Characteristics of individual studies
We identified 407 publications from the electronic databases (Figure 1), of which 86 were excluded as duplicates and 273 were excluded based on selection criteria. This resulted in 48 articles, which were independently read by two authors. Eventually, six studies21–26 involving 1,794 patients were included in our meta-analysis. The characteristics of each individual study are presented in Table 1.
(To view a larger version of Figure 1, click here.)
(To view a larger version of Table 1, click here.)
Quality of the included studies
The risk of bias in the included studies was strictly evaluated. Four studies23–26 describe a random component in the sequence generation process and the concealment of treatment allocation, and the four trials23–26 were designed as double-blind trials. In addition, one study22 lost large amounts of data, which may lend to a certain attrition bias. Details of methodological approch are presented in Table 2.
(To view a larger version of Table 2, click here.)
All studies21–26 including 1,794 patients reported HR for OS. When assessing the effect on OS (as shown in Figure 2), COX-2 inhibitors plus chemotherapy did not significantly differ from chemotherapy alone (HR =1.04, 95% CI: 0.91–1.18, I2=0.0%, P=0.808). Further, we conducted two subgroup analyses according to the type of COX-2 inhibitors (celecoxib, rofecoxib, or apricoxib) and treatment line (first or second). Unfortunately, no clinical benefit in OS was found among the groups: celecoxib (HR =1.05, 95% CI: 0.90–1.22, I2=0.0%, P=0.532), rofecoxib (HR =1.00, 95% CI: 0.75–1.34, I2= not applicable [NA], P=NA), apricoxib (HR =1.04, 95% CI: 0.64–1.69, I2=NA, P=NA), first-line treatment (HR =1.01, 95% CI: 0.88–1.16, I2=0.0%, P=0.819), and second-line treatment (HR =1.19, 95% CI: 0.88–1.60, I2=0.0%, P=0.508).
(To view a larger version of Figure 2, click here.)