Table 2. Mediation (Indirect) effect of Metabolic Biomarker Levels on the relationship between IGFs/IR–relevant SNPs and CRC risk, stratified by exogenous estrogen use status.
| SNP_¥ | Effect allele/Other allele | Nonuser Group | User Group | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total effect£ | Direct effect£ | Indirect effect* | Total effect£ | Direct effect£ | Indirect effect* | ||||||||||
| CRC risk in relation to SNP | CRC risk in relation to SNP through pathways other than biomarker | CRC risk in relation to SNP through biomarker | CRC risk in relation to SNP | CRC risk in relation to SNP through pathways other than biomarker | CRC risk in relation to SNP through biomarker | ||||||||||
| HR | 95% CI | HR | 95% CI | %¶ | HR | 95% CI | HR | 95% CI | %¶ | ||||||
| Nonusers (n = 269) | E only users (n = 199) | ||||||||||||||
| Insulin level | Insulin level | ||||||||||||||
| INSRS689_R | T/A | 2.24 | 0.74 | 6.76 | 3.34 | 1.07 | 10.47 | 89.20 | 0.87 | 0.27 | 2.80 | 0.73 | 0.18 | 2.99 | 15.81 |
| Nonusers (n = 273) | E+P users (n = 178) | ||||||||||||||
| Total IGF-I level | Total IGF-I level | ||||||||||||||
| IGF-IRS10778176_R | T/C | 2.61 | 1.02 | 6.69 | 2.70 | 1.03 | 7.05 | 5.31 | 2.35 | 0.76 | 7.25 | 2.78 | 0.89 | 8.72 | 32.17 |
| IGFBP3 level | IGFBP3 level | ||||||||||||||
| IGFBP3RS2471551_AD | C/G | 0.53 | 0.32 | 0.86 | 0.54 | 0.33 | 0.89 | 2.93 | 1.21 | 0.68 | 2.16 | 1.35 | 0.74 | 2.48 | 64.96 |
| IGFBP3RS2471551_D | C/G | 0.39 | 0.22 | 0.69 | 0.40 | 0.23 | 0.71 | 2.80 | 1.51 | 0.69 | 3.31 | 1.75 | 0.76 | 4.05 | 49.05 |
| IGFBP3RS3110697_D | A/G | 0.50 | 0.28 | 0.88 | 0.53 | 0.30 | 0.97 | 7.70 | 1.62 | 0.69 | 3.78 | 1.87 | 0.74 | 4.73 | 40.23 |
| Insulin level | Insulin level | ||||||||||||||
| INSRS3842763_D | A/C | 0.58 | 0.34 | 0.99 | 0.52 | 0.30 | 0.90 | 9.67 | 1.35 | 0.63 | 2.90 | 1.22 | 0.55 | 2.69 | 37.36 |
| INSRS689_R | T/A | 2.24 | 0.74 | 6.76 | 3.34 | 1.07 | 10.47 | 89.20 | 1.42 | 0.45 | 4.43 | 1.54 | 0.48 | 5.00 | 29.32 |
| HOMA-IR level | HOMA-IR level | ||||||||||||||
| INSRS689_R | T/A | 2.24 | 0.74 | 6.76 | 3.55 | 1.13 | 11.13 | N/A | 1.41 | 0.45 | 4.38 | 1.21 | 0.34 | 4.26 | 48.68 |
CI, confidence interval; CRC, colorectal cancer; E, exogenous estrogen; HOMA-IR, homeostatic model assessment–insulin resistance; HR, hazard ratio; IGF-I, insulin-like growth factor-I; IGFBP3, IGF binding protein 3; IR, insulin resistance; P, progestin; SNP, single-nucleotide polymorphism.
Note: Among SNPs having statistically significant association with cancer in either subgroup, the SNPs with ≥ 30% of indirect effect in this subgroup or its counterpart are included. Numbers in bold face are statistically significant.
¥ Tag attached to the SNP name indicates the SNP-analysis approach (AD: additive; D: minor-allele dominant; and R: minor-allele recessive).
£ Multivariate regression was adjusted by covariates (age, education, family income, family histories of diabetes mellitus or colorectal cancer, heart failure ever, high cholesterol requiring pills ever, smoking status, alcohol intake, and reproductive history [oral contraceptive and history of hysterectomy or oophorectomy, ages at menarche and menopause, and history of pregnancy]); effect-modifier variables (physical activity, BMI, and exogenous estrogen use), when not evaluated as effect modifier variables, were adjusted as a covariate; when stratified via waist circumference or waist-to-hip ratio, body mass index was not adjusted.
* Indirect effect estimated via the proportional difference between the HRs without (total effect) and with (direct effect) accounting for hormone.
¶ Not applicable due to either ≥ 50% difference between small effect sizes or ≥ 100% difference between two effect sizes.