Third-line chemotherapy in advanced gastric cancer: A systematic review and meta-analysis

Abstract

Backgound: Little information regarding to the survival advantage of third-line chemotherapy in advanced gastric cancer patients is available. The current study is designed to systematically review and perform meta-analysis on the effect of third-line chemotherapy on progressive or recurrent gastric cancer treatment.

Methods: After thorough searching of online databases, total 20 articles were included into qualitative systematic review and 6 of them were used to conduct qualitative meta-analysis.

Results: It was found that the third-line chemotherapy was superior to placebo or best supportive care in terms of prolonging median oval survival (OS) length and progress free survival (PFS) length (Hedges's g for OS = -0.315 ± 0.077, P < .001; and for PFS = -0.382 ± 0.098, P < .001). In addition, the third-line chemotherapy was favored (Hedges's g = 0.848, P < .001) in terms of overall survival rate (Hazard ratio = 0.679, 95% confidence interval: 0.565-0.816, P < .001) or tumor free survival rate (Hazard ratio = 0.561, 95% confidence interval: 0.444-0.709, P < .001).

Conclusion: The third-line chemotherapy is superior to the best supportive care in advanced gastric cancer patients who had been pretreated with first-line and second-line chemotherapy.

Figure 1

Flow diagram of literature search and eligible publication selection.

Figure 2

Forest plot for median survival length of overall survival. A fixed effect model was used due to non-significant heterogeneity of publications (I² = 45.6%, P = .067). The effect size was assessed by Hedges's g and 95% CI, and the median survival length was in favors of third-line chemotherapy (Hedges's g = −0.315 ± 0.077, P < .001). 3^rd-line T = third-line chemotherapy, BSC = best supportive care, CI =  confidence interval.

Figure 3

Forest plot for median survival length of progress free survival. A random effect model was used due to non-significant heterogeneity of publications (I² = 87.2%, P = .003). The effect size was assessed by Hedges's g and 95% CI, and the median survival length was in favor of third-line chemotherapy (Hedges's g = −0.382 ± 0.093, P < .001). 3^rd-line T = third-line chemotherapy, BSC = best supportive care, CI = confidence interval.

Figure 4

Forest plot for hazard ratio (HR) of overall survival (OS) rate. A fixed effect model was used due to non-significant heterogeneity of publications (I² = 0.12, P = .424). The effect size was assessed by hazard ratio (HR) and 95% CI, and the OS rate was in favors of third-line chemotherapy (HR = 0.679, 95%CI: 0.565–0.816, P < .001). 3^rd-line T = third-line chemotherapy, BSC = best supportive care, CI = confidence interval, HR = hazard ratio, OS = overall survival.

Figure 5

Forest plot for hazard ratio (HR) of progress free survival (PFS) rate. A random effect model was used due to non-significant heterogeneity of publications (I² = 82.9%, P = .04). The effect size was assessed by hazard ratio (HR) and 95% CI, and the OS rate was in favor of third-line chemotherapy (HR = 0.561, 95%CI: 0.444–0.709, P < .001). 3^rd-line T = third-line chemotherapy, BSC = best supportive care, CI = confidence interval, HR = hazard ratio, OS = overall survival, PFS = progress free survival.