Results of recent studies using the ratio of plasma triglyceride (TG) to high-density lipoprotein (HDL) cholesterol concentration to identify insulin-resistant patients at increased cardiometabolic risk have emphasized that the cut point used for this purpose will vary with race. Because TG and HDL cholesterol concentrations vary with gender, this analysis was initiated to define gender-specific plasma TG/HDL cholesterol concentration ratios that best identified high-risk subjects among women (n = 1,102) and men (n = 464) of primarily European ancestry. Insulin resistance was defined as the 25% of the population with the highest values for fasting plasma insulin concentration and homeostasis model assessment of insulin resistance. Using TG/HDL concentration ratios >2.5 in women and >3.5 in men identified subgroups of men and women that were comparable in terms of insulin resistance and associated cardiometabolic risk, with significantly higher values for fasting plasma insulin, homeostasis model assessment of insulin resistance, blood pressure, body mass index, waist circumference, and glucose and TG concentrations and lower HDL cholesterol concentrations than in women and men below these cut points. The sensitivity and specificity of these gender-specific cut points to identify insulin-resistant subjects were about 40% and about 80%, respectively. In conclusion, the plasma TG/HDL cholesterol concentration ratio that identifies patients who are insulin resistant and at significantly greater cardiometabolic risk varies between men and women.
Plasma triglyceride (TG) and high-density lipoprotein (HDL) cholesterol concentrations are independently related to insulin-mediated glucose disposal, and the plasma TG/HDL cholesterol concentration ratio is significantly related to this measure of insulin action, as well as to plasma insulin concentration, a commonly used surrogate estimate of insulin action. Because measurements of TG and HDL cholesterol are standardized, whereas there is no standard assay of plasma insulin concentration, we suggested that the plasma concentration ratio of TG to HDL cholesterol might be a useful surrogate estimate of insulin action. However, recent studies addressing this issue have come to disparate conclusions as to the utility of this ratio as a surrogate estimate of insulin resistance, with particular emphasis on the impact of differences in racial background of the populations being studied. Whether subsequent evidence will establish the TG/HDL cholesterol concentration ratio as a useful surrogate estimate of insulin resistance and associated cardiometabolic risk remains to be seen, but comparisons in non-Hispanic whites, Mexican Americans, and non-Hispanic blacks have shown that the “best” ratio will vary as a function of racial group. Epidemiologic studies have conventionally taken gender differences into account, and a recent editorial in The Lancet reminded investigators of the importance of “analyzing data by sex, not only when scientifically appropriate, but also as a matter of routine.” Given differences in TG and HDL cholesterol metabolism between men and women, it seemed important to see if the “best ratio” might vary as function of gender as well as racial group. Because we were unaware of any study in which that has been done, the present analysis was performed, in which the ability of the plasma concentration ratio of TG to HDL cholesterol to identify insulin-resistant subjects was analyzed separately for men and women.
Methods
As part of community intervention programs on cardiovascular risk factors, epidemiologic studies on hypertension, renal disease, and other cardiometabolic risk factors were conducted in Rauch, in the province of Buenos Aires, Argentina (the RAUCH project), and San Andrés de Giles, also in the province of Buenos Aires (the PROCER project). According to the last national census available, there were 8,246 and 13,922 inhabitants aged ≥15 years in the urban areas of Rauch and San Andrés de Giles, respectively. The methods of the samples, the socioeconomic features, and the prevalence of cardiovascular risk factors of the 2 populations have been published previously.
In brief, the surveys were performed on simple random samples of subjects aged 15 to 80 years who lived in the chosen blocks (RAUCH n = 1,307, PROCER n = 1,591). Blood pressure was measured sitting, after a minimum rest period of 5 minutes, using a mercury sphygmomanometer. Phase I and V Korotkoff sounds were used to identify systolic blood pressure and diastolic blood pressure, respectively, and values were averages of 3 different measurements separated by 2 minutes from one another. Weight was determined with subjects wearing light clothing and no shoes. Height was also measured without shoes, using a metallic metric tape; waist circumference was measured with a relaxed abdomen using a metallic metric tape on a horizontal plane above the iliac crest. Body mass index was calculated, and concentrations of plasma glucose, TG, HDL cholesterol, and fasting plasma insulin (FPI) were determined after an overnight (12-hour) fast. Plasma for the insulin measurements was extracted by centrifugation (15 minutes at 3,000 rpm) and frozen at −20°C until assayed. FPI concentrations in the Rauch population were determined using an immunoradiometric assay, with 2 monoclonal antibodies against 2 different epitopes of the insulin molecule. The inter- and intra-assay coefficients of variation were 8.0% and 3.8%, respectively, with the lowest detectable level of 1.4 pmol/L. FPI concentrations in the San Andrés de Giles population were determined using a solid-phase chemiluminescent assay, using commercially available kits (Immunolite Diagnostic Products, Los Angeles, California), with an analytic sensitivity of 1.4 pmol/L, inter- and intra-assay coefficients of variation <8%, and proinsulin cross-reactivity <8.5%. The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using the formula {[insulin (μU/ml) × glucose (mg/100 ml)/18]/22.5}. FPI was measured in 1,174 women (mean age 46 ± 1 years, range 15 to 80) and 501 men (mean age 47 ± 1 years, range 15 to 80). As in previously published studies, to avoid the potential confounding impact of extreme outliers, subjects with TG concentrations >500 mg/dl and/or HDL cholesterol concentrations >100 mg/dl were excluded from the analysis, as were participants with positive histories of diabetes or fasting glucose concentrations ≥126 mg/dl. The remaining 1,102 women (mean age 45 ± 1 years) and 464 men (mean age 46 ± 1 years) were those included in the analysis.
Men and women were divided into quartiles on the basis of their TG/HDL cholesterol concentration ratios, and mean ± SD values and ranges were estimated. Values of TG/HDL cholesterol concentration ratios were compared between women and men using Student’s t tests for independent samples. Values for age, FPI, HOMA-IR, systolic blood pressure, diastolic blood pressure, body mass index, waist circumference, glucose, HDL cholesterol, TG, and TG/HDL cholesterol ratio were compared between the quartile with the highest TG/HDL cholesterol ratio and the remaining quartiles using analysis of covariance with age and study site (Rauch and San Andrés de Giles) as covariates.
To define insulin resistance, the sample was divided into FPI quartiles and HOMA-IR quartiles, and subjects in the upper quartiles of the 2 variables were classified as insulin resistant on the basis of a prospective outcome study. To evaluate agreement between the 2 definitions of IR, we used coefficient of concordance (κ). The sensitivity and specificity of TG/HDL cholesterol ratio to identify insulin resistance were calculated using as a cut-off point the value that separated the upper 25% of the TG/HDL cholesterol ratio in women and men, separately.
All significant tests were 2 tailed, and p values <0.05 were considered statistically significant. All statistical analyses were performed using SPSS (SPSS, Inc., Chicago, Illinois).
Results
Table 1 lists the demographic and metabolic characteristics of the experimental population, divided on the basis of study site and gender. In general, the values in the 2 populations were comparable, although the Rauch group was somewhat older, with higher blood pressures. Given the relative comparability of the values at the 2 sites, the experimental data are combined in the tables. It should be noted that the men and women were not different in terms of age, FPI, HOMA-IR, and body mass index, whereas every other experimental variable varied as a function of gender.
| Variable | Rauch | San Andrés de Giles | Both Samples | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Women (n = 465) | Men (n = 209) | p Value ⁎ | Women (n = 637) | Men (n = 255) | p Value ⁎ | Women (n = 1,102) | Men (n = 464) | p Value ⁎ | |
| Age (years) | 52 ± 16 | 53 ± 15 | 0.581 | 40 ± 17 | 40 ± 16 | 0.988 | 45 ± 18 | 46 ± 17 | 0.489 |
| FPI (μU/ml) | 7.5 ± 4.2 | 7.6 ± 5.4 | 0.805 | 8.4 ± 6.3 | 8.1 ± 8.9 | 0.650 | 8.0 ± 5.5 | 7.9 ± 7.5 | 0.731 |
| HOMA-IR | 1.66 ± 0.99 | 1.77 ± 1.37 | 0.307 | 1.99 ± 1.58 | 2.00 ± 2.15 | 0.964 | 1.85 ± 1.37 | 1.90 ± 1.84 | 0.651 |
| Body mass index (kg/m 2 ) | 25.5 ± 4.8 | 26.4 ± 3.7 | 0.010 | 27.9 ± 6.5 | 27.5 ± 5.1 | 0.272 | 26.9 ± 6.0 | 27.0 ± 4.6 | 0.770 |
| Waist circumference (cm) | 92 ± 13 | 96 ± 10 | <0.001 | 92 ± 15 | 94 ± 14 | 0.114 | 92 ± 14 | 95 ± 12 | <0.001 |
| Systolic blood pressure (mm Hg) | 131 ± 19 | 136 ± 18 | <0.001 | 122 ± 20 | 125 ± 18 | 0.068 | 126 ± 20 | 130 ± 19 | <0.001 |
| Diastolic blood pressure (mm Hg) | 81 ± 11 | 85 ± 12 | <0.001 | 73 ± 13 | 74 ± 13 | 0.099 | 76 ± 13 | 79 ± 14 | <0.001 |
| Glucose (mg/dl) | 88 ± 10 | 92 ± 12 | <0.001 | 95 ± 11 | 99 ± 10 | <0.001 | 92 ± 11 | 96 ± 12 | <0.001 |
| TG (mg/dl) | 123 ± 61 | 150 ± 79 | <0.001 | 119 ± 50 | 139 ± 76 | <0.001 | 120 ± 55 | 144 ± 78 | <0.001 |
| HDL cholesterol (mg/dl) | 61 ± 12 | 55 ± 12 | <0.001 | 64 ± 15 | 57 ± 14 | <0.001 | 63 ± 14 | 56 ± 13 | <0.001 |
| TG/HDL cholesterol ratio | 2.17 ± 1.44 | 2.93 ± 1.91 | <0.001 | 2.01 ± 1.19 | 2.75 ± 2.13 | <0.001 | 2.08 ± 1.30 | 2.84 ± 2.03 | <0.001 |
The study population was divided into quartiles on the basis of the TG/HDL cholesterol concentration ratios, and the comparison between men and women is listed in Table 2 . It can be seen that the TG/HDL cholesterol ratios were higher in men (p <0.001), irrespective of quartile. However, the magnitude of the difference between the 2 genders became greater as the TG/HDL cholesterol concentrations increased. Finally, the cut points separating the upper quartile from the other 3 were different, with values of about 2.5 and about 3.5 in women and men, respectively.
| TG/HDL Cholesterol Ratio | Women (n = 1,102) | Men (n = 464) | ||||
|---|---|---|---|---|---|---|
| Mean ± SD | Minimum | Maximum | Mean ± SD | Minimum | Maximum | |
| Quartile 1 | 0.99 ± 0.16 | 0.18 | 1.23 | 1.14 ± 0.21 | 0.64 | 1.47 |
| Quartile 2 | 1.46 ± 0.15 | 1.24 | 1.75 | 1.83 ± 0.23 | 1.47 | 2.26 |
| Quartile 3 | 2.05 ± 0.21 | 1.75 | 2.48 | 2.84 ± 0.36 | 2.28 | 3.52 |
| Quartile 4 | 3.81 ± 1.47 | 2.48 | 14.65 | 5.53 ± 2.27 | 3.54 | 15.43 |
| Quartiles 1–4 | 2.08 ± 1.30 | 0.18 | 14.65 | 2.84 ± 2.03 | 0.64 | 15.43 |
Table 3 compares the cardiometabolic risk profiles of men and women in the highest TG/HDL cholesterol quartile with those of subjects in the lowest 3 quartiles. In women, every experimental variable was significantly worse in those whose TG/HDL cholesterol ratios were >2.5. The findings in men were comparable in that all the cardiometabolic risk factors were worse in those with TG/HDL cholesterol ratios >3.5, although the increase in SBP did not reach statistical significance. Most experimental variables were relatively similar in the women and men in the highest TG/HDL cholesterol quartile ( Table 3 ). Furthermore, the FPI and HOMA-IR values were almost identical between women and men who were in the TG/HDL cholesterol top quartile (FPI p = 0.963, HOMA-IR p = 0.584).
| Variable | TG/HDL Cholesterol Ratio | |||||
|---|---|---|---|---|---|---|
| Women | Men | |||||
| ≤2.5 (n = 834) | >2.5 (n = 268) | p Value ⁎ | ≤3.5 (n = 346) | >3.5 (n = 118) | p Value ⁎ | |
| Age (years) | 44 ± 18 | 50 ± 16 | <0.001 | 44 ± 18 | 50 ± 13 | <0.001 |
| FPI (μU/ml) | 7.1 ± 4.8 | 10.7 ± 6.6 | <0.001 | 6.9 ± 7.5 | 10.7 ± 6.9 | <0.001 |
| HOMA-IR | 1.64 ± 1.18 | 2.53 ± 1.68 | <0.001 | 1.65 ± 1.79 | 2.63 ± 1.80 | <0.001 |
| Body mass index (kg/m 2 ) | 26.1 ± 5.7 | 29.3 ± 6.0 | <0.001 | 26.4 ± 4.7 | 28.8 ± 3.8 | <0.001 |
| Waist circumference (cm) | 90 ± 14 | 98 ± 14 | <0.001 | 93 ± 12 | 100 ± 10 | <0.001 |
| Systolic blood pressure (mm Hg) | 124 ± 19 | 132 ± 20 | <0.001 | 129 ± 18 | 134 ± 20 | 0.151 |
| Diastolic blood pressure (mm Hg) | 75 ± 13 | 80 ± 13 | <0.001 | 78 ± 13 | 83 ± 14 | 0.028 |
| Glucose (mg/dl) | 91 ± 10 | 94 ± 12 | 0.001 | 95 ± 11 | 98 ± 13 | 0.021 |
| TG (mg/dl) | 98 ± 28 | 191 ± 58 | <0.001 | 111 ± 37 | 241 ± 84 | <0.001 |
| HDL cholesterol (mg/dl) | 66 ± 12 | 52 ± 12 | <0.001 | 60 ± 12 | 45 ± 10 | <0.001 |
| TG/HDL cholesterol ratio | 1.51 ± 0.47 | 3.84 ± 1.48 | <0.001 | 1.93 ± 0.74 | 5.50 ± 2.26 | <0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
