Yanke Brent V, Gonen Mithat, Scardino Peter T, Kattan Michael W
Department of Urology, State University of New York Downstate Medical Center, Brooklyn, New York, USA.
J Urol. 2005 Feb;173(2):421-4. doi: 10.1097/01.ju.0000150522.82760.00.
We reported a nomogram and subsequently a corrected version for predicting the probability of positive biopsy in men with 1 or more prior negative biopsies. In this study we assessed the validity of this nomogram when applied to an external dataset.
There were 230 patients from the Brooklyn Veterans Administration Medical Center who underwent 1 or more repeat biopsies after initial negative biopsy from January 1993 to June 2003. Predictor variables studied in the nomogram were patient age, family history of prostate cancer, digital rectal examination, serum prostate specific antigen, prostate specific antigen slope, months from initial negative biopsy session, months from previous negative biopsy session, cumulative number of negative cores previously taken and history of high grade intraepithelial neoplasm or atypical small acinar proliferation. We calculated the nomogram predicted probability in each patient. These predicted outcomes were compared with actual biopsy results. Area under the ROC curve was calculated as a measure of discrimination. Calibration was assessed graphically.
We evaluated a total of 356 repeat biopsies in 230 patients (mean 2.56 total biopsies per patient). The mean number of total cores per patient was 17.9. There were 78 positive biopsies. The area under the ROC curve was 0.71, which was greater than any single risk factor. Nomogram calibration appeared to be good.
Our corrected nomogram for predicting positive repeat biopsy performed well when applied to a sample of men at the Brooklyn Veterans Administration Medical Center. This nomogram can provide important additional information to aid the urologist and patient with a negative biopsy in evaluating clinical options.
我们报告了一种列线图，随后又报告了其校正版本，用于预测有1次或更多次既往活检阴性的男性患者活检阳性的概率。在本研究中，我们评估了该列线图应用于外部数据集时的有效性。
选取1993年1月至2003年6月间在布鲁克林退伍军人事务部医疗中心接受初次活检阴性后又进行1次或更多次重复活检的230例患者。列线图中研究的预测变量包括患者年龄、前列腺癌家族史、直肠指检、血清前列腺特异性抗原、前列腺特异性抗原斜率、距初次阴性活检时间、距上次阴性活检时间、既往阴性活检的累计阴性核心数以及高级别上皮内瘤变或非典型小腺泡增生病史。我们计算了每位患者列线图预测的概率。将这些预测结果与实际活检结果进行比较。计算ROC曲线下面积作为区分度的指标。通过图形评估校准情况。
我们共评估了230例患者的356次重复活检（每位患者平均2.56次活检）。每位患者的平均总核心数为17.9。有78次活检阳性。ROC曲线下面积为0.71，大于任何单一风险因素。列线图校准情况似乎良好。
我们校正后的用于预测重复活检阳性的列线图应用于布鲁克林退伍军人事务部医疗中心的男性样本时表现良好。该列线图可为泌尿外科医生和活检阴性的患者在评估临床选择时提供重要的额外信息。
