寰枢椎转移瘤和脊索瘤的影像学诊断及鉴别诊断

Diagnosis and Differential Diagnosis of Atlantoaxial Metastasis and Chordoma

目的探讨寰枢椎转移瘤和脊索瘤影像学特征以提高对寰枢椎肿瘤的诊断及鉴别诊断能力。方法回顾性搜集经穿刺或手术病理证实为寰枢椎转移瘤(40例)和脊索瘤(29例)患者的临床及影像学资料,比较二者在年龄、发生节段及影像学征象上的差异,连续变量采用独立样本t检验或Mann-Whiney U检验,分类变量采用χ^(2)检验,P<0.05为差异有统计学意义。以寰枢椎转移瘤和脊索瘤作为因变量,年龄、节段及影像学各项评估指标为自变量,进行决策树分析。结果转移瘤患者的平均年龄为(55.0±11.6)岁,脊索瘤为(46.6±15.0)岁,二者年龄差异有统计学意义(t=2.626,P=0.011)。转移瘤发生于寰椎3例,枢椎32例,同时累及寰枢椎3例,累及斜坡2例;脊索瘤分别为0例、23例、0例、6例,二者的发生节段差异有统计学意义(χ^(2)=10.129,P=0.018)。指标是否呈膨胀性改变(χ^(2)=5.141,P=0.023)、骨基质(χ^(2)=9.544,P=0.008)、CT值(t=7.780,P=0.000)、MR信号是否均匀(χ^(2)=7.329,P=0.014)、T2WI信号强度(χ^(2)=53.143,P=0.000)、脂肪抑制序列信号强度(χ^(2)=40.615,P=0.000)、软组织肿块内是否有纤维分隔(χ^(2)=13.860,P=0.000)在转移瘤和脊索瘤的鉴别诊断中有统计学意义。被选入树形图的自变量为:T2WI信号强度(χ^(2)=46.870,P=0.000)、CT值(χ^(2)=7.639,P=0.029)、病变的最大直径(χ^(2)=8.148,P=0.039)及骨基质(χ^(2)=5.850,P=0.016)。决策树模型对于寰枢椎转移瘤预测的准确性为97.5%,对脊索瘤预测的准确性为89.7%,模型整体预测准确性为94.2%。结论寰枢椎转移瘤患者的年龄更大,骨质破坏易呈膨胀性改变,软组织肿块CT值更高,T2WI及脂肪抑制序列中以等信号和稍高信号多见;脊索瘤易累及斜坡,骨质破坏区内可见不规则骨化或钙化,MRI上信号更不均匀,T2WI及脂肪抑制序列中以高信号多见,软组织肿块内多可见纤维分隔。决策树模型对二者的鉴别诊断有一定的帮助。

Objective To explore the imaging features of atlantoaxial metastasis and chordoma in order to improve the ability of diagnosis and differential diagnosis of atlantoaxial tumors.Methods The clinical and imaging data of 40 patients with atlantoaxial metastasis and 29 patients with chordoma confirmed by puncture or surgical pathology in the Third Hospital of Peking University from 2006 to 2019 were retrospectively collected.The differences in age,occurrence segment and imaging signs between the two were compared.The independent sample t test and the Mann-Whiney U test were used for the continuous variables,and theχ^(2)test was used for the categorical variables.P<0.05 for the difference was statistically significant.Taking atlantoaxial metastasis and chordoma as dependent variables and age,segment and imaging features as independent variables,decision tree analysis was applied.Results The mean age of metastasis patients was 55.0±11.6 years old,and the mean age of chordoma was(46.6±15.0)years old.The age difference between them was statistically significant(t=2.626,P=0.011).Metastasis occurred in 3 cases of atlas,32 cases of axis,3 cases involving atlantoaxial spine and 2 cases of clivus;there were 0,23,0 and 6 cases for chordomas respectively.The difference in the segment of occurrence between the 2 diagnoses was statistically significant(χ^(2)=10.129,P=0.018).There was statistical significance in the differential diagnosis of metastasis and chordoma in whether the indexes were swelling(χ^(2)=5.141,P=0.023),bone matrix(χ^(2)=9.544,P=0.008),CT value(t=7.780,P=0.000),MR signal uniformity(χ^(2)=7.329,P=0.014),T2WI signal strength(χ^(2)=53.143,P=0.000),signal strength of fat suppression sequence(χ^(2)=40.615,P=0.000),and whether there was fibrous separation in soft tissue mass(χ^(2)=13.860,P=0.000).The T2WI signal strength(χ^(2)=46.870,P=0.000),CT value(χ^(2)=7.639,P=0.029),maximum diameter of lesion(χ^(2)=8.148,P=0.039)and structure of bone destruction area(χ^(2)=5.850,P=0.016)were selected as the independent variables into the tree diagram.The decision tree model has an accuracy rate of 97.5%for the prediction of atlantoaxial metastases,89.7%for chordoma,and 94.2%for the overall model.Conclusion The patients with atlantoaxial metastases are older,the bone destruction is easy to show expansive change,the CT value of soft tissue mass is higher,equal signals and slightly higher signals are more common in T2WI and fat suppression sequence.Chordomahas a predilictionfor the clivus,andirregular ossification or calcification can be seen in the region of bone destruction.Thesignal on MR is more uneven,the high signal is more common in T2WI and fat suppression sequence,and there is more visible fiber separation in the soft tissue mass.The decision tree model is helpful to the differential diagnosis of the twoentities.