Sheng et al.

Journal of Translational Medicine

https://doi.org/10.1186/s12967-022-03470-z
(2022) 20:266
Journal of
Translational Medicine

RESEARCH Open Access

Evaluation of the value of conventional

and unconventional lipid parameters
for predicting the risk of diabetes
in a non‑diabetic population
Guotai Sheng1, Maobin Kuang2, Ruijuan Yang2,3, Yanjia Zhong3, Shuhua Zhang4 and Yang Zou4*   

Abstract
Background: Conventional and unconventional lipid parameters are associated with diabetes risk, the comparative
studies on lipid parameters for predicting future diabetes risk, however, are still extremely limited, and the value of
conventional and unconventional lipid parameters in predicting future diabetes has not been evaluated. This study
was designed to determine the predictive value of conventional and unconventional lipid parameters for the future
development of diabetes.
Methods: The study was a longitudinal follow-up study of 15,464 participants with baseline normoglycemia. At
baseline, conventional lipid parameters such as low-density lipoprotein cholesterol (LDL-C), triglyceride (TG), total
cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) were measured/calculated, and unconventional lipid
parameters such as non-HDL-C, remnant cholesterol (RC), LDL/HDL-C ratio, TG/HDL-C ratio, non-HDL/HDL-C ratio, TC/
HDL-C ratio and RC/HDL-C ratio were calculated. Hazard ratio (HR) and 95% confidence interval (CI) were estimated by
Cox proportional hazard regression adjusting for demographic and diabetes-related risk factors. The predictive value
and threshold fluctuation intervals of baseline conventional and unconventional lipid parameters for future diabetes
were evaluated by the time-dependent receiver operator characteristics (ROC) curve.
Results: The incidence rate of diabetes was 3.93 per 1000 person-years during an average follow-up period of
6.13 years. In the baseline non-diabetic population, only TG and HDL-C among the conventional lipid parameters
were associated with future diabetes risk, while all the unconventional lipid parameters except non-HDL-C were sig-
nificantly associated with future diabetes risk. In contrast, unconventional lipid parameters reflected diabetes risk bet-
ter than conventional lipid parameters, and RC/HDL-C ratio was the best lipid parameter to reflect the risk of diabetes
(HR: 6.75, 95% CI 2.40–18.98). Sensitivity analysis further verified the robustness of this result. Also, time-dependent
ROC curve analysis showed that RC, non-HDL/HDL-C ratio, and TC/HDL-C ratio were the best lipid parameters for
predicting the risk of medium-and long-term diabetes.
Conclusions: Unconventional lipid parameters generally outperform conventional lipid parameters in assessing and
predicting future diabetes risk. It is suggested that unconventional lipid parameters should also be routinely evalu-
ated in clinical practice.

*Correspondence: jxyxyzy@163.com
4
Jiangxi Provincial People’s Hospital, Jiangxi Cardiovascular Research Institute,
Nanchang 330006, Jiangxi, China
Full list of author information is available at the end of the article

© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecom-
mons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 2 of 13

Keywords: Prediction, Unconventional lipid parameters, Lipid ratios, Diabetes, Conventional lipid parameters

Background predictive value of conventional and unconventional

The high prevalence of diabetes and its extensive compli- lipid parameters for future diabetes risk. The NAGALA
cations have created a huge disease burden for mankind, cohort data has been uploaded, by Professor Okamura, to
which is one of the leading causes of death and disability the Dryad database for public sharing [20], and a detailed
worldwide [1–3]. Diabetes is known to be associated with description of the research design has been published
a series of interrelated abnormalities of plasma lipids previously [21]. In short, NAGALA is a longitudinal
and lipoproteins. Diabetic individuals generally exhibit cohort study based on the health check-up population,
a pattern of atherogenic risk factors compared with indi- which began in 1994 and continuously recruited peo-
viduals without diabetes [4]. Among them, metabolic ple who took health examinations at Murakami Memo-
disturbances of triglyceride-rich lipoproteins are thought rial Hospital (follow-up response rate was approximately
to be key to the pathophysiology of this lipids-induced 60%) to assess non-communicable diseases and their
atherosclerosis, including increased hepatic secretion of related risk factors. In the current research dataset,
VLDL and impaired clearance of VLDL and gut-synthe- 12,498 male and 8446 female participants who registered
sized chylomicrons [4–7]. Clinically, the main manifesta- to participate in the research from 1994 to 2016 were
tions in conventional lipids are decreased HDL-C levels included, and the last visit was on December 31, 2016.
and increased TG levels [4, 8], while that in unconven- According to the purpose of the study, we excluded par-
tional lipids are elevated levels of RC, TC/HDL ratio, ticipants with diabetes, impaired fasting glucose, liver
non-HDL-C, TG/HDL-C ratio, non-HDL/HDL-C ratio, disease, excessive drinking [22], incomplete data, base-
and LDL/HDL-C ratio [9–12]. These dyslipidemia char- line medication, and unexplained withdrawal from the
acteristics are associated with an increased risk of car- survey. Figure 1 shows the current research population
diovascular disease [13–15], which is the leading cause screening process.
of death in patients with type 2 diabetes [16]. Therefore,
early identification of the above-mentioned diabetic Ethical approval and consent to participation
dyslipidemia manifestations can effectively prevent and In the initial survey, the Murakami Memorial Hos-
intervene the risk of diabetes, which is all-important for pital Institutional Ethics Review Board approved the
reducing the incidence of diabetes and diabetes-related NAGALA cohort study, and informed written consent
mortality. In recent years, a series of studies have been was obtained from each participant for the use of their
carried out on the relationship between conventional data (IRB2018-09-01) [21]. This study was a secondary
and unconventional lipid parameters and the risk of dia- analysis of data from the NAGALA cohort. The research
betes [8–12, 17, 18]; on the whole, unconventional lipid scheme has been approved by the Ethics Review Commit-
parameters appear to have good application value, and tee of Jiangxi Provincial People’s Hospital (IRB2021-066),
they reveal excellent performance in the assessment and and the entire process followed the Helsinki Declaration.
prediction of diabetes risk. However, comparative stud- See STROBE statement in Additional file 1: Text S1.
ies on lipid parameters for predicting future diabetes
risk are still relatively limited [10, 18, 19], and the value
of conventional and unconventional lipid parameters in Population and lifestyle data
predicting future diabetes has not been evaluated. To Information on age, sex, smoking and drinking, medi-
address this question, in the current study, we investi- cation history, disease history, and exercise habits was
gated the independent association and predictive value of self-reported. Blood pressure, weight, waist circumfer-
conventional and unconventional lipid parameters (addi- ence, height, and body mass index (BMI) were measured
tion the RC/HDL-C ratio on the basis of previous stud- according to the standard protocols. In terms of exercise
ies) with the future risk of diabetes based on NAGALA habits, participants were assessed for the frequency of
large longitudinal cohort study data. physical exercise per week, and more than one time per
week was considered an exercise habit. Drinking status
was assessed based on the amount and type of weekly
Methods alcohol consumption by participants in the past month
Data sources and research population and participants were classified as non/small drinkers,
In this current longitudinal analysis, we used NAGALA light drinkers, moderate drinkers, and heavy drinkers
cohort data from 1994 to 2016 to investigate the
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 3 of 13

Fig. 1 Study profile

[22]. For smoking status, participants were classified as Determination of diabetes fatty liver, and metabolic score
nonsmokers, past smokers, and current smokers. for insulin resistance (METS‑IR)
The diagnosis of diabetes was based on follow-up physi-
Determination and calculation of biomarkers cal examination information of the participants. Refer-
Venous blood samples were drawn from study partici- ring to the criteria of the American Diabetes Association
pants after fasting for > 8 h, and the concentrations of [26], participants were diagnosed with diabetes if they
HDL-C, gamma-glutamyl transferase (GGT), fasting measured HbA1c ≥ 6.5% or FPG ≥ 7.0 mmol/L dur-
plasma glucose (FPG), TC, aspartate aminotransferase ing follow-up. Additionally, the diabetes diagnosis self-
(AST), TG, hemoglobin A1c (HbA1c), and alanine ami- reported by the participants during follow-up was also
notransferase (ALT) were determined by the automatic included in the analysis.
biochemical analyzer in a standard laboratory. Fatty liver was diagnosed by gastroenterologists based
on the following features of abdominal Doppler ultra-
The lipid parameters were calculated as follows: sound: vascular blurring, liver brightness, deep attenua-
TG/HDL-C ratio = TG/HDL-C [10]; tion, and hepatorenal echo contrast [27].
TC/HDL-C ratio = TC/HDL-C [10];
Non-HDL-C = TC − HDL-C [11]; METS-IR: Ln [BMI × (TG + 2FPG)]/[Ln(HDL-C)] [28].
LDL-C = 90% non-HDL-C − 10%TG [23];
LDL/HDL-C ratio = LDL-C/HDL-C [12]; Statistical analysis
Non-HDL/HDL-C ratio = non-HDL-C/HDL-C [23]; Descriptive data were expressed as the count (percent-
RC = Non-HDL-C − LDL-C [24]; age) of categorical variables and the median (interquar-
RC/HDL-C ratio = RC/HDL-C [25]. tile range) or mean (standard deviation) of continuous
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 4 of 13

variables. Pearson chi-square test for categorical vari- of lipid parameters for predicting the future diabetes
ables and the Mann–Whitney U test or Student’s t-test risk, and calculated the corresponding best threshold.
for continuous variables were used to compare the dif- Additionally, to verify the validity of best thresholds
ferences between groups. In addition, to further quan- of lipid parameters for predicting future diabetes risk,
tify the differences between groups, we also calculated we also fitted the shape of the dose–response correla-
the standardized differences between groups (the differ- tion between lipid parameters and diabetes risk using a
ence > 10% was considered significant) [29, 30]. 4-knots restricted cubic splines (RCS, based on model 4)
Several ancillary analyses were carried out before eval- analysis nested in Cox regression [34]. By visually assess-
uating the association of lipid parameters with diabetes ing the shape of the curve, the value of the lipid param-
risk. Firstly, the Pearson correlation coefficient was calcu- eter corresponding to HR = 1 was determined as the
lated to evaluate the correlation of lipid parameters with best threshold. All analyses were done using the statisti-
METS-IR, in which the non-normal continuous variables cal programming language R (version 3.4.3; www.r-​proje​
were analyzed after the Box-Cox normal transformation ct.​org) and Empower (R) (version 3.4.3; X&Y Solutions,
[31]. Secondly, the collinearity of covariates was checked Inc., Boston, MA, USA). All tests were two-tailed and the
by multiple linear regression, in which covariates with a significance was set to P < 0.05.
variance inflation factor (VIF) > 5 would not be included
in the subsequent multivariate adjustment model [32]. Results
Thirdly, the proportional hazard assumption was checked Baseline characteristics
by visual inspection of the Kaplan–Meier (KM) curve. The study population consisted of 15,464 baseline dia-
The associations and magnitudes of associations betes-free participants, with a mean age of 43.7 (8.9)
between baseline conventional and unconventional lipid years. During a mean follow-up of 6.13 years (min.–max.:
parameter levels and follow-up endpoints were estimated 0.46–13.14 years), 373 participants were diagnosed with
using Cox proportional hazard regression model, and the new-onset diabetes (3.93 per 1000 person-years). Differ-
corresponding HR and 95% CI were recorded. To system- ences in baseline characteristics of the study population
atically consider the influence of confounding factors on were reported in Table 1 according to whether or not dia-
the outcome, we evaluated four multivariate Cox regres- betes will develop in the future. As expected, there were
sion models based on epidemiology [33]. In the first already significant differences at baseline among peo-
model (Model 1), the effects of the principal basic demo- ple with or without diabetes in the future (All P < 0.05).
graphic data (age, sex, and BMI) on the main endpoints Notably, those without diabetes showed the greatest dif-
were considered. On this basis, the second model fur- ferences in baseline glucose metabolism (FPG, HbA1c,
ther incorporated important risk factors such as exercise METS-IR) compared with those who developed diabetes
habits, smoking status, drinking status, and fatty liver during follow-up (Standardized difference > 100%). Sec-
(Model 2). Model 3 further adjusted the related param- ondly, judging from the demographic characteristics of
eters of blood pressure and glucose based on model 2. To the data, there were big differences in weight, waist cir-
fully consider the potential role of covariates, we took the cumference, and BMI between the two groups (Standard-
fourth model as the final model to adjust all non-collinear ized difference: 76–92%). Among the lipid parameters,
variables except lipid parameters (Model 4). except non-HDL-C, LDL-C, and TC, all conventional
We conducted three sensitivity analyses based on and unconventional lipid parameters have shown great
model 4 to assess the robustness of the relationship differences in the early stage of physical examination
between lipid parameters and diabetes risk. Sensitivity-1: (Standardized difference > 70%). Finally, it is also worth
exclude people diagnosed with diabetes within 2 years of mentioning that the difference between the two groups
entry into the study cohort to reduce the potential impact in the population with fatty liver at baseline should also
of reverse causality. Sensitivity-2: to minimize confound- be noted (Standardized difference: 99%). These findings
ing in the relationship between lipid parameters and dia- suggested that early physical examination and assessment
betes due to baseline fatty liver, we limited the analysis to of these widely differing baseline characteristics may be
people without fatty liver. Sensitivity-3: participants with useful for predicting future diabetes risk.
high blood pressure at baseline were excluded because
hypertension was associated with an increased risk of Correlation between baseline conventional
diabetes. and unconventional lipid parameters and METS‑IR
Based on the results of the association analysis, we fur- Pearson correlation analysis showed that all conven-
ther drew time-dependent ROC curves and calculated tional and unconventional lipid parameters were cor-
the area under the 3-year, 6-year, 9-year, and 12-year related with METS-IR at baseline (Additional file 3:
time-dependent ROC curves to quantify the accuracy Table S1). In contrast, the lipid parameter/HDL-C ratio
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 5 of 13

Table 1 Baseline demographic, lifestyle, and laboratory characteristics in participants with and without diabetes
Non-diabetic Diabetes Standardized difference, % P value
(95% CI)

Participants, n 15,091 373

Sex 49 (39, 59) < 0.001
Women 6947 (46.03%) 87 (23.32%)
Men 8144 (53.97%) 286 (76.68%)
Age, years 42.00 (37.00–50.00) 46.00 (41.00–53.00) 40 (30, 51) < 0.001
Height, m 1.65 (0.08) 1.67 (0.09) 19 (9, 29) < 0.001
Weight, kg 60.41 (11.48) 69.84 (13.32) 76 (65, 86) < 0.001
BMI, kg/m2 22.04 (3.07) 25.03 (3.82) 86 (76, 97) < 0.001
WC, cm 76.26 (8.97) 85.08 (10.20) 92 (82, 102) < 0.001
ALT, U/L 17.00 (13.00–23.00) 24.00 (18.00–39.00) 67 (56, 77) < 0.001
AST, U/L 17.00 (14.00–21.00) 20.00 (16.00–26.00) 44 (34, 55) < 0.001
GGT, U/L 15.00 (11.00–22.00) 24.00 (17.00–36.00) 47 (37, 58) < 0.001
HDL-C, mmol/L 1.42 (1.17–1.71) 1.13 (0.96–1.32) 77 (66, 87) < 0.001
TC, mmol/L 5.12 (0.86) 5.43 (0.90) 35 (25, 46) < 0.001
TG, mmol/L 0.72 (0.49–1.11) 1.21 (0.86–1.93) 73 (62, 83) < 0.001
LDL-C, mmol/L 3.03 (2.55–3.55) 3.46 (2.93–3.95) 52 (42, 63) < 0.001
Non-HDL-C, mmol/L 3.59 (3.00–4.23) 4.20 (3.57–4.82) 65 (54, 75) < 0.001
RC, mmol/L 0.53 (0.43–0.67) 0.71 (0.58–0.91) 80 (70, 91) < 0.001
TC/HDL-C ratio 3.50 (2.86–4.39) 4.71 (3.86–5.78) 87 (77, 97) < 0.001
TG/HDL-C ratio 0.50 (0.30–0.89) 1.09 (0.64–1.93) 74 (63, 84) < 0.001
LDL/HDL-C ratio 2.12 (1.59–2.83) 3.01 (2.38–3.85) 83 (73, 93) < 0.001
Non-HDL/HDL-C ratio 2.50 (1.86–3.39) 3.71 (2.86–4.78) 87 (77, 97) < 0.001
RC/HDL-C ratio 0.53 (0.43–0.67) 0.71 (0.58–0.91) 78 (67, 88) < 0.001
FPG, mmol/L 5.15 (0.41) 5.61 (0.36) 121 (111, 132) < 0.001
HbA1c, % 5.16 (0.32) 5.53 (0.37) 107 (97, 118) < 0.001
METS-IR 30.98 (6.36) 38.58 (7.74) 107 (97, 118) < 0.001
SBP, mmHg 114.31 (14.91) 122.03 (15.59) 51 (40, 61) < 0.001
DBP, mmHg 71.44 (10.47) 77.18 (10.23) 55 (45, 66) < 0.001
Fatty liver 2518 (16.69%) 223 (59.79%) 99 (89, 109) < 0.001
Exercise habits 2658 (17.61%) 51 (13.67%) 11 (1, 21) 0.048
Drinking status 21 (11, 31) < 0.001
Non/small 11,539 (76.46%) 266 (71.31%)
Light 1718 (11.38%) 40 (10.72%)
Moderate 1323 (8.77%) 37 (9.92%)
Heavy 511 (3.39%) 30 (8.04%)
Smoking status 45 (35, 55) < 0.001
None 8886 (58.88%) 145 (38.87%)
Past 2875 (19.05%) 77 (20.64%)
Current 3330 (22.07%) 151 (40.48%)
Values were expressed as mean (SD) or medians (quartile interval) or n (%)
BMI body mass index; WC waist circumference; ALT alanine aminotransferase; AST aspartate aminotransferase; GGT​ gamma-glutamyl transferase; HDL-C high-density
lipoprotein cholesterol; TC total cholesterol; Non-HDL-C non-high-density lipoprotein-cholesterol; LDL-C low density lipoprotein cholesterol; TG triglyceride; RC
remnant cholesterol; HbA1c hemoglobin A1c; FPG fasting plasma glucose; SBP systolic blood pressure; DBP diastolic blood pressure; METS-IR metabolic score for
insulin resistance

and HDL-C showed stronger statistical linear correla- ratio; Pearson r = 0.7049 for LDL/HDL-C ratio; Pearson
tions with METS-IR than other lipid parameters (Pear- r = 0.7318 for non-HDL/HDL-C ratio; Pearson r = 0.7139
son r = − 0.7139 for HDL-C; Pearson r = 0.7324 for for RC/HDL-C ratio).
TC/HDL-C ratio; Pearson r = 0.7447 for TG/HDL-C
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 6 of 13

Relationship between baseline conventional Table 3 Adjusted hazard ratios and 95% confidence intervals for
and unconventional lipid parameters and future diabetes diabetes risk associated with the lipid parameters in different test
Before performing multivariate Cox regression analy- populations: sensitivity analysis
sis, our data passed the log-linear assumption and the HR (95% CI)
proportional hazard assumption tests. Additional file 3:
Sensitivity-1 Sensitivity-2 Sensitivity-3
Table S2 shows the collinearity screening procedure
according to VIF, and the collinear variables waist cir- HDL-C 0.63 (0.41, 0.99) 0.49 (0.29, 0.82) 0.50 (0.34, 0.75)
cumference, SBP, and weight will not be included in the TC 0.99 (0.86, 1.14) 0.89 (0.73, 1.08) 0.97 (0.85, 1.10)
following multivariable model. Additional file 2: Fig. TG 1.17 (1.01, 1.34) 1.24 (1.00, 1.52) 1.30 (1.15, 1.46)
S1 takes the ratio of RC/HDL-C as an example to show LDL-C 1.00 (0.85, 1.19) 0.94 (0.75, 1.18) 0.98 (0.84, 1.14)
the KM curve corresponding to the quartiles of the RC/ Non-HDL-C 1.04 (0.90, 1.20) 0.99 (0.81, 1.20) 1.04 (0.92, 1.19)
HDL-C ratio, and the results showed that there was no RC 1.64 (0.98, 2.74) 1.74 (0.82, 3.72) 2.25 (1.44, 3.53)
intersection between the KM curves. TC/HDL-C ratio 1.10 (1.00, 1.21) 1.17 (1.01, 1.35) 1.15 (1.05, 1.26)
Based on epidemiology, we ran four multivariate Cox TG/HDL-C ratio 1.14 (1.02, 1.27) 1.19 (1.03, 1.37) 1.23 (1.13, 1.34)
regression models to evaluate the relationship of base- Non-HDL/HDL-C 1.10 (1.00, 1.21) 1.17 (1.01, 1.35) 1.15 (1.05, 1.26)
line conventional and unconventional lipid param- ratio
eters with future diabetes risk (Table 2). In the models LDL/HDL-C ratio 1.11 (0.98, 1.26) 1.20 (1.00, 1.44) 1.16 (1.03, 1.29)
adjusted for demographic and lifestyle factors (Model 1 RC/HDL-C ratio 3.69 (1.09, 12.47) 14.01 (2.75, 71.47) 7.73 (2.68, 22.28)
and Model 2), all lipid parameters were associated with Note 1: (1) sensitivity-1: subjects with a follow-up of less than 3 years were
future diabetes risk. However, the associations between excluded (n = 11,237); (2) sensitivity-2: subjects diagnosed with fatty liver at
baseline were excluded. (n = 12,823); (3) sensitivity-3: excluding subjects whose
LDL-C, non-HDL-C, and TC with the risk of diabetes SBP ≥ 140 mmHg or DBP ≥ 90 mmHg (n = 14,878);
disappeared after further consideration of the effects of Note 2: sensitivity-1 adjusted for sex, age, BMI, habits of exercise, drinking
blood pressure and glucose (Model 3). In the final model status, smoking status, fatty liver, SBP, FPG, HbA1c, height, ALT, AST and GGT.
Sensitivity-2 adjusted for sex, age, BMI, habits of exercise, drinking status,
(Model 4), we adjusted all non-collinear variables except smoking status, SBP, FPG, HbA1c, height, ALT, AST and GGT. Sensitivity-3
lipid parameters and obtained similar results to Model adjusted for sex, age, BMI, habits of exercise, drinking status, smoking status,
fatty liver, FPG, HbA1c, height, ALT, AST and GGT​
3. Among the conventional lipid parameters, only TG
HR hazard ratios; CI confidence interval; other abbreviations as in Table 1
and HDL-C were related to the risk of diabetes, while
all unconventional lipid parameters except non-HDL-C
were associated with the risk of diabetes. It is worth men- HDL-C ratio can better reflect the risk of developing dia-
tioning that by comparing the HR of lipid parameters betes in the future than other lipid parameters (HR: 6.75,
corresponding to the risk of diabetes, we found that RC/ 95% CI 2.40–18.98).

Table 2 Association of baseline conventional and unconventional lipid parameters and future diabetes
HR (95% CI)
Model 1 Model 2 Model 3 Model 4

HDL-C 0.32 (0.22, 0.47) 0.49 (0.33, 0.73) 0.55 (0.38, 0.81) 0.53 (0.36, 0.78)
TC 1.18 (1.05, 1.33) 1.13 (1.00, 1.28) 0.94 (0.83, 1.06) 0.92 (0.81, 1.04)
TG 1.48 (1.34, 1.63) 1.34 (1.20, 1.50) 1.21 (1.07, 1.36) 1.20 (1.06, 1.35)
LDL-C 1.27 (1.10, 1.46) 1.17 (1.01, 1.35) 0.94 (0.81, 1.09) 0.93 (0.80, 1.08)
Non-HDL-C 1.33 (1.19, 1.50) 1.21 (1.07, 1.37) 1.00 (0.88, 1.13) 0.99 (0.87, 1.12)
RC 4.55 (3.16, 6.55) 3.06 (2.03, 4.60) 1.72 (1.11, 2.67) 1.66 (1.07, 2.58)
TC/HDL-C ratio 1.34 (1.24, 1.44) 1.21 (1.12, 1.32) 1.11 (1.02, 1.21) 1.11 (1.02, 1.21)
TG/HDL-C ratio 1.34 (1.25, 1.42) 1.26 (1.16, 1.36) 1.19 (1.09, 1.30) 1.18 (1.08, 1.29)
Non-HDL/HDL-C ratio 1.34 (1.24, 1.44) 1.21 (1.12, 1.32) 1.11 (1.02, 1.21) 1.11 (1.02, 1.21)
LDL/HDL-C ratio 1.39 (1.26, 1.52) 1.24 (1.11, 1.37) 1.11 (1.00, 1.24) 1.12 (1.00, 1.24)
RC/HDL-C ratio 25.49 (9.69, 67.06) 8.13 (2.84, 23.24) 6.08 (2.17, 17.08) 6.75 (2.40, 18.98)
Model 1 adjusted for sex, age and BMI
Model 2 adjusted for sex, age, BMI, exercise habits, fatty liver, drinking status and smoking status
Model 3 adjusted for sex, age, BMI, exercise habits, fatty liver, drinking status, smoking status, SBP, FPG and HbA1c
Model 4 adjusted for sex, age, BMI, exercise habits, fatty liver, drinking status, smoking status, SBP, FPG, HbA1c, height, ALT, AST and GGT​
HR hazard ratios; CI confidence interval; other abbreviations as in Table 1
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 7 of 13

Sensitivity analysis used to predict future diabetes risk only fluctuated within
After adjusting for covariates based on model 4, the a narrow range (HDL-C: 1.3602–1.5025; TG: 0.8242–
results of the three sensitivity analyses (Table 3) were 0.8580; RC: 0.5578–0.6280; TC/HDL-C ratio: 3.6975–
consistent with the main results of Table 2: there were 3.9741; TG/HDL-C ratio: 0.5983–0.7276; LDL-HDL-C
no significant correlations of LDL-C, non-HDL-C, TC ratio: 2.2192–2.3191; Non-HDL/HDL-C ratio: 2.6975–
with diabetes risk, while other lipid parameters were all 2.9741; RC/HDL-C ratio: 0.3509–0.3681). Moreover,
significantly correlated with diabetes risk, and the HR of through RCS, we confirmed that the thresholds of HDL-
diabetes risk associated with RC/HDL-C ratio was higher C, TG, RC, TC/HDL-C ratio, TG/HDL-C ratio, LDL/
than other lipid parameters. HDL-C ratio, non-HDL/HDL-C ratio, and RC/HDL-C
ratio were 1.46 mmol/L, 0.90 mmol/L, 0.58 mmol/L, 3.75,
Predictive value of conventional and unconventional lipid 0.71, 2.30, 2.75 and 0.36, respectively (Additional file 2:
parameters for future diabetes risk Fig. S2A–H). In the threshold analysis of RCS and ROC,
Time-dependent ROC curves were drawn to evalu- the thresholds of all lipid parameters were within the sta-
ate the predictive value of HDL-C, TG, and unconven- ble fluctuation range, except for the slight differences in
tional lipid parameters for future diabetes risk (Table 4). the threshold levels of TG.
Figure 2 shows the area under the curve of these lipid
parameters over time. Overall, the predictive value of all Discussion
lipid parameters for diabetes risk decreased over time. There were four main findings in this longitudinal cohort
Furthermore, relatively speaking, the predictive value study: (1) In the baseline non-diabetic population, only
of unconventional lipid parameters for the risk of dia- TG and HDL-C among the conventional lipid param-
betes was slightly higher than that of conventional lipid eters were associated with future diabetes risk, while all
parameters. Among them, RC/HDL-C ratio had the best the unconventional lipid parameters except non-HDL-
predictive value for predicting short-term diabetes risk, C were significantly associated with the future risk of
non-HDL/HDL-C ratio and LDL/HDL-C ratio had the diabetes. (2) Compared with other conventional and
best predictive value for the medium-term diabetes risk, unconventional lipid parameters, the RC/HDL-C ratio
while non-HDL/HDL-C ratio RC and TC/HDL-C ratio can better reflect the future diabetes risk. (3) Compared
had the best predictive value for the medium-and long- with conventional lipid parameters, unconventional lipid
term diabetes risk. parameters were of higher value for predicting the diabe-
tes risk in the future. (4) RC/HDL-C ratio had the best
Threshold analysis of conventional and unconventional predictive value for the short-term diabetes risk, non-
lipid parameters for predicting the risk of diabetes HDL/HDL-C ratio and LDL/HDL-C ratio for the short-to
in the future medium-term diabetes risk, while RC, non-HDL/HDL-C
Using the time-dependent ROC curve, we also calculated ratio, and TC/HDL-C ratio for the mid-to long-term and
the best thresholds of conventional and unconventional long-term diabetes risk.
lipid parameters for predicting the risk of diabetes at 3, Diabetes mellitus, one of the most rapidly growing
6, 9, and 12 years. As can be seen from Fig. 3, despite the chronic non-communicable diseases in this century, is a
passage of time, the thresholds of all lipid parameters complex metabolic disease that presents with different

Table 4 Best threshold and areas under the time-dependent receiver operating characteristic curves for each lipid parameters
predicting future diabetes risk
3-years 6-years 9-years 12-years
AUC (best threshold) AUC (best threshold) AUC (best threshold) AUC (best threshold)

HDL-C 0.55142 (1.5025) 0.55667 (1.5025) 0.55579 (1.3602) 0.52807 (1.4792)

TG 0.66443 (0.8242) 0.66771 (0.8468) 0.67733 (0.8242) 0.64812 (0.8580)
RC 0.66802 (0.6280) 0.69109 (0.5832) 0.68682 (0.5705) 0.66872 (0.5578)
TC/HDL-C ratio 0.6753 (3.9741) 0.7185 (3.7134) 0.69438 (3.6975) 0.66934 (3.7351)
TG/HDL-C ratio 0.6731 (0.6567) 0.69201 (0.5983) 0.68738 (0.6118) 0.65279 (0.7276)
LDL/HDL-C ratio 0.67259 (2.3345) 0.71343 (2.3191) 0.6872 (2.2192) 0.6591 (2.3001)
Non-HDL/HDL-C ratio 0.67532 (2.9741) 0.71851 (2.7134) 0.69438 (2.6975) 0.66935 (2.7351)
RC/HDL-C ratio 0.68861 (0.3610) 0.69989 (0.3509) 0.67917 (0.3645) 0.65162 (0.3681)
AUC​area under the curve; other abbreviations as in Table 1
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 8 of 13

Fig. 2 The area under the receiver operator characteristics curve of lipid parameters varying with time to predict the future risk of diabetes. AUC​
area under the curve

subphenotypes according to various etiologies [35, 36]. is still the primary task of lipid management in patients
The most common one is the metabolic syndrome- with diabetes [37], and multifaceted treatment strate-
related phenotype, which is mainly characterized by gies for hyperlipidemia, hyperglycemia, and hypertension
dyslipidemia and obesity. Effective management of these can further improve the health status of diabetic patients
complex lipid metabolic disorders and obesity is an [42]. In the current study, we also observed, from the
important part of comprehensive diabetes prevention, comparison of baseline information, that compared with
treatment, and care, and can reduce the risk of cardio- those who did not develop diabetes during the follow-up
vascular morbidity and mortality [37]. The relationship period, people with new-onset diabetes showed signifi-
between conventional lipid parameters HDL-C, TC, cant differences in all conventional and unconventional
LDL-C, and TG and diabetes has been widely studied lipid parameters except non-HDL-C, LDL-C, and TC
in the past, in which high concentrations of TG and low at baseline. Moreover, in regression analysis, when only
concentrations of HDL-C significantly increased diabetes adjusting for lifestyle and demographic data, all conven-
risk has become the consensus of almost every scholar tional lipid parameters were associated with future dia-
in the field of metabolism [37–39], but there is still some betes risk. In further covariation-adjusted models, the
debate about the direct relationship between LDL-C, TC, associations of LDL-C and TC with diabetes risk disap-
and diabetes [10, 40, 41]. However, it should be men- peared. A similar situation was also described in the work
tioned that in terms of treatment, reducing LDL-C levels of Khaloo et al. [10]. This phenomenon suggested that
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 9 of 13

Fig. 3 Threshold fluctuation of lipid parameters used to predict future diabetes risk

LDL-C and TC may play a certain role in the risk of dia- in many studies [8–12, 17, 18], but comparative studies
betes in the future, but this role may be weak and unsta- on lipid parameters for predicting future diabetes are
ble compared with other lipid parameters. extremely limited. There were still some differences in
Unconventional lipid parameters such as lipid ratio, the results of several longitudinal studies that have been
RC, and non-HDL-C are hot topics in recent years. These published. Hadaegh et al. conducted the first compara-
lipid parameters are calculated by conventional lipid tive study in Iran in 2010 to evaluate the utility of con-
parameters according to certain formulas or direct divi- ventional and unconventional lipid parameters to predict
sion [10–12, 23–25]. At present, a series of epidemiologi- future diabetes risk [19]. They measured/calculated TG/
cal studies have been carried out using unconventional HDL-C ratios, non-HDL-C, HDL-C, TC, and TG in 5,201
lipid parameters as novel markers in the endocrine sys- baseline diabetes-free subjects. After a median follow-up
tem, cardio-cerebrovascular system, digestive system, of 5.6 years, they found that the TG/HDL-C ratio was a
respiratory system, and urinary system diseases [8–12, better indicator of future diabetes risk in women, while
43–48]. In general, unconventional lipid parameters the TC/HDL-C ratio was a better indicator of future dia-
improved the ability of conventional lipid parameters to betes risk in men. Subsequently, in 2011, South Korea’s
assess and identify the risk of most diseases. Seo et al. conducted another longitudinal study of 5577
The relationship between conventional and unconven- subjects without diabetes [18]. During the 4-year fol-
tional lipid parameters and diabetes has been reported low-up, they found that TC/HDL-C ratio had the best
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 10 of 13

performance in reflecting future diabetes risk. It should relative simplicity and accuracy [51, 52]. Although the
be noted that in the study of Professor Hadaegh and Pro- conventional lipid measurement methods have excel-
fessor Seo [18, 19], despite their follow-up, the time fac- lent diagnostic performance, they still require expensive
tor was not taken into account in the final data analysis. instruments, trained operators, and dedicated laborato-
The results of their final analysis were based on multi- ries. Therefore, conventional lipid measurement methods
variate Logistic regression analysis, which may be biased. may not be optimal for primary health care institutions
Recently, a longitudinal study by Khaloo et al. involving and people with limited mobility [53]. Recently, biosen-
5474 non-diabetic subjects showed that Ln-TG/HDL-C sors for measuring lipids and lipoproteins have been
ratio was a better lipid parameter to predict future dia- developed for preventive and therapeutic monitoring of
betes risk [10]. Summing up the results of several similar chronic diseases at the proof-of-concept stage, allowing
studies above, among multiple lipid parameters, the TC/ simultaneous measurement of multiple lipid parameters
HDL-C ratio and TG/HDL-C ratio may be good param- in a home setting by a user with minimal training [54–
eters for predicting future diabetes. In our current study, 56]. Compared with conventional methods, biosensors
we included more unconventional lipid parameters in a for lipids and lipoproteins are more convenient, rapid,
cohort of 15,464 participants with baseline normogly- and easy to operate. In addition, recent studies have
cemia. During a follow-up of up to 13 years, we further shown that new biomaterials (such as nanomaterials)
confirmed their findings: both TC/HDL-C ratio and TG/ have better biocompatibility, stability, and unique physi-
HDL-C ratio were good lipid parameters for predict- cal, electronic and chemical properties [52, 57, 58]. These
ing future diabetes risk. On this basis, our results added new materials and techniques may have great potential
more evidence, and we found that among all lipid param- for early screening of people at high risk of diabetes by
eters, the RC/HDL-C ratio was the best parameter to testing lipid parameters [59].
reflect the risk of developing diabetes in the future. This There are many mechanisms by which lipid param-
finding has not been reported in previous similar studies, eters are associated with diabetes. From a clinical point
and so far as we know, this is the first study to evaluate of view, the atherogenic effect of lipid parameters and
the relationship between diabetes and RC/HDL-C ratio. IR caused by lipid parameters may be the main factors
Several studies have been reported on the thresholds associated with diabetes [4, 60–62]. Among them, the
of conventional versus unconventional lipid parameters vasoactive hormone pathway including the renin–angi-
for the diagnosis/prediction of diabetes [9, 18, 49, 50]. In otensin–aldosterone system (RAAS) seems to play an
summary, the main differences between the results of the important role in this process [63]. The RAAS is known
threshold analysis of lipid parameters in the published to be an important system in the body for maintaining
studies and the analysis in our current study are in the plasma sodium concentration, arterial blood pressure,
TG and the consideration of time variables. Compared and extracellular volume [64]. The imbalance between
with other threshold analysis studies, our current study renin and angiotensin II can lead to a large number of
fully considered time-dependent variables and assessed chronic and acute diseases [64, 65], and plaque forma-
baseline lipid parameters as predictors of future diabetes tion induced by angiotensin II in the early stage is one
risk, while the existing similar studies focused more on of the most important effects of RAAS on atherosclero-
the present [9, 18, 49, 50]. Furthermore, in the current sis [66], while under pathological conditions, RAAS also
study, we also evaluated the predictive value of conven- contributes directly or indirectly to the development of
tional and unconventional lipid parameters for predicting atherosclerosis and its various complications through its
short -, medium-and long-term diabetes risk. Overall, the effects on other systems [67]. In addition to this, when
RC/HDL-C ratio had the best predictive value for pre- renin and angiotensin II are imbalanced, the RAAS detri-
dicting the short-term diabetes risk, non-HDL/HDL-C mental axis will also increase the release of inflammatory
ratio and LDL/HDL-C ratio for the short-to medium- cytokines, and generate and increase oxidative stress;
term diabetes risk, while RC, non-HDL/HDL-C ratio, these pathological changes will further promote the for-
and TC/HDL-C ratio for the mid-to long-term and long- mation of atherosclerosis, exacerbated IR, and decreased
term diabetes risk. insulin secretion [66, 68]. On the other hand, it should
Conventional methods for measuring lipids and lipo- also be noted that serum hepcidin and hepcidin/ferritin
proteins include chemical methods and enzymatic meth- ratio also play an important role in the development of IR
ods. Generally speaking, chemical methods are relatively and diabetes [69, 70]. In the current research, we calcu-
time-consuming, so they are usually only used for cali- lated the IR substitute index METS-IR, and the correla-
bration in the clinic, while the enzyme method has been tion analysis showed that all lipid parameters were closely
used as the most commonly used analysis method in related to METS-IR, among which TG/HDL-C ratio had
automatic biochemical analytical instruments due to its the strongest correlation (Pearson r = 0.7447 for TG/
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 11 of 13

HDL-C ratio). Additionally, it is worth noting that uncon- postprandial glucose. Therefore, the current study may
ventional lipid parameters were also more powerful in underestimate the incidence of diabetes. (4) The types of
identifying atherosclerosis and IR than conventional diabetes have not been distinguished in the current study,
lipid parameters [61, 71]. Combining current research but based on a large number of published research data
findings, we have several simple suggestions for a series [1, 8, 74], the current results are more suitable for type
of possible future work: (1) it is suggested that medical 2 diabetes, and the applicability in other special types of
staff should strengthen their understanding of unconven- diabetes needs further study.
tional lipid parameters. (2) It is suggested that based on
conventional lipid measurement, the inspection center of Conclusion
medical institutions can add unconventional lipid param- In summary, our results demonstrated the importance
eters as detection items to the display list by adding an of unconventional lipid parameters for predicting the
algorithm to the computer. (3) It is suggested that more risk of diabetes in the future. Compared with conven-
comparative studies on conventional and unconventional tional lipid parameters, it is worthwhile to use uncon-
lipid parameters should be carried out to find out the ventional lipid parameters to predict the future risk
value of lipid parameters in the risk assessment of other of diabetes. It is recommended to incorporate uncon-
diseases. (4) It is suggested to add more unconventional ventional lipid parameters as soon as possible in clini-
lipid parameters on the basis of non-HDL-C as the main cal practice for routine assessment of diabetes risk and
target of lipid management [72]. (5) It is suggested that treatment monitoring.
unconventional lipid parameters should be considered as
a potential target for developing high-sensitivity biosen- Abbreviations
sors for diabetes. TC: Total cholesterol; TG: Triglyceride; HDL-C: High-density lipoprotein choles-
terol; LDL-C: Low-density lipoprotein cholesterol; RC: Remnant cholesterol; HR:
Hazard ratio; CI: Confidence interval; ROC: Receiver operator characteristics;
BMI: Body mass index; AST: Aspartate aminotransferase; GGT​: Gamma-gluta-
Advantages and limitations of research myl transferase; ALT: Alanine aminotransferase; FPG: Fasting plasma glucose;
While interpreting the current research results, there are HbA1c: Hemoglobin A1c; METS-IR: Metabolic score for insulin resistance; VIF:
Variance inflation factor; RCS: Restricted cubic splines; KM: Kaplan–Meier.
several research advantages to be mentioned: (1) Com-
pared with similar studies, the current study further
expanded the sample size based on longitudinal design Supplementary Information
The online version contains supplementary material available at https://​doi.​
and included more unconventional lipid parameters. (2) org/​10.​1186/​s12967-​022-​03470-z.
We investigated the relationship of RC/HDL-C ratio with
diabetes for the first time, and through multivariate Cox
regression analysis found that compared with other con- Additional file 1: Text S1. STROBE Statement—checklist of items that
should be included in reports of observational studies.
ventional and unconventional lipid parameters, the RC/
Additional file 2: Figure S1. Proportional hazards assumption checking.
HDL-C ratio could better reflect the future risk of diabe-
Figure S2. A RCS evaluates the best threshold of HDL-C for predict-
tes. (3) The current study compared the predictive value ing future diabetes risk. B RCS evaluates the best threshold of TG for
of lipid parameters for future diabetes by time-dependent predicting future diabetes risk. C RCS evaluates the best threshold of RC
for predicting future diabetes risk. D RCS evaluates the best threshold of
ROC curve analysis for the first time.
TC/HDL-C ratio for predicting future diabetes risk. E RCS evaluates the
Several research limitations need to be acknowledged: best threshold of TG/HDL-C ratio for predicting future diabetes risk. F
(1) the endpoint of the current study is new-onset diabe- RCS evaluates the best threshold of LDL/HDL-C ratio for predicting future
diabetes risk. G RCS evaluates the best threshold of non-HDL/HDL-C ratio
tes events, while the death events during follow-up are
for predicting future diabetes risk. H RCS evaluates the best threshold of
not recorded in the current dataset, which may have a RC/HDL-C ratio for predicting future diabetes risk.
certain competitive risk to the current research results. Additional file 3: Table S1. Pearson correlation analysis of baseline
(2) In this study, we analyzed the predictive value of sev- conventional and unconventional lipid parameters and METS-IR. Table S2.
eral lipid parameters for the risk of developing diabetes Collinearity diagnostics steps.
in 3, 6, 9, and 12 years respectively, but the results of the
current study may be more suitable for the short-and Acknowledgements
I would like to thank my colleagues at the Cardiovascular Institute for their
medium-term risk prediction of diabetes because the
supervision and help in statistical analysis.
body’s metabolic profile gradually deteriorates with age
[73], the previously predicted risk of diabetes may no Author contributions
YZ and GT-S: writing—original draft preparation. YZ, MB-K, GT-S, YJ-Z and
longer apply. (3) The participants of the current study
SH-Z: writing—reviewing and editing. YZ, MB-K, RJ-Y and YJ-Z: formal analysis
were ordinary people who underwent health check-ups. and validation. GT-S and YZ: data curation and validation. SH-Z, GT-S and YZ:
Generally speaking, most people who receive a health Supervision. YZ: conceptualization. All authors read and approved the final
manuscript.
examination do not have a routine measurement of
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 12 of 13

Funding 10. Khaloo P, Hasheminia M, Tohidi M, Abdi H, Mansournia MA, Azizi F, et al.
This study was funded by the Natural Science Foundation of Jiangxi Province Impact of 3-year changes in lipid parameters and their ratios on incident
(No. 20192BAB205007). type 2 diabetes: Tehran lipid and glucose study. Nutr Metab. 2018;15:50.
11. Zhang N, Hu X, Zhang Q, Bai P, Cai M, Zeng TS, et al. Non-high-density
Availability of data and materials lipoprotein cholesterol: high-density lipoprotein cholesterol ratio is an
The data used in this study have been uploaded to the “Dryad” database by independent risk factor for diabetes mellitus: results from a population-
Professor Okamura et al. based cohort study. J Diabetes. 2018;10:708–14.
12. Hong M, Ling Y, Lu Z, Liu Y, Gu P, Shao J, et al. Contribution and interaction
of the low-density lipoprotein cholesterol to high-density lipoprotein
Declarations cholesterol ratio and triglyceride to diabetes in hypertensive patients: a
cross-sectional study. J Diabetes Investig. 2019;10:131–8.
Ethics approval and consent to participate 13. van Wijk DF, Stroes ES, Kastelein JJ. Lipid measures and cardiovascular
In the initial survey, the Murakami Memorial Hospital Institutional Ethics disease prediction. Dis Markers. 2009;26:209–16.
Review Board approved the NAGALA cohort study, and informed written 14. Manickam P, Rathod A, Panaich S, Hari P, Veeranna V, Badheka A, et al.
consent was obtained from each participant for the use of their data (IRB2018- Comparative prognostic utility of conventional and novel lipid param-
09-01). This study was a secondary analysis of data from the NAGALA cohort. eters for cardiovascular disease risk prediction: do novel lipid parameters
The research scheme has been approved by the Ethics Review Committee offer an advantage? J Clin Lipidol. 2011;5:82–90.
of Jiangxi Provincial People’s Hospital (IRB2021-066), and the whole process 15. Zhu L, Lu Z, Zhu L, Ouyang X, Yang Y, He W, et al. Lipoprotein ratios are
follows the Helsinki Declaration. See STROBE statement in Additional file 1: better than conventional lipid parameters in predicting coronary heart
Text S1. disease in Chinese Han people. Kardiol Pol. 2015;73:931–8.
16. Tancredi M, Rosengren A, Svensson AM, Kosiborod M, Pivodic A, Gudb-
Consent for publication jörnsdottir S, et al. Excess mortality among persons with type 2 diabetes.
Not applicable. N Engl J Med. 2015;373:1720–32.
17. Chen Z, Hu H, Chen M, Luo X, Yao W, Liang Q, et al. Association of
Competing interests triglyceride to high-density lipoprotein cholesterol ratio and incident of
All authors have no competing interests to declare. diabetes mellitus: a secondary retrospective analysis based on a Chinese
cohort study. Lipids Health Dis. 2020;19:33.
Author details 18. Seo MH, Bae JC, Park SE, Rhee EJ, Park CY, Oh KW, et al. Association of lipid
1
Department of Cardiology, Jiangxi Provincial People’s Hospital, Nan- and lipoprotein profiles with future development of type 2 diabetes in
chang 330006, Jiangxi, China. 2 Medical College of Nanchang University, nondiabetic Korean subjects: a 4-year retrospective, longitudinal study. J
Nanchang 330006, Jiangxi, China. 3 Department of Endocrinology, Jiangxi Pro- Clin Endocrinol Metab. 2011;96:E2050–4.
vincial People’s Hospital, Nanchang 330006, Jiangxi, China. 4 Jiangxi Provincial 19. Hadaegh F, Hatami M, Tohidi M, Sarbakhsh P, Saadat N, Azizi F. Lipid ratios
People’s Hospital, Jiangxi Cardiovascular Research Institute, Nanchang 330006, and appropriate cut off values for prediction of diabetes: a cohort of
Jiangxi, China. Iranian men and women. Lipids Health Dis. 2010;9:85.
20. Okamura T, et al. Data from: ectopic fat obesity presents the greatest risk
Received: 20 April 2022 Accepted: 30 May 2022 for incident type 2 diabetes: a population-based longitudinal study. 2019.
Dryad Dataset. https://​doi.​org/​10.​5061/​dryad.​8q0p1​92.
21. Okamura T, Hashimoto Y, Hamaguchi M, Obora A, Kojima T, Fukui M.
Ectopic fat obesity presents the greatest risk for incident type 2 diabetes:
a population-based longitudinal study. Int J Obes. 2019;43:139–48.
22. Hashimoto Y, Hamaguchi M, Kojima T, Ohshima Y, Ohbora A, Kato T, et al.
References Modest alcohol consumption reduces the incidence of fatty liver in men:
1. Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, et al. a population-based large-scale cohort study. J Gastroenterol Hepatol.
Global and regional diabetes prevalence estimates for 2019 and projec- 2015;30:546–52.
tions for 2030 and 2045: results from the International Diabetes Federa- 23. Chen Y, Zhang X, Pan B, Jin X, Yao H, Chen B, et al. A modified formula for
tion Diabetes Atlas, 9th edition. Diabetes Res Clin Pract. 2019;157:107843. calculating low-density lipoprotein cholesterol values. Lipids Health Dis.
2. Mortality and Causes of Death Collaborators. Global, regional, and 2010;9:52.
national life expectancy, all-cause mortality, and cause-specific mortality 24. Nordestgaard BG, Varbo A. Triglycerides and cardiovascular disease.
for 249 causes of death, 1980–2015: a systematic analysis for the Global Lancet. 2014;384:626–35.
Burden of Disease Study 2015. Lancet. 2016;388:1459–544. 25. Zou Y, Hu C, Kuang M, Chai Y. Remnant cholesterol/high-density lipopro-
3. GBD 2015 DALYs and HALE Collaborators. Global, regional, and national tein cholesterol ratio is a new powerful tool for identifying non-alcoholic
disability-adjusted life-years (DALYs) for 315 diseases and injuries and fatty liver disease. BMC Gastroenterol. 2022;22(1):134.
healthy life expectancy (HALE), 1990–2015: a systematic analysis for the 26. American Diabetes Association. Standards of medical care in diabe-
Global Burden of Disease Study 2015. Lancet. 2016;388:1603–58. tes—2011. Diabetes Care. 2011;34(Suppl 1):S11-61.
4. Krauss RM. Lipids and lipoproteins in patients with type 2 diabetes. 27. Hamaguchi M, Kojima T, Itoh Y, Harano Y, Fujii K, Nakajima T, et al. The
Diabetes Care. 2004;27:1496–504. severity of ultrasonographic findings in nonalcoholic fatty liver disease
5. Sascău R, Clement A, Radu R, Prisacariu C, Stătescu C. Triglyceride-rich reflects the metabolic syndrome and visceral fat accumulation. Am J
lipoproteins and their remnants as silent promoters of atherosclerotic Gastroenterol. 2007;102:2708–15.
cardiovascular disease and other metabolic disorders: a review. Nutrients. 28. Bello-Chavolla OY, Almeda-Valdes P, Gomez-Velasco D, Viveros-Ruiz T,
2021;13:1774. Cruz-Bautista I, Romo-Romo A, et al. METS-IR, a novel score to evaluate
6. Borén J, Taskinen MR, Björnson E, Packard CJ. Metabolism of triglyceride- insulin sensitivity, is predictive of visceral adiposity and incident type 2
rich lipoproteins in health and dyslipidaemia. Nat Rev Cardiol. 2022. diabetes. Eur J Endocrinol. 2018;178:533–44.
https://​doi.​org/​10.​1038/​s41569-​022-​00676-y. 29. Sato T, Matsuyama Y. Marginal structural models as a tool for standardiza-
7. Krauss RM. Triglycerides and atherogenic lipoproteins: rationale for lipid tion. Epidemiology. 2003;14:680–6.
management. Am J Med. 1998;105:58S-62S. 30. Muanda FT, Weir MA, Bathini L, Blake PG, Chauvin K, Dixon SN, et al. Asso-
8. Yang W, Lu J, Weng J, Jia W, Ji L, Xiao J, et al. Prevalence of diabetes ciation of baclofen with encephalopathy in patients with chronic kidney
among men and women in China. N Engl J Med. 2010;362:1090–101. disease. JAMA. 2019;322(20):1987–95.
9. Xie G, Zhong Y, Yang S, Zou Y. Remnant cholesterol is an independent 31. Box GEP, Cox DR. An analysis of transformations. J R Stat Soc Ser B.
predictor of new-onset diabetes: a single-center cohort study. Diabetes 1964;26:211–52.
Metab Syndr Obes. 2021;14:4735–45.
Sheng et al. Journal of Translational Medicine (2022) 20:266 Page 13 of 13

32. Kim JH. Multicollinearity and misleading statistical results. Korean J Anes- 55. Saxena U, Das AB. Nanomaterials towards fabrication of cholesterol
thesiol. 2019;72:558–69. biosensors: key roles and design approaches. Biosens Bioelectron.
33. Fitchett EJA, Seale AC, Vergnano S, Sharland M, Heath PT, Saha SK, et al. 2016;75:196–205.
Strengthening the reporting of observational studies in epidemiology for 56. Ferreira CE, França CN, Correr CJ, Zucker ML, Andriolo A, Scartezini M.
newborn infection (STROBE-NI): an extension of the STROBE statement Clinical correlation between a point-of-care testing system and labora-
for neonatal infection research. Lancet Infect Dis. 2016;16(10):e202–13. tory automation for lipid profile. Clin Chim Acta. 2015;446:263–6.
34. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat 57. Suhito IR, Koo KM, Kim TH. Recent advances in electrochemical sen-
Med. 1989;8:551–61. sors for the detection of biomolecules and whole cells. Biomedicines.
35. Stidsen JV, Henriksen JE, Olsen MH, Thomsen RW, Nielsen JS, Rungby J, 2020;9(1):15.
et al. Pathophysiology-based phenotyping in type 2 diabetes: a clinical 58. Howes PD, Chandrawati R, Stevens MM. Colloidal nanoparticles as
classification tool. Diabetes Metab Res Rev. 2018;34: e3005. advanced biological sensors. Science. 2014;346:1247390.
36. Wagner R, Heni M, Tabák AG, Machann J, Schick F, Randrianarisoa E, et al. 59. Salek-Maghsoudi A, Vakhshiteh F, Torabi R, Hassani S, Ganjali MR, Norouzi
Pathophysiology-based subphenotyping of individuals at elevated risk P, et al. Recent advances in biosensor technology in assessment of early
for type 2 diabetes. Nat Med. 2021;27:49–57. diabetes biomarkers. Biosens Bioelectron. 2018;99:122–35.
37. Kendall DM. The dyslipidemia of diabetes mellitus: giving triglycerides 60. Kane JP, Pullinger CR, Goldfine ID, Malloy MJ. Dyslipidemia and diabetes
and high-density lipoprotein cholesterol a higher priority? Endocrinol mellitus: role of lipoprotein species and interrelated pathways of lipid
Metab Clin North Am. 2005;34:27–48. metabolism in diabetes mellitus. Curr Opin Pharmacol. 2021;61:21–7.
38. Alexopoulos AS, Qamar A, Hutchins K, Crowley MJ, Batch BC, Guyton JR. 61. Zhang L, Chen S, Deng A, Liu X, Liang Y, Shao X, et al. Association
Triglycerides: emerging targets in diabetes care? Review of moderate between lipid ratios and insulin resistance in a Chinese population. PLoS
hypertriglyceridemia in diabetes. Curr Diab Rep. 2019;19:13. ONE. 2015;10: e0116110.
39. Bitzur R, Cohen H, Kamari Y, Shaish A, Harats D. Triglycerides and HDL cho- 62. Poznyak A, Grechko AV, Poggio P, Myasoedova VA, Alfieri V, Orekhov
lesterol: stars or second leads in diabetes? Diabetes Care. 2009;32(Suppl AN. The diabetes mellitus-atherosclerosis connection: the role of lipid
2):S373–7. and glucose metabolism and chronic inflammation. Int J Mol Sci.
40. Pan W, Sun W, Yang S, Zhuang H, Jiang H, Ju H, et al. LDL-C plays a 2020;21:1835.
causal role on T2DM: a Mendelian randomization analysis. Aging. 63. Jandeleit-Dahm K, Cooper ME. Hypertension and diabetes: role of
2020;12:2584–94. the renin-angiotensin system. Endocrinol Metab Clin North Am.
41. Cui J, Ma P, Sun JP, Baloch Z, Yin F, Xin HL, et al. The ability of baseline 2006;35(469–90):vii.
triglycerides and total cholesterol concentrations to predict incidence of 64. Patel S, Rauf A, Khan H, Abu-Izneid T. Renin-angiotensin-aldosterone
type 2 diabetes mellitus in chinese men and women: a longitudinal study (RAAS): the ubiquitous system for homeostasis and pathologies. Biomed
in Qingdao, China. Biomed Environ Sci. 2019;32:905–13. Pharmacother. 2017;94:317–25.
42. Shi Q, Liu S, Krousel-Wood M, Shao H, Fonseca V, Shi L. Long-term out- 65. Ahmadian E, Pennefather PS, Eftekhari A, Heidari R, Eghbal MA. Role of
comes associated with triple-goal achievement in patients with type 2 renin-angiotensin system in liver diseases: an outline on the potential
diabetes mellitus (T2DM). Diabetes Res Clin Pract. 2018;140:45–54. therapeutic points of intervention. Expert Rev Gastroenterol Hepatol.
43. Taskinen MR, Barter PJ, Ehnholm C, Sullivan DR, Mann K, Simes J, et al. 2016;10:1279–88.
Ability of traditional lipid ratios and apolipoprotein ratios to predict 66. Poznyak AV, Bharadwaj D, Prasad G, Grechko AV, Sazonova MA, Orekhov
cardiovascular risk in people with type 2 diabetes. Diabetologia. AN. Renin-angiotensin system in pathogenesis of atherosclerosis and
2010;53:1846–55. treatment of CVD. Int J Mol Sci. 2021;22:6702.
44. Liu X, Yan L, Xue F. The associations of lipids and lipid ratios with stroke: a 67. Durante A, Peretto G, Laricchia A, Ancona F, Spartera M, Mangieri A, et al.
prospective cohort study. J Clin Hypertens. 2019;21:127–35. Role of the renin-angiotensin-aldosterone system in the pathogenesis of
45. Sheng G, Lu S, Xie Q, Peng N, Kuang M, Zou Y. The usefulness of obesity atherosclerosis. Curr Pharm Des. 2012;18:981–1004.
and lipid-related indices to predict the presence of non-alcoholic fatty 68. Favre GA, Esnault VL, Van Obberghen E. Modulation of glucose metabo-
liver disease. Lipids Health Dis. 2021;20:134. lism by the renin-angiotensin-aldosterone system. Am J Physiol Endo-
46. Yan X, Gao Y, Zhao Q, Qiu X, Tian M, Dai J. Correlation of lipid ratios with crinol Metab. 2015;308:E435–49.
the severity of pulmonary alveolar proteinosis: a cross-sectional study. 69. Aregbesola A, Voutilainen S, Virtanen JK, Aregbesola A, Tuomainen TP.
Front Nutr. 2021;8: 610765. Serum hepcidin concentrations and type 2 diabetes. World J Diabetes.
47. Zhang L, Yuan Z, Chen W, Chen S, Liu X, Liang Y, et al. Serum lipid profiles, 2015;6:978–82.
lipid ratios and chronic kidney disease in a Chinese population. Int J 70. Karamzad N, Eftekhari A, Ashrafi-Asgarabad A, Sullman MJM, Sahebkar
Environ Res Public Health. 2014;11:7622–35. A, Safiri S. Serum Hepcidin, the hepcidin/ferritin ratio and the risk of
48. Yu Y, Lan T, Wang D, Fang W, Tao Y, Li M, et al. The association of lipid ratios type 2 diabetes: a systematic review and meta-analysis. Curr Med Chem.
with hyperuricemia in a rural Chinese hypertensive population. Lipids 2021;28:1224–33.
Health Dis. 2021;20:121. 71. Yang WS, Li R, Shen YQ, Wang XC, Liu QJ, Wang HY, et al. Importance
49. Zhou Y, Yang G, Qu C, Chen J, Qian Y, Yuan L, et al. Predictive performance of lipid ratios for predicting intracranial atherosclerotic stenosis. Lipids
of lipid parameters in identifying undiagnosed diabetes and prediabetes: Health Dis. 2020;19:160.
a cross-sectional study in eastern China. BMC Endocr Disord. 2022;22:76. 72. Osentino F, Grant PJ, Aboyans V, Bailey CJ, Ceriello A, Delgado V, et al. 2019
50. Song Q, Liu X, Wang A, Wang Y, Zhou Y, Zhou W, et al. Associations ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases
between non-traditional lipid measures and risk for type 2 diabetes mel- developed in collaboration with the EASD. Eur Heart J. 2020;41:255–323.
litus in a Chinese community population: a cross-sectional study. Lipids 73. Varghese M, Song J, Singer K. Age and sex: impact on adipose tissue
Health Dis. 2016;15:70. metabolism and inflammation. Mech Ageing Dev. 2021;199: 111563.
51. Nakamura M, Iso H, Kitamura A, Imano H, Kiyama M, Yokoyama S, et al. 74. Neville SE, Boye KS, Montgomery WS, Iwamoto K, Okamura M, Hayes RP.
Total cholesterol performance of Abell-Levy-Brodie-Kendall reference Diabetes in Japan: a review of disease burden and approaches to treat-
measurement procedure: Certification of Japanese in-vitro diagnostic ment. Diabetes Metab Res Rev. 2009;25:705–16.
assay manufacturers through CDC’s Cholesterol Reference Method Labo-
ratory Network. Clin Chim Acta. 2015;445:127–32.
52. Lu S, Yu T, Wang Y, Liang L, Chen Y, Xu F, et al. Nanomaterial-based biosen- Publisher’s Note
sors for measurement of lipids and lipoproteins towards point-of-care of Springer Nature remains neutral with regard to jurisdictional claims in pub-
cardiovascular disease. Analyst. 2017;142:3309–21. lished maps and institutional affiliations.
53. Qureshi A, Gurbuz Y, Niazi JH. Biosensors for cardiac biomarkers detec-
tion: a review. Sens Actuators. 2012;171–172:62–76.
54. Sekretaryova AN, Eriksson M, Turner AP. Bioelectrocatalytic systems for
health applications. Biotechnol Adv. 2016;34:177–97.