Colorectal cancer (CRC) is the third most common cancer worldwide. Predictive biomarkers are needed to predict patients' outcomes and to select a chemotherapy regimen. We assessed whether dihydropyrimidine dehydrogenase (DPD)-related indicators can predict CRC patients' outcomes. We searched the studies in PubMed, EmBase, and the Cochrane Library up to March 4, 2018. We mainly analyzed different CRC patients' outcomes according to specific DPD-related indicators. Twenty-five articles were included in the meta-analysis. The results showed that for disease-free survival (DFS), low DPD expression was significantly superior to high expression (I2 = 72%; HR: 1.59; 95%CI: 1.21-2.09; p = 0.001). However, this result had a potential publication bias (Begg's test: p = 0.007; Egger's test: p = 0.004). Among patients treated with chemotherapy, a high thymidylate phosphorylase (TP)/DPD ratio was advantageous for DFS (I2 = 63.7%; HR: 0.65; 95%CI: 0.46-0.92; p = 0.015), and this result did not have a publication bias. For overall survival (OS), low DPD expression was superior to high expression (I2 = 74.4%; HR: 2.11; 95%CI: 1.48-3.00; p < 0.001), although this result had a publication bias (Egger's test: p = 0.003; Begg's test: p = 0.010). There was no difference in OS according to the TP/DPD ratio (I2 = 0%; HR: 0.92; 95%CI: 0.75-1.13; p = 0.420). DFS and OS were better in CRC patients with low DPD expression than in those with high DPD expression. However, because of publication bias, more DPD indicator-related studies, especially with negative results, are still needed. Patients with a high TP/DPD ratio have better DFS but not OS.
Keywords: Colorectal cancer; Dihydropyrimidine dehydrogenase; Meta-analysis; Thymidylate phosphorylase.