Metaregression analyses and diagnostic performance of 18F-FDG-PET in subgroups of brain tumors
Analysis | No. of Studies (pts) | I2 of Metaregression (95% CI, P Value) | Independent Estimates (95% CI) | LR (95% CI) | ||
---|---|---|---|---|---|---|
Sensitivity | Specificity | LR+ | LR− | |||
Design | ||||||
Retrospective | 9 (333) | 0.00 (0–100, .86) | 0.71 (0.58–0.81) | 0.73 (0.57–0.85) | 2.63 (1.68–4.13) | 0.40 (0.29–0.55) |
Prospective | 15 (524) | 0.71 (0.60–0.79) | 0.80 (0.66–0.89) | 3.45 (1.96–6.07) | 0.37 (0.26–0.52) | |
PET or PET/CT | ||||||
PET | 17 (511) | 0.00 (0–100, .38) | 0.74 (0.64–0.81) | 0.74 (0.60–0.84) | 2.82 (1.74–4.55) | 0.36 (0.25–0.51) |
PET/CT | 7 (346) | 0.65 (0.50–0.78) | 0.84 (0.68–0.92) | 3.97 (2.09–7.54) | 0.42 (0.29–0.59) | |
PET measurement | ||||||
Qualitative | 19 (757) | 57.87 (4.96–100, .09) | 0.69 (0.60–0.76) | 0.79 (0.67–0.87) | 3.24 (2.00–5.24) | 0.40 (0.30–0.52) |
Quantitative | 4 (85) | 0.86 (0.71–0.94) | 0.67 (0.43–0.84) | 2.59 (1.40–4.81) | 0.21 (0.09–0.45) | |
Malignant grade | ||||||
Low | 14 (330)a | 0.00 (0–100, .46) | 0.60 (0.35–0.81) | 0.79 (0.64–0.89) | 2.87 (1.52–5.41) | 0.50 (0.27–0.92) |
High | 17 (473)a | 0.74 (0.68–0.80) | 0.78 (0.65–0.87) | 3.35 (2.09–5.38) | 0.33 (0.26–0.42) | |
NS | 7 (192) | 0.70 (0.50–0.84) | 0.70 (0.54–0.82) | 2.33 (1.47–3.67) | 0.44 (0.26–0.74) | |
Disease status | ||||||
SPBT | 3 (127)b | 77.72 (51.54–100, .01) | 0.43 (0.28–0.59) | 0.74 (0.49–0.90) | 1.67 (0.59–4.76) | 0.77 (0.48–1.24) |
SRBT | 20 (643)b | 0.75 (0.67–0.81) | 0.79 (0.66–0.88) | 3.51 (2.17–5.66) | 0.32 (0.25–0.41) | |
Histology | ||||||
Glioma | 17 (609)c,d | 0.00 (0–100, .47) | 0.75 (0.64–0.83) | 0.78 (0.64–0.87) | 3.36 (2.02–5.59) | 0.33 (0.23–0.47) |
Nonglioma or NS | 8 (250)c | 0.64 (0.52–0.75) | 0.75 (0.63–0.84) | 2.58 (1.69–3.94) | 0.48 (0.35–0.65) |
Note:—NS indicates not separable; pts, patients.
↵a Patients with nonneoplastic lesions in 14 studies2,6⇓–8,11,12,15,16,18,19,21,23⇓–25 were used for both low- and high-grade subgroups.
↵b Patients with nonneoplastic lesions in 2 studies8,25 were used for both SPBT and SRBT subgroups.
↵c Five patients with nonneoplastic lesions from Hong et al 201116 were used for both glioma and nonglioma subgroups.
↵d One patient with lymphoma in Pauleit et al 20098 and 2 patients with esthesioblastoma and renal cell metastasis in Kahn et al 199412 were excluded from subgroup analysis.