四种预测前列腺穿刺阳性风险模型准确性的外部验证与比较

发布时间：2018-08-16 16:37

【摘要】：目的:验证与比较四种用于个体化预测经直肠超声(transrectal ultrasound,TRUS)引导下初次前列腺穿刺阳性风险模型的准确性。方法:收集813例从2010年1月到2014年9月在我院进行经直肠超声引导下前列腺穿刺的患者数据,符合四种模型条件的431例患者最终入选本研究。分别用各模型进行个体化前列腺穿刺阳性风险计算,通过受试者工作特征曲线下面积(area under the receiver operating characteristic curve,AUC)来评估该模型的预测准确性,并用z检验进行各模型AUC之间的比较。结果:前列腺癌组与非前列腺癌组之间,除游离前列腺特异性抗原百分比(percentage of free prostate-specific antigen,%fPSA)差异无统计学意义外(P=0.242),年龄、PSA、直肠指检、前列腺体积以及超声结果之间的差异均具有统计学意义(P0.05)。北美前列腺癌预防试验的前列腺癌风险计算(Prostate Cancer Prevention Trial Derived Cancer Risk Calculator,PCPT-CRC)模型、蒙特利尔模型、国内模型1、国内模型2以及PSA的AUC分别是0.774(95%CI,0.726~0.822),0.765(95%CI,0.714~0.816),0.813(95%CI,0.767~0.858),0.795(95%CI,0.749~0.842)和0.736(95%CI,0.684~0.788),各模型之间AUC的比较示差异无统计学意义(P0.05),而与PSA比较,当PSA范围无限制时,国内模型1的预测准确性提高了7.7%(P0.05)。当PSA为4~10 ng?ml-1时,该4种模型和PSA的AUC分别是0.688(95%CI,0.560~0.816),0.818(95%CI,0.719~0.918),0.830(95%CI,0.740~0.919),0.853(95%CI,0.771~0.935)和0.565(95%CI,0.419~0.710),国内模型2预测准确性最高,较PSA提高了28.8%(P0.05)。结论:4种模型均具有较高的预测准确性,国外PCPT-CRC模型和蒙特利尔模型与国内两种模型的预测准确性无差异,但与PSA比较,当PSA范围无限制时,国内模型1更有优势,而当PSA为4~10ng?ml-1时,国内模型2预测准确性最高。
[Abstract]:Objective: to verify and compare the accuracy of four individual models for predicting the risk of primary prostate puncture under the guidance of transrectal ultrasound (TRUS). Methods: the data of 813 patients who underwent transrectal ultrasound guided prostate puncture in our hospital from January 2010 to September 2014 were collected. The positive risk of individual prostate puncture was calculated with each model. The predictive accuracy of the model was evaluated by the area under the operating characteristic curve (area under the receiver operating characteristic curveAUC), and the AUC of each model was compared with z test. Results: there was no significant difference in the percentage of free prostate specific antigen (percentage of free prostate-specific antigenfPSA) between prostate cancer group and non-prostate cancer group (P < 0. 242). Prostate volume and ultrasound results were statistically significant (P0.05). (Prostate Cancer Prevention Trial Derived Cancer Risk Calculatorus PCPT-CRC Model, Montreal Model, for Prostate Cancer Prevention trial in North America, The AUC of domestic model 1, domestic model 2 and PSA were 0.774 (95CI0.7260.822) 0.765 (95CI0.7140.816) 0.813 (95CI0.7670.858) 0.795 (95CI0.7490.0.842) and 0.736 (95CI0.6840.788), respectively. There was no significant difference in AUC between the two models (P0.05), but compared with PSA, the prediction accuracy of domestic model 1 increased by 7.7% (P0.05). When the PSA was 4 ~ 10 ng?ml-1, the AUC of the four models and PSA were 0.688 (95CI0.5600.816) and 0.818 (95CI0.7190.918), 0.830 (95CI0.7400.9019) 0.853 (95CI0.7710.935) and 0.565 (95CI0.4190.7910), respectively. The prediction accuracy of domestic model 2 was the highest, which was 28.8% higher than that of PSA (P0.05). Conclusion the prediction accuracy of PCPT-CRC model and Montreal model is not different from that of domestic model. However, compared with PSA, domestic model 1 has more advantages when the range of PSA is not limited. When PSA is 4~10ng?ml-1, domestic model 2 has the highest prediction accuracy.
【学位授予单位】：苏州大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R737.25

【相似文献】