HEM- fidence intervals (CIs) for incident T2DM by quartile
770AFuzzy; Omron, Kyoto, Japan) after at least a 5-min for each individual measure, lipid ratio, and the TyG,
rest, with participants in a seated position, according to VAI, and LAP indices. The HRs and 95% CIs for the
the American Heart Association standardized protocol.24 upper three quartiles were calculated using the lowest
The measurements were repeated three times at 30-s inter- quartile as the reference category. Two models were
vals, and the average was used for analysis. fitted: Model 1, controlling for age, family history of
Blood samples were obtained after an overnight fast diabetes, and family history of hypertension; and Model
of at least 8 h. Levels of FPG, TG, TC, and HDL-C 2, additionally adjusted for education level, marital sta-
were measured using an automatic clinical analyzer tus, smoking, alcohol consumption, physical activity,
(Model 7060; Hitachi, Tokyo, Japan). The LDL-C level and systolic blood pressure (SBP). The linear trends
was calculated using the Friedewald formula;25 TyG, across FPG, WC, TG: HDL-C ratio, TyG, VAI, and

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 643
Diabetes prevention in China B. WANG et al.

LAP quartiles were evaluated by a median value within Therefore, we compared TyG, VAI, and LAP indices,
each quartile used as a continuous variable. The area FPG, WC, and the TG: HDL-C ratio to explore
under the receiver operating characteristic (ROC) curve whether TyG, VAI, and LAP could improve the ability
(AUC) was used to test the ability of FPG, WC, TG: to predict incident T2DM.
HDL-C ratio, and TyG, VAI and LAP indices at base- The incidence of T2DM by quartiles of TyG, VAI,
line to predict the risk of T2DM at follow-up. Differ- and LAP was increased for men (Table 2) and women
ences in AUCs were compared by a non-parametric test (Table 3; Ptrend < 0.001 for all). For men, the risk of
developed by DeLong et al.27 There were no missing incident T2DM was increased with Quartiles 3 and 4 of
data for the main outcome variables. The ROCs were TyG versus Quartile 1 (aHR 2.22 [95% CI 1.27–3.88;
calculated using STATA v12.0 (StataCorp, College Sta- χ2 = 7.92, P < 0.01] and aHR 3.54 [95% CI 2.08–6.03;
tion, TX, USA) and other analyses were performed χ2 = 21.67, P < 0.001], respectively), but only with
using SAS 9.1 (SAS Institute, Cary, NC, USA). Two- Quartile 4 (vs Quartile 1) of VAI (aHR 2.89, 95% CI
sided P < 0.05 was considered significant. 1.72–4.87; χ2 = 16.02, P < 0.001) and with Quartiles
3 and 4 (vs Quartile 1) of LAP (aHR 2.12 [95% CI
1.15–3.91; χ2 = 5.75, P = 0.02] and aHR 5.02 [95% CI
2.85–8.85; χ2 = 31.04, P < 0.001], respectively).
Results
The risk of T2DM was even greater for women than
During a median of follow-up of 6.0 years, T2DM men with increasing quartiles of TyG, VAI, and LAP.
developed in 439 people; the overall incidence of The risk of T2DM was increased with Quartiles 2–4 of
T2DM was 6.59/1000 person-years. The T2DM inci- TyG versus Quartile 1 (aHR 2.50 [95% CI 1.36–4.60;
dence for men and women was 7.29 and 6.14/1000 χ2 = 8.67, P < 0.01], aHR 3.12 [95% CI 1.72–5.67;
person-years, respectively. Table 1 summarizes the base- χ2 = 14.00, P < 0.001], and aHR 6.15 [95% CI
line characteristics of participants according to T2DM 3.48–10.85; χ2 = 39.25, P < 0.001], respectively) and
status at follow-up and by sex. Quartiles 3 and 4 of VAI (aHR 2.13 [95% CI 1.22–3.74;
For men, a family history of diabetes was more fre- χ2 = 7.01, P = 0.01] and aHR 4.40 [95% CI 2.61–7.42;
quent with than without T2DM (P < 0.001). Type 2 dia- χ2 = 31.00, P < 0.001], respectively). For LAP, the risk
betes mellitus was associated with increased BMI, WC, of T2DM was increased with Quartiles 2–4 (aHR 2.42
SBP, DBP, FPG, TC, TG, TG: HDL-C, TC: HDL- [95% CI 1.23–4.74; χ2 = 6.59, P = 0.01], aHR 3.65
C, and LDL-C: HDL-C ratios, and TyG, VAI, and [95% CI 1.92–6.92; χ2 = 15.69, P < 0.001], and aHR
LAP values, but reduced HDL-C levels (all P < 0.05). 6.49 [95% CI 3.48–12.12; χ2 = 34.57, P < 0.001],
For women, T2DM was associated with older age, respectively).
increased BMI, WC, SBP, DBP, FPG, TC, TG, LDL- The ROC curves and AUCs for FPG, WC, TG:
C, TG: HDL-C, TC: HDL-C, and LDL-C: HDL-C HDL-C ratio, and TyG, VAI, and LAP indices for dis-
ratios, and TyG, VAI, and LAP values, but reduced criminating T2DM incidence by sex are shown in Fig. 2
HDL-C levels (all P < 0.05). and summarized in Table 4. All these indicators had
After adjusting for variables in Model 2, the risk of high negative predictive value but low positive predic-
incident T2DM was increased with elevated FPG, WC, tive value for predicting T2DM. In both men and
TC, TG, TG: HDL-C and TC: HDL-C ratios, and women, TyG, VAI, and LAP had a higher sensitivity
TyG, VAI, and TyG values for both men and women but lower specificity than FPG. For men, TyG had a
(Fig. 1). Compared with other individual measures, high similar AUC as the TG: HDL-C ratio, FPG, and WC
FPG was associated the most with T2DM for both men (all P > 0.05). The AUCs for VAI and LAP were simi-
(adjusted [a] HR 5.96, 95% CI 3.48–10.23; χ2 = 42.07, lar to those for FPG (all P > 0.05) and WC (all
P < 0.001) and women (aHR 5.05, 95% CI 3.18–8.03; P > 0.05), but larger than that for the TG: HDL-C
χ2 = 47.00, P < 0.001). A strong association between ratio (all P < 0.05). For women, the AUCs for TyG,
WC and T2DM was observed among both men (aHR VAI, and LAP were similar to those for FPG (all
4.25, 95% CI 2.51–7.21; χ2 = 28.85, P < 0.001) and P > 0.05) and WC (all P > 0.05), but larger than that
women (aHR 4.07, 95% CI 2.36–7.03; χ2 = 25.49, for the TG: HDL-C ratio (all P < 0.05).
P < 0.001). Compared with other lipid ratios, the TG:
HDL-C ratio was associated the most with T2DM for
both men (aHR 2.37, 95% CI 1.45–3.86; χ2 = 11.91,
Discussion
P = 0.001) and women (aHR 4.05, 95% CI 2.46–6.67;
χ2 = 30.33, P < 0.001). Receiver operating characteristic In the present population-based prospective cohort
analysis revealed similar results (data not shown). study, we compared three novel IR-related lipid indices

644 © 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
B. WANG et al.

Table 1 Baseline characteristics of the study participants by sex and type 2 diabetes mellitus status at follow-up

Men (n = 4269) Women (n = 6844)

Total (n = 11 113) No T2DM (n = 4081) T2DM (n = 188) P-value No T2DM (n = 6593) T2DM (n = 251) P-value
Age (years) 50.00 (41.00, 59.00) 52.00 (43.00, 61.00) 53.00 (42.00, 60.00) 0.94 48.00 (40.00, 57.00) 53.00 (44.00, 60.00) <0.001
High school or above 1137 (10.23) 650 (15.93) 30 (15.96) 0.99 442 (6.70) 15 (5.98) 0.65
Married or cohabiting 10 230 (92.08) 3713 (91.05) 177 (94.15) 0.14 6108 (92.64) 232 (92.43) 0.90
Smoking 3009 (27.08) 2863 (70.15) 130 (69.15) 0.77 14 (0.21) 2 (0.80) 0.11
Drinking alcohol 1247 (11.22) 1154 (28.28) 45 (23.94) 0.20 45 (0.68) 3 (1.20) 0.26
Physical activity
Low 5623 (50.60) 2407 (58.98) 102 (54.26) 0.21 2997 (45.46) 117 (46.61) 0.42
Moderate 2345 (21.10) 748 (18.33) 33 (17.55) 1515 (22.98) 49 (19.52)
High 3145 (28.30) 926 (22.69) 53 (28.19) 2081 (31.56) 85 (33.86)
Family history
Diabetes* 573 (6.30) 194 (5.89) 22 (14.77) <0.001 339 (6.20) 18 (9.42) 0.07
Hypertension* 3687 (38.84) 1293 (37.69) 61 (40.67) 0.46 2242 (39.27) 91 (44.61) 0.13
BMI (kg/m2) 23.91 (21.62, 26.37) 23.13 (21.08, 25.44) 25.52 (22.29, 27.68) <0.001 24.27 (21.95, 26.81) 25.92 (23.85, 28.73) <0.001
WC (cm) 81.30 (74.50, 88.75) 81.50 (74.90, 89.25) 89.18 (79.50, 96.28) <0.001 80.85 (74.00, 87.75) 87.50 (80.50, 93.75) <0.001
SBP (mmHg) 121.33 (111.67, 135.33) 122.00 (112.33, 133.67) 126.50 (115.50, 135.50) <0.01 120.67 (109.33, 135.67) 130.33 (117.33, 147.33) <0.001
DBP (mmHg) 77.00 (70.33, 85.33) 76.67 (70.33, 84.33) 79.67 (72.50, 85.33) <0.01 77.00 (70.33, 85.33) 83.33 (76.00, 91.67) <0.001
FPG (mmol/L) 5.25 (4.94, 5.56) 5.22 (4.91, 5.53) 5.45 (5.13, 5.76) <0.001 5.25 (4.95, 5.55) 5.55 (5.17, 5.84) <0.001
TC (mmol/L) 4.34 (3.8, 4.97) 4.24 (3.74, 4.85) 4.37 (3.84, 5.05) 0.03 4.39 (3.83, 5.04) 4.69 (4.18, 5.21) <0.001
TG (mmol/L) 1.32 (0.95, 1.90) 1.28 (0.94, 1.83) 1.56 (1.12, 2.22) <0.001 1.33 (0.94, 1.89) 1.81 (1.23, 2.57) <0.001
HDL-C (mmol/L) 1.14 (0.99, 1.32) 1.09 (0.95, 1.26) 1.04 (0.89, 1.17) <0.01 1.18 (1.03, 1.37) 1.13 (0.99, 1.31) <0.01
LDL-C (mmol/L) 2.50 (2.00, 3.00) 2.50 (2.00, 2.90) 2.50 (2.10, 3.10) 0.56 2.50 (2.10, 3.00) 2.60 (2.10, 3.10) 0.03
TG: HDL-C ratio 1.16 (0.77, 1.79) 1.18 (0.79, 1.82) 1.51 (1.00, 2.37) <0.001 1.12 (0.74, 1.73) 1.62 (0.99, 2.47) <0.001

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
TC: HDL-C ratio 3.80 (3.22, 4.48) 3.90 (3.30, 4.56) 4.27 (3.54, 4.95) <0.001 3.71 (0.15, 4.38) 4.11 (3.54, 4.89) <0.001
TC: LDL-C ratio 1.72 (1.60, 1.89) 1.70 (1.57, 1.87) 1.71 (1.57, 1.91) 0.60 1.73 (1.61, 1.91) 1.75 (1.63, 1.97) 0.15
LDL-C: HDL-C ratio 2.17 (1.74, 2.66) 2.25 (1.81, 2.76) 2.32 (1.88, 2.97) 0.04 2.11 (1.70, 2.58) 2.27 (1.78, 2.82) <0.01
TyG 8.61 (8.26, 8.99) 8.60 (8.24, 8.95) 8.82 (8.45, 9.19) <0.001 8.62 (8.26, 8.99) 8.94 (8.54, 9.34) <0.001
VAI 1.83 (1.17, 2.93) 1.47 (0.96, 2.30) 1.98 (1.25, 3.08) <0.001 2.06 (1.34, 3.24) 3.06 (1.92, 4.71) <0.001
LAP 26.78 (13.79, 47.56) 20.50 (9.88, 39.35) 37.53 (17.12, 72.00) <0.001 29.83 (16.32, 50.62) 51.15 (31.80, 83.27) <0.001

Data are given as the median (interquartile range) or n (%).


BMI, body mass index; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein cholesterol; LAP, lipid accumulation product; LDL-C, low-density lipopro-
tein cholesterol; SBP, systolic blood pressure; T2DM, type 2 diabetes mellitus; TC, total cholesterol; TG, triglycerides; TyG, triglyceride glucose index; VAI, visceral adiposity index; WC, waist
circumference. *Partial data deletion.
Diabetes prevention in China

645
Diabetes prevention in China B. WANG et al.

Figure 1 Risk of incident type 2 diabetes mellitus (T2DM) by quartiles of traditional individual measures, lipid ratios and three novel indices in
men and women separately. HR, hazard ratio; CI, confidence interval; FPG, fasting plasma glucose; WC, waist circumference; TC, total choles-
terol; TG, triglycerides; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TyG, triglyceride glucose index;
VAI, visceral adiposity index; LAP, lipid accumulation product. Data are adjusted for age, education level, marital status, smoking, alcohol con-
sumption, physical activity, family history of diabetes, family history of hypertension, and systolic blood pressure.

(TyG, VAI, and LAP) with traditional individual mea- the relationship between lipid ratios and diabetes differs
sures and their ratios to predict T2DM in a rural Chi- depending on the selected study population. Results
nese population. Elevated FPG, WC, TC, TG levels, from the Isfahan Diabetes Prevention Study showed
TG: HDL-C ratio, TC: HDL-C ratio, TyG, VAI, and that the TG: HDL-C and TC: HDL-C ratios were not
LAP were significantly associated with the risk of inci- robust predictors of T2DM in first-degree relatives of
dent T2DM in both sexes. Fasting plasma glucose and patients with T2DM after a 7-year follow-up.32 How-
WC were superior to other individual measures for pre- ever, another cohort study of an Iranian population
dicting T2DM, and the TG: HDL-C ratio was superior suggested a significant association of TG: HDL-C and
to other lipid ratios for predicting T2DM in both men TC: HDL-C ratios with incident T2DM among non-
and women. Compared with FPG, WC, and the TG: T2DM participants over a median follow up of
HDL-C ratio, the AUCs for the three indices (TyG, 6.4 years for both sexes.33 Heredity may be the key fac-
VAI, and LAP) were larger than that for the TG: tor for T2DM development among first-degree relatives
HDL-C ratio, but similar to those for FPG and WC. of people with T2DM. These groups may have a high
The present study suggests that the TG: HDL-C and propensity for T2DM developing at a younger age,
TC: HDL-C ratios are significantly associated with even with low BMI and WC and favorable lipid
T2DM in both sexes, and an elevated TG: HDL-C levels,32 which may weaken the relationship between
ratio is associated the most with T2DM, which is in lipid ratios and T2DM.
agreement with results from studies in other Chinese The TyG index, which combines TG and FPG levels,
populations.28,29 Nevertheless, the utility of these two has been used to identify IR and T2DM in many epide-
lipid ratios to identify T2DM in clinical practice is con- miological studies.9,12,13 Both TG and FPG levels are
troversial in different ethnic groups. In a multiethnic independent risk factors for IR and T2DM.34,35 Pancre-
study, the risk of IR was increased with an elevated atic β-cell dysfunction is considered the most important
TG: HDL-C ratio in White obese subjects, but not in risk factor for T2DM.36 Prolonged exposure to high
Hispanics and African Americans.30 The main reason amounts of free fatty acids is associated with TG accu-
for this finding may be differences in dietary and life- mulation in islets,37 which may alter the function of
style habits and metabolic-related indicators in different pancreatic β-cells, leading to T2DM.38 Long-term expo-
ethnic groups,29,31 which could affect the association sure to abnormally high levels of glucose can lead to
between lipid ratios and IR and diabetes. In addition, T2DM because of toxic effects on pancreatic β-cells.39

646 © 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Table 2 Cox proportional hazards regression analysis of risk of incident type 2 diabetes mellitus by quartiles of fasting plasma glucose, triglycerides: high-density lipoprotein cholesterol ratio,
triglyceride glucose index, visceral adiposity index, and lipid accumulation product for men

Model 1 Model 2
2
No. cases No. person-years Incidence* HR (95% CI) χ P-value HR (95% CI) χ2 P-value
B. WANG et al.

Fasting plasma glucose


Q1 (<4.92 mmol/L) 26 6584.08 3.95 1 1
Q2 (4.92–5.22 mmol/L) 37 6506.17 5.69 1.95 (1.05–3.59) 4.52 0.03 1.89 (1.02–3.49) 4.12 0.04
Q3 (5.23–5.54 mmol/L) 40 6461.17 6.19 2.31 (1.27–4.22) 7.45 <0.01 2.24 (1.23–4.10) 6.88 <0.01
Q4 (≥5.55 mmol/L) 85 6249.08 13.60 6.23 (3.65–10.65) 44.83 <0.001 5.96 (3.48–10.23) 42.07 <0.001
Ptrend <0.001 <0.001 <0.001
Waist circumference
Q1 (<75.00 cm) 26 6341.17 4.10 1 1
Q2 (75.00–81.47 cm) 34 6521.58 5.21 1.05 (0.58–1.91) 0.02 0.88 1.06 (0.58–1.94) 0.03 0.86
Q3 (81.75–89.74 cm) 38 6462.33 5.88 1.39 (0.77–2.50) 1.21 0.27 1.49 (0.82–2.69) 1.72 0.19
Q4 (≥89.75 cm) 90 6468.42 13.91 3.65 (2.21–6.01) 25.72 <0.001 4.25 (2.51–7.21) 28.85 <0.001
Ptrend <0.001 <0.001 <0.001
TG: HDL-C ratio
Q1 (<0.79) 28 6442.25 4.35 1 1
Q2 (0.79–1.18) 37 6458.92 5.73 1.14 (0.66–1.97) 0.21 0.65 1.15 (0.66–2.00) 0.25 0.62
Q3 (1.19–1.83) 48 6434.25 7.46 1.37 (0.81–2.34) 1.36 0.24 1.42 (0.83–2.43) 1.64 0.20
Q4 (≥1.84) 74 6380.58 11.60 2.41 (1.49–3.91) 12.80 <0.001 2.37 (1.45–3.87) 11.91 <0.001
Ptrend <0.001 <0.001 <0.001
Triglyceride glucose index
Q1 (<8.25) 28 6533.75 4.29 1 1
Q2 (8.25–8.57) 33 6451.42 5.12 1.54 (0.85–2.78) 2.04 0.15 1.59 (0.88–2.88) 2.32 0.13
Q3 (8.58–8.95) 51 6423.00 7.94 2.19 (1.26–3.80) 7.71 <0.01 2.22 (1.27–3.88) 7.92 <0.01
Q4 (≥8.96) 75 6327.00 11.85 3.55 (2.10–5.99) 22.39 <0.001 3.54 (2.08–6.03) 21.67 <0.001
Ptrend <0.001 <0.001 <0.001
Visceral adiposity index
Q1 (<0.97) 24 6443.50 3.72 1 1

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
Q2 (0.97–1.47) 42 6429.33 6.53 1.62 (0.93–2.83) 2.85 0.09 1.65 (0.94–2.89) 3.07 0.08
Q3 (1.48–2.34) 44 6450.42 6.82 1.45 (0.82–2.57) 1.65 0.20 1.49 (0.84–2.64) 1.89 0.17
Q4 (≥2.35) 77 6379.75 12.07 2.95 (1.76–4.93) 16.98 <0.001 2.89 (1.72–4.87) 16.02 <0.001
Ptrend <0.001 <0.001 <0.001
Lipid accumulation product
Q1 (<10.11) 22 6448.58 3.41 1 1
Q2 (10.11–20.95) 34 6427.50 5.29 1.45 (0.77–2.73) 1.33 0.25 1.59 (0.84–3.01) 2.03 0.15
Q3 (20.96–40.23) 42 6420.25 6.54 1.98 (1.08–3.62) 4.90 0.03 2.12 (1.15–3.91) 5.75 0.02
Q4 (≥40.24) 89 6386.83 13.93 4.45 (2.57–7.70) 28.46 <0.001 5.02 (2.85–8.85) 31.04 <0.001
Ptrend <0.001 <0.001 <0.001

Model 1 was adjusted for age, family history of diabetes, and a family history of hypertension.
Model 2 was adjusted for the variables in Model 1 as well as education level, marital status, smoking, alcohol consumption, physical activity, and systolic blood pressure.
CI, confidence interval; HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; Q1–Q4, Quartiles 1–4; TG, triglycerides. *Per 1000 person-years.
Diabetes prevention in China

647
Table 3 Cox proportional hazards regression analysis of risk of incident type 2 diabetes mellitus by quartiles of fasting plasma glucose, triglycerides: high-density lipoprotein cholesterol ratio,

648
triglyceride glucose index, visceral adiposity index, and lipid accumulation product for men

Model 1 Model 2
2
No. cases No. person-years Incidence* HR (95% CI) χ P-value HR (95% CI) χ2 P-value

Fasting plasma glucose


Q1 (<4.95 mmol/L) 31 10429.50 2.97 1 1
Q2 (4.95–5.25 mmol/L) 38 10217.17 3.72 1.58 (0.92–2.71) 2.70 0.10 1.52 (0.88–2.62) 2.29 0.13
Q3 (5.26–5.55 mmol/L) 57 10040.67 5.68 2.50 (1.51–4.15) 12.58 <0.001 2.35 (1.41–3.90) 10.83 <0.01
Diabetes prevention in China

Q4 (≥5.56 mmol/L) 125 10168.83 12.29 5.50 (3.47–8.70) 52.88 <0.001 5.05 (3.18–8.03) 47.00 <0.001
Ptrend <0.001 <0.001 <0.001
Waist circumference
Q1 (<74.00 cm) 23 9935.58 2.31 1 1
Q2 (74.00–81.04 cm) 44 10467.83 4.20 1.80 (0.99–3.25) 3.74 0.05 1.74 (0.96–3.16) 3.36 0.07
Q3 (81.05–87.99 cm) 60 10074.67 5.96 2.14 (1.19–3.86) 6.46 0.01 1.97 (1.09–3.56) 5.07 0.02
Q4 (≥88.00 cm) 124 10366.08 11.96 4.67 (2.72–7.99) 31.48 <0.001 4.07 (2.36–7.03) 25.49 <0.001
Ptrend <0.001 <0.001 <0.001
TG: HDL-C ratio
Q1 (<0.75) 30 10298.17 2.91 1 1
Q2 (0.75–1.12) 41 10243.17 4.00 1.46 (0.83–2.57) 1.74 0.19 1.39 (0.79–2.45) 1.32 0.25
Q3 (1.13–1.75) 64 10135.75 6.31 2.04 (1.19–3.50) 6.73 <0.01 1.97 (1.15–3.38) 6.05 0.01
Q4 (≥1.76) 111 10082.08 11.01 4.40 (2.68–7.20) 34.48 <0.001 4.05 (2.46–6.67) 30.33 <0.001
Ptrend <0.001 <0.001 <0.001
Triglyceride glucose index
Q1 (<8.27) 21 10441.67 2.01 1 1
Q2 (8.27–8.62) 48 10213.75 4.70 2.55 (1.39–4.68) 9.05 <0.01 2.50 (1.36–4.60) 8.67 <0.01
Q3 (8.63–9.00) 68 10139.25 6.71 3.34 (1.85–6.04) 15.91 <0.001 3.12 (1.72–5.67) 14.00 <0.001
Q4 (≥9.01) 111 10006.58 11.09 6.67 (3.80–11.71) 43.67 <0.001 6.15 (3.48–10.85) 39.25 <0.001
Ptrend <0.001 <0.001 <0.001
Visceral adiposity index
Q1 (<1.36) 27 10285.08 2.63 1 1
Q2 (1.36–2.08) 44 10218.08 4.31 1.82 (1.03–3.23) 4.22 0.04 1.75 (0.99–3.10) 3.64 0.06
Q3 (2.09–3.28) 61 10153.92 6.01 2.25 (1.29–3.94) 8.09 <0.01 2.13 (1.22–3.74) 7.01 0.01
Q4 (≥3.29) 114 10084.08 11.30 4.80 (2.86–8.06) 35.25 <0.001 4.40 (2.61–7.42) 31.00 <0.001
Ptrend <0.001 <0.001 <0.001
Lipid accumulation product
Q1 (<16.74) 17 10275.83 1.65 1 1
Q2 (16.74–30.41) 42 10239.08 4.10 2.48 (1.27–4.86) 7.04 <0.01 2.42 (1.23–4.74) 6.59 0.01
Q3 (30.42–51.83) 69 10166.92 6.79 3.96 (2.10–7.49) 18.01 <0.001 3.65 (1.92–6.92) 15.69 <0.001
Q4 (≥51.84) 120 10107.42 11.87 7.26 (3.92–13.43) 39.80 <0.001 6.49 (3.48–12.12) 34.57 <0.001
Ptrend <0.001 <0.001 <0.001

Model 1 was adjusted for age, family history of diabetes, and a family history of hypertension.
Model 2 was adjusted for the variables in Model 1 as well as education level, marital status, smoking, alcohol consumption, physical activity, and systolic blood pressure.
CI, confidence interval; HDL-C, high-density lipoprotein cholesterol; HR, hazard ratio; Q1–Q4, Quartiles 1–4; TG, triglycerides. *Per 1000 person-years.
B. WANG et al.

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
B. WANG et al. Diabetes prevention in China

Figure 2 Receiver operating characteristic (ROC) curves in (a) men and (b) women for fasting plasma glucose (FPG), waist circumference
(WC), triglyceride (TG) to high-density lipoprotein cholesterol (HDL-C) ratio, triglyceride glucose index (TyG), visceral adiposity index (VAI), and
the lipid accumulation product (LAP) for predicting type 2 diabetes mellitus.

However, compared with other routine biomarkers, the Adipose tissue is generally considered to play a cen-
ability of TyG to identify T2DM is controversial in dif- tral role in lipid and glucose metabolism.40 Excessive
ferent ethnic groups. In a European population, TyG visceral adipose tissue accumulation is closely associ-
was a better predictor of T2DM development in normo- ated with the risk of incident cardiovascular diseases
glycemic people than either FPG or TG.15 In an and T2DM.4,41 The VAI includes body physical and
Iranian population, TyG did not improve diabetes pre- metabolic factors and may indirectly reflect cardiometa-
diction compared with FPG, 1-h plasma glucose, or 2-h bolic risk factors, including altered production of adipo-
plasma glucose.14 In the present study, the risk of inci- cytokines, increased lipolysis, and increased plasma free
dent T2DM was increased with elevated TyG values in fatty acid content.10 Previous studies showed that VAI
both men and women. For predicting T2DM risk, the was significantly associated with T2DM and more effi-
AUCs for TyG did not differ from those for individual cient than the TG: HDL-C ratio, but not superior to
measures (FPG and WC) for either men or women BMI, WC, or waist-to-height ratio, in detecting T2DM
among rural Chinese people. among non-T2DM people.12,16,18 Consistent with

Table 4 Receiver operating characteristic analysis for fasting plasma glucose, waist circumference, triglycerides: high-density lipoprotein cho-
lesterol ratio, triglyceride glucose index, visceral adiposity index, and lipid accumulation product for predicting type 2 diabetes mellitus at the
6-year follow-up in men and women separately

Cut-off value AUC (95% CI) Sensitivity (%) Specificity (%) PPV (%) NPV (%)
Men
FPG (mmol/L) 5.6 0.638 (0.623–0.652) 42.55 80.30 9.05 96.81
WC (cm) 87.45 0.654 (0.640–0.669) 58.51 69.75 8.18 97.33
TG: HDL-C ratio 1.69 0.613 (0.598–0.628) 47.59 71.93 7.23 96.76
TyG 8.72 0.625 (0.610–0.639) 59.89 60.75 6.55 97.06
VAI 2.25 0.622 (0.607–0.636) 47.06 74.23 7.75 96.82
LAP 36.96 0.653 (0.638–0.667) 51.34 73.02 8.04 97.03
Women
FPG (mmol/L) 5.44 0.664 (0.652–0.675) 60.56 66.77 6.49 97.80
WC (cm) 82.05 0.669 (0.657–0.680) 70.92 55.00 5.66 98.03
TG: HDL-C ratio 1.17 0.644 (0.632–0.655) 69.92 52.90 5.26 97.92
TyG 8.79 0.669 (0.657–0.680) 62.90 62.62 5.96 97.82
VAI 2.50 0.654 (0.642–0.665) 63.41 61.41 5.79 97.82
LAP 37.84 0.693 (0.682–0.704) 69.35 61.49 6.35 98.16

AUC, area under the receiver operating characteristic curve; CI, confidence interval; FPG, fasting plasma glucose; HDL-C, high-density lipopro-
tein cholesterol; LAP, lipid accumulation product; NPV, negative predictive value; PPV, positive predictive value; TG, triglycerides; TyG, triglyc-
eride glucose index; VAI, visceral adiposity index; WC, waist circumference.

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 649
Diabetes prevention in China B. WANG et al.

previous studies, the present study revealed larger VAI, and LAP may not be a convenient or inexpensive
AUCs for VAI than the TG: HDL-C ratio for detect- approach for predicting T2DM in rural Chinese people.
ing T2DM in both sexes. The VAI was equivalent to
FPG and WC for detecting T2DM in both men and
women. Acknowledgements
The LAP has been reported as a reliable visceral adi-
This study was supported by the National Natural Sci-
posity index combining TG levels and WC for predict-
ence Foundation of China (Grant no. 81373074,
ing IR, T2DM, and cardiovascular diseases.12,26 The
81402752, and 81673260) and the Science and Technol-
phenotype of a hypertriglyceridemic waist (large waist
ogy Development Foundation of Shenzhen (Grant no.
with elevated TG levels) could be an inexpensive screen-
JCYJ20140418091413562, JCYJ 20160307155707264,
ing tool to predict the atherogenic metabolic triad
JCYJ 20170302143855721, and JCYJ 20170412110537191).
(hyperinsulinemia, elevated apolipoprotein B levels, and
small, dense LDL-C particles) for men.42 A meta-
analysis has also suggested that the risk of T2DM is
Disclosure
closely associated with the hypertriglyceridemic waist
phenotype in the general population.43 Alternatively, as None declared.
a continuous index, LAP has been developed to reflect
the combined anatomic and physiologic changes associ-
ated with overaccumulation of lipids in adults.26 In the References
present study, we found that the risk of incident T2DM
1. International Diabetes Federation (IDF). IDF Diabetes
was associated with elevated LAP in both sexes. Com-
Atlas – 7th Edition. 2015. Brussels, Belgium: International
pared with FPG, LAP had a higher sensitivity but Diabetes Federation. Available from: http://www.
lower specificity for predicting T2DM risk in both diabetesatlas.org/key-messages.html (accessed 17 February
sexes. Compared with WC, LAP had a lower sensitivity 2017).
and higher specificity for predicting T2DM risk in men 2. Xu Y, Wang L, He J et al. Prevalence and control of
and a similar sensitivity but higher specificity in women. diabetes in Chinese adults. JAMA. 2013; 310: 948–59.
The AUCs for LAP did not differ from those for FPG 3. Wang L, Wang A, Zhou D, FitzGerald G, Ye D,
Jiang Q. An empirical analysis of rural–urban differ-
and WC in men and women.
ences in out-of-pocket health expenditures in a low-
The present study was a well-designed and income society of China. PLoS One. 2016; 11:
population-based prospective cohort study with a low e0154563.
rate of loss to follow-up and standardized anthropomet- 4. Liu PJ, Ma F, Lou HP, Chen Y. Visceral adiposity index
ric, physical, and laboratory measurements to compare is associated with pre-diabetes and type 2 diabetes melli-
the association of traditional individual measures, lipid tus in Chinese adults aged 20–50. Ann Nutr Metab. 2016;
ratios, and three novel indices (TyG, VAI, LAP) with 68: 235–43.
5. Dugani SB, Akinkuolie AO, Paynter N, Glynn RJ,
the risk of incident T2DM in a rural Chinese popula-
Ridker PM, Mora S. Association of lipoproteins, insulin
tion. However, the present study has some limitations. resistance, and rosuvastatin with incident type 2 diabetes
We did not use a 2-h oral glucose tolerance test or mea- mellitus: Secondary analysis of a randomized clinical
sure HbA1c levels to diagnose T2DM, so we may have trial. JAMA Cardiol. 2016; 1: 136–45.
underestimated the incidence of T2DM, and this may 6. McLaughlin T, Lamendola C, Liu A, Abbasi F. Prefer-
affect the association between different indicators and ential fat deposition in subcutaneous versus visceral
T2DM. In addition, the participants in the present depots is associated with insulin sensitivity. J Clin Endo-
crinol Metab. 2011; 96: E1756–60.
study were from a rural population in the middle of
7. Vasques AC, Novaes FS, de Oliveira Mda S et al. TyG
China, which may not represent the characteristics of index performs better than HOMA in a Brazilian popu-
other Chinese people well. lation: A hyperglycemic clamp validated study. Diabetes
In conclusion, elevated TyG, VAI, and LAP are sig- Res Clin Pract. 2011; 93: e98–e100.
nificantly associated with the risk of incident T2DM in 8. Cheung AS, de Rooy C, Hoermann R et al. Correlation
both men and women in rural China. For discriminat- of visceral adipose tissue measured by lunar prodigy dual
ing T2DM, TyG, VAI, and LAP were mostly superior X-ray absorptiometry with MRI and CT in older men.
Int J Obesity. 2016; 40: 1325–8.
to the TG: HDL-C ratio and equal to FPG and WC
9. Guerrero-Romero F, Simental-Mendia LE, Gonzalez-
for predicting T2DM at follow-up. According to eco- Ortiz M et al. The product of triglycerides and glucose, a
nomic benefits, as compound indices, TyG, VAI, and simple measure of insulin sensitivity. Comparison with
LAP did not improve T2DM prediction compared with the euglycemic–hyperinsulinemic clamp. J Clin Endocri-
measurement of FPG or WC alone. Therefore, TyG, nol Metab. 2010; 95: 3347–51.

650 © 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd
B. WANG et al. Diabetes prevention in China

10. Amato MC, Giordano C, Galia M et al. Visceral adipos- 24. Perloff D, Grim C, Flack J et al. Human blood pressure
ity index: A reliable indicator of visceral fat function determination by sphygmomanometry. Circulation. 1993;
associated with cardiometabolic risk. Diabetes Care. 88: 2460–70.
2010; 33: 920–2. 25. Bairaktari E, Hatzidimou K, Tzallas C et al. Estimation
11. Du T, Yuan G, Zhang M, Zhou X, Sun X, Yu X. Clini- of LDL cholesterol based on the Friedewald formula
cal usefulness of lipid ratios, visceral adiposity indicators, and on apo B levels. Clin Biochem. 2000; 33: 549–55.
and the triglycerides and glucose index as risk markers of 26. Kahn HS. The “lipid accumulation product” performs
insulin resistance. Cardiovasc Diabetol. 2014; 13: 146. better than the body mass index for recognizing cardio-
12. Er LK, Wu S, Chou HH et al. Triglyceride glucose–body vascular risk: A population-based comparison. BMC
mass index is a simple and clinically useful surrogate Cardiovasc Disord. 2005; 5: 26.
marker for insulin resistance in nondiabetic individuals. 27. DeLong ER, DeLong DM, Clarke-Pearson DL. Com-
PLoS One. 2016; 11: e0149731. paring the areas under two or more correlated receiver
13. Amato MC, Magistro A, Gambino G, Vesco R, operating characteristic curves: A nonparametric
Giordano C. Visceral adiposity index and DHEAS are approach. Biometrics. 1988; 44: 837–45.
useful markers of diabetes risk in women with polycystic 28. Song Q, Liu X, Wang A et al. Associations between
ovary syndrome. Eur J Endocrinol. 2015; 172: 79–88. non-traditional lipid measures and risk for type 2 diabetes
14. Janghorbani M, Almasi SZ, Amini M. The product of mellitus in a Chinese community population: A cross-
triglycerides and glucose in comparison with fasting sectional study. Lipids Health Dis. 2016; 15: 70.
plasma glucose did not improve diabetes prediction. Acta 29. Lin D, Qi Y, Huang C et al. Associations of lipid param-
Diabetol. 2015; 52: 781–8. eters with insulin resistance and diabetes: A population-
15. Navarro-Gonzalez D, Sanchez-Inigo L, Pastrana- based study. Clin Nutrition. 2017. https://doi.org/10.1016/
Delgado J, Fernandez-Montero A, Martinez JA. Triglycer- j.clnu.2017.06.018.
ide–glucose index (TyG index) in comparison with fasting 30. Giannini C, Santoro N, Caprio S et al. The triglyceride-
plasma glucose improved diabetes prediction in patients to-HDL cholesterol ratio: Association with insulin resis-
with normal fasting glucose: The vascular-metabolic CUN tance in obese youths of different ethnic backgrounds.
cohort. Prev Med. 2016; 86: 99–105. Diabetes Care. 2011; 34: 1869–74.
16. Janghorbani M, Amini M. The visceral adiposity index 31. Gao H, Salim A, Lee J, Tai ES, van Dam RM. Can
in comparison with easily measurable anthropometric body fat distribution, adiponectin levels and inflamma-
markers did not improve prediction of diabetes. Can J tion explain differences in insulin resistance between eth-
Diabetes. 2016; 40: 393–8. nic Chinese, Malays and Asian Indians? Int J Obesity.
17. Wakabayashi I, Daimon T. A strong association 2012; 36: 1086–93.
between lipid accumulation product and diabetes melli- 32. Janghorbani M, Amini M. Utility of serum lipid ratios
tus in japanese women and men. J Atheroscler Thromb. for predicting incident type 2 diabetes: The Isfahan dia-
2014; 21: 282–8. betes prevention study. Diabetes Metab Res Rev. 2016;
18. Zhang M, Zheng L, Li P et al. 4-Year trajectory of vis- 32: 572–80.
ceral adiposity index in the development of type 2 diabe- 33. Hadaegh F, Hatami M, Tohidi M, Sarbakhsh P,
tes: A prospective cohort study. Ann Nutr Metab. 2016; Saadat N, Azizi F. Lipid ratios and appropriate cut off
69: 142–9. values for prediction of diabetes: A cohort of Iranian
19. Li YQ, Sun CQ, Li LL et al. Resting heart rate as a men and women. Lipids Health Dis. 2010; 9: 85.
marker for identifying the risk of undiagnosed type 2 dia- 34. McPhillips JB, Barrett-Connor E, Wingard DL. Cardio-
betes mellitus: A cross-sectional survey. BMC Public vascular disease risk factors prior to the diagnosis of
Health. 2014; 14: 1052. impaired glucose tolerance and non-insulin-dependent
20. Zhang M, Wang B, Liu Y et al. Cumulative increased diabetes mellitus in a community of older adults.
risk of incident type 2 diabetes mellitus with increasing Am J Epidemiol. 1990; 131: 443–53.
triglyceride glucose index in normal-weight people: The 35. Tirosh A, Shai I, Bitzur R et al. Changes in triglyceride
rural Chinese cohort study. Cardiovasc Diabetol. 2017; levels over time and risk of type 2 diabetes in young
16: 30. men. Diabetes Care. 2008; 31: 2032–7.
21. Zhao Y, Zhang M, Luo X et al. Association of obesity 36. Lillioja S, Mott DM, Spraul M et al. Insulin resistance
categories and high blood pressure in a rural adult Chi- and insulin secretory dysfunction as precursors of non-
nese population. J Hum Hypertens. 2016; 30: 613–8. insulin-dependent diabetes mellitus. Prospective studies
22. American Diabetes Association. Standards of medical of Pima Indians. N Engl J Med. 1993; 329: 1988–92.
care in diabetes. Diabetes Care. 2005; 28 (Suppl. 1): 37. Lupi R, Del Guerra S, Fierabracci V et al. Lipotoxicity
S4–36. in human pancreatic islets and the protective effect of
23. Van Holle V, De Bourdeaudhuij I, Deforche B, Van metformin. Diabetes. 2002; 51 (Suppl. 1): S134–7.
Cauwenberg J, Van Dyck D. Assessment of physical 38. Unger RH. Lipotoxicity in the pathogenesis of obesity-
activity in older Belgian adults: Validity and reliability of dependent NIDDM. Genetic and clinical implications.
an adapted interview version of the long International Diabetes. 1995; 44: 863–70.
Physical Activity Questionnaire (IPAQ-L). BMC Public 39. Robertson RP, Harmon J, Tran PO, Poitout V. Beta-cell
Health. 2015; 15: 433. glucose toxicity, lipotoxicity, and chronic oxidative stress

© 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd 651
Diabetes prevention in China B. WANG et al.

in type 2 diabetes. Diabetes. 2004; 53 (Suppl. 1): 42. Lemieux I, Pascot A, Couillard C
S119–24. et al. Hypertriglyceridemic waist: A marker of the ath-
40. Hajer GR, van Haeften TW, Visseren FL. Adipose tissue erogenic metabolic triad (hyperinsulinemia; hyperapoli-
dysfunction in obesity, diabetes, and vascular diseases. poprotein B; small, dense LDL) in men? Circulation.
Eur Heart J. 2008; 29: 2959–71. 2000; 102: 179–84.
41. Yusuf S, Hawken S, Ounpuu S et al. Obesity and the risk 43. Ren Y, Luo X, Wang C et al. Prevalence of hypertrigly-
of myocardial infarction in 27,000 participants from ceridemic waist and association with risk of type 2 diabe-
52 countries: A case-control study. Lancet. 2005; 366: tes mellitus: A meta-analysis. Diabetes Metab Res Rev.
1640–9. 2016; 32: 405–12.

652 © 2018 Ruijin Hospital, Shanghai Jiaotong University School of Medicine and John Wiley & Sons Australia, Ltd

Вам также может понравиться