摘要
卵巢肿瘤由于其内在特性，发生肿瘤时无典型的临床表现，故如何早期正确诊断卵巢肿瘤成为临床医生亟待解决的一个关键问题。目前诊断卵巢癌的方法以彩色多普勒超声应用最为广泛。而单用超声的二维形态学或彩色多普勒血流显像技术均不能正确诊断卵巢肿瘤。利用彩超技术检测其血流阻力，血流速度，血流分布及能量等指标结合肿瘤形态学方面的特征进行综合分析，会大大提高卵巢肿瘤术前早期诊断的正确性。多变量Logistic回归分析法是将二者完美结合的最好的诊断方法。多变量Logistic回归分析法是一个系统的运算程序。这种运算方法通过将灰阶超声和彩色多普勒显像技术检测的变量输入电脑，应用SPSS统计软件包，早期预测患者附件区肿物为恶性的可能性。
目的 通过对患有附件区肿物的患者进行灰阶超声和彩色多普勒超声的扫描，运用多变量Logistic回归分析法来评价患者的肿物为恶性的可能性。
方法 采用回顾性分析方法分析附件肿物患者110例，这些患者均经过经腹部或者经阴道彩色多普勒超声扫描。每个病人记录如下变量：⑴年纪，⑵月经状态，⑶RI（阻力指数），⑷PI（搏动指数），⑸PSV（收缩期最大流速），⑹TAMVX（时间平均最高流速），⑺肿瘤最大直径，⑻肿瘤
体积，⑼肿瘤形态；⑽有无包膜；⑾肿瘤回声；⑿多普勒超声的血流位置；⒀有无腹水及腹水大小。由这些变量建立回归模型，将这13个独立的变量和患者术后的组织病理学结合进行综合分析。
结果 在110例患者中卵巢良性肿瘤49例，炎症包块11例；卵巢恶性肿瘤45例，交界性肿瘤2例，输卵管肉瘤1例，间皮肉瘤1例，伯基特淋巴瘤1例。分别对各个变量进行统计运算，良性肿瘤的平均年龄39岁与恶性肿瘤（52岁）相比差异无显著性（P＝0.417）。闭经后患者良性肿瘤占20％，恶性肿瘤为70％，二者差异有显著性（P﹤0.001）。附件区恶性肿瘤的患者在肿瘤形态方面与良性肿瘤相比，恶性肿瘤形态不规则（P﹤0.001），包膜不完整（P﹤0.001），低回声及强回声较多（P﹤0.001）差异有显著性。而在肿瘤的最大直径（P＝0.867）及体积(P=0.632)方面无统计学意义。腹水的出现在良恶性肿瘤中亦有统计学意义（P﹤0.001）。98％的恶性肿瘤中可检测出中心血流，而在良性肿瘤中有18例（30％）出现中心血流，余为周边血流，二者差异有显著性（P﹤0.001）。阻力指数（RI和PI）在恶性肿瘤中显著低于良性肿瘤。而PSV和TAMVX在良恶性肿瘤中差异无显著性，P值分别为0.829和0.139。对13个变量进行回归分析，有4个变量进入方程，它们分别是RI，肿瘤形态，肿瘤血流位置，腹水。这4个变量能够准确的预测患者的附件肿物是否为恶性。得出恶性概率的公式：1／1＋e-z。z=（-15.034×RI）+(4.423×肿瘤形态评分)+(3.040
×血流评分)+(2.456×腹水)+0.531。通过ROC曲线得出所有连续性变量的最佳诊断点和单一变量与多变量Logistic回归模型的相比较的特征。从ROC曲线图上可见Logistic回归模型的建立比单独应用肿瘤形态，血流位置，腹水，RI能够更好的鉴别诊断附件区的良恶性肿瘤。在本模型中当诊断概率大于25％时，敏感度和特异度分别为98％，91.7% ，阳性预测值为100%，阴性预测值为90%,准确性为95.4%。当诊断概率大于50％时，敏感度和特异度分别为92％，95％，阳性预测值为97.9%，阴性预测值为93.4%，准确性为93.6%。临床应用中当我们选择诊断的概率大于50％时，其诊断的敏感度和特异度最高。在110例附件肿物中，诊断为良性肿瘤的61例患者中，有57例与诊断相符，诊断符合率为96.7%。诊断为恶性肿瘤的49例患者中，有46例与诊断相符，恶性肿瘤的诊断符合率为95％。
结论 多变量Logistic回归法诊断附件区良恶性肿瘤的诊断符合率较高，说明多变量Logistic回归法诊断附件恶性肿瘤的准确性比单独应用灰阶超声或者彩色多普勒超声更高，更具有诊断价值。多变量Logistic回归模型与进入回归方程的单一变量（RI，肿瘤形态评分，肿瘤血流位置，腹水）相比能更准确的预测附件区肿瘤的恶性度，差异有显著性（P<0.0001）。多变量Logistic回归模型的诊断概率大于25％时，对恶性肿瘤的诊断的敏感度和特异度，阳性预测值及准确性最高。多变量联合应用，互相补充来预测附件区恶性肿瘤的概率，才能同时提高其敏感度、特异度及阳性预测
值，并能帮助临床医生对疾病做出早期诊断及治疗，同时可避免不必要的临床干预。
Objective：The aim of the study was to assign a probability of malignancy for any patient with an adnexal mass by the application of multivariate logistic regression analysis to variables recorded at the time of transvaginal and transabdominal ultrasonography.
Methods: One hundred and eleven women with known adnexal masses were examined using transabdominal B-mode and color Doppler imaging. For each patient the variables included:⑴age, ⑵menstruation status,⑶resistance index (RI)，⑷pulsatility index (PI)，⑸highest peak systolic (PSV)，⑹time-averaged maximum velocity (TAMVX)，⑺maximum tumor diameter, ⑻tumor volume, ⑼tumor morphology, ⑽the presence or absence of tumor membrane, ⑾tumor echogenicity, ⑿the location of blood flow, ⒀ascite. These 13 independent variables and the final histological diagnosis for each patient (the dependent variable) were used for the regression analysis.
Result In the entire dataset there were 49 women with benign ovarian tumors, 11 with inflammation mass, two with borderline
and 41 with primary invasive ovarian tumors , 6 with metastatic invasive tumor , 1 with fallopian sarcoma. The mean age of the patients with benign tumors was 39 years compared with 52 years for those malignant or borderline tumors（P＝0.417）,there was no significant difference .Of the postmenopausal women, 20% is the benign tumor and 70% is the malignant tumor. There was significant difference(P﹤0.001). For the entire dataset the irregular morphology of the tumor (P﹤0.001) ,the unentire membrane of the tumor (P﹤0.001) and the tumor with low or strong echo (P﹤0.001) were significantly higher in the malignant tumors. In contrast, there were no significant difference in the largest diameter and the volume for each lesion among the benign and malignant tumors(p=0.867 and 0.632 respectively). There was significant difference of the presence or absence of the ascite between the benign and malignant tumors (P﹤0.001). The 98 percent of the malignant tumors were checked with the central blood flow, while 18 (30%) benign tumors were checked with the central blood stream, the others of the benign tumors had peripheral blood stream, there was significant difference (P﹤0.001).Both impedance
parameters (RI and PI) were significant lower in malignant tumors(p=0.011 and <0.0001,respectively).While both velocity parameters (PSV and TAMVX) were not significant (p=0.829 and 0.139 respectively).Regression analysis on the 13 variables resulted in the retention of only “RI”, “tumor morphology score”, “flow score” and “ascite” as significantly contributing to predicting the presence and absence of malignancy. The probability of malignancy for any patient was given by solving the equation:1/(1+e-z),where e is the base value for natural logarithms and z=（-15.034×RI）+(4.423×tumor morphology score)+(3.040×tumor flow score)+(2.456×ascite)+0.531.Through the ROC curve , the optimum cut-off of the continuous variables can be made and the character that the multivariate logistic regression compare with the univariable. Multivariable logistic regression model discriminates between benign and malignant tumors better than univariable (RI, tumor morphology score , flow score ,ascite) from the area of ROC curve. With this model ,a probability greater than 25% gave a sensitivity and specificity of 98% and 91.7%,postivie prediction value 100%,negetive prediction value 97%,accurancy
95.4%. A probability greater than 50% gave a sensitivity and specificity of 92% and 95%, positive prediction value 97.9%,negetive prediction value 93.4%,accurancy 93.6%.Of the entire patients, 57 women were diagnosed exactly ,of the whole 61 women diagnosed as the benign tumor, diagnosis coincidence ratio 96.7%. 46 women were diagnosed exactly as malignant tumor of the whole 49 women diagnosed malignant tumor, the diagnosis coincidence ratio 95%.
Conclusion The diagnosis value of multivariable logistic regression model appears to be better than independently use gray ultrasound and color Doppler ultrasound .Th
引文
Fleischer A，Brader K. Sonographic depiction of ovarian vascularity and flow: current improvements and future applications. J Ultrasound Med 2001,20(3):241-250
Buy J，Hugol D，Ghossain M，et al. Malignant tumors of the ovaries：role of imaging. J Radiol，2000，81：1833－1843
Fleischer AC，Shappell HW，Parker LP，et al. Color Doppler sonography of endometrial masses. J Ultrasound Med 2002 Aug;21(8):861-5;quiz 867-8
Guerriero S, Alcazar JL, Coccia ME et al. Complex pelvic mass as a target of evaluation of vessel distribution by color Doppler sonography for the diagnosis of adnexal malignancies:results of a multicenter European study. J Ultrasound Med 2002 Oct;21(10):1105-11
Orden MR, Jurvelin JS, Kirkinen PP. Kinetics if a US contrast agent in benign and malignant adnexal tumors. Radiology.2003 Feb;226(2):405-10
BenjapibalMida M，Maruyama R，et al. Prognostic significance
of ultrasound derived intratumoral peak systolic in epithelia ovarian cancer. Ultrasound Obstet Gynecol，2002 ；20：186-191
Sawicki W，Spiewankiewicz B，Cendrowski K ，et al. Preoperative discrimination between malignant and benign adnexal masses with transvaginal ultrasonography and colour blood flow imaging. Eur J Gynaecol Oncol 2001;22(2):137-42
Tailor A, Jurkovic D, Bourne TH,et al. Sonographic prediction of malignancy in adnexal mass using multivariate logistic regression analysis. Ultrasound Obstet Gynecol 1997 10,41-47
Tailor A，Jurkovic D，Bourne TH，et al. Sonographic prediction of malignancy in adnexal masses using an artificial neural network. Br J Obstet Gynaecol 1999 Jan;106(1):21-30
Rampone B，Rampone A，Tirabasso S，et al. Ovarian cancer screening by transvaginal color Doppler ultrasonography. Minerva Ginecol 2001 Feb;53(1):125-8
Berlanda N，Ferrari MM，Mezzopane R，et al. Impact of a multi-parameter, ultrasound-based triage on surgical management of adnexal masses. Ultrasound Obstet Gynecol
2002 Aug;20(2):181-5
Emoto M，Obama H，Horiuchi S，et al. Transvaginal color Doppler ultrasonic characterization of benign and malignant ovarian cystic teratomas and comparison with serum squamous cell carcinoma antigen. Cancer，2000 May 15;88(10)2298-304
Angeid-Backman E，Coleman BG，Arger PH，et al. Comparison of resistive index versus pulsatility index in assessing the benign etiology of adnexal masses. Clin Imaging 1998 Jul-Aug;22(4):284-91
Benjapibal M，Sunsaneevitayakul P，Boriboonhirunsarn D，et al. Color Doppler ultrasonograph for prediction of malignant ovarian tumors. J Med Assoc Thai 2002 Jun;85(6):709-15
Timmerman D,Bourne HT,Tailor A et al. A comparison of methods for preoperative discrimination between malignant and benign adnexal masses:The development of a new logistic regression model. Am J Obstet Gynaecol 1999 July; 18(1): 57-65
16.Marret H,Ecochard R,Giraudeau B,et al. Color Doppler energy prediction of malignancy in adnexal masses using logistic
regression models. Ultrasound Obstet Gynecol. 2002 Dec; 20(6):597-604
17.Boll D,Geomini PM,Brohnann HA et al. The pre-operative assessment of the adnexal mass:the accuracy of clinical estimates versus clinical prediction rules. BJOG. 2003 May;110(5):519-23