TABLE 4.

Regression coefficients (β) and P-values for the interaction term comparing the abundance of pre-specified faecal bacterial taxa in aspirin to placebo arm after 3 weeks (upper row) and 6 weeks (lower row) of treatment vs baseline, the ASMIC study

Bacterial taxa	Regression coefficient βa	P-value	Association with CRC reported in literature
Bacterial taxa differentially abundant in CRC cases and controls in the meta-analysis18
Akkermansia	0.30b 0.22	0.002 0.009	↑18,50
Blautia	0.15 0.31	0.40 0.10	↑18
Dialister	0.13 0.52	0.65 0.27	↑18
Dorea	−0.16 −1.57c	0.77 0.02	↑18
Faecalibacterium	0.21c 0.16	0.05 0.16	↓18
Parabacteroides	−0.42c −0.43c	0.0003 0.0002	↑18,40
Ruminococcus
1	0.01 −0.63 c	0.96 0.03	↓18,40
2	−0.42 −0.30	0.39 0.56
Streptococcus	0.59 0.41	0.38 0.49	↑18,50,51
Bacterial taxa that discriminated aspirin and non-aspirin users17 (direction was not reported)
Bacteroides	−0.31b,c −0.39	<2.0 e-16 <2.00e-16	↑↓18,40,50,51
Barnesiella	0.23 0.12	0.37 0.61	Unclear17
Prevotella	0.87 bc 0.51	< 2.0e-16 4.66e-10	↑↓18,40
Ruminococcaceae	0.25c 0.33	<6.16e-13 < 2e-16	↓18,40
Bacterial taxa that were differentially abundant in the aspirin vs placebo in our study
Clostridium XlVa	0.44 b	0.0008	↓39
Clostridium XVIII	0.27b	0.70	↓64
Prevotella	0.51b,c	4.66e-10	↑↓18,40
Veillonella	Model did not converge		↑65

^aPositive regression coefficient (β) implies an increase in the abundance of the bacterial taxon, while negative, a decrease in the taxon abundance for 6 weeks (or three weeks) of treatment vs baseline in the aspirin compared to the placebo arm.

^bThe association between the taxon and aspirin in our study is in the direction opposite to the association between that taxon and CRC status as reported in previous studies.

^cThe association between the taxon and aspirin in our study is in the direction consistent with the association between the taxon and CRC status as reported in previous studies.