1 Department of Kinesiology, University of Maryland, College Park 20742
2 Division of Endocrinology, Diabetes, and Nutrition, University of Maryland School of Medicine, Baltimore 21201
3 Geriatric Research Education and Clinical Center, Baltimore Veterans Affairs Medical Center, Baltimore, Maryland 21201
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
insulin sensitivity; insulin responsiveness; body composition; physical activity; aging; diet
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Among the many candidate genes for diabetes is the intestinal fatty acid binding protein (FABP2) gene. The FABP2 gene belongs to a family of more than 20 FABP genes that are expressed in a tissue-specific manner (6). The FABP2 gene is expressed only in the absorptive simple columnar epithelial cells of the small intestine (54). In these cells, FABP2 transports hydrophobic fatty acids (FAs) from the plasma membrane, through the aqueous cytosol, to the endoplasmic reticulum (ER). In the ER, FAs are esterified with glycerol-3-phosphate (G3P) to form triglycerides (TGs). The TGs are packaged into chylomicrons that are transported into the circulation (19). The FABP2 gene is a candidate gene for insulin resistance because its product is involved in FA absorption and because defects in FA regulation have been hypothesized to play a role in insulin resistance (44).
It is important to keep in mind that type 2 diabetes is clearly multifactorial in origin, being determined by both genetic and environmental factors. Body composition, habitual physical activity (PA) levels, and diet are factors that are at least partly environmentally determined, and all are known to exert their own substantial and independent effects on insulin resistance and type 2 diabetes. Environmental factors clearly must play a substantial role in determining type 2 diabetes prevalence, because type 2 diabetes has become dramatically more common in the last few decades. During this same time, factors such as habitual PA levels and body composition have changed substantially, while negligible genetic alterations could have occurred over this time frame.
It is beyond the scope of this paper to review all of the evidence supporting the effects of body composition, habitual PA levels, and diet on insulin resistance/diabetes. Recent reviews have concluded that total body weight, overall body fatness, abdominal obesity, and increased visceral fat mass substantially increase a persons risk for developing insulin resistance/type 2 diabetes (27, 47). Furthermore, there is also substantial evidence that lifestyle factors such as PA and diet affect a persons risk of developing insulin resistance/type 2 diabetes (13, 27, 40, 47, 56). Data supporting these relationships have been derived from epidemiological, clinical, and intervention studies (27) including recently published results from the Finnish Diabetes Prevention Study (56) and the findings from the Diabetes Prevention Program in the United States (13). Insulin resistance/type 2 diabetes prevalence is also highly age related. However, it is possible that this age-related increase in prevalence of insulin resistance/type 2 diabetes may be due to age-related changes in body composition, habitual PA levels, and diet, rather than aging per se. Thus it is very possible that if not controlled or accounted for, body composition, PA, diet, and age could potentially confound the interpretation of the effects of FABP2 genotype on glucose and insulin metabolism or FABP2 genotype distribution differences between type 2 diabetic and control subjects.
We present in this paper an overview of the physiological role of FABP2 and the relationship between FABP2 gene variants and insulin resistance.
![]() |
Structure and Function of the FABP2 Gene and FABP2 |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Structure of the FABP2 protein.
FABP2 consists of 131 amino acid residues and has a molecular mass of 15 kDa (6, 54). The tertiary structure resembles a semi-flattened cylinder that is open at one end and closed at the other (Fig. 1). The body of the cylinder consists of 10 anti-parallel ß-strands. The ß-strands form a pair of ß-sheets that, in turn, form a partially flattened ß-barrel known as a ß-clam. The ß-clam is closed off at one end by two -helices. The internal cavity of the ß-barrel is the ligand-binding site and is capable of binding one molecule of ligand. Approximately half of the residue side chains that line the cavity are hydrophobic, which gives the protein specificity for binding saturated and unsaturated long-chain FA. The protein-ligand binding forces consist of both hydrophobic and electrostatic interactions. Ligand affinity varies with the identity of the FA (6, 54).
|
![]() |
FABP2 Variants and Their Relationship to Insulin Resistance |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Intron 2 trinucleotide repeat sequence variant.
Among the seven alleles found at the intron 2 repeat sequence, only two occur with a sufficiently high frequency to assess for associations with diabetes traits. Prochazka et al. (43) examined linkage of the two common alleles with fasting insulin concentration and maximum insulin-stimulated glucose uptake (Max M) determined in vivo using the euglycemic-hyperinsulinemic clamp procedure in nondiabetic Pima Indians. The base sequences for the alleles were not specified but were numbered " allele 1" through "allele 7." The common alleles were allele 1 and allele 3, which had frequencies of 0.28 and 0.67, respectively. Allele 1 homozygotes had lower fasting insulin concentrations and higher Max M values than allele 3 homozygotes, indicating a clear association between the FABP2 microsatellite variants and insulin responsiveness. It should be noted that no association was found between this polymorphism and submaximal insulin-stimulated glucose disposal. Only subjects with >25% body fat were included in this study; percent body fat was not linked to the FABP2 locus, but was corrected for in the linkage analyses.
Humphreys et al. (24) studied whites from Finland, the United Kingdom, and Wales for associations between diabetes status and FABP2 intron 2 alleles. A log-linear analysis revealed a positive association between the 108-bp allele and diabetes. On the other hand, the 114-bp allele was negatively associated with diabetes. BMI was measured in this study, but was not accounted for in the statistical analysis. Because the allele nomenclature differed between the studies by Humphreys et al. (24) and Prochazka et al. (43), comparing their findings is difficult. However, if the alleles are matched based on the frequencies from both studies, then the findings agree. The second most frequent allele in the study by Humphreys et al. was associated with diabetes, whereas the second most frequent allele in the study by Prochazka et al. was associated with low Max M and high fasting insulin concentrations. Humphreys et al. failed to find an association between specific metabolic parameters and the microsatellite alleles. However, the only metabolic data in this study were fasting insulin and glucose, 2-h post glucose challenge insulin and glucose, and the homeostasis model assessment (HOMA) index of insulin resistance.
Yagi et al. (60) studied a sample of 475 Japanese for associations between the intron 2 repeat sequence alleles and diabetes status. Allele frequencies were compared among groups of individuals with normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and type 2 diabetes. No association between these gene variants and glucose tolerance status was detected. These three groups also had very similar and very low BMI values.
Vionnet et al. (57) studied 182 sib-pairs from 172 French families to test for linkage between FABP2 microsatellite alleles and the presence/absence of insulin resistance. No linkage was identified; however, 74 of the subjects with overt type 2 diabetes were <45 yr old, while an unspecified number of the non-insulin-resistant subjects were <45 yr old. Because the type 2 diabetes phenotype has age-dependent penetrance, it is possible that this study may be confounded by the inclusion of younger affected individuals with maturity-onset diabetes in the young (MODY). Statistical analyses were performed separately on those with BMI less than vs. greater than 27 kg/m2, and body weight was included in some of the linkage analyses.
Rather than searching for an association between categorical intron 2 alleles and diabetes status, Rissanen and coworkers (46) looked for an association between the number of trinucleotide repeats and diabetes status. No relationship was found. As mentioned above, the diabetic subjects were older and had higher BMI values than controls.
In summary, the results concerning the relationship between FABP2 intron 2 variants and insulin resistance/diabetes status are equivocal. The only study to quantify insulin responsiveness (Max M) found an association with two of the more common alleles (43). The same study did not find a relationship between submaximal insulin-stimulated glucose uptake and intron 2 variation. The two studies that investigated an association between fasting insulin and intron 2 variation produced conflicting results (24, 43), with no obvious reason for the discrepancy. Last, three of the four studies that reported the relationship between intron 2 variation and diabetes status found no association. Overall, it appears that intron 2 variation does not have a major effect on insulin sensitivity or diabetes risk. However, many of the studies did not account for the potentially confounding influences of body composition and age, which could have obscured or masked possible associations. In addition, none of these studies accounted or controlled for the potentially confounding effects of habitual PA levels and diet. Furthermore, associations with this variant have not been addressed with substantial statistical power in most studies, and especially not within specific ethnic groups. The possibility also exists for stratification bias, which could result in both false-positive and false-negative results.
Codon 54 threonine for alanine variant.
The most extensively studied FABP2 variant is the Ala54Thr polymorphism. This variant is particularly intriguing because the variant gene product has been characterized and found to have altered function. Additionally, results from the functional studies are consistent with recent hypotheses concerning the physiological mechanisms thought to underlie insulin resistance (4, 27).
Hypothetical link between Ala54Thr FABP2 and insulin resistance.
One current hypothesis concerning the development of type 2 diabetes was presented recently by Ivy et al. (27). This hypothesis is consistent with the mechanisms presented by Baier et al. (4) relative to how the Ala54Thr variant causes insulin resistance. A schematic of the combined hypotheses of Ivy et al. and Baier et al. is presented in Fig. 2. Ivy et al. (27) proposed that the initial step in the development of insulin resistance is hypertrophy of adipocytes resulting from a positive caloric balance. This hypertrophy results in a lower density of insulin receptors on the adipocyte membrane that in turn causes adipocyte insulin resistance. Secondary to adipocyte insulin resistance, glucose uptake by adipocytes is impaired which results in inadequate adipocyte glycolysis and, consequently, G3P deficiency. If the adipocyte has adequate amounts of G3P, then FA from intracellular lipolysis and from the import of extracellular FA will be esterified with intracellular G3P to form TG, which cannot leave the cell. However, in insulin-resistant adipocytes, G3P levels are low, which allows FA to remain unesterified and free to diffuse into the bloodstream. The increase in circulating FA results in increased hepatic gluconeogenesis, because the liver takes up FA but needs little for energy. This de novo glucose is either stored as glycogen or released to the bloodstream. Additionally, lipid can accumulate in myocytes, which has been associated with muscle cell insulin resistance (17, 34, 39). The combination of increased hepatic glucose production and reduced muscle uptake of glucose results in hyperglycemia. Furthermore, euglycemia is commonly maintained in insulin-resistant individuals by means of hyperinsulinemia. The overproduction of insulin may subsequently exhaust the pancreatic ß-cells, resulting in insulin insufficiency, which greatly exacerbates glucoregulatory dysfunction.
|
![]() |
FABP2 Ala54Thr Gene Variant Studies |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Glucose clamp studies of insulin sensitivity.
Four studies used euglycemic-hyperinsulinemic clamps to quantify insulin sensitivity. The first was by Baier and coworkers (4) who studied insulin sensitivity at a medium insulin infusion rate of 40 mU·m-2·min-1 in 137 nondiabetic, obese Pima Indians. Subjects with at least one Thr54 allele were less insulin sensitive as indicated by 8% lower insulin-stimulated glucose uptake than Ala54 homozygotes. Baier et al. also found that the Thr54 carriers had larger integrated insulin responses to an oral glucose load and to a mixed meal, higher fasting fat oxidation rates, and higher fasting insulin concentrations. These differences were evident after adjusting for BMI and age.
Rissanen et al. (46) studied 100 diabetic and nondiabetic Finns and found no difference in clamp-measured insulin sensitivity between those with and without a Thr54 allele. It was not clear whether the gene associations were analyzed separately in diabetic and nondiabetic groups or whether all subjects were pooled for the analysis. The nondiabetic group had a mean BMI of 27 kg/m2, indicating that most individuals in this group were not obese by current World Health Organization standards and were only marginally overweight. This is in sharp contrast to the Pima Indians in the Baier et al. (4) study, who were much more obese, on average (fat mass = 34% of total body mass). If it is assumed that Finns studied by Rissanen et al. (46) live a healthier lifestyle than the Pima Indians studied by Baier et al. (4), then the results of these studies could indicate that the penetrance of insulin resistance associated with the Thr54 allele may depend on behavioral factors such as exercise and diet. The fact that the Finns were closer to healthy weight suggests that they are more physically active and/or eat healthier diets. In support of this is the fact that the prevalence of physical inactivity among US American Indians is 37%, whereas that of Finns is only 15% (52). This potential dependency of insulin resistance on PA and diet in individuals with Thr54 is consistent with the combined Baier and Ivy hypotheses discussed earlier.
Another study on Finns by the same researchers found no association between Ala54Thr genotype and clamp-measured insulin sensitivity (41). The focus of this study was on combined familial hyperlipidemia (FCHL), and therefore the subjects were probands with FCHL and first-degree relatives of the probands. As in the study by Rissanen et al. (46), these Finns were not obese and were only marginally overweight (BMI = 26 kg/m2).
The last investigation to examine the relationship between FABP2 genotype and insulin sensitivity by the glucose clamp methodology was conducted in Japanese type 2 diabetic subjects (26). Insulin action was not different between groups with and without a Thr54 allele. It is important to note that 60% of the subjects were undergoing diet therapy at the time of the clamp study. Since diet is an effective means of reducing insulin resistance, this may well have confounded the study results. Additionally, the mean BMI of the subjects was 22 kg/m2, indicating that most subjects were not obese. Lean individuals with type 2 diabetes are not representative of the common diabetic phenotype in non-Asians. Also, it has been hypothesized that the etiology of type 2 diabetes in Japanese may differ from that in non-Asian racial groups and is thought to stem from a defect in insulin secretion (11). These studies suggest that in lean Japanese, Ala54Thr is not associated with insulin resistance.
Insulin sensitivity estimates from intravenous glucose tolerance test data.
We recently assessed the association between FABP2 Ala54Thr genotype and insulin sensitivity in nondiabetic, postmenopausal white women while accounting for the potentially confounding effects of habitual PA levels, body composition, and hormone replacement therapy (HRT) status (10). Insulin sensitivity (SI) was estimated by Bergman minimal modeling of insulin-assisted frequently sampled intravenous glucose tolerance test glucose and insulin data. SI was correlated with maximal O2 uptake (r = 0.35, P < 0.01), as a surrogate measure of habitual PA levels, and with percent body fat (r = -0.37, P < 0.01) (9). SI was 26% lower (P < 0.05) in women with at least one Thr54 allele compared with Ala54 homozygotes after accounting for the effects of body composition, habitual PA levels, and HRT on SI. Although some previous studies reported significant associations between Ala54Thr FABP2 genotypes and insulin resistance/type 2 diabetes, they generally did not account for or report body composition or habitual PA level data that could have confounded the results. In our study, these factors were accounted for in the statistical analyses and Ala54Thr FABP2 genotypes were independently associated with insulin sensitivity.
Insulin sensitivity estimates from insulin suppression test.
Lopez-Miranda et al. (33) demonstrated that FABP2 Ala54Thr is associated with insulin sensitivity but only when subjects are on a high-fat diet. Somatostatin-induced insulin suppression and subsequent steady-state plasma glucose levels (SSPG) were used to assess insulin resistance in young healthy men and women after 4 wk on each of three diets. Ala54 carriers (n = 53) and Thr54 homozygotes (n = 6) had similar SSPG values after 4 wk on a low-fat diet. However, after a diet high in saturated fat and after a diet high in monounsaturated fat, Thr54 homozygotes had 28% and 44% higher SSPG values in response to somatostatin, respectively. These differences were independent of glycogen synthase XbaI and APO-C3 SstI polymorphisms, which were also suspected to associate with SSPG. Because the sample of Thr54 homozygotes was small, it is important that this finding is confirmed. However, this is the only study to date in which dietary fat (an environmental factor) was experimentally controlled and found to interact with Ala54Thr FABP2 genotype to associate with insulin sensitivity.
Insulin resistance estimates from oral glucose tolerance test data.
Studies examining the associations of genes with insulin resistance often use variables extracted from fasting blood samples and oral glucose tolerance tests (OGTTs) as surrogate measures for insulin resistance. Among these measures are indices of insulin sensitivity based on OGTT glucose and insulin, glucose and insulin areas under the curve (AUC) from OGTTs, 2-h glucose and insulin concentrations, fasting glucose and insulin concentrations, and glucose tolerance status. These measures are more economical and less time consuming to collect and are, therefore, often the only feasible option for cross-sectional studies with large sample sizes. However, it should be emphasized that none of these measures is a tightly controlled index of insulin resistance.
Chiu et al. (12) evaluated two OGTT-based estimates of insulin resistance (35, 53) for associations with Ala54Thr FABP2 genotype in 71 white men and women aged 1939 yr. Only individuals with normal blood pressure and NGT were studied. Additionally, a multivariate analysis was performed to identify and account for other potential covariates that affect insulin sensitivity. Both estimates of insulin sensitivity were associated with FABP2 Ala54Thr genotype. Based on the index of Matsuda and Defronzo (35), insulin sensitivity was lower in Thr54 carriers vs. Ala54 homozygotes after adjusting for waist-to-hip ratio and diastolic blood pressure; FABP2 genotype accounted for 4% of the variation in insulin sensitivity. Likewise, based on the index of Stumvoll et al. (53), insulin sensitivity was lower in Thr54 carriers after accounting for variance from waist-to-hip ratio; 5% of the variance in insulin sensitivity was due to FABP2 genotype.
HOMA studies of insulin resistance.
Other studies have attempted to quantify insulin resistance using the HOMA index, which is derived from the logarithmic relationship between fasting glucose and insulin levels (36). Among the four studies that used this index, one found a significant association with Ala54Thr FABP2 genotypes, whereas three did not (Table 2). All four studies were on Japanese individuals. Caution must be used in interpreting these studies, however, because HOMA was originally designed to differentiate between ß-cell defects and peripheral insulin resistance. The relatively low reproducibility of HOMA estimates of insulin resistance (coefficient of variation = 3040% between measurements made on two separate days) (Ref. 36) make extremely large sample sizes a requirement if associations are to be found between gene variants and insulin resistance.
|
More recent studies in Japanese subjects did not reproduce these results; however, none of the studies had as large a sample size as Yamada et al. (61). Ishii et al. (25) studied the effects of FABP2 Ala54Thr on HOMA measured insulin resistance in separate groups of 196 young men, 186 middle- to older-aged normoglycemic men, and 122 middle- to older-aged hyperglycemic men (including 77 subjects with diabetes). BMI, age, and blood pressure were similar among genotype groups within each of the three study groups. HOMA index did not differ among the three genotypes in any of the study groups. Hayakawa et al. (21) studied 258 individuals and pooled heterozygotes with Thr54 homozygotes for statistical analyses. Although the mean HOMA insulin resistance index was highest in the group with at least one Thr54 allele, the difference was not significant (Table 2). Age, BMI, and waist-to-hip ratio were similar between the Ala54Thr genotype groups. Similarly, in a study (26) of 88 diabetic Japanese, there was a trend for individuals with at least one Thr54 allele to be more insulin resistant, but the difference was not significant (Table 2). Age, BMI, and cross-sectional visceral fat area did not differ between Ala54Thr genotype groups.
The reason for the discrepancy among the studies that used HOMA as a measure of insulin resistance may be due to statistical power limitations in those that did not find a significant gene-phenotype association. Alternatively, it is possible that the study with positive results was a false positive. It is interesting to note that if the HOMA index values are averaged within genotypes across the four studies, Thr54 homozygotes and heterozygotes are 10% and 9% more insulin resistant than the Ala54 homozygotes, respectively (Table 2). Furthermore, these investigations all studied lean Japanese individuals. As discussed earlier, if leanness reflects a healthy lifestyle with a good diet and moderate-to-high habitual PA levels, these individuals may have represented a population in which insulin resistance and type 2 diabetes penetrance would be minimal. Additionally, as discussed before, it is possible that the etiology of type 2 diabetes in Japanese may not be initially dependent on insulin resistance.
Studies evaluating insulin, glucose, and C-peptide AUC from OGTTs.
Tahvanainen et al. (55) reported findings from the European Atherosclerosis Research Study II (EARS II) on 666, mostly lean, healthy, 18- to 28-yr-old male university students from 11 European countries. The three Ala54Thr FABP2 genotype groups had similar values for BMI, waist-to-hip ratio, age, and blood lipids. Neither insulin AUC nor glucose AUC differed among genotype groups. The lack of a genotype effect suggests that FABP2 does not influence insulin sensitivity in young, lean, healthy males.
In middle- to older-aged Chinese men and women with NGT (n = 115) or IGT (n = 54), Xiang et al. (59) found similar OGTT glucose areas between Thr54 carriers and noncarriers. Furthermore, fasting and 2-h postchallenge glucose levels did not differ between the genotype groups. However, in IGT individuals, but not those with NGT, postchallenge C-peptide AUCs were 19% lower in Thr54 carriers vs. Ala54 homozygotes, and insulin AUC tended (P = 0.063) to be lower in Thr54 carriers. In Ala54 homozygotes, individuals with IGT had 20% greater C-peptide AUC than NGT individuals, as would be expected in a state of insulin resistance. In Thr54 carriers, however, means for C-peptide AUC were identical between IGT and NGT groups (P = 0.98), suggesting that inappropriate insulin secretion may be contributing to the impaired glucoregulatory state in Thr54 carriers with IGT.
Two-hour insulin and glucose concentration studies.
Mitchell et al. (38) examined the relationship between short tandem repeats near the FABP2 gene locus and OGTT 2-h insulin concentrations using linkage analysis in 28 Mexican-American families including 288 sib-pairs. The LOD score for linkage between the putative locus and 2-h insulin values was 2.80. The authors determined that 33% of the variability in 2-h insulin concentrations was associated with this polymorphism. Additionally, if the effect of this locus is removed, then the heritability of 2-h insulin values was zero, which, as concluded by the authors, suggests "a single major gene effect." These results were based on analyses that accounted for the effects of age and BMI. From these studies it is not clear that the linkage identified by Mitchell et al. (38) was due to the Ala54Thr FABP2 variant or another nearby gene variant.
Data from the Framingham Offspring Study (18) were used to evaluate men (n = 762) and women (n = 922) separately for Ala54Thr polymorphism associations with fasting and 2-h postchallenge plasma glucose and insulin levels. Most subjects were white, and all lived in the United States. Five percent of the men and 9% of the women were diabetic. Although age and BMI varied widely among individuals, the mean values (56 yr and 27 kg/ m2, respectively) are suggestive of mostly middle- to older-aged, overweight individuals. A sexual dimorphism was found, as women Thr54 carriers had 12% higher 2-h post glucose insulin levels compared with their female Ala54 homozygous counterparts; however, no such association was found in men. Furthermore, the genotype association in women remained significant after adjusting 2-h insulin values for the potentially confounding effects of familial relationships, age, BMI, TG, APO E genotype, smoking, alcohol consumption, ß-blocker use, menopausal status, and HRT status. Although no further associations were found in women or men, fasting insulin level in women Thr54 carriers tended (P = 0.070) to be higher than those for the Ala54 homozygous women. This is the only study that analyzed women and men separately. Several other studies exclusively selected men for their research samples (25, 30, 55, 61), and their findings are equivocal, as half produced positive results (30, 61) while the others yielded negative results (25, 55). Only one study analyzed exclusively women (10) and it demonstrated an association between Ala54Thr genotype and insulin sensitivity.
Kim et al. (30) evaluated 96 lean Korean men with NGT for FABP2 Ala54Thr polymorphism associations with fasting and 2-h post glucose plasma insulin and glucose levels. Age and BMI were similar between genotype groups. Men with at least one Thr54 allele had 53% higher fasting plasma insulin levels than Ala54 homozygotes. Additionally, 2-h postchallenge insulin level for Thr54 carriers was 203 ± 24 pmol/l (mean ± SE), whereas that for the Ala54 homozygotes was 152 ± 22 pmol/l; however, these values were not statistically different (P > 0.05). Kim et al. (30) also found that the contribution of lipid oxidation to basal metabolic rate was 57% higher in Thr54 carriers vs. noncarriers. This finding supports that from studies on Pima Indians (4), which revealed greater fasting fat oxidation rates in Thr54 carriers vs. Ala54 homozygotes. Despite these genotype-dependent differences reported by Kim et al. (30), it is important to note that digestive oral fat absorption, measured using tritiated oleic acid in 25 ml of corn oil, was not different between genotype groups. That fat absorption was not different between groups appears contrary to the proposed mechanistic hypothesis for the connection between Ala54Thr FABP2 genotype and insulin resistance.
Hegele and coworkers (23) studied fasting and 2-h postchallenge blood glucose levels in 175 Inuit (Eskimos) from northern Canada for associations with Ala54Thr genotype. The Inuit are a unique group of people in that they have an unusually low prevalence of coronary heart disease (CHD) compared with other Canadians. This is despite higher frequencies of "disease-associated alleles" than their white counterparts for AGT codon 235, FABP2 codon 54, PON codon 192, and APO E exon 4 polymorphisms (23). Although fasting glucose did not differ between FABP2 genotype groups, 2-h postchallenge glucose for Thr54 homozygotes was 20% and 18% lower than in Ala54 homozygotes and heterozygotes, respectively. This outcome is counterintuitive with regard to the hypothetical connection between FABP2 genotype and insulin resistance. Hegele et al. (23) suggested that the effects of the Thr54 allele might not be uniform across all types of diets. The authors note that the diet of the Inuit is distinct from that of other populations studied in that it is heavily marine based with high levels of -3 FAs. Perhaps the higher affinity of the threonine-containing FABP2 for FAs provides a glucoregulatory benefit if the dietary fatty acids are
-3 unsaturated. Although these findings are generally opposite of previous findings, they do emphasize the importance of considering population, diet, and other environmental and behavioral factors when evaluating various genotype-phenotype associations.
Fasting glucose and insulin studies.
Hegele et al. (22) studied 507 Canadian aboriginals in a small isolated community in northern Canada. No association of Ala54Thr FABP2 genotype with either fasting insulin or glucose was found. Unlike most other studies, these authors did describe the activity pattern of the subjects as generally sedentary, although the method for determining this was not described. A major weakness of this study is that all volunteers >18 yr old were accepted as subjects, which resulted in a mean age of 36 yr for the study sample. Because the penetrance of insulin resistance is low in those <45 yr, most individuals would not yet be expected to demonstrate signs of insulin resistance. Age did not differ among Ala54Thr genotype groups, but BMI and percent body fat were significantly higher in the heterozygotes than Ala54 homozygotes.
Diabetes and glucose tolerance status studies.
Although diabetes and glucose tolerance status are less precise indices of insulin resistance, they have been widely studied because of their obvious direct implications to health. Since most of the studies discussed previously also assessed the association between Ala54Thr FABP2 genotype and diabetes status, they will all be reviewed in this section. Additionally, studies that used diabetes status as their sole phenotype will be reviewed. Before proceeding, however, it is important to consider the etiologic heterogeneity of type 2 diabetes. Until recently, diabetes mellitus was thought to develop from either one of two conditions: autoimmune destruction of pancreatic ß-cells (type 1) or insulin resistance (type 2). Although most cases of diabetes mellitus can be classified as type 1 or type 2, several phenotypes are known to exist. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus recently suggested that diabetes mellitus etiologies be classified as type 1, type 2, gestational, or "other specific types" (16). Although the etiology for type 1, gestational, and other specific types is relatively objective, the guidelines for classification as type 2 diabetes are quite general and therefore include cases for which a specific etiology cannot be assigned. The research implications based on this are twofold. First, it is implicit that many etiologies can result in a case being classified as type 2 diabetes. Studies on type 2 diabetes are therefore not studies on one phenotype, but many phenotypes. Secondly, if a study does not specifically exclude individuals with "other specific types" of diabetes, these cases will erroneously be included in the group of type 2 diabetic subjects, further broadening the range of phenotypes being analyzed. The common forms of type 2 diabetes, therefore, result from a variety of genetic and environmental factors, and it is unlikely that a single gene variant will explain most cases of type 2 diabetes.
Among 10 studies published on the association of Ala54Thr FABP2 genotype with diabetes status, only one showed a significant genotype-phenotype association. Boullu-Sanchis et al. (8) studied migrant Indian workers in Guadeloupe. Genotype frequencies were compared between 89 type 2 diabetic subjects and 100 age- and sex-matched controls. Thr54 homozygotes and heterozygotes were more likely to be diabetic than Ala54 homozygotes, with odds ratios of 2.54 and 1.53, respectively (P = 0.02). After adjusting for the presence or absence of TG 200 mg/dl, Thr54 homozygotes were still more likely to be diabetic (odds ratio = 3.11, P < 0.01), but heterozygotes were not. The authors did not find sedentary lifestyle to be a predictor of diabetes in their multivariate analysis; however, no description of "sedentary" was given. Age, BMI, waist-to-hip ratio, and central obesity were all significantly greater in diabetic subjects compared with controls. It appears that the Ala54Thr distribution differences between diabetic subjects and nondiabetic subjects continued to exist after accounting for the effect of BMI and central obesity.
Baier et al. (4) did not detect Thr54 allele frequency differences between diabetic and nondiabetic Pima Indians despite their finding that the allele predicts various indices of insulin resistance including in vivo insulin-stimulated glucose uptake. Hegele et al. (22) found no difference in diabetes prevalence between Ala54 homozygotes and heterozygotes (Thr54 homozygotes were not studied) in northern Canadian aborigines. In a study on Finns (46), Ala54 and Thr54 allele frequencies were not different between 19 diabetic patients and 81 controls; however, the small sample resulted in low statistical power. In white Americans studied by Galluzzi et al. (18), diabetes rates were 5% and 9% for 922 women and 762 men, respectively. However, the prevalence of diabetes between Thr54 carriers and Ala54 homozygotes was not different for either sex. Ito et al. (26) found similar allele frequencies between 150 diabetic and 147 nondiabetic Japanese. Ishii et al. (25) found no genotype or allele frequency differences between 186 normoglycemic and 122 hyperglycemic Japanese men. Xiang et al. (59) did not find differences in Ala54Thr genotype and allele frequencies across groups of 116 NGT, 54 IGT, and 61 type 2 diabetic Chinese men and women. Lei et al. (32), in a large study on blacks, found similar allele frequencies between 321 diabetic subjects and 992 nondiabetic subjects.
A common limitation of case-control studies such as those presented above is that case and control groups may be genetically homogeneous within groups and genetically heterogeneous between groups, thereby resulting in spurious associations for many polymorphic markers. This population stratification problem is generally credited with generation of false-positive associations and might have contributed to the association described by Boullu-Sanchis et al. (8). However, the subjects in the case and control groups of Boullu-Sanchis et al. (8) were ethnically homogeneous, which reduces the risk of population stratification.
The only study that formally accounted for the possible stratification effects was that by Altshuler et al. (2). The study used Scandinavian family trios in attempt to identify an association between Thr54 and "cases" (individuals with type 2 diabetes, IGT, or impaired fasting glucose). In the 333 family trios evaluated, the number of Thr54 alleles transmitted from a homozygous parent was similar to the number of alleles not transmitted, indicating no association between FABP2 Thr54 and diabetes/impaired glucoregulatory status.
To summarize, only one study among 10 has identified an association between Ala54Thr and type 2 diabetes status. Based on this, it is reasonable to conclude that FABP2 is not a major "diabetes gene." What cannot be ruled out, however, is whether FABP2 genotype has a minor effect by altering the propensity for type 2 diabetes. Perhaps, as the mystery of type 2 diabetes genetics is solved, we will realize that major "disease genes" do not exist; rather, many polymorphic loci will explain the genetic contribution to diabetes. Alternatively, as our understanding of the sub-phenotypes of "type 2 diabetes" becomes more complete, "major genes" may be found that predict disease risk in specific subtypes of type 2 diabetes.
Among the studies that evaluated FABP2 Ala54Thr relationships with insulin sensitivity, the findings are equivocal. Several studies identified associations, whereas others did not. This scenario of unconfirmed association study findings has received recent attention in the form of a number of editorials (14, 58). The editorials suggest that association studies are prone to bias; however, this bias can be minimized if caution is used in designing studies and interpreting results. A reason for inconsistent results may be that some of the studies used indirect indices of insulin resistance. A second problem is that few studies accounted or controlled for body composition, habitual PA levels, or diet, all of which are known to have large effects on insulin resistance. Furthermore, study samples were drawn from a variety of cultural and racial populations, which could produce varied results. The studies with positive associations might well have been false-positive results or spurious associations due to factors such as population stratification, observation bias, or type 1 statistical errors. With regard to type 1 statistical errors, it is noteworthy that all but one of the nine studies with positive findings identified the Thr54 allele as the " high-risk" allele. If these studies were false positives, then it would be expected that half of the studies would identify Thr54 as the "high-risk allele," whereas the others would find Thr54 as the "low-risk allele." The unilateral pattern of the positive findings in addition to the theoretical biological relationship between Thr54 and insulin sensitivity is suggestive of a true-positive genotype association (48). Conversely, studies with negative findings might have failed to detect associations due to the use of insensitive phenotype measures, inadequate sample sizes (type 2 errors), or failure to account for confounding genetic and/or environmental factors. To date, a complete map of SNPs in the regions flanking the FABP2 locus has not been generated. It is quite possible that any associations found between FABP2 polymorphisms and type 2 diabetes/insulin resistance are due to linkage disequilibrium with a nearby putative gene polymorphism. Finally, a limitation of this review is that only published studies are presented here. The scientific literature tends to be biased toward studies with positive findings. Although the extent of a publication bias cannot be known, the conclusions presented in this review reflect such a bias if a bias is present in the literature.
![]() |
Summary and Conclusion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
ACKNOWLEDGMENTS |
---|
E. P. Weiss was supported by National Institutes of Health (NIH) Grant T32-AG-00268. M. D. Brown was supported by the American Heart Association. A. R. Shuldiner was supported by NIH Grants K24-DK-02673 and RO1-DK-54261 and by the American Heart Association, the American Diabetes Association, and the Baltimore VA Geriatric Research, Education, and Clinical Center. J. M. Hagberg was supported by NIH Grants RO1-AG-17474 and RO1-AG-15389.
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: J. M. Hagberg, Dept. of Kinesiology, Univ. of Maryland, College Park 20742-2611 (E-mail: jh103{at}umail.umd.edu).
10.1152/physiolgenomics.00070.2001.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|