Native American ancestry and breast cancer risk in Colombian and Mexican women: ruling out potential confounding through ancestry-informative markers

Table 2 Results from main, sensitivity and stratified instrumental variable analyses on the association between Mexican Native American ancestry and breast cancer risk

	Method¹	# IV	Exp. var.² (%)	OR	95% CI		p-val	Q p-val³	intercept p-val⁴
Main analysis
Two-sample MR	IVW	134	38.5	0.988	0.987	0.990	1.4 × 10^–44	> 0.99
Sensitivity analyses
Two-sample MR	Egger	134	38.5	0.990	0.983	0.996	2.6 × 10^–03	> 0.99	0.66
Two-sample MR	WM	134	38.5	0.988	0.986	0.991	5.0 × 10^–22
expl. Variance ~ 8%⁵	IVW	17	8.4	0.990	0.986	0.993	6.7 × 10^–09	0.94
HGDP as reference⁶	IVW	133	37.9	0.988	0.987	0.990	7.4 × 10^–44	> 0.99
refined Mexican reference⁷	IVW	134	38.6	0.988	0.987	0.990	7.7 × 10^–44	> 0.99
Stratified analyses
BC diagnosed ≤ 45 y⁸	IVW	136	39.3	0.994	0.991	0.997	1.4 × 10^–04	> 0.99
Familial BC⁹	IVW	132	37.4	0.973	0.969	0.978	1.0 × 10^–33	> 0.99
ER-positive BC¹⁰	IVW	135	38.5	0.989	0.986	0.991	6.6 × 10^–18	> 0.99
ER-negative BC	IVW	139	40.2	1.004	1.001	1.008	1.5 × 10^–02	> 0.99
Triple-negative BC¹¹	IVW	142	40.3	1.004	0.999	1.009	0.11	> 0.99

Bold values indicate that the probability value is below 0.05
¹IVW: Inverse variance weighted; Egger: MR-Egger regression; WM: weighted median estimates
²Exp. var.: Variance explained by the instrumental variables
³Cochran’s Q statistic p-values higher than 0.05 are suggestive of no instrument heterogeneity as a proxy for pleiotropy
⁴MR-Egger intercept p-values higher than 0.05 are consistent with no horizontal pleiotropy
⁵Subset of 17 IV explaining a cumulative variance in global proportion of Native American ancestry similar to the one in the Colombian study (8%)
⁶Individuals from the Human Genome Diversity Project were used instead of the 177 Native Americans from Mexico as the reference panel for estimation of global ancestry proportions
⁷The Maya, Mixe, Mixtec, Purepecha and Zapotec populations in the Human Genome Diversity Project were used as the reference panel for estimation of global ancestry proportions
⁸Breast cancer diagnosed before the age of 45 years
⁹Breast cancer patients with first-degree relatives affected by breast and/or ovarian cancer
¹⁰Oestrogen receptor-positive breast cancer
¹¹Oestrogen receptor-negative, progesterone receptor-negative and human epidermal growth factor receptor-negative breast cancer

ISSN: 1465-542X