患者年龄及直肠超声特征联合血清PSA水平的logistic回归模型对前列腺癌的预测

Prediction of prostate cancer by logistic regression model about patient's age and transrectal ultrasound characteristics combined with serum PSA level

目的探讨logistic回归模型的相关参数(患者年龄及直肠超声特征联合血清PSA水平)对前列腺癌的预测价值分析。方法选取2011年1月~2014年1月聊城市人民医院收治的424例前列腺疾病患者的临床资料进行回顾性分析,对患者的年龄及直肠超声检查结果与穿刺病理检查结果进行统计分析,并建立logistic回归模型。结果在前列腺癌患者分年龄段比较中,70岁以上的患者的PSA水平明显高于其他年龄段的患者(P<0.05);前列腺癌患者与非癌性前列腺疾病患者在各个PSA水平的比较中差异均有统计学意义(P<0.05);2组患者的前列腺形状、前列腺边界、内外腺边界、前列腺和精囊腺分离线的清晰度、前列腺和直肠分离线清晰度、淋巴结转移、非正常血流信号及弱回声之间的比较差异均有统计学意义(P<0.05),而结节和不均匀回声2个因素的比较差异无统计学意义;经直肠超声声像图联合年龄及PSA诊断前列腺癌的准确度为83.46%,特异性为92.73%,敏感度为65.01%,阳性预测值为81.89%,阴性预测值为83.95%。结论 logistic回归模型的相关参数如患者年龄及直肠超声特征联合血清PSA水平对前列腺癌的预测结果较佳。

Objective To construct a logistic regression model about the patients'' age and transrectal ultrasound characteristics combined with serum PSA level and to predict prostate cancer.Methods 424 clinical data of patients with prostate disease in our hospital from January 2011 to January 2014 were retrospectively analyzed,the patients'' age ,transrectal ultrasound pathological findings and biopsy findings in patients were used to establish a logistic regression model.Results Compared to patients with prostate cancer by age segment,>70 years of age or PSA levels in patients was significantly higher than in patients with other age groups;prostate cancer patients and patients with non-cancerous prostate disease were significant differences in the comparison of the various PSA levels (P<0.05);shape of two groups of patients with prostate,prostate boundaries,both inside and outside the boundaries gland,prostate and seminal vesicles separation line clarity,prostate and rectum separation line definition,lymph node metastasis, abnormal blood flow signals and weak echo comparison with a statistically significant difference between two groups(P<0.05 ),whereas no significant difference between nodules and uneven echo two factors (P>0.05 );transrectal ultrasonographic combined age and PSA prostate cancer diagnosis sensitivity of 65.01% and a specificity of 92.73% and accuracy was 83.46%,81.89% positive predictive value,negative predictive value 83.95%. Conclusion Age and characteristics of patients with rectal ultrasound combined with serum PSA level logistic regression model to predict the results of the prostate is better,and should be introduced.