Table 1. Associations of breast cancer among postmenopausal women with microbiome metrics in faeces, before and after sorting for immunoglobulin A (IgA) staining.

	P valuea
	Unsorted	IgA-positive	IgA-negative	IgA-positive vs IgA-negative
Richness/alpha diversity metric
Observed species	0.021*	0.007*	0.046*	0.054*
Chao1	0.023*	0.007*	0.052*	0.045*
PD_whole_tree	0.039*	0.018*	0.027*	0.480
Shannon	0.103	0.041*	0.129	0.222
Beta diversity metric
Unweighted UniFrac	0.060	0.023*	0.050*	NE
Unweighted UniFrac_PC1	0.066	0.037*	0.027*	0.660
Unweighted UniFrac_PC2	0.370	0.815	0.828	0.940
Unweighted UniFrac_PC3	0.517	0.370	0.527	0.054*
Weighted UniFrac	0.253	0.197	0.195	NE
Weighted UniFrac_PC1	0.147	0.161	0.444	0.225
Weighted UniFrac_PC2	0.233	0.465	0.309	0.096
Weighted UniFrac_PC3	0.281	0.056	0.023*	0.550
Bray Curtis	0.390	0.078	0.209	NE
Bray Curtis_PC1	0.080	0.079	0.137	0.389
Bray Curtis_PC2	0.847	0.471	0.380	0.745
Bray Curtis_PC3	0.137	0.050*	0.280	0.316

Abbreviations: IgA=immunoglobulin A; PC=principal component; PD=Phylogenetic diversity.

^aP-values by Wald test in logistic regression models, else by microbiome regression-based kernel association test (MiRKAT) for UniFrac and Bray Curtis, each adjusted for age, body mass index, and total urinary oestrogen level; permutation test for difference between IgA-positive vs IgA-negative for effect size (not evaluable (NE) by MiRKAT). *P⩽0.05.