ROC曲线评价18F—FDGPET／CT、99Tcm—MDP骨显像及二者联合对骨转移的检出效能

目的采用ROC曲线比较18F-FDGPET／CT、99TcmMDP骨显像及二者联合对骨转移患者的检出效能。方法296例恶性肿瘤患者在2个月内同时接受了18F-FDGPET／CT和99TcmMDP骨显像，对2种显像结果按5分法（0分：骨转移阴性，1分：可能阴性，2分：不能确定，3分：可能阳性，4分：肯定阳性）分别评分，两者之和为联合评分值。以病理诊断或临床随访为确诊“金标准”，采用。检验比较ROC曲线下面积，以评价骨显像、PET／CT及联合评分法对骨转移患者的检出效能，采用r检验比较不同方法在各自最佳诊断阈值下的灵敏度、特异性、准确性、阳性预测值、阴性预测值。结果296例患者中，确诊骨转移阳性61例（占20．6％）、阴性235例（占79．4％）。骨显像、PET／CT及联合评分诊断骨转移的ROC曲线下面积（95％可信区间）分别为0．919（0．867—0．971）、0．949（0．906～0．991）、0．994（0．988～0．999），联合评分法的曲线下面积明显大于骨显像（z=2．866，P=0．004）和PET／CT（z=2．027，P=0．043）各自单独评分法，骨显像和PET／CT法曲线下面积差异没有统计学意义（z=0．881，P=0．378）。最佳阈值点下，骨显像和PET／CT单独检出骨转移患者的灵敏度、特异性、准确性、阳性预测值、阴性预测值分别为90．2％（55／61）、85．1％（200／235）、86．1％（255／296）、61．1％（55／90）、97．1％（200／206）和88．5％（54／61）、97．0％（228／235）、95．3％（282／296）、88．5％（54／61）、97．0％（228／235），而联合评分检出的结果分别为98．4％（60／61）、95．7％（225／235）、96．3％（285／296）、85．7％（60／70）、99．6％（225／226）。PET／CT对骨转移患者检出的特异性（X2=19．600，P〈0．001）、准确性（X2=13．755，P〈0．001）、阳性预测值（x2=13．608，P〈0．001）均高于骨显像，灵敏度（r=0，P=1．000）差异无统计学意义；与骨显像、PET／CT单独评分比较，联合评分法检出的特异性（X2=19．862，P〈0．001）、准确性（x2=23．361，P〈0．001）和阳性预测值（x2=11．791，P=0．001）均明显高于骨显像，灵敏度明显高于PET／CT（x2=4．167，P=0．031）。结论18F—FDGPET／CT对骨转移患者的检出效能优于99Tcm—MDP骨显像，二者联合明显提高了对骨转移患者的检出率。

The efficacy of 18F-FDG PET/CT, 99Tcm-MDP bone scintigraphy, and combination of the two techniques for detecting malignant bone metastasis

Objective To compare the efficacy of 18F-FDG PET/CT, 99Tcm-MDP bone scintigraphy (BS), and combination of the two techniques (PET/CT + BS) for detecting bone metastasis by ROC curve analysis. Methods All 296 patients with various cancers, who underwent both 99Tcm-MDP BS and 18F-FDG PET/CT within two months, were retrospectively analyzed. These images were interpreted according to 5-point scale (0: definitely negative, 1: probably negative, 2: equivocal, 3: probably positive, 4:definitely positive for bone metastasis), and the scale of PET/CT + BS was the sum of PET/CT and BS. In light of the confirmed diagnosis derived from pathology or follow-up, ROC curve analysis was performed.The area under the ROC curve (AUC) was compared by z-test. Results Of 296 cases, 61 (20.6%) were confirmed as bone metastases and 235 (79.4%) were negative. The AUC were 0. 919 (95% confidence interval (95% CI) :0. 867 - 0. 971) for BS, 0. 949 (95% CI: 0. 906 - 0. 991) for PET/CT, and 0. 994 (95% CI: 0.988-0.999) for PET/CT + BS, rctrospectively. The AUC of PET/CT + BS was statistically significantly larger than that of BS (z=2. 866, P=0.004) or PET/CT (z =2.027, P=0.043), while the AUC of PET/CT was larger than that of BS, but no statistically significance (z = 0. 881, P = 0. 378) was showed. The optimal sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value(NPV) were 90. 2% (55/61), 85. 1% (200/235), 86. 1% (255/296), 61. 1% (55/90), 97. 1%(200/206) for BS, 88.5% (54/61), 97.0% (228/235), 95.3% (282/296), 88.5% (54/61), 97.0% for PET/CT, and 98.4% (60/61), 95.7% (225/235), 96.3% (285/296), 85.7% (60/70) for PET/CT + BS,respectively. The specificity (χ2 = 19.862, P<0. 001), accuracy (χ2 = 23. 361, P<0.001) and PPV (χ2 =11. 791, P =0.001) of PET/CT + BS were significantly higher than those of BS, the sensitivity of PET/CT +BS was significantly higher than that of PET/CT (χ2 =4.167, P=0.031). Compared with BS, PET/CT had a higher specificity (χ2 = 19.600, P<0. 001), accuracy (χ2 = 13. 755, P <0. 001), PPV (χ2 = 13. 608, P <0. 001), but their sensitivity showed no statistically significant difference (χ2 = 0, P = 1. 000). Conclusions The efficacy of 18F-FDG PET/CT for detecting malignant bone metastasis was superior to that of 99Tcm-MDP BS alone. The detection ability can be obviously improved by combination of the two techniques.