Patients with diabetes mellitus suffer higher morbidity crease in ... found that the diabetic patients had higher serum lipid levels, SCBR values were greater ...
D
ecreased Serum Cholesterol-binding Reserve in Diabetes Mellitus S. L. HSIA, LAWRENCE M. FISHMAN, FRANKLIN W. BRIESE, GEORGE CHRISTAKIS, JANICE BURR, AND LEE ALAN BRICKER
Serum cholesterol-binding reserve (SCBR), the capacity of a serum sample to solubilize additional cholesterol in excess of its cholesterol content, was measured in 43 white male patients with maturityonset diabetes in the age range of 35-59 years who were under treatment with insulin. The values were compared with those of 194 nondiabetic controls of the same race, sex, and age range. The mean ±S.D. of SCBR of patients (71.9 ± 29.3 mg./dl.) was lower than that of controls (88.9 ± 30.9mg./dl.) (p < 0.001). Age in the range of 35 to 59 years had no correlation with SCBR in either patients or controls. SCBR was positively correlated with serum levels of cholesterol (SC) and triglycerides (TG) in both patients and controls. After adjustment for SC and TG, the difference in SCBR between patients and controls persisted (p < 0.001). In 15 of 20 (75 per cent) patient-control pairs matched for SC and TG to within 5 per cent, the patient had lower SCBR (paired t-test, p < 0.002). In 16 patients without elevation of serum lipid levels (SC below 250 and TG below 150 mg./dl.), the mean ± S.D. of SCBR (59.1 ± 17.7 mg./dl.) was lower than that of 49 controls having serum lipids in the same range (77.4 ± 3 1 . 7 mg./dl.) (p < 0.03). These results indicate an association of decreased SCBR with diabetes and are consistent with the hypothesis that low SCBR is associated with accelerated atherosclerosis and enhanced risk for coronary heart disease, DIABETES CARE I-. 89-93, MARCH-APRIL 1978.
P
atients with diabetes mellitus suffer higher morbidity
crease in cardiovascular morbidity and mortality experienced
and mortality from cardiovascular causes due to accelerated atherosclerosis than that expected of the general population.1"6 In a 1967 report, 5 Epstein estimated the risk of cardiovascular mortality of diabetic women between the ages of 15 and 44 to be 6.4 times higher and that of diabetic men to be 4.6 times higher than that of the white population in New England; the corresponding ratios at ages 45 to 74 years were 2.0 and 3.2, respectively. Garcia et al.6 recently reported similar ratios of cardiovascular mortality between diabetic and nondiabetic men and women in the Framingham population. The excessive susceptibility of diabetic patients to cardiovascular complications has not been satisfactorily explained. In a 16year follow-up of the Framingham population, Garcia et al.6 found that the diabetic patients had higher serum lipid levels, more hypertension, and more obesity. Yet these known risk factors for coronary heart disease, when considered either individually or combined, could not entirely explain the in-
by the diabetic population. The authors commented that an as yet unknown factor appeared to be present in the diabetic patients that might be responsible for their higher incidence of cardiovascular complications. We recently observed that human serum could solubilize a measurable amount of additional cholesterol and that this capability, designated serum cholesterol-binding reserve (SCBR), was lower in patients with premature myocardial infarction than in control subjects who had no signs or symptoms of coronary heart disease.7'8 The data showed that SCBR values increased with rising levels of cholesterol and triglycerides in the serum of control subjects, but this increase was not observed in the serum of patients with premature myocardial infarction. Consequently, the differences in SCBR values were greater between the hyperlipidemic patients and hyperlipidemic controls.8 Experiments with serum lipoproteins indicated that SCBR could be attributed to the cholesterol-solubilizing capacities of high-density lipo-
DIABETES CARE, VOL. 1 NO. 2, MARCH-APRIL 1978
89
CHOLESTEROL-BINDING RESERVE IN DIABETES / S. L. HSIA AND ASSOCIATES
proteins (HDL) and very-low-density lipoproteins (VLDL).7 In the light of the suggestion of Miller and Miller9 that HDL cleans the arterial wall of cholesterol deposits, it seems possible that the capability of the serum to solubilize additional cholesterol may play a role in facilitating cholesterol efflux from the arterial intima and thereby retard atherogenesis; accordingly, low SCBR values reflect a deficiency of this capability and are signals of accelerated atherosclerosis. To test the hypothesis that decreased SCBR is correlated with an enhanced risk of atherosclerosis, the present study was undertaken to measure SCBR of diabetic patients and to compare the values with those of nondiabetic controls. The results demonstrated significantly lower SCBR values among the diabetic patients. SUBJECTS AND METHODS
T
he study subjects included 43 diabetic patients and 194 nondiabetic controls. The patients were seen in the outpatient clinics of Jackson Memorial Hospital and the Miami Veterans Administration Hospital. They were white men between the ages of 35 and 59 who had maturity-onset diabetes and were under treatment with insulin. Control subjects were participants in the Miami Multiple Risk Factor Intervention Trial (MRFIT) Program. They were white men between the ages of 35 and 59 who were placed in the upper 10 per cent of the estimated risk for coronary heart disease by the combination of their cigarette-smoking habit, serum cholesterol level, and diastolic blood pressure, based on data from the middle-aged male population of the Framingham study.10 All control subjects were free of clinical signs and symptoms of coronary heart disease and diabetes mellitus at the time of the study. Blood specimens were obtained from the antecubital vein following an overnight fast. All blood samples were drawn with the subject resting in the sitting position. After the blood clotted, the serum was separated by centrifugation and was used for measurements of cholesterol, triglycerides, and SCBR. Blood samples from patients and controls were collected and analyzed concurrently to minimize possible bias in the laboratory procedures. Cholesterol and triglycerides were determined on a Technicon AutoAnalyzer II with precision of ±5 per cent, and SCBR was measured by the previously described procedure7 with slight modifications. In brief, the serum (0.5 ml.) was mixed with crystalline cholesterol (7 mg.) that had been pulverized by sonication so that the particle size was reduced to 10—60 jum. (measured under a microscope). After incubation at 37° C. for 16 hours on a rotator operating at 40 rpm, the undissolved cholesterol was removed by filtering the serum through Whatman filter paper no. 42 in a filtering tube 7 mm. in diameter. The cholesterol content of the serum was 90
measured before the incubation and again after the filtration. SCBR was calculated from the difference of the two determinations and expressed in mg. of cholesterol/dl. of serum. The variation of results in repeated determinations was less than ±11 per cent. It was shown previously7 by the incubation of serum with sonicated cholesterol for various lengths of time that the amount of cholesterol solubilized by serum reaches a plateau in 10 to 12 hours, which remains up to 20 hours. Solubilization of cholesterol by serum is influenced by the amounts of cholesterol and serum incubated, the speed of mixing, and the temperature and time of incubation. The conditions selected for SCBR determination as described above were shown to be optimal. In control experiments, the cholesterol preparation was incubated with saline solution in the same manner. The amount of cholesterol in the final filtrate was undetectable and could not have exceeded 3 mg./dl. RESULTS
The data collected on age, serum levels of cholesterol (SC) and triglycerides (TG), and values of SCBR from the 43 diabetic patients and 194 controls are summarized in table 1. The mean ± S.D. for SCBR in the diabetic group (71.9 ± 29.3 mg./dl.) was significantly lower than that for the control group (88.9 ± 30.9 mg./dl.) (two-tailed t-test, p < 0.001). The frequency distributions of SCBR among patients and controls are compared in figure 1. The distribution among patients is skewed toward lower SCBR values; 67 per cent of the patients but only 41 per cent of the controls had SCBR below 80 mg./dl. At each level of SCBR below 80 mg./dl., a greater percentage of patients than controls is distributed, whereas the reverse is true at SCBR levels above 81 mg./dl. Although the patients and controls were in the same age range of 35 to 59 years, the mean ± S.D. of age of patients (53.0 ± 5.3 years) was higher than that of controls (48.8 ± 7.1 years) (p < 0.001). To
TABLE 1 Mean ± S.D. of serum cholesterol-binding reserve (SCBR), serum cholesterol (SC), fasting triglycerides (TG), and age of diabetic and nondiabetic men
N
SCBR (mg./dl.)
SC (mg./dl.)
TG (mg./dl.)
Age (years)
Diabetic patients
43
71.9 ± 29.3
217.0 ± 41.4
228.3 ± 167.5
53.0 ± 5.3
Nondiabetic controls
194
88.9 ± 30.9
243.1 ± 34.9
195.2 ± 108.4
48.8 ± 7.1
0.2), but the mean ± S.D. for SC of patients (217.0 ± 41.4 mg./dl.) was significantly lower than that of the controls (243.1 ± 34.9 mg./dl.) (p < 0.001) (table 1). Since our previous study showed that SCBR increased with rising levels of SC and TG among subjects who had no signs or symptoms of coronary heart disease, it was important to determine whether the bias in SC contributed to the observed difference in SCBR. Covariance analysis showed that SCBR was positively correlated with both SC and TG in the patients (r = 0.58 and 0.49, respectively) and also in the controls (r = 0.37 and 0.34, respectively). After adjustment for SC and TG, the means for SCBR of patients and controls were 72.7 and 88.7 mg./dl., respectively, and the difference between them remained statistically significant (p < 0.001). The SCBR of the diabetic patients and nondiabetic controls was further compared by matching each patient, when possible, with a control subject whose SC and TG fell within 5 per cent of the patient's. The available data permitted the selection of 20 patient-control pairs meeting this criterion. In these pairs, the SC ranged from 152 to 340 mg./dl. and TG from 83 to 1,145 mg./dl. In 15 of the 20 (75 percent) patient-control pairs, the diabetic patient had lower SCBR,
and paired t-test showed that the differences of SCBR in these pairs (mean ± S.D., —38.7 ± 49.2 mg./dl.) were significantly different from 0 (p < 0.002). These data are presented in figure 2, in which the SCBR values of each of the patient-control pairs is graphed as one point, with the control value on the horizontal axis and the patient's value on the vertical axis. On line m = 1, the SCBR values of the patient-control pair would be equal. It is seen that five points fall above and 15 fall below this line, indicating that in 15 pairs the patient had lower SCBR than the control. Patients and controls with normal levels of SC and TG were also compared. Within the study cohorts, 16 patients and 49 controls had SC below 250 and TG below 150 mg./dl. An examination of their SCBR revealed that the mean ± S.D. for patients (59.1 ± 17.7 mg./dl.) was significantly lower than that of the controls (77.4 ± 31.2 mg./dl.) (two-tailed t-test, p < 0.03). Thus, decreased SCBR was an abnormality present in diabetic patients with or without elevated lipid profiles. Finally, SCBR values of the patients were compared according to the duration of diabetes. Table 2 shows that the distribution of SCBR values changed with the duration. In the first decade after the onset of the disease, 13 of 20 patients (65 per cent) had SCBR below 71.9 mg./dl., which was the mean value of SCBR for the 43 study patients (see table 1). The percentage of patients with SCBR < 71.9 mg./ dl. decreased with the years of the disease, such that two of six patients (33.3 per cent) who survived 20 years or longer
25
50
75
100
125
150
175
200
225
SCBR OF NONDIABETIC CONTROLS, mg/dl
FIG. 2. SCBR of patient-control pairs. Each point represents SCBR values of a patient-control pair whose SC and TG levels were matched to within 5 per cent. The SC levels in these pairs ranged from 152 to 340 mg./dl. and TG from 83 to 1,145 mg./dl. The points falling below the line of identity indicate the preponderance of lower SCBR values in the diabetic patients.
DIABETES CARE, VOL. 1 NO. 2, MARCH-APRIL 1978
91
CHOLESTEROL-BINDING RESERVE IN DIABETES / S. L. HSIA AND ASSOCIATES
Since hyperlipidemia is a common complication in diabetes, it was of interest to determine whether decreased SCBR in the diabetic patients was secondary to Patients hyperlipidemia. A comparison was made, therefore, of SCBR Duration with SCBR
