To investigate the inhibitory effect of different doses of metformin on colorectal cancer (CRC) and its mechanism involving the regulation of bile acid metabolism and the cyclooxygenase-2 (COX-2)/prostaglandin E2 (PGE2) pathway.

A total of 60 BALB/c mice were divided into control, model, low-dose metformin, medium-dose metformin and high-dose metformin groups. A CRC model was induced using azoxymethane (AOM)/dextran sodium sulfate (DSS). Tumor number and diameter were measured with a caliper; fecal bile acid levels were detected by liquid chromatography-mass spectrometry (LC-MS); and protein expression of COX-2 and PGE2 in colon tissue was determined by immunohistochemistry.

Tumor numbers in the control, model, low-, medium- and high-dose groups were 0, 11.58±7.13, 7.83±3.38, 3.33±2.06 and 2.83±1.40, respectively; tumor diameters were 0, (2.91±0.32), (2.43±0.31), (2.11±0.37) and (1.98±0.38) mm, respectively; total bile acid concentrations were (20 989.00±306.69), (43 256.50±187.08), (32 265.26±325.08), (23 035.72±136.33) and (22 351.51±145.36) ng·g^-1, respectively; deoxycholic acid (DCA) concentrations were (3 160.14±113.31), (16 684.58±4.39), (11 544.47±7.86), (5 472.60±25.03) and (5 201.36±79.95) ng·g^-1, respectively; lithocholic acid (LCA) concentrations were (2 140.43±76.23), (14 452.6±47.82), (9 085.87±11.14), (3 145.09±54.20) and (2 955.51±38.02) ng·g^-1, respectively; cholic acid (CA) concentrations were (6 917.71±51.12), (2 051.41±88.47), (3 181.64±23.79), (5 564.01±35.45) and (5 866.10±63.80) ng·g^-1, respectively; chenodeoxycholic acid (CDCA) concentrations were (5 720.33±53.07), (1 135.43±48.38), (1 530.56±37.57), (4 060.14±37.09) and (4 542.83±27.39) ng·g^-1, respectively; taurocholic acid (TCA) concentrations were (497.24±59.29), (1 560.15±26.03), (958.13±28.00), (705.21±3.66) and (669.24±33.00) ng·g^-1, respectively; taurochenodeoxycholic acid (TCDCA) concentrations were (1 027.96±84.28), (2 918.60±122.90), (2 518.23±70.50), (1 411.36±49.15) and (1 389.87±5.80) ng·g^-1, respectively; COX-2 staining scores in the model, low-, medium- and high-dose groups were (5.75±0.97), (4.50±1.78), (2.75±1.66) and (2.67±0.89) points, respectively; PGE2 staining scores were (5.75±0.97), (4.50±1.93), (2.67±1.67) and (2.50±0.90) points, respectively; compared with the control group, the aforementioned indicators in the model group showed statistically significant differences; compared with the model group, the indicators in the low-, medium- and high-dose groups showed statistically significant differences (all P<0.05). The medium- and high-dose groups were significantly more effective than the low-dose group in reducing tumor number and diameter and regulating bile acid metabolism (P<0.05). The inhibition of COX-2/PGE2 expression was negatively correlated with the dose of metformin.

Metformin exerts anti-CRC effects through dose-dependent regulation of bile acid metabolism (reducing DCA, LCA, TCA, TCDCA and restoring CA and CDCA levels) and inhibition of the COX-2/PGE2 signaling pathway. The medium- and high-dose groups (250-500 mg·kg^-1·d^-1) show more significant effects.