非创伤性股骨头缺血性坏死塌陷预测的MRI研究

MRI study on predicting the collapse of avascular necrosis of the femoral head

目的探讨预测股骨头缺血性坏死塌陷的MRI诊断标准。方法经MR检查确诊为股骨头缺血性坏死患者22例（39髋）作为研究对象。对39个坏死股骨头的MRI表现进行评定。内容包括：骨髓水肿、关节腔积液、坏死区信号特点及发病部位，测量并计算出坏死体积比例和坏死表面积比例。患者随访18～84个月，中位时间25个月。前瞻性对上述评定内容进行Logistic回归分析，得出股骨头塌陷的危险因素；对坏死灶表面积比例及坏死体积比例之间的诊断实验采用ROC曲线。计算ROC曲线下面积；取坏死表面积比例约登指数最大时点，作为预测股骨头塌陷的最佳临界点。确定其敏感度、特异度和准确率。结果39髋中21髋发生股骨头塌陷，18髋未发生塌陷。5例双侧髋关节塌陷。其中坏死表面积大于25％的25髋中21髋（84．O％）发生塌陷；坏死体积大于30％的8髋全部塌陷；坏死区位于股骨头外上象限的33髋中21髋（63．6％）塌陷；坏死区表现为混杂信号22髋中18髋（81．8％）发生塌陷；出现中等及大量关节腔积液25髋中16髋（64．O％）发生塌陷；18髋骨髓水肿的患者中13髋发生塌陷（72．2％）。股骨头塌陷结果的Logistic回归分析结果表明：关节腔积液、信号特点、部位、坏死体积比例（％）、坏死面积比例（％）是危险因素。骨髓水肿的OR值小于l，相对于其他因素为低风险因素（按“=0．10的水准）。坏死表面积比例及坏死体积比例的ROC曲线分析：坏死表面积比例ROC曲线下面积（0．987）大于坏死体积比例ROC曲线下面积（0．902）。坏死表面积比例的最佳临界点为26．7％，诊断的敏感度为95．2％，特异度为94．4％。约登指数为0．896。结论股骨头坏死后塌陷是多种因素的综合作用。当坏死范围足够大，具有混杂信号特点、大量关节腔积液、骨髓水肿且坏死部位于股骨头外上象限时易发生股骨头塌陷。坏死表面积比例达26．7％时，是股骨头塌陷的危险临界指标。预测股骨头坏死塌陷时坏死表面积比例较坏死体积比例更准确。

Objective To study the risk factors of MRI for the prediction of collapse in patients with avaseular necrosis of the femoral head. Methods Twenty-two patients （39 hips） diagnosed avascular necrosis of femoral head by MR were enrolled in our study. The following MR appearances were evaluated： bone marrow edema ,joint fluids, signal intensity and location of the lesion. The volume and surface area of the necrosis zone were calculated. The time of follow-up was 18--84 months （median,25 months）. Logistic regression analysis was used to predict the risk factors by SPSS 13.0. The maximum value of Youden index was selected as the critical point to predict the collapse of femoral head and to define the sensitivity, specificity and accuracy. Results In the 39 hips with femoral head necrosis,21 hips had collapse. Bilateral collapse occurred in 5 cases. In 25 hips with the necrosis surface larger than 25% , collapse occurred in 21 （84%）; In 8 hips with the volume of femoral head necrosis larger than 30% ,collapse occurred in all cases; In 33 hips with the necrosis locating at the superolateral quadrant, collapse occurred in 21 （63.6%）; In 22 hips with necrotic areas showing heterogeneous signal intensity, collapse occurred in 18 （81.8%） ;In 25 hips with large amount of joint effusion, collapse occurred in 16 （64%） ;in 18 hips with bone marrow edema, collapse occurred in 13 （65%）. Joint fluid, heterogeneous signal intensity and lesions in the superolateral quadrant, volume ratio, and area ratio were the high risk factors, while bone marrow edema was a relatively low risk factor. The area under ROC curves for area ratio of NASA was greater than that for volume ratio （0. 987 vs 0. 902）. When the critical value for area ratio was 26. 7% ,the true positive rate was 95.2% ,true negative rate was 94. 4% , and Youden＇ s index was 0. 896. Conclusions The collapse of necrosis of femoral head may result from many factors. The femoral head was easy to collapse when it had large enough area of necrosis and mixed signal intensity, a large amount of joint effusion, bone marrow edema, and superolateral quadrant location. The critical value for area ratio to predict the collapse of femoral head was about 26. 7%. The area ratio is more accurate than volume ratio in predicting the collapse of necrosis of femoral head.