OBJECTIVE: The objective of this study was to perform a meta-analysis of the diagnostic test accuracy of Glasgow Prognostic Score (GPS) as a prognostic factor for renal cell carcinoma (RCC).
MATERIALS AND METHODS: Studies were retrieved from PubMed, Cochrane, and Embase databases, and we performed comprehensive searches to identify studies that evaluated the prognostic impact of pretreatment GPS in RCC patients. We assessed sensitivity, specificity, summary receiver operating characteristic curve, and area under the curve (AUC).
RESULTS: Totally, studies were searched under the prespecified criteria, and 8 studies with a total of 1191 patients were included to evaluate the prognostic impact of GPS in RCC finally. They indicated a pooled sensitivity of 0.785 (95% confidence interval [CI]: 0.705-0.848), specificity of 0.782 (95% CI: 0.656-0.871), diagnostic odds ratio of 13.089 (95% CI: 7.168-23.899), and AUC of 0.83 (95% CI: 0.79-0.86). Heterogeneity was significant, and meta-regression revealed that the presence of metastasis might be the potential source of heterogeneity. Subgroup analysis also demonstrated that the presence of metastasis might be the source of heterogeneity.
CONCLUSION: GPS demonstrated a good diagnostic accuracy as a prognostic factor for RCC and especially in the case of nonmetastatic RCC.