OBJECTIVE—The provariant of the Pro12Ala polymorphism in peroxisome proliferator–activated receptor (PPAR)γ has been identified as a risk allele for type 2 diabetes. The purpose of the present study was to reveal a significant association with pre-diabetic phenotypes in nondiabetic individuals based on a systematic meta-analysis of all available published evidence.

RESEARCH DESIGN AND METHODS—We performed a classical meta-analysis of data from ∼32,000 nondiabetic subjects in 57 studies to assess the effect of the Pro12Ala polymorphism on pre-diabetic traits.

RESULTS—In the global comparison, there were no differences in BMI, glucose, insulin, or homeostasis model assessment of insulin resistance between the Pro/Pro and X/Ala genotype. However, in the Caucasian subgroup, the X/Ala genotype was associated with significantly increased BMI. In the obese subgroup (BMI >30 kg/m2), fasting glucose (P = 0.041) and insulin resistance (by homeostasis model analysis) (P = 0.020) were significantly greater in the Pro/Pro group. In subjects with the homozygous Ala/Ala genotype, fasting insulin was significantly lower compared with the Pro/Pro genotype (P = 0.040, NAla/Ala = 154).

CONCLUSIONS—Across all studies, the Pro12Ala polymorphism had no significant effect on diabetes-related traits. Only in selected subgroups, such as Caucasians and obese subjects, did we see an association of the Ala allele with greater BMI and greater insulin sensitivity. This demonstrates the importance for appropriate stratification of analyses by environmental or other genetic factors. Meta-analysis of Ala/Ala homozygotes more clearly demonstrated the association with greater insulin sensitivity of carriers of the Ala allele.

The Pro12Ala polymorphism in peroxisome proliferator–activated receptor (PPAR)γ, caused by a missense mutation in exon B of the adipocyte-specific γ2 isoform, was identified in 1997 and is thought to confer reduced transcriptional activity (1). Several more genetic variants in PPARγ are known but are much less frequent (2). The prevalence of the Ala allele varies from ∼4% in Asian (3) to ∼28% in Caucasian (4) populations. Since PPARγ2 is exclusively expressed in adipose tissue, the prevalent Pro12Ala polymorphism was originally studied for an association with obesity. Soon, evidence emerged suggesting that the Ala variant is associated with increased insulin sensitivity in nondiabetic Caucasians (5). Consequently, association with type 2 diabetes was examined, and an early meta-analysis (6) strongly suggested that the Pro variant is a risk allele. This analysis included 16 genetic association studies and used a family-based design to control for population stratification. A more recent meta-analysis (7) confirmed these findings, and the Pro12Ala polymorphism is now considered the best replicated genetic risk factor of common type 2 diabetes.

The key components in the pathogenesis of type 2 diabetes are insulin resistance and β-cell dysfunction (8,9). To further clarify the mechanisms underlying the association of Pro12Ala with type 2 diabetes, however, it is necessary to examine nondiabetic populations, because hyperglycemia itself can induce insulin resistance and secretory dysfunction (10,11). Therefore, we performed a meta-analysis of 57 suitable studies containing data related to the insulin resistance of cohorts with normal or impaired glucose tolerance. The hypothesis is that the Pro/Pro genotype is associated with a more “diabetic” phenotype in this population (e.g., greater insulin resistance).

Appropriate studies were collected by a PubMed search (before July 2005) using the phrase “(pro12ala or p12a) and (stud* or diabetes or insulin sensitivity or insulin secretion).” Additionally, reference lists from articles were searched for further suitable studies. Two articles could not be obtained, one from a Chinese journal (12) and the other one from a Polish journal (13). The retrieval contained 151 articles.

We used the following exclusion criteria: review articles without new data; multifold data or interim data of studies, except when interim analyses contained data not presented in the final analysis; studies only including patients with type 2 diabetes; patients with polycystic ovarian syndrome; hyperlipidemia; studies including children; missing case numbers or measurements of deviation; and missing genotype-specific data of the quantities of interest (BMI, fasting plasma glucose, and insulin; 2-h glucose and insulin during oral glucose tolerance test [OGTT]; and homeostasis model assessment of insulin resistance [HOMA-IR]). The 57 studies suitable for meta-analysis are shown in Table 1.

### Data analysis

We assumed normal distribution for fasting glucose and 2-h glucose during OGTT and a log normal distribution for all other quantities, compatible with the original data from Tschritter et al. (14). Hence, means and SDs of BMI, fasting insulin, 2-h insulin during OGTT, and HOMA-IR were transformed to means and SDs of corresponding logarithmic quantities using standard formulas.

Logarithmic transformation provided the mean of log insulin when geometric means were presented. In the study of Erikson et al. (15), means and SDs for log glucose were calculated under the assumption of log normally distributed glucose. We calculated the corresponding means and SDs for glucose but treated the later quantities as parameter estimates for a normally distributed quantity according to our distribution assumption.

If only quartile range or CI as a measure of deviation were available, SD was estimated under the above distribution assumptions. More precisely, for normally distributed quantities, we used the differences of quartile range or CI boundaries to calculate SD. For log normally distributed quantities, the differences of the logarithmic boundaries of quartile range or CI were taken to calculate SD.

Altshuler et al. (6) did not provide genotype-specific case numbers after a written request. In this case, we used the available allele frequencies to estimate a lower boundary of the case numbers. Hence, the influence of this study in our analysis is underestimated.

Subjects with Pro/Ala and Ala/Ala were referred to as X/Ala for the pooled analysis. Where possible, a separate analysis of the Ala/Ala genotype in comparison with Pro/Pro was performed. In some studies, sex and age subgroups were pooled by calculating the common mean and SD. Analysis was globally performed for all eligible studies and within the subgroups of Caucasian, Asian, lean, and obese individuals. To avoid overlap of subgroups, instead of via a cutoff point, we determined obesity by classifying mean BMI <25 kg/m2 as lean, mean BMI >30 kg/m2 as obese, and discharged groups in between.

### Statistics

For statistical analysis, we applied the general approach for meta-analyses proposed by Whitehead (16,17). Hence, test statistic is the χ12-distributed U:

$U{=}\ \left(\ {{\sum}_{i}}{\hat{{\theta}}}_{i}w_{i}\right)^{2}/{{\sum}_{i}}w_{i}$

where θ̂i is the effect size and wi is the weight of the i-th study.

We calculated dimensionless standardized effect sizes θ̂i, which are the differences of the means of the compared groups divided by their common SD. Consequently, the weights depend only on case numbers. This method is advised when different unit systems are used, consistency among laboratories cannot be guaranteed, or if absolute study results cannot be directly compared (e.g., because of large differences between study populations and/or different adjustment methods). Results are shown in Figs. 13, providing standardized effect sizes and corresponding CIs.

We tested for the null hypothesis of no difference between the genotypes. The effect of the meta-analysis and its CI is also shown in comparison with the effects of the single studies. Negative values for the meta-effect size indicate that the corresponding quantity is smaller in Pro/Pro genotypes compared with X/Ala genotypes.

Inhomogeneity is analyzed by Q statistics. We calculated fixed-effects models and, in the case of significant inhomogeneity, random-effects models. Studies inconsistent with the other results were identified by radial plots (18) (data not shown). Regression of these plots were used to detect a possible publication bias (65).

Finally, robustness of significant results was investigated by a jackknife procedure. If significance was lost by leaving out single studies, we interpreted the result as having insufficient evidence for an effect. A substantial effect is indicated in cases of improvement of homogeneity and significance by leaving out a single study. All calculations except the power calculations were performed using MATLAB 7.0.4.365 (The MathWorks, Natick, MA). The powers were calculated with nQuery Advisor 6.0 (Statistical Solutions, Saugus, MA).

The Eggers test did not indicate any obvious publication bias.

### X/Ala versus Pro/Pro (Table 2)

BMI was not different in the global analysis. In the Caucasian subgroup, however, BMI was greater in X/Ala compared with Pro/Pro (P = 0.015), which held up to jackknifing (Fig. 1). The BMI in Asians was also greater in the X/Ala group, but after leaving out one of the studies of Tai et al. (19), Hara et al. (20), Kawasaki et al. (21), or Wang et al. (22), the effect became nonsignificant. We found no effect of the Pro12Ala polymorphism with respect to fasting insulin, neither globally nor in any of the subgroups.

Fasting glucose was not different in the global comparison. In the obese subgroup, however, fasting glucose was higher in the Pro/Pro group (P = 0.041), and leaving out the study of Lindi et al. (23) substantially increased significance (standardized effect size 0.127, P = 0.0033) and homogeneity (P = 0.76). There was no significant effect on 2-h insulin and glucose during OGTT, neither globally nor in any subgroup.

HOMA-IR was significantly higher in Pro/Pro compared with X/Ala in the obese subjects, though not globally, indicating greater insulin sensitivity in obese carriers of the Ala allele. This effect became stronger (standardized effect size 0.148, P = 0.0025) with higher homogeneity of the remaining studies after leaving out the study of Lindi et al. (23) (Fig. 2).

### Analysis of Ala/Ala

In 12 studies, the authors presented full genotype information, which allowed the analysis of the homozygous Ala/Ala genotype. Because of insufficient data quantity, the parameters HOMA-IR, 2-h glucose, and insulin were not analyzed in this part. Furthermore, only global testing of all pooled subgroups was performed. A total of up to 160 subjects with the Ala/Ala genotype were available for meta-analysis. Neither BMI (standardized effect size −0.125, P = 0.353) nor fasting glucose (standardized effect size −0.088, P = 0.282) were significantly different between the genotypes after global testing.

Fasting insulin was significantly lower in Ala/Ala compared with the Pro/Pro genotype (standardized effect size 0.168, P = 0.040). However, this effect vanished after jackknifing. The effect also vanished after calculating a random-effects model, which might have been more appropriate in view of the borderline study homogeneity (P = 0.052 for test of inhomogeneity). On the other hand, excluding the study of Rosmond et al. (24), which reported extremely high fasting insulin values in the Ala/Ala group discrepant from all other studies, makes the effect highly significant (standardized effect size 0.227, P = 0.0067) and the remaining studies highly homogeneous (P = 0.81 for inhomogeneity test; Fig. 3).

The BMI of individuals with the Pro/Pro genotype did not significantly differ from individuals with the X/Ala genotype in the global analysis. Interestingly, in the Caucasian subgroup, the X/Ala genotype was associated with significantly increased BMI. The absence of a statistical effect in the Asian subgroup might be explained by gene-gene and gene-environmental interactions (25). Also, the lower frequency of the Ala allele in Asians and the lower sample size in comparison with the Caucasian subgroup may play a role. Masud et al. (64) found a greater BMI in Ala carriers in obese subjects, but this study included patients with diabetes.

Surrogate markers of insulin resistance, such as fasting insulin and HOMA-IR, were not significantly different between the Pro/Pro and X/Ala genotypes globally. In the obese subgroup (BMI >30 kg/m2), however, fasting glucose was significantly higher in the Pro/Pro group. Analogously, HOMA-IR was also significantly higher in the obese subgroup with the Pro/Pro genotype. These effects became even stronger after leaving out the study of Lindi et al. (23), which only included obese subjects with impaired glucose tolerance. Two-hour postchallenge glucose and insulin concentrations were not significantly different between genotypes, neither in the global nor in the subgroup analysis, but the case numbers are much lower.

Subsequently, we investigated the influence of Ala/Ala homozygosity when full genotype information was given. Fasting insulin was higher in the Pro/Pro compared with the Ala/Ala group. Jackknifing revealed that the study of Rosmond et al. (24) was responsible for significant inhomogeneity among the studies. That study reported extremely high fasting insulin values in the Ala/Ala group, in discordance with all other studies. After excluding the study of Rosmond et al. (24), which contained only six individuals with Ala/Ala, the effect became highly significant and the remaining studies highly homogeneous. It is important to note that these six individiuals had a significantly higher BMI, which would easily explain the differences in fasting insulin and insulin sensitivity. Therefore, we interpret the results of the Ala/Ala meta-analysis as an indication of greater insulin sensitivity of this genotype.

To ultimately assess the role of greater insulin resistance in the Pro/Pro group in comparison with the Ala/Ala group, a meta-analysis of euglycemic-hyperinsulinemic clamp data would be desirable. This could not be performed because the four available studies (2629) were too heterogenous and could not be subjected to jackknifing. Two-hour glucose and insulin were not analyzed in Ala/Ala individuals due to insufficient data.

Another shortcoming of our meta-analysis is the classical meta-analysis approach with respect to the quantities considered. Because several studies used different adjustment methods, a meta-regression of the data might have been more appropriate. However, this would require access to original data of the individual studies, which was not possible. Because of different adjustment methods or different study populations, we always calculated standardized effect measures, which are more robust than absolute effect measures.

To evaluate the sensitivity of our meta-analysis, we estimated the statistical power and the absolute effects under the assumption of variance homogeneity among the studies. With a power of 80% for log(BMI), a global effect size of 0.8% difference (related to the group with the greater value) between the Pro/Pro and X/Ala groups could be discovered. On the other hand, with calculation of the common variance, the estimated effect was only a 0.7% difference. For fasting glucose, the discoverable effect was 0.045 mmol/l but the estimated effect was only 0.023 mmol/l. For logarithm of fasting insulin, the discoverable effect was a 2.7% difference but the estimated effect was only 2%. Consequently, even very small effects could be detected with our analysis but were not found globally. Furthermore, the estimates of the global effects were very precise (narrow CI) despite some inhomogeneity of the studies.

In conclusion, across all studies the Pro12Ala polymorphism had no significant effect on diabetes-related traits within a nondiabetic population. Only in selected subgroups, such as Caucasians and obese subjects, did we see an association of the Ala allele with greater BMI and greater insulin sensitivity. This demonstrates the importance for appropriate stratification of analyses by as many factors as possible. Meta-analysis of Ala/Ala homozygotes more clearly demonstrated the association with greater insulin sensitivity of the Ala allele.

Figure 1—

BMI on eligible studies (Pro/Pro versus X/Ala). Estimated standardized effect sizes and CIs are given for the single studies for global comparison and Caucasian subgroup. Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Figure 1—

BMI on eligible studies (Pro/Pro versus X/Ala). Estimated standardized effect sizes and CIs are given for the single studies for global comparison and Caucasian subgroup. Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Close modal
Figure 2—

HOMA-IR on eligible studies (Pro/Pro versus X/Ala). Estimated standardized effect sizes and CIs are given for the single studies, for global comparison and obese subgroup. Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Figure 2—

HOMA-IR on eligible studies (Pro/Pro versus X/Ala). Estimated standardized effect sizes and CIs are given for the single studies, for global comparison and obese subgroup. Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Close modal
Figure 3—

Fasting insulin on eligible studies (Pro/Pro versus Ala/Ala). Estimated standardized effect sizes and CIs are given; the meta-analysis effect and its CI is shown for global comparison and separately after excluding the study of Rosmond et al. (24). Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Figure 3—

Fasting insulin on eligible studies (Pro/Pro versus Ala/Ala). Estimated standardized effect sizes and CIs are given; the meta-analysis effect and its CI is shown for global comparison and separately after excluding the study of Rosmond et al. (24). Positive standardized effect size indicates that the corresponding quantity is greater in Pro/Pro than in X/Ala.

Close modal
Table 1—

Studies included in meta-analysis

Author (ref.)RaceLean/obesenAge (years)Sex distribution (% female)X/Ala frequency (%)Comments
Altshuler (6Caucasian Mixed 349 31 51 25 NGT, estimated minimal case numbers
Andrulionyte (30Caucasian Obese 770 54.8 50 23 IGT
Baratta (31Caucasian Mixed 338 37.9 56 12 —
Beamer (32) a Caucasian Mixed 517 64.4 39 21 BLSA
Beamer (32) b Caucasian Obese 169 43.3 66 17 JHU-WMC
Buzzetti (33Caucasian Obese 1,215 42.7 70 17 —
Clement (34Caucasian Lean 295 53 66 17 SU.VI.MAX, only control subjects taken
Deeb (5) a Caucasian Mixed 973 70 63 29 Elderly group
Deeb (5) b Caucasian Mixed 333 44.8 50 23 Middle-aged group
Douglas (35Caucasian Mixed 191 61.4 68 36 Spousal control subjects, only significant results presented
Ek (26) a Caucasian Lean 364 25.2 49 26 Danish group
Ek (26) b Caucasian Mixed 616 70 — 26 Swedish group
Eriksson (15Caucasian Mixed 476 69.6 64 32 —
Eurlings (36Caucasian Mixed 124 51.1 60 25 Familial combined Hyperlipidemic spouses
Evans (37Caucasian Obese 429 45.7 — 25 —
Fornage (38) a Caucasian Lean 1,954 25.6 59 24 CARDIA study
Fornage (38) b African Mixed 1,844 24.4 54 CARDIA study
Franks (39Caucasian Mixed 506 53.1 55 21 —
Frederiksen (4Caucasian Mixed 2,245 53.3 51 26 —
Ghoussaini (40) a Caucasian Lean 865 51.0 62 22 NGT
Ghoussaini (40) b Caucasian Obese 507 47.2 68 22 Obese group
Hara (20Asian Lean 541 69.1 53 Unselected population
Hegele (41Indian Mixed 171 30.4 100 13 Canadian Oji-Cree
Kahara (42Asian Lean 123 45.2 —
Kawasaki (21Asian Lean 116 53.4 61 —
Kolehmainen (43Caucasian Obese 22 43.8 67 27 —
Li (44mixed Mixed 686 20–38 — 20 Bogalusa Study
Lindi (23Caucasian Obese ca. 490* 55.2 67 31 IGT
Lindi (45Caucasian Mixed 150 49.2 49 24 KANWU study
Luan (46Caucasian Mixed 592 53.9 56 21 Isle of Ely Study
Mancini (47Caucasian Mixed 312 45.2 — 18 NGT
Meirhaeghe (48Caucasian Mixed 1,133 35–64 50 21 MONICA study
Mori (3Asian Lean 1,212 71.3 57 —
Mousavinasab (49Caucasian Mixed 252 17–28 31 Soldiers study
Muller (28Indian Obese 241 27.5 44 14 Pimas
Nicklas (50Caucasian Obese ca. 50* 60.2 100 21 Postmenopausal women, before weight loss
Oh (51Asian Mixed 229 48.0 36 —
Ostergard (27Caucasian Mixed 54 33.6 44 41 Only control subjects taken
Pisabarro (52Caucasian Obese 45 39.0 71 13 —
Poirier (53Caucasian Lean 675 18–28 25 Family history of myocardial infarction
Poulsen (54Caucasian Mixed 553 66.5 — 22 Twin study
Robitaille (55Caucasian Mixed ca. 600* 43.1 57 19 Quebec Family Study
Rosmond (24Caucasian Mixed 268 58 31 —
Sanchez (56) a Caucasian Mixed ca. 460* 48.7 55 17 All subjects group
Sanchez (56) b Caucasian Obese ca. 140 52.0 65 18 Obese group
Stumvoll (29Caucasian Mixed 318 33 60 24 —
Tai (19Asian Lean 3,618 37.1 55 NGT + IGT group
Takata (57Asian Lean 247 24.2 41 —
Thamer (58Caucasian Mixed 648 36 63 23 —
Tschritter (14Caucasian Lean 229 32.0 51 23 Clamp data
Vaccaro (59) a Caucasian Lean 280 45.6 42 15 Lean group
Vaccaro (59) b Caucasian Obese 141 45.6 44 23 Obese and obese with early onset
Valve (2Caucasian Obese 141 43.0 100 24 —
Wang (22Asian Mixed 153 — — NGT + IGT
Weiss (60Caucasian Mixed 73 58.4 56 16 Sedentaries, baseline
Yamamoto (61Asian Lean 595 48.1 20 —
Yamamoto (62Asian Lean 81 43.8 Patients with hypertension
Yang (63Asian Mixed 1,708 50.8 56 —
Author (ref.)RaceLean/obesenAge (years)Sex distribution (% female)X/Ala frequency (%)Comments
Altshuler (6Caucasian Mixed 349 31 51 25 NGT, estimated minimal case numbers
Andrulionyte (30Caucasian Obese 770 54.8 50 23 IGT
Baratta (31Caucasian Mixed 338 37.9 56 12 —
Beamer (32) a Caucasian Mixed 517 64.4 39 21 BLSA
Beamer (32) b Caucasian Obese 169 43.3 66 17 JHU-WMC
Buzzetti (33Caucasian Obese 1,215 42.7 70 17 —
Clement (34Caucasian Lean 295 53 66 17 SU.VI.MAX, only control subjects taken
Deeb (5) a Caucasian Mixed 973 70 63 29 Elderly group
Deeb (5) b Caucasian Mixed 333 44.8 50 23 Middle-aged group
Douglas (35Caucasian Mixed 191 61.4 68 36 Spousal control subjects, only significant results presented
Ek (26) a Caucasian Lean 364 25.2 49 26 Danish group
Ek (26) b Caucasian Mixed 616 70 — 26 Swedish group
Eriksson (15Caucasian Mixed 476 69.6 64 32 —
Eurlings (36Caucasian Mixed 124 51.1 60 25 Familial combined Hyperlipidemic spouses
Evans (37Caucasian Obese 429 45.7 — 25 —
Fornage (38) a Caucasian Lean 1,954 25.6 59 24 CARDIA study
Fornage (38) b African Mixed 1,844 24.4 54 CARDIA study
Franks (39Caucasian Mixed 506 53.1 55 21 —
Frederiksen (4Caucasian Mixed 2,245 53.3 51 26 —
Ghoussaini (40) a Caucasian Lean 865 51.0 62 22 NGT
Ghoussaini (40) b Caucasian Obese 507 47.2 68 22 Obese group
Hara (20Asian Lean 541 69.1 53 Unselected population
Hegele (41Indian Mixed 171 30.4 100 13 Canadian Oji-Cree
Kahara (42Asian Lean 123 45.2 —
Kawasaki (21Asian Lean 116 53.4 61 —
Kolehmainen (43Caucasian Obese 22 43.8 67 27 —
Li (44mixed Mixed 686 20–38 — 20 Bogalusa Study
Lindi (23Caucasian Obese ca. 490* 55.2 67 31 IGT
Lindi (45Caucasian Mixed 150 49.2 49 24 KANWU study
Luan (46Caucasian Mixed 592 53.9 56 21 Isle of Ely Study
Mancini (47Caucasian Mixed 312 45.2 — 18 NGT
Meirhaeghe (48Caucasian Mixed 1,133 35–64 50 21 MONICA study
Mori (3Asian Lean 1,212 71.3 57 —
Mousavinasab (49Caucasian Mixed 252 17–28 31 Soldiers study
Muller (28Indian Obese 241 27.5 44 14 Pimas
Nicklas (50Caucasian Obese ca. 50* 60.2 100 21 Postmenopausal women, before weight loss
Oh (51Asian Mixed 229 48.0 36 —
Ostergard (27Caucasian Mixed 54 33.6 44 41 Only control subjects taken
Pisabarro (52Caucasian Obese 45 39.0 71 13 —
Poirier (53Caucasian Lean 675 18–28 25 Family history of myocardial infarction
Poulsen (54Caucasian Mixed 553 66.5 — 22 Twin study
Robitaille (55Caucasian Mixed ca. 600* 43.1 57 19 Quebec Family Study
Rosmond (24Caucasian Mixed 268 58 31 —
Sanchez (56) a Caucasian Mixed ca. 460* 48.7 55 17 All subjects group
Sanchez (56) b Caucasian Obese ca. 140 52.0 65 18 Obese group
Stumvoll (29Caucasian Mixed 318 33 60 24 —
Tai (19Asian Lean 3,618 37.1 55 NGT + IGT group
Takata (57Asian Lean 247 24.2 41 —
Thamer (58Caucasian Mixed 648 36 63 23 —
Tschritter (14Caucasian Lean 229 32.0 51 23 Clamp data
Vaccaro (59) a Caucasian Lean 280 45.6 42 15 Lean group
Vaccaro (59) b Caucasian Obese 141 45.6 44 23 Obese and obese with early onset
Valve (2Caucasian Obese 141 43.0 100 24 —
Wang (22Asian Mixed 153 — — NGT + IGT
Weiss (60Caucasian Mixed 73 58.4 56 16 Sedentaries, baseline
Yamamoto (61Asian Lean 595 48.1 20 —
Yamamoto (62Asian Lean 81 43.8 Patients with hypertension
Yang (63Asian Mixed 1,708 50.8 56 —
*

Different number of cases for different quantities.

Part a completely includes part b; therefore, they were only separately included in the analysis. Studies included into analysis in alphabetical order. BLSA, Baltimore Longitudinal Study of Aging; CARDIA, Coronary Artery Risk Development in Young Adults; IGT, impaired glucose tolerance; JHU-WMC, Johns Hopkins University Weight Management Center; KANWU, Kuopio, Arthus, Naples, Wallongog, Uppsala; MONICA, Monitoring of Trends and Determinants in Cardiovascular Diseases; NGT, normal glucose tolerance; SU.VI.MAX, Supplementation en Vitamines et Mineraux Antioxydants Study.

Table 2—

Standardized effect sizes for comparison Pro/Pro versus X/Ala

Pro/Pro-X/Ala (standardized effect size)
GlobalCaucasiansAsiansLean (mean BMI <25 kg/m2)Obese (mean BMI >30 kg/m2)
Log (BMI) −0.037 (−0.087 to 0.012); P = 0.139; r, 50; NPro/Pro = 24,074; NX/Ala= 5,350 −0.062 (−0.111 to −0.012); P = 0.015; r, 37; NPro/Pro = 15,129; NX/Ala = 4,570* −0.087 (−0.174 to −0.001); P = 0.048; f, 10; NPro/Pro = 6,354; NX/Ala = 561
Fasting glucose 0.022 (−0.019 to 0.062); P = 0.296; r, 49; NPro/Pro = 23,615; NX/Ala = 4,930 0.011 (−0.041 to 0.063); P = 0.673; r, 34; NPro/Pro = 13,015; NX/Ala = 3,965 0.069 (−0.009 to 0.147); P = 0.085; f, 11; NPro/Pro = 7,937; NX/Ala = 686 0.015 (−0.040 to 0.070); P = 0.602; f, 14; NPro/Pro = 9,416; NX/Ala = 1,550 0.081 (0.003–0.158); P = 0.041; f, 12; NPro/Pro = 3,099; NX/Ala = 824
2-h glucose 0.018 (−0.079 to 0.115); P = 0.713; r, 12) NPro/Pro =5,121; NX/Ala =1,163 −0.019 (−0.131 to 0.094); P = 0.745; r, 8; NPro/Pro = 2,976; NX/Ala = 959 — — −0.010 (−0.124 to 0.104); P = 0.864; f, 4; NPro/Pro = 1,251; NX/Ala = 395
Log(fasting insulin) 0.031 (−0.024 to 0.085); P = 0.268; r, 51; NPro/Pro = 23,467; NX/Ala = 5,145 0.017 (−0.037 to 0.071); P = 0.536; r, 37; NPro/Pro = 13,981; NX/Ala = 4,278 −0.076 (−0.160 to 0.008); P = 0.078; f, 10; NPro/Pro = 6,823; NX/Ala = 588 −0.055 (−0.114 to 0.003); P = 0.062; f, 12; NPro/Pro = 8,056; NX/Ala = 1,403 0.163 (−0.009 to 0.334); P = 0.064; r, 12; NPro/Pro = 3,075; NX/Ala = 813
Log(2-h insulin) 0.087 (−0.031 to 0.205); P = 0.150; r, 9; NPro/Pro = 4,520; NX/Ala = 1,046 0.078 (−0.038 to 0.194); P = 0.186; r, 7; NPro/Pro = 2,734; NX/Ala = 883 — — 0.038 (−0.256 to 0.333); P = 0.798; r, 4; NPro/Pro = 1,251; NX/Ala = 395
Log(HOMA-IR) 0.079 (−0.077 to 0.234); P = 0.322; r, 23; NPro/Pro = 13,667; NX/Ala = 2,919 0.158 (−0.059 to 0.376); P = 0.154; r, 13; NPro/Pro = 8,038; NX/Ala = 2,454 −0.030 (−0.159 to 0.100); P = 0.652; f, 8; NPro/Pro = 3,317; NX/Ala = 247 −0.065 (−0.144 to 0.015); P = 0.111; f, 8; NPro/Pro = 3,746; NX/Ala = 776 0.102 (0.016 to 0.188); P = 0.020; f, 5; NPro/Pro = 2,453; NX/Ala = 674
Pro/Pro-X/Ala (standardized effect size)
GlobalCaucasiansAsiansLean (mean BMI <25 kg/m2)Obese (mean BMI >30 kg/m2)
Log (BMI) −0.037 (−0.087 to 0.012); P = 0.139; r, 50; NPro/Pro = 24,074; NX/Ala= 5,350 −0.062 (−0.111 to −0.012); P = 0.015; r, 37; NPro/Pro = 15,129; NX/Ala = 4,570* −0.087 (−0.174 to −0.001); P = 0.048; f, 10; NPro/Pro = 6,354; NX/Ala = 561
Fasting glucose 0.022 (−0.019 to 0.062); P = 0.296; r, 49; NPro/Pro = 23,615; NX/Ala = 4,930 0.011 (−0.041 to 0.063); P = 0.673; r, 34; NPro/Pro = 13,015; NX/Ala = 3,965 0.069 (−0.009 to 0.147); P = 0.085; f, 11; NPro/Pro = 7,937; NX/Ala = 686 0.015 (−0.040 to 0.070); P = 0.602; f, 14; NPro/Pro = 9,416; NX/Ala = 1,550 0.081 (0.003–0.158); P = 0.041; f, 12; NPro/Pro = 3,099; NX/Ala = 824
2-h glucose 0.018 (−0.079 to 0.115); P = 0.713; r, 12) NPro/Pro =5,121; NX/Ala =1,163 −0.019 (−0.131 to 0.094); P = 0.745; r, 8; NPro/Pro = 2,976; NX/Ala = 959 — — −0.010 (−0.124 to 0.104); P = 0.864; f, 4; NPro/Pro = 1,251; NX/Ala = 395
Log(fasting insulin) 0.031 (−0.024 to 0.085); P = 0.268; r, 51; NPro/Pro = 23,467; NX/Ala = 5,145 0.017 (−0.037 to 0.071); P = 0.536; r, 37; NPro/Pro = 13,981; NX/Ala = 4,278 −0.076 (−0.160 to 0.008); P = 0.078; f, 10; NPro/Pro = 6,823; NX/Ala = 588 −0.055 (−0.114 to 0.003); P = 0.062; f, 12; NPro/Pro = 8,056; NX/Ala = 1,403 0.163 (−0.009 to 0.334); P = 0.064; r, 12; NPro/Pro = 3,075; NX/Ala = 813
Log(2-h insulin) 0.087 (−0.031 to 0.205); P = 0.150; r, 9; NPro/Pro = 4,520; NX/Ala = 1,046 0.078 (−0.038 to 0.194); P = 0.186; r, 7; NPro/Pro = 2,734; NX/Ala = 883 — — 0.038 (−0.256 to 0.333); P = 0.798; r, 4; NPro/Pro = 1,251; NX/Ala = 395
Log(HOMA-IR) 0.079 (−0.077 to 0.234); P = 0.322; r, 23; NPro/Pro = 13,667; NX/Ala = 2,919 0.158 (−0.059 to 0.376); P = 0.154; r, 13; NPro/Pro = 8,038; NX/Ala = 2,454 −0.030 (−0.159 to 0.100); P = 0.652; f, 8; NPro/Pro = 3,317; NX/Ala = 247 −0.065 (−0.144 to 0.015); P = 0.111; f, 8; NPro/Pro = 3,746; NX/Ala = 776 0.102 (0.016 to 0.188); P = 0.020; f, 5; NPro/Pro = 2,453; NX/Ala = 674

Data are standardized effect sizes (95% asymptotic CI); P values; choice of a random-effects model (r) in case of significant inhomogeneity or a fixed-effects model (f) in case of no significant inhomogeneity; and the number (N) of studies included into the corresponding analysis as well as number of individuals per genotype subgroup are provided.

*

Result remains significant after jackknifing;

significant result.

1
Yen CJ, Beamer BA, Negri C, Silver K, Brown KA, Yarnall DP, Burns DK, Roth J, Shuldiner AR: Molecular scanning of the human peroxisome proliferator activated receptor gamma (hPPAR gamma) gene in diabetic Caucasians: identification of a Pro12Ala PPAR gamma 2 missense mutation.
Biochem Biophys Res Commun
241
:
270
–274,
1997
2
Valve R, Sivenius K, Miettinen R, Pihlajamaki J, Rissanen A, Deeb SS, Auwerx J, Uusitupa M, Laakso M: Two polymorphisms in the peroxisome proliferator-activated receptor-gamma gene are associated with severe overweight among obese women.
J Clin Endocrinol Metab
84
:
3708
–3712,
1999
3
Mori H, Ikegami H, Kawaguchi Y, Seino S, Yokoi N, Takeda J, Inoue I, Seino Y, Yasuda K, Hanafusa T, Yamagata K, Awata T, Kadowaki T, Hara K, Yamada N, Gotoda T, Iwasaki N, Iwamoto Y, Sanke T, Nanjo K, Oka Y, Matsutani A, Maeda E, Kasuga M: The Pro12→Ala substitution in PPAR-γ is associated with resistance to development of diabetes in the general population: possible involvement in impairment of insulin secretion in individuals with type 2 diabetes.
Diabetes
50
:
891
–894,
2001
4
Frederiksen L, Brodbaek K, Fenger M, Jorgensen T, Borch-Johnsen K, Madsbad S, Urhammer SA: Comment: studies of the Pro12Ala polymorphism of the PPAR-gamma gene in the Danish MONICA cohort: homozygosity of the Ala allele confers a decreased risk of the insulin resistance syndrome.
J Clin Endocrinol Metab
87
:
3989
–3992,
2002
5
Deeb SS, Fajas L, Nemoto M, Pihlajamaki J, Mykkanen L, Kuusisto J, Laakso M, Fujimoto W, Auwerx J: A Pro12Ala substitution in PPARgamma2 associated with decreased receptor activity, lower body mass index and improved insulin sensitivity.
Nat Genet
20
:
284
–287,
1998
6
Altshuler D, Hirschhorn JN, Klannemark M, Lindgren CM, Vohl MC, Nemesh J, Lane CR, Schaffner SF, Bolk S, Brewer C, Tuomi T, Gaudet D, Hudson TJ, Daly M, Groop L, Lander ES: The common PPARgamma Pro12Ala polymorphism is associated with decreased risk of type 2 diabetes.
Nat Genet
26
:
76
–80,
2000
7
Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN: Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease.
Nat Genet
33
:
177
–182,
2003
8
Chiasson JL, Rabasa-Lhoret R: Prevention of type 2 diabetes: insulin resistance and β-cell function.
Diabetes
53(Suppl. 3)
:
S34
–S38,
2004
9
Kahn SE: The relative contributions of insulin resistance and β-cell dysfunction to the pathophysiology of type 2 diabetes.
Diabetologia
46
:
3
–19,
2003
10
Hartling SG, Roder ME, Dinesen B, Binder C: Proinsulin, C-peptide, and insulin in normal subjects during an 8-h hyperglycemic clamp.
Eur J Endocrinol
134
:
197
–200,
1996
11
Kahn SE, Bergman RN, Schwartz MW, Taborsky GJ Jr, Porte D Jr: Short-term hyperglycemia and hyperinsulinemia improve insulin action but do not alter glucose action in normal humans.
Am J Physiol
262
:
E518
–E523,
1992
12
Fu M, Chen H, Li X, Li J, Wu B, Cheng L, Cai M, Fu Z: [Association of Pro12Ala variant in peroxisome proliferator-activated receptor-gamma2 gene with type 2 diabetes mellitus].
Zhonghua Yi Xue Yi Chuan Xue Za Zhi
19
:
234
–238,
2002
[Article in Chinese]
13
Kobalava ZD, Nosikov VV, Tolkacheva VV, Chernysheva GV, Volkova EG, Shustov SB, Ignat’ev IV, Moiseev VS: [Clinical genetic determinants of carbohydrate metabolism disturbances in patients with hypertension and excessive weight.].
Kardiologiia
45
:
37
–43,
2005
[Article in Russian]
14
Tschritter O, Fritsche A, Stefan N, Haap M, Thamer C, Bachmann O, Dahl D, Maerker E, Teigeler A, Machicao F, Haring H, Stumvoll M: Increased insulin clearance in peroxisome proliferator-activated receptor gamma2 Pro12Ala.
Metabolism
52
:
778
–783,
2003
15
Eriksson J, Lindi V, Uusitupa M, Forsen T, Laakso M, Osmond C, Barker D: The effects of the Pro12Ala polymorphism of the PPARgamma-2 gene on lipid metabolism interact with body size at birth.
Clin Genet
64
:
366
–370,
2003
16
Whitehead A, Whitehead J: A general parametric approach to the meta-analysis of randomized clinical trials.
Stat Med
10
:
1665
–1677,
1991
17
Hedges LV: Statistical Methods for Meta-Analysis. Orlando, FL, Academic Press,
1985
18
Galbraith RF: A note on graphical presentation of estimated odds ratios from several clinical trials.
Stat Med
7
:
889
–894,
1988
19
Tai ES, Corella D, Deurenberg-Yap M, Adiconis X, Chew SK, Tan CE, Ordovas JM: Differential effects of the C1431T and Pro12Ala PPARgamma gene variants on plasma lipids and diabetes risk in an Asian population.
J Lipid Res
45
:
674
–685,
2004
20
Hara K, Okada T, Tobe K, Yasuda K, Mori Y, Kadowaki H, Hagura R, Akanuma Y, Kimura S, Ito C, Kadowaki T: The Pro12Ala polymorphism in PPAR gamma2 may confer resistance to type 2 diabetes.
Biochem Biophys Res Commun
271
:
212
–216,
2000
21
Kawasaki I, Tahara H, Emoto M, Shoji T, Shioji A, Okuno Y, Inaba M, Nishizawa Y: Impact of Prol2Ala variant in the peroxisome proliferator-activated receptor (PPAR) gamma2 on obesity and insulin resistance in Japanese type 2 diabetic and healthy subjects.
Osaka City Med J
48
:
23
–28,
2002
22
Wang C, Zhai F, Chi Y, Wang G: [Association of Pro12Ala mutation in peroxisome proliferator-activated receptor gamma 2 with obesity and diabetes in Chinese population].
Wei Sheng Yan Jiu
33
:
317
–320,
2004
[Article in Chinese]
23
Lindi VI, Uusitupa MI, Lindstrom J, Louheranta A, Eriksson JG, Valle TT, Hamalainen H, Ilanne-Parikka P, Keinanen-Kiukaanniemi S, Laakso M, Tuomilehto J: Association of the Pro12Ala polymorphism in the PPAR-γ2 gene with 3-year incidence of type 2 diabetes and body weight change in the Finnish Diabetes Prevention Study.
Diabetes
51
:
2581
–2586,
2002
24
Rosmond R, Chagnon M, Bouchard C: The Pro12Ala PPARgamma2 gene missense mutation is associated with obesity and insulin resistance in Swedish middle-aged men.
Diabetes Metab Res Rev
19
:
159
–163,
2003
25
Ochoa MC, Marti A, Azcona C, Chueca M, Oyarzabal M, Pelach R, Patino A, Moreno-Aliaga MJ, Martinez-Gonzalez MA, Martinez JA: Gene-gene interaction between PPAR gamma 2 and ADR beta 3 increases obesity risk in children and adolescents.
Int J Obes Relat Metab Disord
28(Suppl. 3)
:
S37
–S41,
2004
26
Ek J, Andersen G, Urhammer SA, Hansen L, Carstensen B, Borch-Johnsen K, Drivsholm T, Berglund L, Hansen T, Lithell H, Pedersen O: Studies of the Pro12Ala polymorphism of the peroxisome proliferator-activated receptor-gamma2 (PPAR-gamma2) gene in relation to insulin sensitivity among glucose tolerant caucasians.
Diabetologia
44
:
1170
–1176,
2001
27
Ostergard T, Ek J, Hamid Y, Saltin B, Pedersen OB, Hansen T, Schmitz O: Influence of the PPAR-gamma2 Pro12Ala and ACE I/D polymorphisms on insulin sensitivity and training effects in healthy offspring of type 2 diabetic subjects.
Horm Metab Res
37
:
99
–105,
2005
28
Muller YL, Bogardus C, Beamer BA, Shuldiner AR, Baier LJ: A functional variant in the peroxisome proliferator–activated receptor γ2 promoter is associated with predictors of obesity and type 2 diabetes in Pima Indians.
Diabetes
52
:
1864
–1871,
2003
29
Stumvoll M, Stefan N, Fritsche A, Madaus A, Tschritter O, Koch M, Machicao F, Haring H: Interaction effect between common polymorphisms in PPARgamma2 (Pro12Ala) and insulin receptor substrate 1 (Gly972Arg) on insulin sensitivity.
J Mol Med
80
:
33
–38,
2002
30
Andrulionyte L, Zacharova J, Chiasson JL, Laakso M: Common polymorphisms of the PPAR-gamma2 (Pro12Ala) and PGC-1alpha (Gly482Ser) genes are associated with the conversion from impaired glucose tolerance to type 2 diabetes in the STOP-NIDDM trial.
Diabetologia
47
:
2176
–2184,
2004
31
Baratta R, Di Paola R, Spampinato D, Fini G, Marucci A, Coco A, Vigneri R, Frittitta L, Trischitta V: Evidence for genetic epistasis in human insulin resistance: the combined effect of PC-1 (K121Q) and PPARgamma2 (P12A) polymorphisms.
J Mol Med
81
:
718
–723,
2003
32
Beamer BA, Yen CJ, Andersen RE, Muller D, Elahi D, Cheskin LJ, Andres R, Roth J, Shuldiner AR: Association of the Pro12Ala variant in the peroxisome proliferator–activated receptor-γ2 gene with obesity in two Caucasian populations.
Diabetes
47
:
1806
–1808,
1998
33
Buzzetti R, Petrone A, Ribaudo MC, Alemanno I, Zavarella S, Mein CA, Maiani F, Tiberti C, Baroni MG, Vecci E, Arca M, Leonetti F, Di Mario U: The common PPAR-gamma2 Pro12Ala variant is associated with greater insulin sensitivity.
Eur J Hum Genet
12
:
1050
–1054,
2004
34
Clement K, Hercberg S, Passinge B, Galan P, Varroud-Vial M, Shuldiner AR, Beamer BA, Charpentier G, Guy-Grand B, Froguel P, Vaisse C: The Pro115Gln and Pro12Ala PPAR gamma gene mutations in obesity and type 2 diabetes.
Int J Obes Relat Metab Disord
24
:
391
–393,
2000
35
Douglas JA, Erdos MR, Watanabe RM, Braun A, Johnston CL, Oeth P, Mohlke KL, Valle TT, Ehnholm C, Buchanan TA, Bergman RN, Collins FS, Boehnke M, Tuomilehto J: The peroxisome proliferator–activated receptor-γ2 Pro12A1a variant: association with type 2 diabetes and trait differences.
Diabetes
50
:
886
–890,
2001
36
Eurlings PM, van der Kallen CJ, Vermeulen VM, de Bruin TW: Variants in the PPARgamma gene affect fatty acid and glycerol metabolism in familial combined hyperlipidemia.
Mol Genet Metab
80
:
296
–301,
2003
37
Evans D, de Heer J, Hagemann C, Wendt D, Wolf A, Beisiegel U, Mann WA: Association between the P12A and c1431t polymorphisms in the peroxisome proliferator activated receptor gamma (PPAR gamma) gene and type 2 diabetes.
Exp Clin Endocrinol Diabetes
109
:
151
–154,
2001
38
Fornage M, Jacobs DR, Steffes MW, Gross MD, Bray MS, Schreiner PJ: Inverse effects of the PPAR(gamma)2 Pro12Ala polymorphism on measures of adiposity over 15 years in African Americans and whites: the CARDIA study.
Metabolism
54
:
910
–917,
2005
39
Franks PW, Luan J, Browne PO, Harding AH, O’Rahilly S, Chatterjee VK, Wareham NJ: Does peroxisome proliferator-activated receptor gamma genotype (Pro12ala) modify the association of physical activity and dietary fat with fasting insulin level?
Metabolism
53
:
11
–16,
2004
40
Ghoussaini M, Meyre D, Lobbens S, Charpentier G, Clement K, Charles MA, Tauber M, Weill J, Froguel P: Implication of the Pro12Ala polymorphism of the PPAR-gamma 2 gene in type 2 diabetes and obesity in the French population.
BMC Med Genet
6
:
11
,
2005
41
Hegele RA, Cao H, Harris SB, Zinman B, Hanley AJ, Anderson CM: Peroxisome proliferator-activated receptor-gamma2 P12A and type 2 diabetes in Canadian Oji-Cree.
J Clin Endocrinol Metab
85
:
2014
–2019,
2000
42
Kahara T, Takamura T, Hayakawa T, Nagai Y, Yamaguchi H, Katsuki T, Katsuki K, Katsuki M, Kobayashi K: PPARgamma gene polymorphism is associated with exercise-mediated changes of insulin resistance in healthy men.
Metabolism
52
:
209
–212,
2003
43
Kolehmainen M, Uusitupa MI, Alhava E, Laakso M, Vidal H: Effect of the Pro12Ala polymorphism in the peroxisome proliferator-activated receptor (PPAR) gamma2 gene on the expression of PPARgamma target genes in adipose tissue of massively obese subjects.
J Clin Endocrinol Metab
88
:
1717
–1722,
2003
44
Li S, Chen W, Srinivasan SR, Boerwinkle E, Berenson GS: The peroxisome proliferator–activated receptor-γ2 gene polymorphism (Pro12Ala) beneficially influences insulin resistance and its tracking from childhood to adulthood: the Bogalusa Heart Study.
Diabetes
52
:
1265
–1269,
2003
45
Lindi V, Schwab U, Louheranta A, Laakso M, Vessby B, Hermansen K, Storlien L, Riccardi G, A Rivellese A, the KANWU Study Group: Impact of the Pro12Ala polymorphism of the PPAR-gamma2 gene on serum triacylglycerol response to n-3 fatty acid supplementation.
Mol Genet Metab
79
:
52
–60,
2003
46
Luan J, Browne PO, Harding AH, Halsall DJ, O’Rahilly S, Chatterjee VK, Wareham NJ: Evidence for gene-nutrient interaction at the PPAR γ locus.
Diabetes
50
:
686
–689,
2001
47
Mancini FP, Vaccaro O, Sabatino L, Tufano A, Rivellese AA, Riccardi G, Colantuoni V: Pro12Ala substitution in the peroxisome proliferator–activated receptor-γ2 is not associated with type 2 diabetes.
Diabetes
48
:
1466
–1468,
1999
48
Meirhaeghe A, Tanck MW, Fajas L, Janot C, Helbecque N, Cottel D, Auwerx J, Amouyel P, Dallongeville J: Study of a new PPARgamma2 promoter polymorphism and haplotype analysis in a French population.
Mol Genet Metab
85
:
140
–148,
2005
49
Mousavinasab F, Tahtinen T, Jokelainen J, Koskela P, Vanhala M, Oikarinen J, Keinanen-Kiukaanniemi S, Laakso M: Common polymorphisms in the PPARgamma2 and IRS-1 genes and their interaction influence serum adiponectin concentration in young Finnish men.
Mol Genet Metab
84
:
344
–348,
2005
50
Nicklas BJ, van Rossum EF, Berman DM, Ryan AS, Dennis KE, Shuldiner AR: Genetic variation in the peroxisome proliferator–activated receptor-γ2 gene (Pro12Ala) affects metabolic responses to weight loss and subsequent weight regain.
Diabetes
50
:
2172
–2176,
2001
51
Oh EY, Min KM, Chung JH, Min YK, Lee MS, Kim KW, Lee MK: Significance of Pro12Ala mutation in peroxisome proliferator-activated receptor-gamma2 in Korean diabetic and obese subjects.
J Clin Endocrinol Metab
85
:
1801
–1804,
2000
52
Pisabarro RE, Sanguinetti C, Stoll M, Prendez D: High incidence of type 2 diabetes in peroxisome proliferator–activated receptor γ2 Pro12Ala carriers exposed to a high chronic intake of trans fatty acids and saturated fatty acids.
Diabetes Care
27
:
2251
–2252,
2004
53
Poirier O, Nicaud V, Cambien F, Tiret L: The Pro12Ala polymorphism in the peroxisome proliferator-activated receptor gamma2 gene is not associated with postprandial responses to glucose or fat tolerance tests in young healthy subjects: the European Atherosclerosis Research Study II.
J Mol Med
78
:
346
–351,
2000
54
Poulsen P, Andersen G, Fenger M, Hansen T, Echwald SM, Volund A, Beck-Nielsen H, Pedersen O, Vaag A: Impact of two common polymorphisms in the PPAR γ gene on glucose tolerance and plasma insulin profiles in monozygotic and dizygotic twins: thrifty genotype, thrifty phenotype, or both?
Diabetes
52
:
194
–198,
2003
55
Robitaille J, Despres JP, Perusse L, Vohl MC: The PPAR-gamma P12A polymorphism modulates the relationship between dietary fat intake and components of the metabolic syndrome: results from the Quebec Family Study.
Clin Genet
63
:
109
–116,
2003
56
Gonzalez Sanchez JL, Serrano Rios M, Fernandez Perez C, Laakso M, Martinez Larrad MT: Effect of the Pro12Ala polymorphism of the peroxisome proliferator-activated receptor gamma-2 gene on adiposity, insulin sensitivity and lipid profile in the Spanish population.
Eur J Endocrinol
147
:
495
–501,
2002
57
Takata N, Awata T, Inukai K, Watanabe M, Ohkubo T, Kurihara S, Inaba M, Katayama S: Pro12Ala substitution in peroxisome proliferator-activated receptor gamma 2 is associated with low adiponectin concentrations in young Japanese men.
Metabolism
53
:
1548
–1551,
2004
58
Thamer C, Machicao F, Fritsche A, Stumvoll M, Haring H: No influence of the PPARgamma2 Pro12Ala genotype on serum adiponectin concentrations in healthy Europeans (Letter).
Metabolism
52
:
798
,
2003
59
Vaccaro O, Mancini FP, Ruffa G, Sabatino L, Colantuoni V, Riccardi G: Pro12Ala mutation in the peroxisome proliferator-activated receptor gamma2 (PPARgamma2) and severe obesity: a case-control study.
Int J Obes Relat Metab Disord
24
:
1195
–1199,
2000
60
Weiss EP, Kulaputana O, Ghiu IA, Brandauer J, Wohn CR, Phares DA, Shuldiner AR, Hagberg JM: Endurance training-induced changes in the insulin response to oral glucose are associated with the peroxisome proliferator-activated receptor-gamma2 Pro12Ala genotype in men but not in women.
Metabolism
54
:
97
–102,
2005
61
Yamamoto Y, Hirose H, Miyashita K, Nishikai K, Saito I, Taniyama M, Tomita M, Saruta T: PPAR(gamma)2 gene Pro12Ala polymorphism may influence serum level of an adipocyte-derived protein, adiponectin, in the Japanese population.
Metabolism
51
:
1407
–1409,
2002
62
Yamamoto J, Kageyama S, Nemoto M, Sasaki T, Sakurai T, Ishibashi K, Mimura A, Yokota K, Tajima N: PPARgamma2 pro12Ala polymorphism and insulin resistance in Japanese hypertensive patients.
Hypertens Res
25
:
25
–29,
2002
63
Yang WS, Hsiung CA, Ho LT, Chen YT, He CT, Curb JD, Grove J, Quertermous T, Chen YD, Kuo SS, Chuang LM: Genetic epistasis of adiponectin and PPARgamma2 genotypes in modulation of insulin sensitivity: a family-based association study.
Diabetologia
46
:
977
–983,
2003
64
Masud S, Ye S, the SAS group: Effect of the peroxisome proloferator activated receptor-γ gene Pro12Ala variant on body mass index: a meta-analysis.
J Med Genet
40
:
773
–780,
2003
65
Egger M, Smith G, Schneider M, Minder C: Bias in meta-analysis detected by a simple, graphical test.
BMJ
315
:
629
–634,
1997

A.T. and M.S. contributed equally to this work.

A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C Section 1734 solely to indicate this fact.