The incidence of and mortality rate associated with prostate cancer (PCa) have been increasing worldwide over the past decades,1,2 making PCa a major health concern worldwide. The diagnosis rate of localized or early-stage PCa has been increasing since the adoption of evaluations based on prostate-specific antigen (PSA). Patients with localized PCa are often offered definitive treatment, including radical prostatectomy (RP) or radiotherapy (RT).3,4 RP can provide effective oncological control but may also result in erectile dysfunction.5 The nerve-sparing RP (NSRP) has gained popularity after anatomical studies,6 and international guidelines have emphasized the importance of NSRP for preserving erectile function postoperatively whenever feasible.3,4
To optimize the functional outcomes and decrease the surgical margin-positive rate,7 accurate preoperative prediction of extraprostatic extension (EPE) is vital for the operating surgeon.8 Although digital rectal examinations (DREs) and transrectal ultrasound (TRUS) evaluations have been historically used to assess the prostatic capsule status, both tools have shown low accuracy in previous studies.9
Several studies have assessed the possibility of the presence of EPE by using preoperative clinical and pathologic factors. Sebo et al10 analyzed 207 patients and proposed that the percent positive cores and percent cancer in systematic biopsies were the strongest predictors of EPE. Tosoian et al11 analyzed 4459 patients and proposed a model that included clinical stage, serum PSA levels, and biopsy Gleason score to predict EPE. However, the traditional prediction model, which might lack of site-specific information, did not integrate with multiparametric magnetic resonance imaging (mpMRI) findings, which was used more widely in evaluation and localization of PCa. Advancements in imaging technology have improved the accuracy of mpMRI in detection, lesion characterization, and staging of PCa. mpMRI consists of anatomical T2-weighted images and at least one other sequence, such as diffusion-weighted imaging or dynamic contrast-enhanced imaging. The use of mpMRI has been rapidly growing worldwide over the last decade,12 but its performance in predicting EPE differed among studies,13 and these variations may influence the surgeon’s decision to perform NSRP.14
However, several novel biomarkers, including p2PSA, have been recently proposed for detecting PCa. Previous studies have shown that levels of p2PSA and its derivative measurements, including the percentage of p2PSA to free PSA (%p2PSA: p2PSA/free PSA) and the prostate health index (PHI: [p2PSA/fPSA] × √tPSA) (Beckman Coulter; Brea, CA, USA) were more significantly associated with poor post-prostatectomy pathological outcomes than PSA levels in European and Chinese populations.15–18
Furthermore, our group and other groups have shown that the integration of these serum biomarkers and imaging techniques may improve the prediction of clinically significant cancer in biopsy-naive and repeated-biopsy settings.16,19,20 Therefore, we conducted this prospective study to investigate the accuracy of a combination of mpMRI and PHI in the prediction of EPE after RP.
2.1. Study design
This is a retrospective study generated from a prospectively registered cohort. The study was approved by the Institutional Review Board/Ethics Committee of Taipei Veterans General Hospital (No. 201708017A), and all patients were informed about the procedures and possible complications before enrollment. Eligible patients who had undergone RP for localized PCa were enrolled in our study between February 2017 and July 2019 after obtaining informed consent. All methods were performed in accordance with the relevant guidelines and regulations.
Patients who had undergone RP for PCa at our hospital were included. PCa was confirmed by TRUS biopsy, with or without MRI-ultrasound fusion prostate biopsy.
All the patients received MRI before RP. The exclusion criteria were treatment with any dosage of androgen deprivation therapy or 5-alpha-reductase inhibitors before enrollment in this study or a diagnosis of PCa by transurethral resection of the prostate.
2.3. Data collection
We collected clinical data, including age, body mass index, and DRE findings. Blood samples were collected before surgery to measure the tPSA level, percentage of free PSA (%fPSA), p2PSA level, and percentage of p2PSA (%p2PSA). PHI was calculated as (p2PSA/fPSA) × √tPSA. The blood samples were processed using a Beckman Coulter DxI 800 UniCel Immunoassay system (Beckman Coulter Taiwan, Inc., Taipei, Taiwan).
All patients in our hospital underwent 3-T multiparametric MRI (MR750; GE Medical Systems, Milwaukee, WI, USA). The mpMRI protocol included T2-weighted imaging, diffusion-weighted imaging, and dynamic contrast-enhanced imaging.21 All the MRI images were interpreted by a single uroradiologist (S.-H.S.) with over 15 years of experience in prostate MRI. The uroradiologist was not blinded to the clinical characteristics of individual cases but blind to PHI value. According prostate imaging reporting & data system (PIRADS) 2.0, the standard report format, including lesion location with PIRAD value, capsule status, bladder neck status, apex status, seminal vesicle status, lymph node status, and tentative stage, was used.
An evaluation of the surgical specimen, including determination of pathologic tumor stage, pathologic Gleason score (pGS), and pathologic nodal status, was performed by experienced genitourinary pathologists who were blinded to the blood sample results and the mpMRI results.
The extraprostatic extension (EPE) grade system proposed by Mehralivand et al22 was used for the evaluation of EPE on mpMRI. EPE grade 1 and above was thought to be EPE on mpMRI. Histopathological EPE was defined as cancer invading beyond the prostatic capsule, regardless of the status of the seminal vesicles.23 PCa was staged and graded according to the 2010 version of the American Joint Committee on Cancer.
2.4. Statistical analysis
Continuous variables are reported as mean, median, and interquartile range (IQR), and were compared using the Mann-Whitney U test. Categorical variables are reported as proportions and were compared using the chi-square test. The area under the receiver operating characteristic (ROC) curve was used to assess the accuracy in predicting the EPE of each variable. Univariate and multivariable logistic regression models were used to predict EPE outcomes. All statistical analyses were conducted using SPSS software for Windows version 22 (IBM, Armonk, NY, USA). A two-sided p value of less than 0.05 was considered statistically significant.
3.1. Baseline demographics and findings
Our study enrolled 185 patients who had undergone RP for PCa between February 2017 and July 2019. Four patients had received 5-alpha-reductase inhibitors within 3 months before blood samples were taken, and PCa was diagnosed by transurethral resection of the prostate in 14 patients. Among the 167 patients, four were excluded due to poor quality of mpMRI data, which was defined as difficulty to interpret PIRAD or tentative stage due to diffuse hemorrhage after TRUS biopsy, and the data from the remaining 163 patients were used to analyze the accuracy of mpMRI, PHI, and a combination of both, to predict EPE after prostatectomy.
Table 1 shows the baseline demographics and findings of the prostate evaluations. Among the 163 patients, 97 (59.5%) were found to have a pathological T3 (pT3) stage. Patients with the pT3 stage disease had a higher median PHI value (65.7 vs 42.6; p < 0.001) than those with the pathological T2 (pT2) stage disease. The pathological characteristics of the two groups are also listed in Table 1.
Table 1 -
Demographic and pathologic characteristics of the patients undergoing radical prostatectomy
||Total (n = 163)
||pT3 (n = 97)
||pT2 (n = 66)
|Age (y), mean ± SD
||66.4 ± 5.9
||66.8 ± 6.1
||65.9 ± 5.6
|BMI (kg/m2), mean ± SD
||25.3 ± 3.21
||25.6 ± 3.1
||24.9 ± 3.4
|Abnormal digital rectal examination findings (%)
|Preoperative TRUS-BX Gleason score (%)
|tPSA (ng/mL), median (IQR)
|PSA density (ng/mL) median (IQR)
|%fPSA (%), fPSA/tPSA, median (IQR)
|PHI median (IQR)
|MRI tumor stage (%)
|Postoperative Gleason score (%)
|Positive core ratio, median (IQR)
|Maximum % of tumor in any core, median (IQR)
%fPSA = free prostate-specific antigen to total prostate-specific antigen; BMI = body mass index; fPSA = free prostate-specific antigen; IQR = interquartile range; MRI = magnetic resonance imaging; PHI = prostate health index; PIRADS = prostate imaging reporting & data system; PSA = prostate-specific antigen; pT2 = pathological T2 stage; pT3 = pathological T3 stage; tPSA = total prostate-specific antigen; TRUS-Bx = transrectal ultrasound prostate biopsy.
3.2. Prediction analysis of EPE
Table 2 shows the findings from univariate logistic regression analyses of EPE and mpMRI were significant predictors (p < 0.001). In the ROC curve analysis, the areas under the curve (AUCs) of mpMRI, PHI, and the combination of mpMRI and PHI were 0.717 (95% CI, 0.635-0.799), 0.722 (95% CI, 0.643-0.8), and 0.785 (95% CI, 0.714-0.857), respectively (Fig. 1 and Table 2). The AUC of the combination of mpMRI and PHI was significantly higher than those of mpMRI (0.785 vs 0.717; p = 0.0007) or PHI (0.785 vs 0.722; p = 0.0236) alone.
Table 2 -
Univariate analyses and area under the curve for pT3 stage
||Univariate OR (95% CI); p
||AUC (95% CI of AUC); p
||1.025 (0.972-1.082); 0.361
||0.526 (0.408-0.645); 0.669
||1.069 (1.019-1.122); 0.006
||0.612 (0.496-0.728); 0.068
||2.292 (1.205-4.357); 0.011
||0.640 (0.524-0.756); 0.023
||1.618 (1.162-2.254); 0.004
||0.642 (0.522-0.761); 0.021
|% Free PSA
||0.995 (0.963-1.028); 0.759
||0.421 (0.332-0.510); 0.088
||1.030 (1.015-1.044); <0.001
||0.722 (0.643-0.8); <0.001
||6.413 (3.205-12.832); <0.001
||0.722 (0.641-0.803); <0.001
|PHI + mpMRI
||0.786 (0.715-0.857); <0.001
AUC = area under the receiver operating characteristic curve; DRE = digital rectal examination; GS = Gleason score; mpMRI = multiparametric magnetic resonance imaging; OR = odds ratio; PHI = prostate health index; PSA = prostate-specific antigen; pT3 = pathological T3 stage.
3.3. Performance of prediction to EPE in mpMRI
In our cohort, the accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of mpMRI to predict EPE were 72.4%, 73.2%, 71.2%, 78.9%, and 64.4%, respectively. Twenty-six patients showed no evidence of EPE preoperatively on mpMRI but were found to have pT3 disease after the operation. However, these patients had a higher PHI median preoperatively than the other patients without evidence of EPE by mpMRI (52.62 vs 45.1; p = 0.006).
3.4. PHI thresholds
In the overall cohort, the median (IQR) PHI value was 55.96 (40.56-80.0). We investigated the 25th and 50th percentile values as diagnostic thresholds combined with mpMRI for prediction of EPE (Table 3). When we used the criteria of PHI >40 as the diagnostic threshold, undiagnosed EPE could be avoided in 21 out of 26 patients.
Table 3 -
Diagnostic performance of mpMRI and different median PHI cutoffs for EPE of prostate cancer
||No. EPE missed
|PHI > 40
|PHI > 56
|mpMRI or PHI > 40
|mpMRI or PHI > 56
EPE = extraprostatic extension; mpMRI = multiparametric magnetic resonance imaging; NPV = negative predictive value; PHI = prostate health index; PPV = positive predictive value.
EPE is associated with the potential risk of a positive surgical margin, resulting in adverse outcomes of PCa. Furthermore, EPE is also considered an independent prognostic factor for biochemical recurrence after RP.24 Conversely, patients with organ-confined disease were thought to be good candidates for NSRP, even though the selection criteria for NSRP have not been well established.25 Therefore, preoperative determination of the existence of EPE is particularly important to allow surgeons to tailor the surgical approach appropriately. To the best of our knowledge, this is the first study to evaluate PHI, mpMRI, and the combination of these in predicting the existence of EPE preoperatively. We found that both PHI and mpMRI could be used together to predict the presence of EPE, and the area under the ROC curve of this combination was higher than those of PHI or mpMRI alone. Moreover, a PHI threshold of >40 could be used to prevent undiagnosed EPE before RP in most patients.
Chung et al26 analyzed 1031 Korean patients and proposed a model to predict EPE whose AUC was 0.777. Tosoian et al11 proposed Partin table 2017, which enrolled up to 4459 patients, and the AUC of the predictive model was 0.724. These predictive tools were developed to predict EPE, and they were based on large amounts of data with substantial statistical power.11,26,27 However, these tools were developed many years before novel biomarkers such as PHI and prostate cancer antigen 3, or mpMRI became available. Therefore, none of these tools can provide cancer site-specific information. Furthermore, the AUC of our predictive model was 0.786, which is equal to or even better than those models.
Surgical techniques such as extra-, inter-, and intrafascial dissection can be planned to achieve NSRP. As the dissection plane moves closer to the true capsule, better functional outcomes may be achieved, increasing the importance of anatomical details. In this regard, most prostate surgeons utilize mpMRI for surgical planning, especially for NSRP.28 However, the sensitivity of mpMRI for the detection of EPE is not satisfactory, based on the literature. de Rooij et al13 analyzed 75 studies in a diagnostic meta-analysis and reported that the sensitivity and specificity of MRI were only 0.57 (95% CI, 0.49-0.64) and 0.91 (95% CI, 0.88-0.93), respectively. They concluded that MRI has high specificity but poor and heterogeneous sensitivity for local PCa staging.
Guazzoni et al17 reported that PHI levels were significantly higher in patients with pT3 disease (p < 0.001) in an analysis of 350 men who underwent RP for clinically localized PCa. Chiu et al15 reported similar results in a cohort of 135 Chinese patients: the mean PHI of patients with ≥pT3 disease was 71.0, and that of patients with <pT3 disease was 41.8 (p = 0.001). The mean PHIs of patients with ≥pT3 disease and <pT3 disease in our cohort were 79.01 and 48.3, respectively (p < 0.001). Our cohort included 26 patients with false-negative mpMRI results for EPE, and the median PHI in these patients was higher than that in the patients with true-negative mpMRI results (52.62 vs 45.1; p = 0.006). This is because the advanced pathological T stage was associated with a higher PHI value. Using a PHI threshold >40, undiagnosed EPE was prevented in 21 of the 26 patients with false-negative mpMRI results for EPE. Furthermore, the surgical positive-margin rate was higher in patients with false-negative mpMRI results than in those whose EPE was forecast correctly by MRI (23.1% vs 14.3%). These data suggest that a supplementary tool, such as PHI, can reduce the false-negative rate of mpMRI in diagnosing EPE, thereby further reducing the positive-margin rate.
There are three main strengths to the current study. First, we collected plasma samples before RP, and ensured adherence to the manufacturers’ recommendations. Second, our mpMRI technique was based on modern standards, that is, the PIRADS recommendations, and the results were reviewed by a single experienced, dedicated genitourinary radiologist.
However, there are also several limitations to this study. First, our sample size was relatively small. Second, our cohort appeared to show more advanced-stage cases of PCa, which cannot represent the general population. Third, our study was based on a Taiwanese cohort from a single institution, and external validation was needed to confirm the quality. Fourth, our uroradiologist was not blinded to the clinical characteristics of participants. However, the current study is the first to combine mpMRI and newer biomarkers in predicting EPE and has shown encouraging results. Further large-scale prospective studies with various populations are needed to validate the findings.
In conclusion, the combination of PHI and mpMRI showed greater predictive power for preoperative detection of pathological EPE in comparison with PHI or mpMRI alone. For surgeons, this combination may be useful in preoperative evaluation and may help in tailoring NSRP.
1. Center MM, Jemal A, Lortet-Tieulent J, Ward E, Ferlay J, Brawley O, et al. International variation in prostate cancer incidence and mortality rates. Eur Urol 2012;61:1079–92.
2. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin 2017;67:7–30.
3. Mottet N, Bellmunt J, Bolla M, Briers E, Cumberbatch MG, De Santis M, et al. EAU-ESTRO-SIOG guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent. Eur Urol 2017;71:618–29.
4. Sanda MG, Cadeddu JA, Kirkby E, Chen RC, Crispino T, Fontanarosa J, et al. Clinically localized prostate cancer: AUA/ASTRO/SUO guideline. Part II: recommended approaches and details of specific care options. J Urol 2018;199:990–7.
5. Alemozaffar M, Regan MM, Cooperberg MR, Wei JT, Michalski JM, Sandler HM, et al. Prediction of erectile function following treatment for prostate cancer. JAMA 2011;306:1205–14.
6. Walsh PC. Anatomic radical retropubic prostatectomy. Philadelphia: W.B. Saunders; 1998.
7. Druskin SC, Liu JJ, Young A, Feng Z, Dianat SS, Ludwig WW, et al. Prostate MRI prior to radical prostatectomy
: effects on nerve sparing and pathological margin status. Res Rep Urol 2017;9:55–63.
8. Rabbani F, Stapleton AM, Kattan MW, Wheeler TM, Scardino PT. Factors predicting recovery of erections after radical prostatectomy
. J Urol 2000;164:1929–34.
9. Jones D, Friend C, Dreher A, Allgar V, Macleod U. The diagnostic test accuracy of rectal examination for prostate cancer diagnosis in symptomatic patients: a systematic review. BMC Fam Pract 2018;19:79.
10. Sebo TJ, Bock BJ, Cheville JC, Lohse C, Wollan P, Zincke H. The percent of cores positive for cancer in prostate needle biopsy specimens is strongly predictive of tumor stage and volume at radical prostatectomy
. J Urol 2000;163:174–8.
11. Tosoian JJ, Chappidi M, Feng Z, Humphreys EB, Han M, Pavlovich CP, et al. Prediction of pathological stage based on clinical stage, serum prostate-specific antigen, and biopsy Gleason score: Partin Tables in the contemporary era. BJU Int 2017;119:676–83.
12. Turkbey B, Merino MJ, Gallardo EC, Shah V, Aras O, Bernardo M, et al. Comparison of endorectal coil and nonendorectal coil T2W and diffusion-weighted MRI at 3 Tesla for localizing prostate cancer: correlation with whole-mount histopathology. J Magn Reson Imag 2014;39:1443–8.
13. de Rooij M, Hamoen EH, Witjes JA, Barentsz JO, Rovers MM. Accuracy of magnetic resonance imaging for local staging of prostate cancer: a diagnostic meta-analysis. Eur Urol 2016;70:233–45.
14. Park BH, Jeon HG, Jeong BC, Seo SI, Lee HM, Choi HY, et al. Influence of magnetic resonance imaging in the decision to preserve or resect neurovascular bundles at robotic assisted laparoscopic radical prostatectomy
. J Urol 2014;192:82–8.
15. Chiu PK, Lai FM, Teoh JY, Lee WM, Yee CH, Chan ES, et al. Prostate health index and %p2PSA predict aggressive prostate cancer pathology in Chinese patients undergoing radical prostatectomy
. Ann Surg Oncol 2016;23:2707–14.
16. Fossati N, Buffi NM, Haese A, Stephan C, Larcher A, McNicholas T, et al. Preoperative prostate-specific antigen isoform p2PSA and its derivatives, %p2PSA and prostate health index, predict pathologic outcomes in patients undergoing radical prostatectomy
for prostate cancer: results from a multicentric european prospective study. Eur Urol 2015;68:132–8.
17. Guazzoni G, Lazzeri M, Nava L, Lughezzani G, Larcher A, Scattoni V, et al. Preoperative prostate-specific antigen isoform p2PSA and its derivatives, %p2PSA and prostate health index, predict pathologic outcomes in patients undergoing radical prostatectomy
for prostate cancer. Eur Urol 2012;61:455–66.
18. Huang YP, Lin TP, Cheng WM, Wei TC, Huang IS, Fan YH, et al. Prostate health index density predicts aggressive pathological outcomes after radical prostatectomy
in Taiwanese patients. J Chin Med Assoc 2019;82:835–9.
19. Fan YH, Pan PH, Cheng WM, Wang HK, Shen SH, Liu HT, et al. The prostate health index aids multi-parametric MRI in diagnosing significant prostate cancer. Sci Rep 2021;11:1286.
20. Hsieh PF, Li WJ, Lin WC, Chang H, Chang CH, Huang CP, et al. Combining prostate health index and multiparametric magnetic resonance imaging in the diagnosis of clinically significant prostate cancer in an Asian population. World J Urol 2020;38:1207–14.
21. Lai WJ, Wang HK, Liu HT, Park BK, Shen SH, Lin TP, et al. Cognitive MRI-TRUS fusion-targeted prostate biopsy according to PI-RADS classification in patients with prior negative systematic biopsy results. J Chin Med Assoc 2016;79:618–24.
22. Mehralivand S, Shih JH, Harmon S, Smith C, Bloom J, Czarniecki M, et al. A grading system for the assessment of risk of extraprostatic extension of prostate cancer at multiparametric MRI. Radiology 2019;290:709–19.
23. Rocco B, Sighinolfi MC, Sandri M, Eissa A, Elsherbiny A, Zoeir A, et al. Is extraprostatic extension of cancer predictable? A review of predictive tools and an external validation based on a large and a single center cohort of prostate cancer patients. Urology 2019;129:8–20.
24. Ball MW, Partin AW, Epstein JI. Extent of extraprostatic extension independently influences biochemical recurrence-free survival: evidence for further pT3 subclassification. Urology 2015;85:161–4.
25. Graefen M, Haese A, Pichlmeier U, Hammerer PG, Noldus J, Butz K, et al. A validated strategy for side specific prediction of organ confined prostate cancer: a tool to select for nerve sparing radical prostatectomy
. J Urol 2001;165:857–63.
26. Chung JS, Choi HY, Song HR, Byun SS, Seo S, Song C, et al. Preoperative nomograms for predicting extracapsular extension in Korean men with localized prostate cancer: a multi-institutional clinicopathologic study. J Korean Med Sci 2010;25:1443–48.
27. Satake N, Ohori M, Yu C, Kattan MW, Ohno Y, Miyakawa A, et al. Development and internal validation of a nomogram predicting extracapsular extension in radical prostatectomy
specimens. Int J Urol 2010;17:267–72.
28. Tewari AK, Srivastava A, Huang MW, Robinson BD, Shevchuk MM, Durand M, et al. Anatomical grades of nerve sparing: a risk-stratified approach to neural-hammock sparing during robot-assisted radical prostatectomy
(RARP). BJU Int 2011;108(6 Pt 2):984–92.