Table 2.

Mendelian randomization analysis for the effect of IR on colorectal cancer risk.

Subgroup	Fasting glucose			SNP	Fasting insulin			SNP	HOMA-IR			SNP
	HR (95% CI)¶*	P	Phat	n	HR (95% CI)¶*	P	Phat	n	HR (95% CI)¶	P	Phat	n
Overall	0.76 (0.47–1.21)	0.244	N/A	1					0.99 (0.32–3.07)	0.903	0.664	2€
BMI ≥ 30					0.82 (0.43–1.54)	0.531	N/A	1
Active (MET ≥ 10)	0.75 (0.002–232.37)	0.638	0.050	2†
Inactive (MET <10)					0.12 (0.02–0.62)	0.011	N/A	1
% calories from SFA <7.0					1.73 (0.28–10.50)	0.552	N/A	1	1.07 (0.26–4.35)	0.655	0.125	2¥
% calories from SFA ≥ 7.0	0.59 (0.37–0.94)	0.027	N/A	1					0.82 (0.37–1.81)	0.191	0.826	2§
Pooled estimate	0.70 (0.41–1.19)	0.120	0.207	4	0.70 (0.09–5.52)	0.538	0.060	3	1.06 (0.87–1.30)	0.429	0.442	6

BMI, body mass index; CI, confidence interval; HOMA-IR, homeostatic model assessment–insulin resistance; HR, hazard ratio; IR, insulin resistance; MET, metabolic equivalent; SFA, saturated fatty acids; SNP, single–nucleotide polymorphism. Numbers in bold face are statistically significant.

P_hat was estimated on the basis of Cochran's Q.

^¶The Mendelian randomization HR was estimated by adjusting for correlation rho between each phenotype and colorectal cancer risk within the same population.

^*The Mendelian randomization effects of single SNPs on colorectal cancer risk were estimated via the ratio of ß-coefficients (=ß_{colorectalcancer}/ß_IR) (20).

^€Two SNPs were PABPC1P2 rs77772624 and LINC00460 rs17254590.

^†Two SNPs were MKLN1 rs117911989 and NKX2-2 rs7273292.

^¥Two SNPs were MSC rs13277245 and DOCK1 rs113847670.

^§Two SNPs were PABPC1P2 rs77772624 and LINC00460 rs17254590.