了前列腺磁共振检查。我们对T2加权序列及DWI弥散序列进行研究。PI-RADS v2被应运于双参数前列腺磁共振研究。ISUP评分法用来组织病理学分级。有三种情况：独立的ISUP评分检测病变（isup≥1），介于独立与临床显著病变检测之间的（isup≥2），单独检测临床显著病变（isup≥3）。通过回归曲线和逻辑回归分析，评价了bp-MR和psa-d的预测值。P<0.05具有统计学意义。

Results: In all evaluated scenarios, bp-MR showed a significantly higher predictive power (AUC =0.87-0.95) compared to the performance of PSA-D (AUC= 0.73–0.79), while their combination (AUC =0.91-0.95) showed no statistically significant improvement compared to bp-MR alone.

结果：bp-mr显示出与psa-d（auc=0.73–0.79）相比，预测能力（auc=0.87-0.95）要高得多。并且他们的组合（auc=0.91-0.95）与单独bp-mr相比没有显著的统计上的改善。

Conclusion: Our results confirm that bp-MR is a powerful tool in detection of clinically significant PCa. Contrary to findings in the recent literature, PSA-D does not appear to significantly improve its diagnostic performance.

结论：bp-mr对于临床前列腺癌的诊断是一个有意义的工具。并且psa-d似乎并没有显著提高其诊断性能。

1、 简介

Prostate cancer (PCa) is the most frequent malignant tumor in males and it is estimated that one out of seven men will be diagnosed with PCa during his lifetime [1]. While clinical suspicion of PCa is typically based on digital rectal examination (DRE) and/or raising serum levels of prostate-specific antigen (PSA), the European Association of Urology suggests the use of transrectal ultrasound-guided biopsy (TRUS-biopsy) for initial diagnosis and relegates the use of multiparametric MR (mp-MR) to target or guide any repeat biopsy [2]. However, this approach could change in the years to come as it has already been demonstrated that using mp-MR as a triage test, before the first prostate biopsy, could reduce unnecessary biopsies while improving the detection of clinically significant prostate cancer(csPCa) [3]. Known and recognized limitations of MRI are its high costs and scan duration.In recent studies, an abbreviated version of the mp-MRreferred to as bi-parametric MR (bp-MR) has been proposed. It excludes the dynamic contrast enhanced sequences, resultingin a faster, cheaper and more tolerable exam, giving the opportunity to caretakers to refer more patients to MR. Bp-MR proved to be a valuable tool for the detection of PCa with a diagnostic accuracy comparable to the standard mp-MR protocol [4,5].

前列腺癌（pca）是男性最常见的恶性肿瘤之一，据估计，七分之一的男性一生中会被诊断出前列腺癌【1】。而临床工作中我们对怀疑前列腺癌的病人通常采用前列腺指诊或血清PSA进行检测，欧洲泌尿学协会建议使用经直肠超声引导的活检进行初步诊断，并放弃使用多参数前列腺磁共振来针对或指导任何重复活检【2】。然而，已经有研究证明，在第一次前列腺活检之前，使用双参数前列腺磁共振，可以减少不必要的活检，同时提高临床工作中前列腺癌的诊断率【3】。目前已知和公认的磁共振的局限性在于它的高成本和长的扫描时间。在最近的研究中，一个缩减版的多参数前列腺磁共振，我们将其称为双参数前列腺磁共振。它排除了动态对比增强的序列，导致更快，更便宜，这被证明是检测前列腺癌的一个有价值的工具，其诊断精度相当于标准的多参数前列腺磁共振。

While serum PSA levels have a role in the diagnosis and follow-up of PCa, it has a low specificity alone; thus, a quotient of serum PSA and prostate volume defined as PSA-density (PSA-D) was introduced to enhance the accuracy and has shown to have high value in the prediction of negative outcome of prostate biopsy while being more strongly associated with the presence of Gleason score ≥ 7 cancers [6–9]. Recent works have suggested that the combination of PSA-D andprostate MR findings may improve the detection of clinically significant prostate lesions [5–7,10].

血清psa水平对前列腺癌的诊断和随访有一定作用，但其特异性较低。引入血清PSA和前列腺体积的比值定义为前列腺特异性抗原密度（psa-d），以提高准确性，并已证明在预测前列腺活检结果方面有很高的价值，同时与gleason评分≥7的存在有较强的相关性[6-9]。最近的研究表明，PSAD和前列腺磁共振的联合检测可以提高前列腺病变检测的准确度。【5-7,10】

The aim of this study was to assess the value of combined bp-MR and PSA-D for prediction of PCa in a biopsy-naïve patient population.

这项研究的目的是评估bp-mr和psa-d联合检测来预测进行前列腺活检患者中的前列腺癌病人。

2. Materials and methods

This retrospective study was approved by the local Institutional Review Board, and the need for informed consent was waived. All procedures performed in the study were in accordance with the ethical standards of the institutional and/or national research committee and conformed to the principles of the Declaration of Helsinki on human research.

这项回顾性研究已获本地机构检讨委员会批准，并免除了知情同意的要求。研究中执行的所有程序都符合机构和/或国家研究委员会的道德标准，并符合赫尔辛基人类研究宣言的原则。

2.1. Patient population

We evaluated 344 consecutive patients who underwent prostate MR exams, between January 2015 and September 2017, at our Institution. All patients had clinical suspicion of PCa, derived from either elevated PSA levels or abnormal findings on DRE.The exclusion criteria were: absence of TRUS-biopsy at our Institution within 30days following the MR exam, previous prostate biopsy, MR images affected by artifacts, incomplete MR exam. The study population flowchart is shown in Fig. 1.

在2015年1月至2017年9月期间，我们对344名连续接受前列腺MR检查的患者进行了评估。所有患者临床上都怀疑有前列腺癌，这种怀疑来自于psa水平的升高或dre的异常发现。排除的标准是：在测试后的30天内，我们的医院没有进行测试活检，以前的前列腺活检，受人工制品影响的图像，不完全的测试。研究人口流程图如图1所示。

2.2. MR acquisition protocol 

All MR scans were acquired on a 3T scanner (Magnetom Trio,Siemens Medical Solutions, Erlangen, Germany) using the standard surface phased-array coil and integrated spine phased-array coil (Body Matrix, Siemens Medical Solutions).

所有的mr扫描都是在3t扫描仪（磁控三进制，西门子医疗解决方案，德国埃尔朗根）上获得的，使用标准的表面相控阵线圈和集成的脊柱相控阵线圈（身体矩阵，西门子医疗解决方案）。

Acquisition protocol included for all subjects the following sequences: T2-weighted turbo spin echo (T2w) imaging acquired with axial, sagittal and coronal orientations (repetition time: 4000 ms; echo time: 101ms; field of view: 200× 200mm; thickness: 3mm, no gap) and a diffusion weighted sequence (DWI) with an axial orientation (repetition time: 4900 ms; echo time: 89ms; b-values: 0, 400, 1500; field of view: 200× 200mm; thickness: 3mm, no gap).

所有科目的采集协议包括以下序列：以轴向、矢状和日冕方向获取的t2加权Turbo自旋回波（t2w）成像（重复时间：4000MS；回波时间：101ms；视场：200×200mm；厚度：3mm，无间隙）和轴向方向的扩散加权序列（dwi）（重复时间：4900MS；回波时间：89ms；b值：0、400、1500；视场：200×200mm；厚度：3mm，无间隙）。

2.3. Image analysis

MR criteria suggestive for malignancy were based on the PI-RADS v2 criteria [11]. Specifically, exams were classified by giving a Likert scale score (referred to as bpPI-RADS) from 1 to 5, where 1 was considered very low probability of csPCa (highly unlikely to be present); 2 as low probability (csPCa was unlikely to be present); 3 as intermediate probability (the presence of csPCa disease was equivocal); 4 as high probability (csPCa was likely to be present) and 5 as very high (csPCa was highly likely to be present) [12]. PSA-D values were obtained for each patient from the ratio between their last PSA serum concentration and the prostate volume, calculated on MR images.

MR提示为恶性肿瘤的标准是基于pi-rads V2标准[11]。具体地说，检测是通过给likert标度分数（简称bpi-rads）从1到5进行分类，其中1被认为是非常低的cspca概率（极不可能存在）；2为低概率（cspca不可能存在）；3为中间概率（cspca疾病的存在不确定）；高概率为4（cspca可能存在），高概率为5（cspca很可能存在）[12]。每一患者的psa-d值是根据他们最后的psa血清浓度与前列腺体积的比率，根据mr mids计算得出的。

Two radiologists with respectively 16 years (X. X.) and 12 years (X.X.) of experience in prostate MR imaging, both blinded to the clinical indication, PSA and PSA-D values,analyzedin consensus bp-MR images assigning scores to theexams.

235/5000

由两位分别有16年和12年前列腺诊断经验的放射科医生，采用盲法对临床指标、psa和psa-d值进行分析，分析bp-m.m.影像中的评分至意见统一。

2.4. Prostate biopsy procedure前列腺活检程序

After MR examination, all patients underwent a standard 12-core TRUS-guided biopsy. All prostate biopsies in this study were operated by one of two urologists, with over 10 years of experience in TRUS- guided biopsy.They were performed by administrating peri-prostatic local anesthetic (10 ml Lidocaine 1%) and pre-procedural antibiotic prophylaxis (Ciprofloxacin), using a BK ultrasound scanner, an endorectal probe, a needle guide and an 18-Gauge 25-cm biopsy needle.Biopsies were obtained from each of 6 sagittal regions (3 per side) moving from prostatic base to apex with 2 cores from each zone for a total of 12 cores. 10mm in length TRUS-guided biopsies cores were taken from 12 prostatic regions and marked separately. Additionally, cognitive-targeted biopsy was used to determine the exact location for biopsy sampling, after careful review of the patient’s bp-MR exam, as follows. The probe was angled as needed in the sagittal plane such that the needle trajectory included the area of concern. For lesions located more cranially, the biopsy needle was advanced 10–20 mm into the prostate gland prior to deploying the biopsy device. For more caudally located lesions, the biopsy device was activated from the prostate surface or with minimal advancement into the gland prior to triggering the device. Samples were taken more medially or laterally as needed based on the pre-procedure bp-MR images. At least two cores of tissue were obtained from the target area.All biopsy samples were sent to the pathology laboratory of our Institution, where they were analyzed and reported according to the International Society of Urological Pathology (ISUP) 2014 modified Gleason score grading system [13]. Patients with a negative histopathologic result were assigned a score of 0.

在前列腺磁共振检查之后，所有病人都要进行经直肠前列腺穿刺活检术。这项研究中所有前列腺活检都是由两位泌尿科医生之一进行的，具有超过10年的经直肠超声引导前列腺穿刺活检经验。使用bk超音波扫描仪、直肠内探针、引针和18毫米25厘米活检针，分别进行前列腺周围局部麻醉（10毫升利多卡因1%）和预处理抗生素预防（环丙沙星）。从6个矢状区（每边3个）中的每个区（从前列腺基到先端，每个区有2个岩心，共12个岩心）进行活检。10 Mm长度的三引导活检核心取自12个前列腺区，并单独标记。另外，在仔细检查病人的bp-mr检查后，使用认知定向活检来确定活检取样的确切位置，如下所示。在矢状平面上，探针的角度是按需要的，因此针的轨迹包括了关注的区域。对于位于前列腺内部的病变，在使用活检装置之前，活检针被推进到前列腺10–20毫米。对于更多位于外周带位置的病变，活检装置在触发该装置之前从前列腺表面或以最小推进进入腺体的方式被激活。根据手术前的bp-mr图像，在需要的时候，更多地在中间或侧面采集样本。从目标区域获得至少两条活检组织。所有活检样本都被送到本机构的病理实验室，并根据国际泌尿学病理学学会（isup）2014年改良的Gleason评分系统进行分析和报告【13】。组织病理学结果为阴性的患者评分为0。

2.5. Statistical analysis

Continuous variables are presented as mean ± standard deviation and categorical variables as count and percentage. The ISUP score was dichotomized using three different cutoff values to assess the diagnostic performance of the considered variables for different purposes:

1) lesion detection independently of ISUP score (ISUP≥ 1)

(threshold1);

2) detection of both intermediate and clinically significant lesions

(ISUP ≥2) (threshold2);

3) detection of clinically significant lesions alone (ISUP≥ 3)

(threshold 3).

连续变量以平均值±标准差表示，分类变量以计数和百分比表示。采用三个不同的截止值对isup评分进行二分法，以评估为不同目的考虑的变量的诊断性能： 

1) 独立于isup评分的病变检测(isup≥1)（阈值1）；
2)中度或临床显著性病变（isup≥2）（阈值2）；
3）临床显著性病变（isup≥3）(阈值3)。

Univariable and multivariable logistic regression analysiswere per-formed for each of the three ISUP score dichotomizations.

对三个isup分数二分法分别进行了单变量和多变量逻辑回归分析。

Receiver operating characteristic (ROC) curves [14]were calculated to determine the discriminative ability of PSA-D, bpPI-RADS and their combined performance in identifying presence in each of the 3 ISUP thresholds.The area under the curve (AUC) was calculated for each ROC curve and DeLong’s test was used to evaluate statistically significant differences between them. Youden’s index was used to determine ideal cutoff values. Ap value< 0.05was considered as statistically sig-nificant, usingthe Bonferroni correction for multiple comparison when appropriate.All statistical analysis was performed with a dedicated software(Stata Statistical Software: Release 15, College Station, TX, StataCorp LP).

计算了ROC曲线[14]，以确定psa-d、bppi-rads的分辨能力及其在三个阈值中的综合识别性能。计算了曲线下的面积，并利用DeLong的试验对两者之间的统计显著性差异进行了评价。Youden的指数被用来确定理想的截止值。在适当的情况下，使用bonferroni校正进行多次比较。所有的统计分析都是用专用软件进行的（stata统计软件：发布15，学院台，tx，statacorp lp）。

Receiver operating characteristic (ROC) curves were calculated after univariable logistic regression to determine the discriminative ability of PSA-D and bpPI-RADS, separately, in predicting each of the 3 ISUP thresholds, and after multivariable logistic analysis including both parameters to assess their combined performance. The area under the curve (AUC) was calculated for each ROC curve and DeLong’s test was used to evaluate statistically significant differences between them [14]. For univariable ROC curves, Youden’s index was used to determine ideal cutoff values.

在单变量逻辑回归后，分别计算了ROC曲线，以确定psa-d和bppi-rads在预测三个阈值中的每一个阈值时的判别能力，并通过包含这两个参数的多变量逻辑分析来评估其综合性能。

3. Results

   The final study population included 114 patients(mean age= 65.75 years; SD ± 7.93; range = 44-85 years) with a mean prostate volume calculated on MR images of61.37 ml (SD ± 27.73 ml; range = 19–170ml); mean PSA serum level was 7.76ng/ml (SD ± 6.46 ng/ml; range =1.5–63ng/ml) and mean PSA-D value was 0.14 (SD ± 0.12; range= 0.02-0.9).Out of 114 patients included in this study, 59 had an equivocal or positive lesion identified using bp-MR (bpPI-RADS ≥ 3), all of which were object of additional cognitive targeted biopsies. The contingency table of bp-MR is reported in Table 1. In our population, in 6 patients the targeted biopsies proved positive while all the systematic ones resulted negative, while TRUS has not identified any additional lesions were not shown on MR. Finally, 61 (54%) had a histopathologically proven PCa. The ISUP score distribution, in relation to the bpPI-RADS score assigned, is shown in Table 2.

   在我们最终研究的114例患者中（平均年龄=65.75岁；sd±7.93岁；范围=44-85岁），根据磁共振图像计算的平均前列腺体积为61.37毫升（sd±27.73毫升；范围=19-170毫升）；平均psa血清水平为7.76纳克/毫升（sd±6.46纳克/毫升；范围=1.5-63纳克/毫升），平均psa-d值为0.14（sd±0.12；范围=0.02-0.9）。在这114例患者中，59例的病变是由bp-mr（bpi-rads≥3）确定的，这些都是新增的被认为需要性前列腺穿刺活检的病人。表1报告了bp-mr的应急表。在我们的人群中，有6名患者的目标活检结果为阳性，而所有的系统性活检结果均为阴性，并且bp-mr也没有发现任何病变。最后，61例（54%）有组织病理学证明的pca。相对于所分配的bppi-rads分数的isup分数分布情况见表2。

The results of the univariable and multivariable analyses for eachISUP score dichotomizations arereported in Tables 3–5. As shown, atmultivariable analysis, only bpPI-RADS and PSA-D were significant independent predictorsfor all considered ISUP thresholds.

表3-5报告了对每位ISUP分数进行单变量和多变量分析的结果。如图所示，在所有考虑的isup阈值中，只有bpi-rad和psa-d是显著的独立预测值。

Following these analysis, ROC curves were calculated for bpPI-RADS, PSA-D and their combination for each of the three ISUP score thresholds (Figs. 2–4). ROC curve analysis and DeLong’s tests showed a significantly higher predictive power in all casesfor bpPI-RADS alone when compared to the performance of the PSA-D alone.The combination of bpPI-RADS and PSA-D showed no statistically significant improvement in any case when compared to bpPI-RADS alone. Figs. 5 and 6 show two examples of bp-MR evaluations

在这些分析之后，计算了bppi-rads、psa-d的ROC曲线及其三个isup分数阈值的组合（图2-4）。roc曲线分析和DeLong的测试表明，仅在bppi-rads的所有情况下，与psa-d的单独性能相比，预测能力要高得多。与单纯的bppi-rad相比，bppi-rad和psa-d的结合在任何情况下都没有明显的统计学上的改善。图5和图6展示了bp-mr评估的两个例子。

Using Youden’s index, a cutoff value ≥3 was found for bpPI-RADSfor threshold 1, with a sensitivityof 93%, a specificity of 96%, a positive predictive value (PPV) of 97%, and a negative predictive value (NPV) of 93%.A bpPI-RADS cutoff value of ≥4 was found for the remaining thresholds, with a sensitivity of 92%, a specificity of 80%, a PPV of 70%, and a NPV of 95%for threshold 2 and a sensitivity of 93%, a specificity of 74%, a PPV of 56%, and a NPV 97% for threshold3, respectively. For PSA-D a cutoff value ≥ 0.13 was found for threshold 1, with a sensitivity of 70%, a specificity of 74%, a PPV of 70%, and a NPV of 68%. A cutoff ≥0.14 was found for the other thresholds with a sensitivity of 76%, a specificity of 73%, a PPV of 59%, and a NPV of 86% for threshold 2 and a sensitivity of 77%, a specificity of 73%, a PPV of 47%, and a NPV of 89% for threshold 3, respectively.

使用Youden指数，阈值1的bppi-rads值≥3被发现，灵敏度为93%，特异性为96%，正预测值（ppv）为97%，负预测值（npv）为93%。其余阈值的bppi-rads截止值≥4，灵敏度为92%，特异性为80%，正预测值为70%，阈值2的负预测值为95%，灵敏度为93%，阈值3的特异性分别为74%、正预测值为56%和负预测值为97%。对于psa-d，阈值为≥0.13，灵敏度为70%，特异性为74%，ppv为70%，npv为68%。在其他阈值上发现截止≥0.14，灵敏度为76%，特异性为73%，ppv为59%，阈值2的npv为86%，灵敏度为77%，阈值3的特异性分别为73%，ppv为47%，npv为89%。

4. Discussion

In recent years, an increasing number of studies have shown the value of mp-MR,which combines anatomic and functional imaging,as the technique of choice for studying patients with clinical suspicion of PCa, improving disease localization and allowing accurate staging [15–18].Ahmed et al. demonstrated that the use ofmp-MR to triage men with suspicion of prostate cancer, allowed to avoid primary biopsy in 27% of patients and diagnosis of 5% fewer clinically insignificant cancers [3]. However, despite these results, some major limitations of mp-MR are now becoming apparent, with particular reference to the relatively high cost and the length of the exam.Therefore, several groups have proposed a more feasible and simple imaging method. In a biopsy-naïve patient population,the clinical feasibility and similar diagnostic accuracy of a bp-MR protocolwas demonstrated when compared to the conventional mp-MR technique for the detection of PCa by using a high field 3T MR scanner without the use of an endorectal coil [4].Similarly, Kuhl et al. evaluated542 patients with elevated PSA who had undergone one or several rounds of TRUS-guided biopsywith negative results,proving that a simplified bp-MR protocol, obtained in under 9min acquisition time, without injection of a contrast agent or use of an endorectal coil, allows the detection of csPCa with comparable diagnostic accuracy and cancer detection rate compared with that of a full contrast-enhanced mp-MR imaging protocol [19]. Specifically, they found that the bp-MR protocol helped detect csPCa as missed by previous TRUS-guided biopsyin a substantial number of individuals (25.6%; 138 of 542), offering a high positive predictive value (73.4%; 138 of 188). Barth et al. found that a short imaging protocol, consisting of only T2w and DWI transverse images had no significant difference in diagnostic accuracy for detection of csPCawhen compared to a full mp- MR protocol [20].Finally, a recently published paper has shown how unenhancedprostate MRfollowed by cognitively guided MR biopsy is the most cost-effective approach in biopsy-naïve patients, when compared to the standard biopsy approach [21].

近年来，越来越多的研究显示了多参数磁共振的价值，多参数磁共振结合了解剖和功能成像技术，对于临床怀疑有前列腺癌的病人，具有提高病灶定位的准确性和精确分期的作用[15–18]。ahmed等人证明，使用多参数前列腺磁共振对怀疑患有前列腺癌的男性进行检查，可以避免27%的患者进行不必要活检，并且诊断出约5%不具有临床诊断意义的癌症【3】。 

然而，尽管有这些结果，mpmr的一些主要局限性现在变得很明显，特别是考相对较高的成本和较长的测试时间。因此，有几个小组提出了一种更可行和更简单的成像方法。在活组织检查患者中，与常规的使用高场3t mr相比，bp-mr的临床可行性和得到相似的诊断准确性得到了证明【4】。 

类似地，kuhl等人对542名psa升高的患者进行研究，他们均进行一次或多次超声引导的前列腺穿刺活检，且结果为阴性，证明我们通过简化的bp-mr检查（在不注射造影剂的情况下9分钟内就可以获得）就可以检测出与mpmr具有相似诊断精度和检测率的具有临床意义的前列腺癌【19】。特别是，他们发现bp-mr检查有助于检测cspca，就像以前经直肠前列腺穿刺忽略了一部分病人（25.6%；138/542人），而提供了较高的阳性预测值（73.4%；138/188）。barth等人发现，与完整的MP-mr相比，仅由t2w和dwi组成的成像检测在cspca的诊断精度上没有显著差异[20]。最后，最近发表的一篇论文表明，与标准的活检方法相比，未增强的前列腺磁共振再加上其指导下的前列腺活检是具有低成本效益的检测方法【21】。

PSA-D has been found useful not only in predicting biopsy outcome but also for suggesting the presence of csPCa [22–24]. Indeed, Kund et al. reported the correlation between PSA-D, higher pathological staging and PCa aggressiveness, resulting in a decreased progression-free survival rate after radical prostatectomy [23]. In addition, Corcoran et al. reported that PSA-D was the strongest predictor of tumor upgrading in the Gleason score between initial prostate biopsy and prostatectomy [24].

研究发现psa-d不仅在预测活检结果方面有用，而且对于暗示具有临床意义的前列腺癌的存在也有用[22–24]。确实，Kund等人报告了psa-d、更高病理分期的前列腺癌之间的相关性，psa-d导致根治性前列腺切除术[23]后生存期下降。此外，Corcoran等人报告说，在前列腺活检和前列腺切除术[24]病人的gleason评分中，psa-d是肿瘤升级的最有力的预测指标。

In agreement with these previous works, in our study for each of the consideredISUP score thresholds, PSA-D values showed a directcorre-lation with lesion presence, with abetter performance and an im-provement in terms of specificitywhen considering an ISUP score-threshold ≥3. Nevertheless, we want to highlight thatno statistically significant improvement was observed in the diagnostic performance of bp-MR when combined with PSA-D in all evaluated scenarios: lesion detection independently of ISUP score, detection of both intermediate and high-grade lesions and detection of high grade lesions alone. Hence, our results suggest a pivotal role of bp-MR in PCa patients, due to its good diagnostic accuracy when used alone that is notsignifi-cantlyincreased when combined with clinical parameters such as PSA-D.

与之前的研究结果一致的是，在我们对每个考虑分数阈值的研究中，psa-d值显示有病变存在，有较好的表现，并且在考虑isup分数阈值≥3时在特异性方面有改进。然而，我们要强调的是，当与psa-d结合时，采用独立于isup评分检测病变，同时检测中、高级病变或者是单独检测高级病变时，bp-mr诊断性能在所有评估方案中没有显著的改善。因此，我们的研究结果表明bp-mr在诊断pca患者中有着关键作用，因为它在单独使用时具有良好的诊断准确性，而当与psa-d等临床参数结合时，它的诊断精度并没有显著提高。

These findings are somewhat in conflict withthose found in some other recentworks that have evaluated the performance of MR and PSA-D for detection of only ISUP ≥2 lesions. Rais-Bahrami et al.evaluated the diagnostic accuracy of the number of significant lesions detected with bp-MR and PSA-derived parameters, showing that while bp-MR outperforms PSA alone (AUC 0.80 and 0.66 respectively), its coupling with PSA and PSA-D significantly improves diagnostic accuracy (AUC 0.83 and 0.87 respectively) [5]. Similarly, Distler et al. reported that using PSA-D in combination with mp-MR improves the NPV of PI-RADS scoring, with an AUC of 0.75 for mp-MR alone [7]. Fascelli et al. also showed, in a biopsy-naïve patient population, that the number of le-sions on bp-MR was highly predictive of the presence of cancer (AUC 0.80), but this parameter was outperformed by PSA-D alone (AUC 0.86). Similarly, the combination of bp-MR with PSA and PSA-D using previously derived composite formulas did not significantly increase the diagnostic accuracy (AUC 0.86) [10]. Washino et al. have shown that bp-MR and PSA-D are both valuable tools for detection of ISUP ≥2 lesions, with respective AUCs of 0.84 and 0.82, and their combination significantly improves their accuracy in predicting biopsy outcome[6].Of the cited works, two used PI-RADSv2 as a scoring system for the interpretation of MR images [6,7], while the others used a system based on the number of “screen-positive” lesions [5,10]. The differences in reported AUC values for MR and PSA-D alone and their combination could be due at least in part to the variability in MR scoring systems, patient population characteristics and sample power.

这些发现与其他一些新近的发现有些冲突，这些发现评估isup≥2的病人的mr和psa-d表现。Rais-Bahrami等人评估了bp-mr和psa的诊断准确性，显示bp-mr单独优于psa（分别为auc 0.80和0.66），它与psa和psa-d的联合显著提高了诊断精度（分别为auc 0.83和0.87）【5】. 类似地，Distler等人报告说，将psa-d与mpmr结合使用可以提高pi-rads评分的阴性预测值，仅mpmr的AUC为0.75【7】。fascelli等人还表明，在一个活组织检查中，bp-mr对于前列腺肿瘤具有的较高的预测值（auc 0.80），但psa-d单独就能达到0.86。[10] washino等人已经表明bp-mr和psa-d都是检测isup≥2病变的有价值的工具，其分别是0.84和0.82，它们的结合大大提高了预测活检结果的准确性【6】。在被引用的作品中，有两部使用pi-radsv 2作为图像的解释的评分系统[6,7]，而其他的则使用基于“屏幕阳性”病变数量的系统[5,10]。与报道的MR和单独PSAD的AUC价值不同，这两者的结合至少在一定程度上是由于评分系统的多变性、患者的群体特征和抽样能力。

In our opinion, the major difference between our results and previous reports is due to the higher diagnostic accuracy of bp-MR reached in our workwhen compared to that reported in the cited papers.However, our diagnostic accuracy is in line with that found in a recent meta-analysis of reported PI-RADS v2 diagnostic performances (AUC 0.83-0.89) [25]. Several reasons may explain this difference.In our study, all images were interpreted in consensus by a dedicated prostate team, composed of two different readers, with extensive experience in prostatic MR imaging.The cited works either did not report number and experience level of individual readers [5,10], had a single reader [6] or had multiple readers of different specialties under supervision of a radiologist (7 years of experience) [7]. Furthermore, some of them used a PI-RADS or Likert scale≥ 3 cutoff. As we found in our analysis, and as reported in literature [25], this is not the best value for detection of ISUP ≥3 lesions. The low number of bpPI-RADS 3 lesions in our manuscript, only 9 (8%),may have contributed as well to the diagnostic performance of bp-MR. This may as well be due to the high quality of the bp-MR technique on a 3T scanner and, again, to the experience of the radiologists.In fact, it is well known that there is a learning curve in prostate imaging and the rate of lesions classified as PI-RADS 3 tends to progressively decrease [26,27].

在我们看来，我们的结果和以前的报告之间的主要差异是由于在我们的工作中所达到的bp-mr的诊断准确性比引用的论文所报道的要高。.然而，我们的诊断准确性与最近对pi-rads V2诊断性能的数据分析（auc 0.83-0.89）中发现的一致【25】。有几个原因可以解释这种差异。在我们的研究中，所有的图像都被一个由两个不同的读者组成的前列腺团队以一致的方式解释，他们在前列腺成像方面有着丰富的经验。被引用的作品要么没有报告医师的数量和经验水平[5,10]，要么只有一个医师[6]，要么在放射学家的监督下有多个不同专业的医师（7年经验）[7]。此外，其中一些使用了pi-rad或likert标度≥3截止。正如我们在分析中所发现的，以及在文献[25]中所报道的，这并不是对isup≥3病变的最佳检测值。我们的手稿中bppi-rad 3病变的数量很少，只有9个（8%），这也可能对bp-mr的诊断效果有贡献。这也可能是由于3t扫描仪上的bp-mr技术的高质量，也可能是由于放射学家的经验。事实上，众所周知，前列腺成像有一条学习曲线，被归类为pi-rads 3的病变发生率趋于逐渐下降[26,27]。

A major limitation of this study is represented by the use of TRUS-guided biopsy as the reference standard, that is less accurate than radical prostatectomy, as recently shown in the PROMIS study [3]. Some further limitations of our study should be acknowledged: the patient population was relatively small, the data collection was retro-spective,and this was a single-institution experience without any patient randomization.

这项研究的局限性在于使用了前列腺穿刺活检作为参考标准，最近的promis研究[3]表明这比根治性前列腺切除术更不准确。我们研究的一些的局限性是：患者人数相对较少，数据收集是回顾性的，这是一个没有进行患者随机化分组的研究。

5. Conclusions

Our results confirm that in patients with a clinical suspicion of PCa, bp-MR is a valuable and non-invasive diagnostic tool in correctly se-lecting patients who must undergo biopsy. In addition, we showed that PSA-D, while useful in arousing clinical suspicion and follow up of patients, has not demonstrated an added value when compared to bp-MR alone or in combination with it. These findings reinforce the notion that bp-MR should have a pivotal role in the future management of patients with PCa, especially before any biopsy is performed, as it may help avoid unnecessary invasive exams and reduce false negative sys-tematic biopsies.

我们的结果证实，在临床怀疑有pca的患者中，bp-mr是一种有价值的、非侵入性的诊断工具，可以正确检测必须进行活检的患者。此外，我们的结果还表明，psa-d虽然在临床提示前列腺肿瘤和随访病人方面很有用，但与bp-mr单独或结合使用相比，它并没有增加价值。这些发现强化了bp-mr在未来对pca患者的管理中的关键作用，特别是在任何活检之前，因为它可能有助于避免不必要的侵入性检查和减少假阴性的活检。