Main

Diabetes appears to have been rare among Native Americans before 1930(1). However, since then disease prevalence has increased(1, 2) with associated high mortality(3). In Oklahoma, a state represented by over 30 tribes, diabetes is a leading cause of death(4). Furthermore, recent evidence suggests that the onset age for NIDDM in Native Americans is decreasing(5), possibly because of the relationship between obesity and NIDDM(6) and the increase in the numbers of obese children(5). Evidence that obesity in childhood may lead to the development of diabetes and its cardiovascular complications(7) led us to suppose that early detection of predisposing factors might be particularly relevant in our Plains Indian population with a view to eventual reversal of risk.

The study was designed to explore the interrelationships between weight, BMI, lipids, apolipoproteins, insulin, and glucose in Native American children. We were particularly interested in the relationship of these factors to family history of diabetes and atherosclerosis, and degree of Native American ancestry or quantum. Also, owing to known associations of NIDDM and obesity with derangements in lipoprotein transport(7, 8), we elected to examine whether apoA-I, apoB, apoC-III, and Lp(a) could be additional markers of inherited risk. ApoA-I and apoB, the apolipoproteins associated, respectively, with HDL and LDL, have been good descriminators for atherosclerosis in adults(9, 10) and the apoA-I:apoB ratio has been shown to reflect parental myocardial infarction in children(11) implying a genetic influence on the levels. ApoC-III, a correlate of plasma triglyceride(12), has been found to be a useful predictor of atherosclerosis regression when assayed in the heparin-precipitated and supernatant fractions after the addition of heparin-manganese for the determination of HDL cholesterol(13). Also the finding that this apolipoprotein is transferred to HDL-cholesterol during lipolysis of VLDL(14) has supported use of the ratio of apoC-III in the heparin supernate to precipitate (apoC-III ratio) as an index of VLDL catabolism(15).

Lp(a), on the other hand, is classed as a lipoprotein which contains a unique protein, closely resembling plasminogen(16). Because high levels of Lp(a) predispose to atherosclerosis(17) and are genetically determined by molecular size polymorphism(18, 19), studies in Native Americans are of particular interest. A relationship of Lp(a) to diabetes had been previously suggested by the finding that decreased insulin secretion was related to high levels of Lp(a)(20, 21). We therefore added Lp(a) to a collection of candidate risk factors which we proposed to test for a relationship to a family history of either NIDDM or atherosclerosis.

METHODS

Research design. Parents of 103 Plains Indian children residing in rural and metropolitan Oklahoma were approached about the possible participation of their children in this study. With the assistance of American Indian lay workers of the Anadarko Catholic Church and the staff of the Oklahoma City Indian Clinic, parents and their children were recruited to form the study group.

After approval by the University of Oklahoma and the Indian Health Services Institutional Review Boards, respective tribal approval was granted, and informed consent was obtained at the research sites. An important part of the initial stage was the team's efforts to clearly indicate to the parents of potential participants the importance of this study and to inform them that a report on the findings of the study would be provided to them by mail, prepared without the use of overly technical or professional jargon. In addition, a brief summary of the protocol was presented to each participant and the parent(s) at the time of blood sample collection.

Subjects included 103 American Indian children (48 female and 55 male) between the ages of 4 and 19 y. The children represented 12 different Indian tribes, such as the Apache, Comanche, Cheyenne, Kiowa, Ponca, and Wichita. Over half (n = 52) of the study group were members of the Kiowa Tribe. Blood quantum levels, the proportion of Indian blood of each study group member, ranged from 12.5% to 100%. The mean blood quantum level was 68.2%.

The data collected for this study included height and weight for the calculation of BMI [weight (kg) ÷ height (m2)], waist and hip circumference, family history of cigarette smoking, obesity, diabetes, cardiovascular disease (myocardial infarction, peripheral vascular disease, and hypertension), Indian blood quantum, serum cholesterol, triglyceride, HDL cholesterol, derived LDL cholesterol, apoA-I, apoB, apoC-III, and fasting serum glucose and insulin.

Serum samples were analyzed by the Laboratory of Lipid and Lipoprotein Studies at the Oklahoma Medical Research Foundation for lipid and apolipoprotein levels using Lipid Research Clinics methodology for total cholesterol, triglyceride, and HDL cholesterol(22), followed by calculation of the LDL cholesterol using the Friedewald formula(23). Apolipoproteins were assayed by electroimmunoassay(24–26) using antibodies prepared on site. The apoC-III ratio was determined from the direct measurement of apoC-III in the supernate and precipitate after heparin-manganese precipitation(27) and computing the respective ratio. Insulin was measured with RIA using a double antibody system(28). Glucose was measured by glucose oxidase using a Beckman autoanalyzer.

Study group data were then compared with similar data from Caucasian children collected during the Lipid Research Clinics survey in 1984(29).

Statistical methods. BMI was grouped in 5-y age brackets from age 5 to 20 y and compared with predominantly Caucasian children studied by the Lipid Research Clinics using t testing. Lipid values were grouped in the same age brackets and separated by quartiles for BMI so as to compare lipid values in the highest BMI quartile with those in the lower quartiles by t testing. The lipid values were also compared with Caucasian children from Oklahoma. Fasting insulin values were log-transformed and compared with all other parameters. Pearson correlation coefficients were used to test the associations between continuous variables. Univariate analysis was used to assess distribution of the variables. Log transformation was conducted if skewness was observed. Spearman correlation coefficients were used for correlating nonparametric variables such as quantum of Indian blood and Lp(a) with the continuous variables.

Insulin, glucose, insulin resistance, triglyceride, BMI, VLDL cholesterol, and apoC-III ratio values were logtransformed due to skewness. The Pearson correlation coefficient was used to test for associations between the continuous variables and glucose, insulin, insulin:glucose ratio, BMI, and lipid values.

RESULTS

The mean BMI increased with age when examined in 5-y brackets from age 5 to 20 y in male subjects (Table 1). The mean BMI tended to be higher in the Native American children than in the Lipid Research Clinic children who were predominantly Caucasian. The difference was significant in the 5-9-y-old and 10-14-y-old boys (p < 0.05) and 10-14-y-old girls (p < 0.001) (Table 1). Ten to 14 y olds tended to have a higher triglyceride (p < 0.05) and a lower HDL cholesterol (p < 0.001) when in the highest quartiles for BMI compared with the other three (Fig. 1,Table 2). In contrast, the 15-19 y olds tended to have a higher triclyceride, LDL cholesterol, and apoB (p < 0.001) when in the highest quartile for BMI (Fig. 2,Table 2).

Table 1 BMI in the Native American children and adolescents compared with predominantly Caucasian children in the Lipid Research Clinics (LRC) study
Full size table
Figure 1
figure 1

Triglyceride and HDL cholesterol in the highest quartile for BMI compared with the lower three quartiles showing that adolescents aged 10-14 y tend to have a higher triglyceride and lower HDL cholesterol when obese. *p < 0.001.

Full size image
Table 2 Lipid and apolipoprotein levels in Native American and local Caucasian children
Full size table
Figure 2
figure 2

Cholesterol, LDL cholesterol, and apoB in the highest quartile for BMI compared with the lower three quartiles, showing that adolescents aged 15-19 y tend to have higher cholesterol, LDL cholesterol, and apoB when obese. *p < 0.05; **p < 0.001.

Full size image

The mean log-transformed fasting insulin levels for Native American children (derived from Table 3) correlated with glucose(r = 0.35, p < 0.001) and with the insulin:glucose ratio (r = 0.95, p < 0.0001), but did not correlate with BMI when adjusting for age using published BMI norms and percentiles for age(30).

Table 3 Glucose, insulin, and insulin:glucose ratios in Native American children
Full size table

There were no significant associations with Lp(a) using nonparametric testing.

CONCLUSIONS

The tendency for Native American children to have a higher BMI earlier than children in the much larger Lipid Research Clinics population suggests that obesity in the Oklahoma Plains Indian population begins at an early age. A similar observation in American Indians and Alaskan Natives(5) supports the hypothesis that prevailing cultural influences on genetic predisposition to obesity are operative in childhood. North American Indian tribes are thought to have undergone a secular change in body composition in the earlier part of this century judging from photographs which showed them to be lean(1). It follows that cultural changes known to have been in effect during this same period may have contributed to both the increased prevalence of obesity in both children(5) and adults(1). Because NIDDM in Native American tribes increased substantially during this period(1, 6), obesity can be implicated as a major predisposing factor(2, 3). However, it should be pointed out that increased weight may in part be attributed to gains in lean body mass, meaning that BMI may not be an ideal index of adiposity for all growing children and adolescents.

The mean insulin levels in Pima Indian children were found to be increased relative to a control Caucasian population(31), a finding which may predict risk for NIDDM(32), but in this cross-sectional study, fasting insulin failed to correlate with BMI when adjusted for age. We were unable to compare insulin levels with previously published studies owing to known variation in insulin assay methodology and lack of reliable standards. However, it is likely that cross-cultural comparison would show differences in fasting insulin(31), and insulin level could be a predictor of NIDDM in Native American children, because ongoing studies on adults in the same population have estimated the prevalence of NIDDM in adults aged 45-74 y to be 38% in men and 42% in women(33, 34).

We also investigated whether early and excessive weight gain in adolescence is associated with adverse lipid and apolipoprotein distribution among lipoproteins. Our finding, that adolescents in the highest quartile for BMI, aged 10-15 y, had a higher triglyceride level but a lower HDL cholesterol level, suggests that the early phases of adolescent weight gain are associated with defective triglyceride disposal(15). These findings indicate that these adolescents become at increased risk for atherosclerosis when they become obese, a phenomenon which is known to occur in adults(35–37). Specifically, the lipoprotein distribution appears to change with age from one characterized by predominant hypertriglyceridemia in the younger obese adolescents to a hypercholesterolemic pattern in the older obese adolescents. The obese younger adolescents were relatively hypertriglyceridemic with a lower HDL cholesterol, but the older obese adolescents aged 15-20 y had a higher LDL cholesterol and apoB values compared with their nonobese counterparts. This observation is comparable to the finding of a correlation of obesity with LDL cholesterol in older adolescents and young adults of Caucasian origin residing in Bogalusa, in contrast to neighboring black children(38). However, the total sample size was smaller in this pilot study, and after stratification reduced statistical power could have been insufficient to demonstrate significance. It should also be pointed out that the significant differences occurred within the desirable ranges for age and therefore should not be considered abnormal but suggesting a trend for relatively increased lipid levels in these Native American adolescents.

Our question, whether a family history of diabetes or heart attacks was associated with biochemical markers of lipoprotein transport, was unanswered because there were no associations with either a family history of diabetes or cardiovascular disease. Previous associations of Lp(a) with insulin secretion(20, 21) have been observed, suggesting that Lp(a) may have a relationship to β cell failure, a recognized sequel to insulin resistance(39); however, larger studies may be required to support or disprove these earlier findings. Many studies have shown Lp(a) to be significantly associated with risk for atherosclerosis(17, 19, 40), whereas some studies have had negative conclusions(41, 42). Our failure to relate Lp(a) to a family history of heart attacks may have been due to a relatively low rate of cardiovascular disease as reported in some American Indian tribes(1) or the impression among Native American tribes that diabetes is always the primary illness despite the presence of underlying primary defects in lipoprotein transport which may have resulted in heart attacks.

Because adverse health outcomes associated with both obesity and early development of atherosclerosis are potentially reversible at an early age(43, 44), the results of our study provide a rationale for investigation of serial change in lipid profiles in longitudinal studies which take into account the influence of age-dependent and culture-specific life style trends. Because the onset and progression of obesity precedes and may predict NIDDM and is associated with the development of high risk lipoprotein profiles, intervention strategy should be aimed at early assessment followed by attempts to modify life style-related factors such as diet and exercise. Subsequent decreases in body fat and plasma lipid levels may then result in improved long-term cardiovascular health.