摘要
目的探讨18^F-脱氧葡萄糖（FDG）PET／CT半定量分析方法预测乳腺癌腋窝淋巴结转移的最佳阈值及诊断效能。方法将56例接受腋窝淋巴结清扫术（ALND）的乳腺癌患者分成腋窝淋巴结转移阳性组（简称阳性组）和对照组（阴性组），回顾其术前18^F-FDG PET／CT结果，测量乳腺癌原发灶（T）、腋窝淋巴结（L）及淋巴结镜像区本底（B）的最大标准摄取值（SUV-T、SUV-L、SUV-B），计算L/B比值（L／B=SUV-L／SUV-B）。采用受试者工作特征（ROC）曲线确定SUV-T、SUV-L、L/B预测腋窝淋巴结转移的最佳阈值，采用Kappa法分别将最佳阈值下的判断结果与病理诊断结果进行一致性检验。结果56例患者共56个原发病灶，长径为（30±19）mm，SUV-T为5．5±3．5，其中阳性者38例（68％），阳性组病灶长径[（36±19）mm]及SUV-T（6．3±3．5）明显高于对照组（t值分别为4．060和2．774，P〈0．001和P=0．008）。对628枚淋巴结进行病理活组织检查，检出阳性（转移性）淋巴结182枚（29％）、阴性（非转移性）淋巴结446枚（71％）。PET／CT所示与手术基本匹配的腋窝淋巴结共145枚（Ⅰ区132枚、Ⅱ区11枚、Ⅲ区2枚），其中转移性淋巴结126枚（Ⅰ区103枚、Ⅱ区11枚、Ⅲ区2枚）。SUV-T、SUV-L、L/B比值ROC曲线下面积分别为0．756，0．821，0．765，最佳阈值分别为3．2，1．0，2．4。采用SUV-T≥3．2为阈值判断56例患者腋窝淋巴结转移的灵敏度、特异性、阳性预测值、阴性预测值、准确性分别为84％（32／38）、67％（12／18）、84％（32／38）、67％（12／18）、79％（44／56）；采用SUV-L≥1．0和L/B≥2．4的判断结果完全一致，分别为92％（35／38）、67％（12／18）、85％（35／41）、80％（12／15）、84％（47／56）；而目测法分析结果分别为95％（36／38）、50％（9／18）、80％（36／45）、82％（9／11）、80％（45／56）。SUV-T法和目测法与病理诊断的一致性中等（Kappa值分别为0．509、0．498，P均〈0．001），SUV-L法和L/B比值法与病理诊断的一致性良好（Kappa=0．615，P〈0．001）。结论SUV-T≥3．2、SUV-L≥1．0、L/B≥12．4是18^F-FDG PET／CT半定量分析预测乳腺癌腋窝淋巴结转移的最佳阈值，与目测法一样均可用于腋窝淋巴结评估，其中SUV-L≥1．0法和L/B≥2．4法与病理诊断的一致性更好。
Objective The presence of axillary lymph nodes （ALN） metastasis was the most important prognostic factor in women with breast cancer. The ability of 18^F-fluorodeoxyglucose （FDG） PET determining axillary status was argued. The aim of this study was to explore the optimal threshold and ability of semiquantitative analysis to determine axillary status by 18^F-FDG PET/CT. Methods Fifty-six patients with invasive breast cancer who underwent ALN biopsy （ALNB） were divided into 2 groups： positive group （with axillary metastasis）, control group （no axillary metastasis）. All the patients underwent 18^F-FDG PET/CT scanning for their preoperative evaluation. The maximum standardized uptake values of primary tumor （ SUVT） of ALN （ SUV-L ） and of mirror imaging background of ALN （ SUV-B ） were measured. The ratio of L/B was calculated （ L/B = SUV-L/ SUV-B ）. With the histopathologic standards of reference for each ALN, the optimal threshold of SUV-T, and SUV-L L/B were identified by receiver operating characteristic （ROC） curve analysis. Using the optimal threshold, the sensitivities, specificities, positive predictive values, negative predictive values and accuracies were calculated and compared with pathologic diagnosis （Kappa test was performed）, respectively. Results Of 56 cases, there were 56 breast primary tumors. The average of size was （30±19） mm and SUV-T was 5.5±3.5. There were 38 cases （68%） in positive group and 18 cases （32%） in control group. The values in positive group [ primary tumor size： （36 ± 19） mm, SUV-T ： （6.3 ± 3.5） ] were higher significantly than control group （ t = 4. 060, P 〈 0. 001 for size and t = 2. 774, P = 0. 008 for SUV-T）. Of all 628 ALN （565 in level Ⅰ ,49 in level Ⅱ and 14 in level Ⅲ ） were histopathologic dissected, 182（29% ） ALN were metastasis and 446 （71%） were negative. Only 145 （23%） were found in PET/CT, and of them 116 （ 103, 11 and 2 in level Ⅰ , Ⅱ and Ⅲ, respectively） were metastasis. In level Ⅰ , the size, SUV-L, L/B of metastasis nodes were higher significantly than benign nodes （t = 6. 701, 7. 020, 1. 381, respectively, P 〈0.001, all）. The SUV-B in level Ⅱ and Ⅲ was higher significantly than in level Ⅰ （t = 10. 172, P 〈 0. 001 ）. The area under ROC curve of SUV-T, SUV-L, L/B were 0. 756, 0. 821, 0.765, respectively. The optimal thresholds were 3.2, 1.0 and 2.4, respectively. The sensitivities, specificities, positive predictive values, negative predictive values, accuracies were 84% （32/38）, 67% （12/18）, 84% （32/38）, 67% （ 12/18）, 79% （44/56） for SUV-t≥3.2, and 92% （35/38）, 67% （ 12/18 ）, 85% （35/41）, 80% （ 12/15 ）, and 84% （47/56） for SUV-L ≥ 1.0 and same for L/B ≥2.4, respectively. The values were 95% （36/38） , 50% （ 9/18 ） , 80% （ 36/45 ） , 82% （ 9/11 ） , and 80% （45/ 56） for visual analyzing. Comparing with pathologic diagnosis, the value of Kappa were 0. 509 for SUV-T, 0. 615 for SUV.L and L/B, 0. 498 for visual, respectively, and P 〈 0. 001, all. Conclusions The optimal threshold of SUV-T, SUV-L, and L/B were 3.2, 1.0 and 2.4, respectively. It could be used to predict ALN metastasis in women with breast cancer for SUV-T≥3.2, SUV-L≥ 1.0, L/B ≥2.4 or visual analyzing by 18^F-FDG PET/CT. Comparing with histopathology, there were better agreement by using SUV-L≥ 1.0 or L/ B≥2.4 than using SUV-T ≥3.2 or visual analyzing.
出处
《中华核医学杂志》
CAS
CSCD
北大核心
2009年第2期73-77,共5页
Chinese Journal of Nuclear Medicine
