Overview and Preliminary Metaanalysis of Available Studies on Differential Diagnosis of Parkinsonism by 18F-FDG PET, Including Multiple Disease Groups (Group of APSs as Target Condition; i.e., Positive Case)
Study | PD | MSA | PSP | CBD | Total | Med | Method | TP | TN | FP | FN | Sens | Spec | LR+ | LR− | Prev | PPV | NPV |
Juh, 2004 | 8 | 9 | 7 | NA | 24 | ND | 1.1 | 16 | 6 | 2 | 0 | 100% | 75% | 4.0 | 0.00 | 66.7% | 89% | 100% |
Eckert, 2005 | 79 | 25 | 20 | 11 | 135 | D | 1.1 | 51 | 76 | 3 | 5 | 91% | 96% | 24.0 | 0.09 | 41.5% | 94% | 94% |
Tang, 2010 | 96 | 41 | 30 | NA | 167 | D | 2.1 | 58 | 94 | 2 | 13 | 82% | 98% | 39.2 | 0.19 | 42.5% | 97% | 88% |
Hellwig, 2012 | 34 | 13 | 23 | 8 | 78 | D | 1.2 | 38 | 31 | 3 | 6 | 86% | 91% | 9.8 | 0.15 | 56.4% | 93% | 84% |
Garraux, 2013 | 42 | 31 | 26 | 21 | 120 | C | 2.2 | 65 | 39 | 3 | 13 | 83% | 93% | 11.7 | 0.18 | 65.0% | 96% | 75% |
Tripathi, 2013 | 84 | 30 | 20 | 2 | 136 | D | 1.1 | 50 | 76 | 8 | 2 | 96% | 90% | 10.1 | 0.04 | 38.2% | 86% | 97% |
Tripathi, 2016 | 81 | 20 | 28 | NA | 129 | D | 2.1 | 30 | 77 | 4 | 18 | 63% | 95% | 12.7 | 0.39 | 37.2% | 88% | 81% |
Brajkovic, 2017 | 29 | 20 | 21 | 2 | 72 | C | 1.1 | 40 | 27 | 2 | 3 | 93% | 93% | 13.5 | 0.07 | 59.7% | 95% | 90% |
Metaanalysis 1* | 234 | 97 | 91 | 23 | 445 | Misc | 1.1&1.2 | 195 | 216 | 18 | 16 | 91.4%‡ (71.6%–97.8%) | 90.6% (70.1%–97.5%) | 9.7 | 0.09 | 47.4% | 89.8% | 92.1% |
Metaanalysis 2† | 219 | 92 | 84 | 21 | 416 | Misc | 2.1&2.2 | 153 | 210 | 9 | 44 | 76.5%‡ (64.6%–85.3%) | 94.7% (90.5%–97.1%) | 14.4 | 0.25 | 47.4% | 92.9% | 81.7% |
↵* Observer-dependent visual reads supported by voxel-based statistical analyses.
↵† Observer-independent automated classifications.
↵‡ P < 0.01.
Med = parkinsonism medication; TP = true-positive for APSs (including MSA, PSP, and CBD); TN = true-negative for APSs; FP = false-positive for APSs; FN = false-negative for APSs; sens = sensitivity; spec = specificity; LR+ = positive likelihood ratio; LR− = negative likelihood ratio; prev = prevalence of APSs in study population; PPV = positive predictive value; NPV = negative predictive value; NA = not applicable; ND = not defined; D = discontinued before PET; C = continued; misc = miscellaneous; 1.1 = observer-dependent visual analysis supported by SPM; 2.1 = observer-independent automated classification based on metabolic covariance patterns; 1.2 = observer-dependent visual analysis supported by NeuroSTAT/3D-SSP; 2.2 = observer-independent automated classifications based on relevance vector machine.
Data in parentheses are 95% confidence intervals.