Subgroup Analyses for Diagnostic Performance of 18F-FDG PET of Bone Marrow Infiltration in Lymphoma
| Analysis | No. of studies (patients) | Independent estimates (95% CI) | Likelihood ratio (95% CI) | ||||
|---|---|---|---|---|---|---|---|
| Sensitivity (%) | Specificity (%) | LR+ | LR− | ||||
| Prospective design | |||||||
| Yes | 7 (361) | 57 (42–70) | 91 (81–96) | 7.26 (3.15–16.7) | 0.56 (0.44–0.72) | ||
| No/not specified | 6 (226) | 39 (20–62) | 92 (83–96) | 4.65 (2.45–8.85) | 0.81 (0.67–0.97) | ||
| Type of lymphoma | |||||||
| HD | 5 (191)* | 76 (47–92) | 92 (79–97) | 9.02 (3.52–23.2) | 0.33 (0.14–0.77) | ||
| NHL | 6 (239) | 43 (28–60) | 88 (75–94) | 3.53 (1.88–6.63) | 0.68 (0.57–0.81) | ||
| Both/not separable | 3 (121) | 52 (24–79) | 97 (91–99) | 13.3 (4.02–44.4) | 0.61 (0.29–1.26) | ||
| Disease status | |||||||
| Primary | 7 (297) | 72 (57–83) | 93 (85–97) | 8.93 (4.31–18.5) | 0.41 (0.25–0.68) | ||
| Recurrent | NA | NA | NA | NA | NA | ||
| Both/not separable | 6 (270) | 38 (27–50) | 90 (79–96) | 3.91 (1.97–7.75) | 0.15 (0.06–0.37) | ||
| Type of BMB | |||||||
| Unilateral | 3 (114) | 75 (53–89) | 87 (74–95) | 5.73 (2.84–11.5) | 0.30 (0.14–0.61) | ||
| Bilateral† | 4 (183) | 46 (32–60) | 87 (70–95) | 4.07 (1.50–11.1) | 0.61 (0.47–0.80) | ||
| Not specified | 6 (290) | 42 (20–67) | 95 (90–98) | 8.39 (3.04–23.1) | 0.07 (0.02–0.26) | ||
| Blinding | |||||||
| Yes | 9 (348) | 54 (41–66) | 89 (80–94) | 5.06 (2.97–8.63) | 0.60 (0.48–0.75) | ||
| No | 4 (239) | 44 (15–79) | 94 (86–98) | 7.23 (2.09–25.1) | 0.80 (0.62–1.05) | ||
| PET measurement | |||||||
| Qualitative | 11 (394) | 54 (41–67) | 89 (82–93) | 4.63 (3.00–7.14) | 0.64 (0.50–0.82) | ||
| Quantitative | 2 (193) | 36 (19–57) | 96 (92–99) | 14.1 (1.54–1.30) | 0.62 (0.24–1.59) | ||
↵* Thirty-two patients from Elstrom et al. (9) and 4 patients from Hong et al. (11) with HD were excluded from this subgroup analysis since all BMBs in these patients with HD were negative.
↵† Moog (8) includes 8 patients with unilateral BMB, but these could not be separated.
NA = not available.
All analyses are based on random-effects calculations.