Table 3. Measures of insulin sensitivity and beta-cell function by genotype for three SNPs of FFAR1.

	rs2301151					rs1573611					rs16970264
				Difference	P				Difference	P				Difference	P
	AA	AG	GG			TT	TC	CC			AA	AG	GG
Fasting glucose (mmol/L)
mean	5.46	5.47	5.58	0.01	0.91	5.78	5.45	5.47	−0.07	0.18	3.57	5.52	5.47	−0.02	0.78
SE	0.04	0.05	0.52			0.13	0.06	0.04			0.20	0.08	0.03
n	233	155	16			20	137	247			3	47	349
Fasting insulin (pmol/L)
Geometric mean	56.8	60.1	60.5	1.03	0.44	55.6	56.6	59.3	1.00	0.91	53.0	59.9	57.4	0.90	0.11
95% CI	[53.3, 64.4]	[56.1, 64.4]	[48.6, 75.4]			[46.9, 65.9]	[52.2, 61.4]	[55.9, 62.9]			[3.22, 873]	[57.8, 74.1]	[54.7, 60.2]
n	227	153	13			20	135	238			3	45	340
Si (×10−4 mL.µU−1.min−1)
Geometric mean	2.81	2.69	3.20	1.03	0.49	3.02	2.74	2.80	1.04	0.36	1.68	2.53	2.84	1.15	0.09
95% CI	[2.60, 3.05]	[2.43, 2.97]	[2.74, 3.74]			[2.29, 3.98]	[2.44, 3.07]	[2.61, 3.00]			[0.00, 904]	[2.11, 3.03]	[2.66, 3.02]
n	211	146	14			17	125	230			2	43	322
AIRg (mL.µU−1.min−1)
Geometric mean	443	319	521	1.01	0.73	255	354	335	1.01	0.83	997	320	337	0.96	0.42
95% CI	[403, 483]	[277, 368]	[398, 681]			[155, 418]	[311, 403]	[299, 374]			[0.51, 9185]	[241, 424]	[309, 369]
n	211	146	14			17	125	230			2	43	322
Disposition Index (Arbitrary Units)
Geometric mean	972	864	1667	1.06	0.27	768	984	933	1.04	0.7	1673	808	960	1.08	0.51
95% CI	[864, 1093]	[737, 1012]	[1236, 2249]			[440, 1342]	[843, 1149]	[829, 1050]			[462, 6057]	[598, 1092]	[598, 1093]
n	211	146	14			17	125	230			2	43	322
Sg (×10−3/min)
mean	17.21	17.94	20.67	1.23	0.18	20.32	18.04	17.06	−1.17	0.20	9.75	17.84	17.60	0.37	0.81
SE	0.42	1.15	1.76			2.56	1.24	0.45			3.96	1.63	0.55
n	211	146	14			17	125	230			2	43	322

Data are presented as mean, SEM, n (glucose, Sg) or geometric mean, 95% confidence intervals, n (insulin, Si, AIRg, DI) stratified by genotype for each of the three SNPs. For each genotype the risk allele was defined as the BMI-increasing allele and the data are presented as the additive model. The differences in trait by genotype were assessed by linear regression analysis coding the number of risk alleles as 0,1,2. Data for insulin, Si, AIRg and disposition index were logged for regression analysis. The beta-coefficient from the regression was exponentiated which approximates to the percentage difference. The P-value for the regression is presented and no associations reached statistical significance. The recessive and dominant models, defined according to the risk allele, are presented in Supplementary Table S2.