Background: Proton pump inhibitors (PPIs) reportedly can impair the absorption of oral anticancer agents by potent acid suppression. Concomitant use of a PPI with capecitabine (Cap) was suggested to be associated with poor outcome in gastrointestinal cancers, however, the potential interaction has not been studied yet in a prospective randomized clinical trial comparing Cap with 5-fluorouracil (FU). We analyzed the differential impact of PPI use on Cap and FU using dataset from AXEPT trial, a phase III randomized trial that demonstrated non-inferiority of a modified XELIRI (mXELIRI; Cap plus irinotecan) with FOLFIRI (FU, leucovorin and irinotecan), both either with or without bevacizumab in patients (pts) with metastatic colorectal cancer (mCRC). Methods: From the per-protocol set (n = 620), pts with available information on concomitant medications (n = 482) were eligible for this sub-study. PPI use was defined as concomitant exposure of Cap and any PPI for 20% or more of the study period. The treatment-by-PPI-use interaction was examined adjusting to stratification factors including age, sex, country, performance status, number of metastatic sites, previous use of oxaliplatin, and concurrent bevacizumab treatment. Results: 49 (10.1%) pts were PPI users. Clinical characteristics were well balanced between the two groups differing in PPI use. In PPI users, the mXELIRI group tended to have poorer OS (hazard ratio [HR], 1.83; 95% confidence interval [CI], 0.96–3.48; p = 0.0644) compared with the FOLFIRI group. In contrast, within PPI non-users, OS of mXELIRI was better than that of FOLFIRI (HR, 0.76; 95% CI, 0.61- 0.95; p = 0.0162). Similarly, a trend of worse PFS with mXELIRI than with FOLFIRI was observed in PPI users (HR, 1.73; 95% CI, 0.94-3.21; p = 0.0798), but not in PPI non-users (HR, 0.90; 95%CI, 0.73 – 1.10; p = 0.2871). Treatment-by-PPI-use interaction was significant for OS (p = 0.0116) and PFS (p = 0.0415). No significant interactions were found between treatment and PPI use in terms of treatment failure, overall response, disease control, and grade 3–4 toxicities. Conclusions: There was a significant interaction between PPI use and treatment (Cap vs FU) in terms of OS and PFS in AXEPT dataset. This suggests that PPI use could impair the efficacy of Cap, but not that of FU. PPIs should be used with caution in pts with mCRC taking Cap. Clinical trial information: NCT01996306.