TABLE 3

Diagnostic Accuracy of PET Parameters to Detect Osseous Metastases

CharacteristicCutoffSensitivitySpecificityPPVNPV
Patient-based analysis
SUVmax≥4.498% [0.87, 1.0] n = 5462% [0.32, 0.86] n = 5491% [0.78, 0.97] n = 5490% [0.55, 1.0] n = 54
SUVmax ratio
 Lesion to  blood pool≥2.297% [0.87, 1.0] n = 5369% [0.39, 0.91] n = 5391% [0.78, 0.97] n = 5390% [0.55, 1.0] n = 53
 Lesion to  liver≥1.3378% [0.62, 0.89] n = 5377% [0.46, 0.95] n = 5391% [0.76, 0.98] n = 5353% [0.29, 0.76] n = 53
 Lesion to bone≥7.11100% [0.91, 1.0] n = 5331% [0.09, 0.61] n = 5382% [0.68, 0.91] n = 53100% [0.40, 1.0] n = 53
Biologic volume≥0.6293% [0.80, 0.98] n = 5323% [0.05, 0.54] n = 5379% [0.64, 0.89] n = 5350% [0.12, 0.88] n = 53
Size≥1.850% [0.33, 0.67] n = 4880% [0.44, 0.97] n = 4890% [0.70, 0.99] n = 4830% [0.14, 0.50] n = 48
PSMA RADS≥493% [0.80, 0.98] n = 5369% [0.39, 0.91] n = 5390% [0.77, 0.97] n = 5375% [0.43, 0.95] n = 53
Lesion-based analysis
SUVmax≥4.193% [0.86, 0.98] n = 10773% [0.45, 0.92] n = 10796% [0.89, 0.99] n = 10765% [0.38, 0.86] n = 107
SUVmax ratio
 Lesion to  blood pool≥2.1189% [0.80, 0.94] n = 10473% [0.45, 0.92] n = 10495% [0.88, 0.99] n = 10452% [0.30, 0.74] n = 104
 Lesion to liver≥0.5521% [0.13, 0.31] n = 10493% [0.68, 1.0] n = 10495% [0.75, 1.0] n = 10417% [0.09, 0.26] n = 104
 Lesion to bone≥4.489% [0.80, 0.94] n = 10460% [0.32, 0.84] n = 10493% [0.85, 0.97] n = 10447% [0.24, 0.71] n = 104
Biologic volume≥0.5289% [0.81, 0.95] n = 9821% [0.05, 0.51] n = 9887% [0.78, 0.93] n = 9825% [0.05, 0.57] n = 98
Size≥1.839% [0.28, 0.51] n = 8580% [0.44, 0.97] n = 8594% [0.79, 0.99] n = 8515% [0.07, 0.27] n = 85
PSMA RADS≥480% [0.70, 0.88] n = 10473% [0.45, 0.92] n = 10495% [0.87, 0.99] n = 10438% [0.21, 0.58] n = 104

  • PPV = positive predictive value; NPV = negative predictive value.

  • Units for size are in centimeters, where clearly measurable lesion was present on anatomic imaging. SUV could not be accurately quantified in 2 cases because of technical error. Data in brackets are 95% CI.