The efficiency of 18F- FDG PET for glioma grading: a Meta-analysis / 中华核医学杂志

ABSTRACT

Objective To systematically review the efficiency of 18 F-FDG PET in glioma grading by using Meta-analysis. Methods Retrieval in PubMed and China National Knowledge Infrastructure (CNKI)was performed. Relevant papers concerning with glioma diagnoses with 18 F- FDG PET were selected. Paper quality was evaluated according to the standard of diagnostic test recommended by Cochrane Workshop. The data of glioma malignancy degree defined as semi-quantitatively and qualitatively were extracted from the papers. Meta-analysis was conducted with the Meta-Disc software to calculate pooled weighted sensitivity and specificity with 95% confidence interval (CI). Summary receiver operating characteristic (SROC) curve was performed and the areas under the curve (AUC) were calculated. Results Seven hundred and fifty-three patients from 17 papers ( 16 in English, 1 in Chinese) were included. Two hundred and seventy-two patients from 11 papers were using semi-quantitative (tumor to cortex ratio, T/C; tumor to white matter ratio,T/W) method and 481 patients from 9 papers were using qualitative method (visual observation, some of the papers had 2 or more methods). After heterogeneity test was done, different effect models were selected. The pooled weighted sensitivity, specificity and diagnostic odds ratio (DOR) with 95% CI for T/C group was 0. 952 (95% CI 0. 903 -0. 980), 0. 409 (95% CI 0. 318-0. 504) and 11. 746 (95% CI5. 368-25. 702) respectively. The pooled weighted sensitivity, specificity and DOR with 95% CI for T/W group was 0. 857 (95% CI 0. 768-0. 922), 0. 538 (95% CI 0. 431 -0. 642) and 22. 066 (95% CI7. 077-68. 800) respectively. The pooled weighted sensitivity, specificity and diagnostic odds ratio (DOR)with 95% CI for qualitative method was 0.810 (95%CI 0.757-0.855), 0.870 (95%CI 0. 819-0.911 ) and 15.282 (95% CI 3. 716-62. 851 ) respectively. The AUC for T/C group, T/W group and qualitative method was 0.8604, 0. 8373 and 0. 8724 respectively. Conclusions Grading glioma by 18 F-FDG PET with semi-quantitative method may provide high diagnostic sensitivity. If qualitative method is used, the diagnostic specificity may be higher.