The Measurement of Lipids Currently and 9 Years Ago—Which Is More Associated With Carotid Intima-Media Thickness?




Massive evidence supports that increase of lipids bring more risk of atherosclerosis. However, it is not clear if lipids measured a long time ago bear more risk than the current measurement.


Lipids measured currently is more associated with carotid atherosclerosis than lipids measured long time ago.


A cohort of 1195 participants age 35 to 64 years was examined in both 1993–1994 and 2002 for serum lipids, and in 2002 for carotid intima-media thickness (CIMT) with B mode ultrasound. The associations of lipids at baseline and at reexamination with CIMT were analyzed and compared using multiple linear regressions.


All lipid variables, except for high-density lipoprotein cholesterol (HDL-C) both at baseline and reexamination, were significantly associated with age-adjusted CIMT in both males and females (all Ptrend <0.01). The age-adjusted mean of CIMT in all of the population was 0.696 mm in those having low low-density lipoprotein cholesterol (LDL-C) at both examinations, 0.719 mm in those having high LDL-C only at baseline, 0.706 mm in those having high LDL-C only at reexamination, and 0.727 mm in those having high LDL-C at both examinations. Further analysis showed that lipids measured at baseline remained significant, whereas lipids at reexamination became not significant in all models, except those for HDL-C and total cholesterol (TC)/HDL-C, which allow the lipids at different times to compete in association with CIMT.


Both the current measurement of lipids (TC, LDL-C, non-HDL-C, TC/HDL-C, and LDL-C/HDL-C) and the measurement from 9 years ago are significantly associated with CIMT, but the measurement from 9 years ago had an even stronger association. Clin. Cardiol. 2012 doi: 10.1002/clc.22015

Additional Supporting Information may be found in the online version of this article.

The authors have no funding, financial relationships, or conflicts of interest to disclose.


Serum lipids have been identified not only as risk factor of cardiovascular disease but also determinants of atherosclerosis and its progression across a broad range of heterogeneous populations.1–7 It is also well known and understood that the persistent increase of lipids would bring more risk for the individual to develop atherosclerosis than occasional increase.8–10 However, it is not clear if lipids measured a long time ago bear more risk than the current measurement, although both are occasional. In clinical practice, only the current measurements are considered by the physician when a judgment on the patient's diagnosis, treatments, prevention, and prognosis needs to be made. The patient's previous measurements of lipids, particularly those measured a long time ago, are never or seldom considered by physicians in practice. Thus, clarifying the question bears important theoretical and clinical practice significance.

In previous studies on cholesterol and cardiovascular risk, low-density lipoprotein cholesterol (LDL-C) and total cholesterol (TC) were the most commonly used variables. Several other variables of cholesterol have been proposed such as non–high-density lipoprotein cholesterol (HDL-C), TC/HDL-C, and LDL-C/HDL-C.11–13 However, which of these variables have the strongest association with cardiovascular risk is not entirely clear.

Carotid intima-media thickness (CIMT) measured by noninvasive B-mode ultrasound has been accepted as a marker of atherosclerosis14,15 and has been used as a surrogate end point on the progression of atherosclerosis.16,17 More recently, CIMT has been reported as a predictor of subsequent cardiovascular events independent of traditional risk factors18–20 and has been used with other established risk factors in risk assessments of cardiovascular disease.20,21

In this article, we used data from the People's Republic of China/United States of America (PRC-USA) Collaborative Study of Cardiovascular and Cardiopulmonary Epidemiology to compare the associations between CIMT with present lipid levels (cross-sectional) and with lipid levels 9 years earlier (longitudinal). We also compared the sizes of the associations between different lipids to understand which lipid variable is the most strongly associated with CIMT.


Study Population

The original cohort of the PRC-USA Collaborative Study of Cardiovascular and Cardiopulmonary Epidemiology was initiated in 1981–1982 and consisted of 4 study populations: urban and rural populations in Beijing, and urban and rural populations in Guangzhou. A detailed description of the goals, design, and methods has been reported elsewhere.22 Data from the study's third survey in the rural Beijing population during 1993–1994 were used in this article.

This survey was a cluster random sample of 2313 subjects selected from all 11 villages of the Beijing Shijingshan district. Of these 2313 participants, 1626 surviving subjects free of myocardial infarction, stroke, and diabetes, and with complete baseline data were invited to the reexamination in 2002.23 Of the 1626 participants, a total of 1394 (85.7%) participated in the reexamination in 2002. Out of those 1394 participants, 1202 (86.2%) underwent carotid artery ultrasound. Ultimately, 1195 subjects with complete data on both serum lipids and ultrasonography were used in the final analysis.

The Fuwai Hospital ethics committee approved all of the studies. Informed written consent was obtained from all of the participants for all surveys and examinations.


Carotid Intima-Media Thickness:

The protocol for CIMT measurement has been published elsewhere.24,25 Briefly, the CIMT was measured in 2002 by 2 trained ultrasonographers who were blinded to participant clinical characteristics. The trained ultrasonographers used 128 systems (Acuson Inc., Mountain View, CA) with a high-resolution 7.5–10.0 MHz transducer for all CIMT measurements.

The CIMT was measured at the far wall of three 10-mm segments of the bilateral carotid arteries: the bifurcation, the distal common carotid artery, and the proximal common carotid artery. In every segment, separate measurement was attempted at 3 points (0, 0.5, and 1.0 cm), provided that these points were in the absence of plaque. The overall average value of the CIMT at these 18 points were calculated and used in the final analysis.

The reliability of the CIMT measurements was conducted in 20 randomly selected subjects with an initial scan and 2 repeat scans. The interobserver coefficient of variation was 4.42% for 2 observers. The intraobserver variability, described as the mean of the absolute difference between the first and second observations, was 0.037 mm. A paired t test showed no statistically significant difference between the 2 measures.25

Serum Lipids

Blood samples were drawn after a minimum of 12 hours of fasting, and were stored on ice and then sent to the lab within 3 hours. TC and triglycerides (TG) were assessed by enzymatic methods. HDL-C was measured by dextran-magnesium precipitation. LDL-C values in 2002 were calculated by the Friedwald formula for patients with triglyceride values of lower than 4.5 mmol/L.26 In addition to these original lipid variables, derivative lipids (including non–HDL-C, TC/HDL-C, and LDL-C/HDL-C) were also used in the final analysis. Non–HDL-C was calculated as TC - HDL-C. The lipid laboratory in the study has participated in the National Heart, Lung and Blood Institute/Centers for Disease Control (NHLBI/CDC) Lipid Standardization Program.27

Lipids were categorized into high and not high groups, according to the medians of the lipids of the study population when baseline and reexamination data were pooled together. High lipids were defined as the following: high TC defined as TC ≥180 mg/dL, high LDL-C defined as LDL-C ≥110 mg/dL, high TG defined as TG ≥99 mg/dL, low HDL-C defined as HDL-C ≤46 mg/dL, high non–HDL-C defined as non–HDL-C ≥133 mg/dL, high TC/HDL-C defined as TC/HDL-C ≥3.9, and high LDL-C/HDL-C defined as LDL-C/HDL-C ≥2.3.

Statistical Analysis

Comparisons of lipid variables between baseline and reexamination were done with a paired t test. Because the distribution of TG was not normal, natural logarithmic transformation of the variables was done before analysis.

To evaluate the trend of CIMT against the increase of lipids at baseline and at reexamination, we first classified the lipids into 4 groups by quartiles of the lipids. Then, age-adjusted means of CIMT were calculated for each group by analysis of covariance. P values for trend were calculated using linear regression. Strength of association was evaluated with the standardized regression coefficients after adjusting for age.

To compare the association of lipids at baseline and at reexamination with CIMT, backward linear regression models (removal: 0.10) were introduced, in which CIMT was introduced as the dependent variable, and lipids at both baseline and reexamination were introduced as independent competing variables. The one remaining in the final model indicates a stronger association of CIMT with the lipid measured at that time.

All analyses were performed with SPSS 19.0 statistical package (IBM, Armonk, NY). A 2-sided test with a P value of 0.05 was considered statistically significant.


Subject Characteristics

Of the 1195 participants, 424 (35.4%) were male. Mean CIMT at reexamination was 0.710 ± 0.096 mm (0.732 ± 0.099 mm in males and 0.698 ± 0.092 mm in females). At the time of the baseline survey in 1993–1994, the average age of the cohort was 47.8 ± 8.1 years (47.4 ± 9.1 years for males and 48.0 ± 7.5 years for females) (Table 1). Comparison of baseline characteristics in 1993–1994 between participants included and excluded from the final analysis showed no significant difference, with the exception of excluded participants being older males (Supplementary Appendix 1).

Table 1. The Mean Carotid Intima-Media Thickness and Lipids (Mean ± Standard Deviation) of the Study Cohort in 1993–1994 and 2002
 Male (n = 424)Female (n = 771)
1993–19942002P Value1993–19942002P Value
  1. Abbreviations: HDL-C, high-density lipoprotein cholesterol; IMT, intima-media thickness; LDL-C, low-density lipoprotein cholesterol; non–HDL-C, non–high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; TG(ln), natural logarithmic transformation of TG.

Age (y) 47.4 ± 9.156.0 ± 9.1<0.00148.0 ± 7.556.7 ± 7.4<0.001
Lipid variables      
 TC (mg/dL)172.3 ± 32.6185.7 ± 30.9<0.001174.7 ± 32.1196.3 ± 35.7<0.001
 LDL-C (mg/dL)102.5 ± 28.2115.4 ± 28.0<0.001104.4 ± 27.4122.7 ± 31.1<0.001
 TG (ln) (mg/dL)4.6 ± 0.64.7 ± 0.6<0.0014.5 ± 0.54.8 ± 0.5<0.001
 HDL-C (mg/dL)47.6 ± 11.544.8 ± 11.3<0.00149.3 ± 10.847.2 ± 10.7<0.001
 non–HDL-C (mg/dL)124.7 ± 32.4140.9 ± 31.8<0.001125.4 ± 32.1149.1 ± 35.8<0.001
 TC/HDL-C3.8 ± 1.14.4 ± 1.2<0.0013.7 ± 1.04.4 ± 1.2<0.001
 LDL-C/HDL-C2.3 ± 0.82.5 ± 0.8<0.0012.2 ± 0.72.6 ± 0.9<0.001
IMT (mm)Not measured0.732 ± 0.099 Not measured0.698 ± 0.092 

Association of Serum Lipids With CIMT

When we classified the lipids into 4 groups by quartiles of the lipids, we found that TC, LDL-C, TG, non–HDL-C, TC/HDL-C, and LDL-C/HDL-C, at both baseline and reexamination, were significantly and positively associated with age-adjusted CIMT for both males and females (all Ptrend < 0.01) (Supplementary Appendix 2).

The Joint Effect of Baseline and Reexamination Lipids Increase on CIMT

Figure 2 shows the pattern of the joint effect of increased LDL-C on CIMT. The age-adjusted mean of CIMT in all population was 0.696 mm in those having low LDL-C at baseline and reexamination, 0.719 mm in those having high LDL-C only at baseline, 0.706 mm in those having high LDL-C only at reexamination, and 0.727 mm in those having high LDL-C at both examinations. Other lipid variables showed very similar patterns.

Figure 1.

The relationship between the time of measurement of the cause (serum lipids) and the time of the measurement of the outcome (carotid intima-media thickness [CIMT]). It illustrates the chronic nature of the development of atherosclerosis demonstrated by our findings in this study. It takes a long time for the increased lipids to actually change the CIMT. The current CIMT is thus largely a result of the past experience of increased lipids. The effect of the current increased lipids takes years to be seen. IMT = intima-media thickness.

Figure 2.

Age-adjusted mean carotid intima-media thickness (IMT) (mm) according to the presence of high low-density lipoprotein cholesterol (LDL-C) (*) at baseline and reexamination. The median of pooled LDL-C data was used as the cutoff to divide the study population into high and not high.

Comparison of the Same Lipids Variable Between Baseline and Reexamination in Association With CIMT

As shown in Table 2, lipids measured at baseline remained in all the final models in which the lipids at reexamination were also introduced to compete in association with CIMT, except for HDL-C and TC/HDL-C.

Table 2. Comparison of Lipid Variables Between Baseline and Reexamination in Association With Carotid Intima-Media Thickness
ModelVariables Introduced in the ModelMale (n = 424)Female (n = 771)Total (N = 1195)
Variables Remained in the ModelStandardized βP ValueVariables Remained in the ModelStandardized βP ValueVariables Remained in the ModelStandardized βP Value
  1. Abbreviations: HDL-C, high-density lipoprotein cholesterol; IMT, intima-media thickness; LDL-C, low-density lipoprotein cholesterol; non–HDL-C, non–high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride.

  2. TG(ln), natural logarithmic transformation of TG.

  3. TC (93) is the total cholesterol measured in 1993–1994. TC (02) is the total cholesterol measured in 2002. The same expression was used in other lipid variables.

TCTC (93), TC (02)TC (93)0.274<0.001TC (93)0.257<0.001TC (93)0.259<0.001
LDL-CLDL-C (93), LDL-C (02)LDL-C (93)0.1870.002LDL-C (93)0.233<0.001LDL-C (93)0.237<0.001
TG (ln)TG (ln)(93), TG (ln) (02)TG (ln) (93)0.1040.032TG (ln) (93)0.240<0.001TG (ln) (93)0.185<0.001
HDL-CHDL-C (93), HDL-C (02)HDL-C (93) HDL-C (02)0.143 −0.154 0.031 0.020HDL-C (02)−0.0870.016HDL-C (02)−0.0750.009
non–HDL-Cnon–HDL-C (93), non–HDL-C (02)non–HDL-C (93)0.262<0.001non–HDL-C (93)0.273<0.001non–HDL-C (93)0.265<0.001
TC/HDL-CTC/HDL-C (93), TC/HDL-C (02)TC/HDL-C (02)0.189<0.001TC/HDL-C (93)0.211<0.001TC/HDL-C (02) TC/HDL-C (93) 0.104 0.117 0.010 0.004
LDL-C/HDL-CLDL-C/HDL-C (93), LDL-C/HDL-C (02)LDL-C/HDL-C (93)0.189<0.001LDL-C/HDL-C (93)0.211<0.001LDL-C/HDL-C (93)0.200<0.001

Comparison Between Different Lipid Variables for Their Association With CIMT

Both for baseline and reexamination, non–HDL-C, TC/ HDL-C, and LDL-C/HDL-C were either better associated with CIMT or at least equivalent to the original lipid variables including TC, LDL-C, TG, and HDL-C (Table 3).

Table 3. Comparison Among Lipid Variables at Baseline and at Reexamination in Association With Carotid Intima-Media Thickness by Sex After Adjusting for Age and Sex
Lipid VariablesMale (n = 424)Female (n = 771)Total (n = 1195)
Standardized βP ValueStandardized βP ValueAdjusted Standardized βP Value
  1. Abbreviations: HDL-C, high-density lipoprotein cholesterol; IMT, intima-media thickness; LDL-C, low-density lipoprotein cholesterol; non–HDL-C, non–high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride.

  2. Standardized β is the standardized regression coefficient.

Lipids at baseline (1993–1994)
TG (ln)0.1260.0040.1170.0010.122<0.001
Lipids at reexamination (2002)
TG (ln)0.1070.0150.0700.0350.0870.001


Our study found that participants with higher lipids at both the baseline and reexamination had the highest mean CIMT, those with lower lipids at both examinations had the lowest, and those with only either higher had a CIMT in between, which was consistent with previous studies that showed the similar pattern between lipids and cardiovascular events.28 This study increased the evidence for long-term or lifetime rather than short-term lipid-lowering strategies for prevention of atherosclerosis. It also implies that (1) the earlier the intervention starts the better, because those who never had lipids increased had the lowest mean CIMT; and (2) it is never too late to initiate the intervention, because those who had lipids increased at baseline but not at reexamination had a lower mean CIMT in comparison with those having lipids increased at both examinations.

In addition, we found that in comparison to the present lipid levels, lipid levels measured 9 years ago were more strongly associated with CIMT. This finding suggest that our current practice where only the present or recent lipids level are considered for clinical decision-making on diagnosis, treatment, prevention, and prognosis should probably be improved by adding asking, checking, and considering the patient's lipids level in history. In those risk-prediction models,29–31 the historical measurements of lipids should be included in addition to the present measurement.

These findings bear important messages for the scientific community, policy makers, and the public in terms of developing strategies for prevention and control of cardiovascular disease through a lipid-lowering strategy. It actually indicates that lipid-lowering strategies targeting current lipid levels and lifetime levels are both important, but long-term strategies are more likely to succeed in reducing the risk for atherosclerotic conditions. Persistent and effective long-term strategies should be heavily emphasized in clinical practice for preventing subclinical atherosclerosis and subsequent atherosclerotic cardiovascular disease. To this end, the current recommended lipid-lowering strategies in clinical practice guidelines,32–34 regardless if they are lifestyle changes or pharmaceutical treatment strategies, should all emphasize the patient's long-term compliance to treatment if they are to reduce their risk for cardiovascular disease.

We also compared different lipids and their derived variables in association with CIMT to understand whether the recently proposed variables in the literature are more associated with atherosclerosis. We found that these new derived variables, non–HDL-C, TC/HDL-C, and LDL-C/HDL-C, had an equal or stronger association with CIMT in both males and females. Non–HDL-C contains all known and potential atherogenic lipid particles.13 Several previous studies have found that non–HDL-C level is a somewhat better predictor of cardiovascular mortality and coronary heart disease than LDL-C level.13,35 There are a few studies that suggest a higher ratio of TC/HDL-C or LDL-C/HDL-C is associated with a greater risk of cardiovascular events in patients with coronary heart disease.36,37 However, the role of these 2 lipid ratios during early-stage atherosclerosis has only recently been paid attention to.12 Our study indicates that an increased TC/HDL-C and LDL-C/HDL-C ratio may also be associated with development of carotid atherosclerosis. Assessment of these lipid indexes may thus facilitate early management of atherosclerotic risks rather than assessment of LDL-C alone.

One of the main limitations of this study is the lack of a measurement of IMT at baseline, so we cannot examine associations of lipids in relation to the development or progression of atherosclerosis. Future prospective studies with repeated measures of intima-media thickness and serum lipids/lipoproteins would be valuable. Second, we only measured the common carotid artery and bifurcation but not the internal carotid artery. This was simply because the common carotid artery and bifurcation are easier to measure than the internal carotid artery, which is very difficult to visualize in many circumstances, and intraobserver and interobserver reproducibility is large.38


CIMT was better associated with TC, LDL-C, non–HDL-C, TC/HDL-C, and LDL-C/HDL-C measured 9 years earlier in comparison to those measured currently, suggesting that lipids level from a long time ago should be used in clinical considerations, together with the current level. The chronically harmful lipid levels cause atherosclerosis, much more than the occasional abnormal levels. Any lipid-lowering strategy, whether for a population or on an individual basis, should be a long-term rather than a short-time plan.


The authors are grateful to the field team who implemented the fieldwork and to the ultrasonic technicians from the Institute of Aeromedical of the Air Force who carried out the examination work. The authors are also grateful to all of the participants involved in this study.